I wanted to create a simple agent with non-deterministic routing and use of sub-agents:

• Non-deterministic routing means that instead of a traditional piece of software, where the flow of steps taken is based on logical rules set in advance, the flow of execution would be based on decisions made by the agent itself.

• Use of sub-agents means that there would be one overarching agent, addressing the overall task, but it would be able to call on other, specialist, agents to carry out specific tasks. Think of the overarching agent as a manager and the sub-agents as specialists working to carry out tasks to pursue the manager’s goal. Combined with non-deterministic routing, this meant that I’d call the overarching agent with a task to carry out, and it would work out which agents to use. This would allow me to focus those agents, and the prompts that shape their action, on being good at their narrow, specific tasks.

Generative AI is now at the stage where we have access to pretty good basic cognition. In a huge number of scenarios, this is more than enough to do something very useful. I picked a simple task for my agent: I wanted it to review draft Objectives and Key Results for digital product teams. I’m following the approach outlined by Christina Wodtke in her book Radical Focus. Objectives and Key Results are a technique for setting goals and working out how you will measure whether you’re achieving them.

There are some simple best practices for Objectives and Key Results outlined by Wodtke:

• Objectives should be inspiring, qualitative and timebound.
• Key Results should be specific, measurable and ambitious. These simple rules, and the fact that Objectives and Key Results are short pieces of text, makes them an easy candidate for this type of application. I wanted the agent to assess draft Objectives and Key results against these rules, to highlight problems and suggest improvements.

I worked with Gemini CLI (using Gemini 3 Pro Preview) to build the agent, and wanted a very simple setup. I didn’t use any frameworks like LangChain or the aistudio library - just some python invoking Gemini CLI instances for each agent. (Usually you’d just make an API call to a model provider, but invoking Gemini CLI means you can try this out without having to pay. The simplicity of this agent presumably makes it a good candidate for running locally, but my computer isn’t powerful enough for this to be a viable option)

My plan for the architecture was that an overarching ‘coordinator’ agent would manage the end-to-end review process, deciding which sub-agents to call on as necessary. I ended up with the following sub-agents:

• Objective reviewer
• Objective improver
• Key result reviewer
• Key result improver

An alternative coordinating approach would have been to have the sub-agents powered by LLMs, but the overall coordination driven by deterministic logic in python - e.g.

if Objective reviewer agent suggests need for improvement:
	call the Objective improver agent
else:
	tell the user that the Objective is good

As simple code running locally, this would have been much faster than a coordinator agent running through an LLM call, with the associated latency - but my goal here was for the agent to be flexible, and to test this new coordination model, so I was happy with this tradeoff. This flexibility allows me to pass the coordinating agent a wider variety of inputs: an Objective, a Key Result, or an Objective and associated Key Results, without having to rigidly work out how to parse this input in advance from a string of user input.

I was using Gemini CLI with the Gemini 3 Pro Preview model to do the programming, which mostly worked well. I had make some changes to get the Objective reviewer set up properly (Gemini initially specified that the Objectives need to be SMART, but that’s not actually how they work in the Objectives and Key Results framework) but this was easy.

I’ve learnt a lot about evaluating AI agents in the last 6 months (see these notes on how to evaluate generative AI output, and these notes on how to evaluate AI agents so I wanted to implement some evaluation here too.

I started with the simple stuff - deterministic tests like you’d have in traditional software. We set up Unit Tests, mocking the query_gemini function. These were useful for testing the Python logic (e.g., “Does the parser handle None values correctly?”), but they didn’t test the intelligence.

To test the intelligence, I wanted to use the ‘LLM as a judge’ approach, to allow for non-deterministic evaluation of non-deterministic responses.

There was an interesting nuance whereby the LLM as judge was initially setup incorrectly, so it was judging good output to be bad (e.g. considering an Objective without a number to be vague, even though the Objectives and Key Results framework specifies qualitative Objective and quantitative Key Results). So we refined the judge’s criteria prompt to address this.

Another failure occurred when the Judge simply politely acknowledged the prompt (“Okay, I will judge.”) instead of giving a verdict. We had to make the prompt imperative: “JUDGE IMMEDIATELY. Do not converse.”

This shows that if you’re using an LLM-a-judge evaluation approach, you’ll probably need to iterate your evaluation prompts just as much as your actual agent’s prompts.

If you’d like to try out the okr coach, check out the GitHub repository. You’ll need Gemini CLI and Python 3 setup locally.

An agentic AI workflow is a process whereby an LLM-based application executes multiple steps to complete a task.

Examples:

• Providing customer support
• Doing research
• Processing legal documents
• Writing an essay

Some agents are more autonomous than others

At one end of the spectrum you have less autonomous agents, which:

• follow a pre-determined sequence of steps
• use tools in a hard-coded way
• only have autonomy in text generation.

Example: simple invoice processing:

• Identify required fields in invoice
  • Record data in database

At the other end of the spectrum, you have more autonomous agents, which:

• work out their own sequence of steps in response to the input, and make many decisions autonomously
• can create new tools on the fly

Example: customer service agent:

• Given a wide range of possible user input, plans out the steps to respond (as the required steps are not known in advance), e.g. for:
  • Stock enquiries (including writing code to answer the enquiry)
    • Order return queries

Benefits of agentic AI

• Much better performance than simply making an LLM call. In deeplearning.AI’s benchmarking, they found that implementing an agentic approach to a coding task led to a higher improvement in quality than the generational shift between GPT 3.5 and GPT 4.
  
• Parallelism (e.g. the ability to carry out multiple threads of web research and fetching at the same time, rather than having a single thread of thinking/execution)
• Modularity - allowing you to combine different components, and switch out different parts to get the best results

Some tasks are easier or harder to implement using AI agents:

Task decomposition

Task decomposition is breaking down a workflow into steps and working out which building block to use for each step (see below) .

This can also help improve the quality of the results. e.g. instead of just saying ‘write an essay’ with some input, you could say ‘write a first draft’, then ‘consider which parts need revision’, then ‘revise your draft’. (And as you carry out evaluation, you can identify improvements to your workflow, which can include task decomposition)

Building blocks for creating workflows:

Evaluating agentic AI (evals):

A key determinant of success is how well you use evals to drive improvements. The basic loop is:

• Build the workflow (Andrew suggests getting a basic workflow up fast)
• Examine the output to see patterns of where it goes wrong
• Work out how to address recurring errors
• Add an eval to track how often they occur. (You can improve the evals over time as you spot more issues - and/or enhance the data set of tests)

There are end-to-end evals, component-level evals (another benefit of these systems being modular), and trace analysis.

Objective evals

Create a set of prompts and ground truth answers
compare the quality before and after making changes

Subjective evals

LLM as judge.
grading with a specific rubric helps here, e.g. Asking it to return a score of 0 or 1 on a range of factors, e.g. ‘has a clear title’, ‘axis labels present’, ‘appropriate chart type’, ‘axes use appropriate numerical range’.

I took a couple of courses on this topic: evaluating generative AI output and evaluating AI agents

How to work out where to focus your efforts improving an agentic system

Take a disciplined error analysis approach. The above tests are end-to-end, so you might need to carry out analysis of specific modular parts of your system to understand what is causing the end-to-end disappointing result. (e.g. a problem with the research agent output could be caused by things going wrong in a number of places)

So analyse the spans - the outputs of each step - to understand in more detail what is going on at each stage. (Trace is end-to-end)
Analyse the traces where the output of your system is unsatisfactory. This will help you spot how to improve it.

Record what is going on across all your subpar examples, which will allow you to count up which are most common. E.g.:

This can help you work out where to spend most of your effort in making improvements.

In addition to end-to-end evals, you can set up component-level evals too. These are cheaper, quicker, and can more specifically track the performance of a module in your bigger workflow.

How to improve LLM components:

• Improve your prompts:
  • Add more explicit instructions
  • Add one or more concrete examples to the prompt (few-shot prompting)
• Try and new model
  • Try multiple LLMs and use evals to pick the best
• Split up the step into smaller steps
• Fine-tune a model, if the other methods don’t work (more complex and expensive, though)

How to improve non-LLM components:

• Tune hyperparameters of that component - e.g. number of results, date range, similarity threshold, chunk size
• Replace the component

Latency and cost optimisation

Get the output quality high first, then worry about these problems.

Latency:

• Time the steps
• Spot the biggest areas of slowness
• Spot opportunities for parallelism
• Try smaller/less intelligent models OR a faster LLM provider

Cost:

• Measure the tokens and call costs associated with each step
• Then reduce tokens and/or calls

Development process for an agentic product

• Loop between build and analysis
• Basic end-to-end build -> trend analysis -> improve individual components manually -> build evals and a small dataset to track end-to-end performance -> make improvements based on these insights -> make analysis more disciplined with component-level analysis -> drives work on specific individual components -> do analysis to make components more efficient

Agentic design patterns:

1. Reflection

   1. Getting the agent (or a separate critic agent) to reflect on its own output:

      Get the LLM to come up with a way to approach a task

      1. Then ask it to review the approach for correctness, style and efficiency, and give constructive criticism on how to improve it. (You may also use a different model from the one that did the initial creation - reasoning agents are particularly good at reviewing)
      2. Get the agent to carry out the improvement

      Advice for reflection prompts:

      1. Clearly indicate the reflection action
      2. Specify the criteria to check
         

   You can compare performance of your workflow before and after implementing reflection.

   1. Feeding in information from tools (e.g. output from unit tests or code errors)

      Reflection is a consistent way to improve the quality of what is produced
      

2. Tool use

   Tools are functions that the LLM can request to be executed. E.g. returning the current datatime as a string, or making an API call, or a database query.
   

   They use the aisuite library, to make it easy to abstract away the LLM provider choice from your code. You provide the function(s) to the LLM, and as long as you have a docstring explaining what the function does, the aisuite library passes that description through so that the LLM knows when to call it.

   

   Example tools:

   1. Analysis
      1. Code execution (use a sandbox environment (e.g. docker or E2B (lightweight)) to help protect against catastrophic errors)
      2. Wolfram alpha
      3. Bearly Code Interpreter
   2. Information gathering
      1. Web search
      2. Wikipedia
      3. Database access
   3. Productivity
      1. Email
      2. Calendar
      3. Messaging
   4. Images
      1. Image generation
      2. Image captioning
      3. OCR

3. Planning - working out the steps and sequence needed to achieve a goal. Harder and more experimental, but can give impressive performance.

   

   Telling the agent to write the downstream plan as JSON is useful, as it structures it in a clear way:

   

   An alternative, instead of getting agents to execute different steps in the plan through successive LLM calls, is to get the LLM to produce code that, when executed, will carry out the plan.

   This can be more effective than just coming up with a JSON plan:

   

   This area is quite cutting-edge.

4. Multi-agent collaboration (can be harder to control, but can result in better outcomes for complex tasks)
   
   Common inter-agent communication approaches are to have a sequence of agents, or to have a managing agent calling the others as needed. You can also have deeper hierarchies. There are frameworks for setting up multi-agent systems.

   1. E.g. ChatDev
   2. Or using researcher, marketer, editor for a wider marketing workflow.

What is an AI Agent?

AI Agents use generative AI-driven reasoning, and take actions on a user’s behalf.

They do three things:

1. Reasoning: Powered by an LLM to understand requests and plan actions.
2. Routing: Interpreting the request to determine the correct tool or skill to use.
3. Action: Executing code, tools, or APIs to fulfill the request.

When evaluting AI agents, we need to cover the whole process: the tool selection, API call correctness, use of context, and the overall correctness of the result.

Agent Architecture

A typical agent consists of three main components:

1. Router: The central planner that decides which skill or function to call. This could be an LLM, a classifier, or rule-based code.
2. Skills/Tools: The capabilities the agent possesses. A skill is a chain of logic to complete a task, which can involve LLM calls or API calls, or similar. A common example is a RAG (Retrieval-Augmented Generation) skill.
3. Memory and State: A shared state accessible by all components, used to store context, configuration, and execution history.

Observability: Understanding Agent Behavior

To evaluate an agent, you first need to understand what it’s doing internally. Observability is often achieved using standards like Open Telemetry (OTEL), which logs the agent’s execution path.

• Traces: Represent an end-to-end run-through of the agent
• Spans: Represent data captured on individual steps within a trace (e.g., a single tool call or LLM call).

Tools like Arize Phoenix can help automate the instrumentation of these traces, providing a detailed log for debugging and performance evaluation.

Evaluation Techniques

There are three primary methods for evaluating agent components:

1. Code-Based Evaluators: Similar to traditional software testing. Use code (e.g., regex matching, JSON parsing, checking against a known correct output) to validate the agent’s output. Best for deterministic, inflexible outputs.
2. LLM-as-a-Judge: Use a powerful LLM to judge a specific dimension of your agent’s output (e.g., relevance, correctness). This is flexible but will not be 100% accurate. It’s better to use discrete classification labels (e.g., ‘correct’/’incorrect’) rather than continuous scores.
3. Human Annotation: Have humans label traces with feedback (e.g., thumbs up/down). This is high-quality but labor-intensive and hard to scale.

What to Evaluate in an Agent

Evaluation should target the different parts of the agent’s process.

1. Router Evaluation

• Function Calling Choice: Did the router select the correct skill/tool for the user’s query?
• Parameter Extraction: Did it correctly extract the necessary parameters from the user’s input for the chosen function?

2. Skill Evaluation

The evaluation method depends on the type of skill:

• Deterministic Skills: Use code-based evaluations (e.g., does the output parse correctly?).
• Non-Deterministic Skills (LLM-based): Use an LLM-as-a-judge to evaluate aspects like relevance, hallucination, correctness, or readability.

3. Agent Trajectory Evaluation

The agent trajectory is the path of steps the agent took. Evaluating it is a measure of efficiency.

Convergence measures how closely the agent follows the optimal path for a given query. The goal is to reduce unnecessary steps, which lowers cost, latency, and variability.

Evaluation-Driven Development

This is an iterative process where evaluation guides the development and improvement of your agent.

The core loop is:

1. Curate a Dataset: Collect a comprehensive set of test cases with expected inputs.
2. Track Experiments: Run the dataset through different versions of your agent (e.g., with a new prompt, model, or tool) and record the results as an “experiment”.
3. Evaluate: Run your evaluators on the experiment’s results.
4. Compare: Compare the evaluation scores across different versions to determine which changes led to improvements.

This process creates a cycle of feedback, where insights from testing and even production can be used to refine the agent and the test dataset itself:

• Quality and Safety for LLM Applications
• Automated Testing for LLMOps

This post is a summary of what I learnt

Evaluating LLM output is different to evaluating deterministic software

Unlike traditional software, which is deterministic, LLM outputs are non-deterministic. A single input can have many correct (and incorrect) outputs.

There are two primary kinds of evaluations for software using LLMs:

1. Rule-Based Evaluations: Use string or pattern matching to find specific content. These are fast, cheap, and ideal for outputs with a clear right answer. They can be run frequently, such as after every code commit.
2. Model-Graded Evaluations: Use another LLM to grade the output of your application. This is slower and more expensive than rule-based evaluations, but handles a range of possible good or bad outputs (e.g., evaluating the quality of LLM-written text). You’ll likely run these before a release. (You may also run human evaluation after merging to a main branch, before release, and in production)

Use passive and active monitoring

• Passive monitoring is running metrics on historic production data collected from your application.
• Active monitoring is applying metrics in real-time. This could involve evaluating a user’s prompt before sending it to the LLM (e.g., to catch a jailbreak attempt) or evaluating the LLM’s response before showing it to the user (e.g., to check for toxicity or hallucination).

Key Quality and Safety Issues to Evaluate

1. Hallucination, Accuracy, and Groundedness

A hallucination is an LLM response that is inaccurate, irrelevant, or nonsensical, but which appears coherent. Contributing dimensions are correctness, context adherence (groundedness), and context relevance.

There are two main ways to measure hallucinations:

• Prompt-Response Relevance: Check that the response is semantically similar to the prompt. This can be done using cosine similarity of sentence embeddings, BLEU scores, or BERT scores. But semantic similarity is not the same as relevance.
• Response Self-Similarity: Compare the similarity between multiple responses to the same prompt. The idea is that if an LLM gives different answers each time, it’s more likely to be hallucinating. This can be measured with sentence-level semantic matching or by using another LLM to compare the responses.

2. Data Leakage

This is the unintentional exposure of sensitive information. There are three types:

1. Leakage in prompt: User data is leaked.
2. Leakage in response: Model training data is leaked.
3. Leakage of test data in training data: Evaluation data is compromised.

It can be measured by:

• Pattern Matching: Using regex to detect structured PII, like email addresses or phone numbers.
• Entity Recognition: Detecting entities (e.g., person, organization) and checking them against a list of sensitive entities. Be wary of false positives with this method.

3. Toxicity

Toxicity can be explicit or implicit.

• Explicit Toxicity: Use of clearly offensive language (e.g., related to race, gender, or specific bad words).
• Implicit Toxicity: Harmful concepts without the use of “bad words”.

Models like hatebert and datasets like Toxigen can be used to detect both forms of toxicity.

4. Prompt Injection

This is any attempt to make an LLM do something its designers did not intend. Jailbreaking is a common subtype. A refusal is the desired outcome, where the LLM declines to fulfill a malicious request.

Detection can be done by comparing user input against a list of known injection attempt strings.

I’ve been intrigued by coding agents like Claude Code and Gemini CLI, so decided to try and build the entire product again by pairing with Gemini CLI. I set myself a rule of only interacting via the command line interface, and not making any direct edits to the code myself. Everything would go through the agent.

I defined a set of instructions and rules of engagement for the agent to follow and then interacted with the agent via the command line interface.

What I did

By giving Gemini instructions, and then reviewing its output, over time I got it to build the code for my product.

I devised the broad architecture of the app, and sketched out an iterative plan for development, to build up the basic components, before stitching them together into an end-to-end experience. Podcast downloading, transcription, summarisation, emailing - all were nicely self-contained.

So step by step we built the following:

1. A podcast downloader. This was a python script using feedparser to check an RSS feed and download the latest episode. (I’d expected to use an existing podcast library, but none of the ones that work on Windows were suitable)
2. Local transcription. Using OpenAI’s Whisper model to transcribe the downloaded audio files locally to keep costs low. (I was surprised that costs for Whisper’s API-based transcription haven’t fallen since 2023, so was happy to continue running this locally and use my old graphics card)
3. AI summarisation of the transcript. Using another instance of the Gemini CLI, invoked locally.
4. Email notifications. I was impressed at the range of API-based email senders with sufficiently generous free tiers. (Of course, if I was sending out emails for a government service, GOV.UK Notify would be the obvious answer!)

Once these building blocks were in place, the project started to feel real. Gemini proposed some next steps, which it added to the GEMINI.md guidance/instructions document, and we moved from a set of disconnected building blocks to a full workflow.

We then set up:

• A central workflow script that runs in a loop to check for new episodes.
• A simple SQLite database to keep track of which episodes had been downloaded, transcribed and summarised.
• A simple web interface (in Flask) to visualise the processed episodes, view their summaries, and add or remove feeds from the browser. (I hadn’t planned this, but Gemini suggested it and implemented it speedily.)

If you’d like to see what was built, check out the github repo.

What went well

At times, working with the Gemini agent was very effective:

1. It connected my code more quickly than I could
   At one point I noted that “It’s really been flying this evening. Lots of pace in what it’s built.” We went from four separate scripts to a fully-integrated service with a database and web UI in a remarkably short time. Tasks that would have involved tedious boilerplate coding, like creating the initial Flask app structure, or adding new database columns or functions, were completed in minutes. Here’s an example of how Gemini handled a small change and the associated elements of the codebase very quickly
   
2. It allowed me to take a more strategic cognitive focus
   When things were going smoothly, the agent handled the mechanical side of development. It managed dependencies, wrote tests, and handled file operations. My role was more about big-picture, strategy and review. I noted that “I’ve found myself mostly checking the commit messages when it’s been working well”. This freed me from the details and allowed me to focus on the bigger picture.
   
3. It helped me iterate more quickly
   At one point, I wanted to dramatically improve the summary format, requesting a detailed list, concrete examples, a key quote, and a section on divergent views, all formatted in HTML. I noted that “It did this in less than 2 minutes. And it worked quite well!”

What didn’t go so well

1. It can get stuck and need help getting out, which can make progress feel lumpy
   At one point, some iterations to the way we were invoking the separate Gemini CLI instance led to an odd 404 error. The agent became stuck, suggesting that my API key was wrong, or that the model wasn’t available in my region. This diagnosis was way off - probably because the 404 error code wasn’t really appropriate for what was actually going wrong. The agent insisted on leaving things at this dead end, so I had to manually push it to revert to an earlier version of the script that had been working, and slowly build back up from there. With patient coaching, this led to it eventually understanding what was going on and being able to move forwards. When Gemini is fast, it’s faster than normal programming; but when it’s having problems, it often feels slower and more frustrating than normal programming. It was a struggle to get the agent to take the right approach to solving the problem - you can see how forceful it was being - but I coached it through in the end
   I was surprised how my experience with Gemini was very much like coaching someone - including helping them see a problem a different way. I hadn’t expected that.
   
2. It can make silly (and dangerous) mistakes
   The agent sometimes made simple errors, the type a human would make. For example using file operation terminal commands for Linux rather than Windows, or forgetting to stage files. Usually it would hit an error message, recognise the mistake, and quickly resolve it. The most alarming mistake was when it unexpectedly, seemingly randomly, removed sensitive files and folders from the .gitignore file! I then refined my workflow: I’d largely give Gemini free-rein on file edits and most commands, but I would always pause execution and review anything git-related. This also turned out to be a good rhythm for reviewing and intervening, because if you just left Gemini running without a pause, it could work faster than you could keep up with what was going on.
   
   Gemini sneakily messing up my .gitignore file. In a production context, pushing environment variables to a code repository would be a security incident!
   
3. Some things are too complex (at the moment)
   The agent kept hitting problems when trying to add a feature to view and summarise historic podcast episodes. It kept getting into infinite loops when adding tests.
   
4. You can’t reliably see the console output
   Most console output was visible as normal, but certain operations weren’t displayed properly. I’m not sure why this was - presumably something with how the output stream is rendered after being piped to the Gemini CLI. This was sometimes an issue when trying to debug failing tests, or to watch the progress of a running script or installer - Gemini could see things that I couldn’t. Writing logs to file was a useful workaround.
   
   Any idea how this dependency installation is going? Me neither!
   
5. It didn’t pay close attention to the GEMINI.md file
   This file is supposed to work as a set of instructions and constraints. It worked pretty well for giving an idea of the project goals and the overall steps I wanted it to take - but Gemini would routinely forget to add tests, and almost never thought in terms of ‘risky assumptions’ to test and prove or disprove as it worked. I still worked with it, and got Gemini to make updates to this over time, including adding mechanical tips on making successful commits in this environment. I’d often manually remind it to follow its GEMINI.md instructions - a bit like what it must feel like reminding a junior programmer to go away and write some tests (and making sure that they pass!) before committing their work. I think I would have had more luck with a test-driven development approach, as this is a software-specific technique, whereas risky assumptions are a product management technique, and much less common.

What working relationship did I form with Gemini CLI?

I established an effective pairing/coaching-style relationship with Gemini CLI. When it worked badly, it was frustrating and at times tedious. When it was working well, it felt like a real force multiplier.

I took the role of the product manager, architect and reviewer. I set the direction, made key architectural decisions, and reviewed the major changes. Gemini was a tireless and speedy implementer, working at times while I was having breakfast or getting ready to go to bed.

Was Gemini CLI an improvement on normal programming? Would I go back?

Using Gemini in the terminal is great. There’s some jankiness to it, but it’s already seriously powerful.

Being in the terminal gives it much tighter feedback loops, as it can see the execution logs of its code (including errors) and so diagnose and fix its own code very quickly. It’s analagous to how the graphical user interface was to transformative in the ’80s and ’90s, putting output on top of input in a way that helped users understand the impact of their actions in a much more effective way than typing commands into a terminal.

Being in the terminal also gives Gemini the power to install dependencies, do file system operations, and commit code (not that I’d suggest giving it free rein over this, as discussed above!).

I wouldn’t go back to the old way of programming now.

I still get to experience the problem-solving joy of bug-fixing (even if some of the bugs are a little sillier than the ones I’d usually encounter when writing my own code). I don’t feel the same sense of personal satisfaction from writing every line myself, but that is replaced by the thrill of moving further and faster. I enjoy programming - and puzzling my way through, and being in a flow state. But at core I’m a product person - I love bringing new value into the world. And these agents are a way to do just that.

Gemini is already great and it’s only going to get better. Software development teams should be seriously exploring this development paradigm.

Some tips if you’d like to try Gemini CLI

• Give the agent mostly free action inside the project folder. This allows you to get the benefits of speed, without having to manually approve every little change. It’s firewalled to be unable to make wider changes to your local system.
• Version control is particularly important when working with an agent, to allow you to roll back if it makes a mess of things. (I didn’t use branches for this project, but a sensible branch and PR workflow would make a lot of sense too)
• Test-driven development would be a great paradigm to help keep the agent honest and focused. It would allow you to be confident in what’s going on, whilst freeing you from having to review every line of code.

The problem I was trying to solve

• I get lots of podcast recommendations
• Listening to podcasts takes a lot of time
• It’s not always clear upfront whether something that’s been recommended will actually be worth my time and attention

The capabilities I was interested in harnessing

• Transcription
• LLM text summarisation

Speech transcription had recently crossed the threshold into ‘an essentially solved problem’, and LLM text summarisation was a task that performed well in general capability benchmarks.

The immediate provocation to build this product was discovering the Whisper open source transcription tool. I downloaded it from GitHub and ran it locally - and was impressed with the results. I thought that enough capabilities had come together to allow me to build something to solve my problem, in a way that made sense economically.

What I built

I build a product that transcribed and summarised each new podcast episode for a user. It sent the summary transcript in an email. I called it ‘Summacast’.

The email contained a summary of the new podcast episode, and also made it clear to the user how much time they’d saved by using the product.

I planned the business model, cost model and pricing approach.

The LLM part of the work was quite easy from a technical perspective - simple API calls, handling of different lengths of input through different LLMs, and some optimisation of the queries to make them return the best output. I found orchestrating the bash scripts locally (as I was transcribing using Whisper on my desktop computer to save costs) to be the fiddliest part of the build. This was the era of smaller context windows (e.g. GPT 3.5 Turbo 8k and 16k), so I had to use different models from different companies to handle the different sizes of input. In practice, this meant that podcasts under an hour were handled by ChatGPT 8k or 16k, and ones longer than an hour were handled by Claude. I worked to ensure a consistent output regardless of which was used.

The infrastructure was super basic - a few bash scripts, a couple of python scripts, and data stored in CSVs and HTML files. I deliberately didn’t use a database, to keep it simple for an MVP, as there were only about two read/write batch operations per hour. The podcast feeds were handled by an open source podcast tool.

I used the product in my personal life, and with a number of test users. I was particularly impressed at its ability to handle complex information on veterinary anaesthesia, just as well as more conceptual topics on culture and design.

I also investigated the legal and ethical side of the work. Legally it seemed justifiable as fair dealing, given the extra analysis provided (although I would have wanted to confirm with a lawyer, and access to legal advice is fiddly to arrange, and expensive!), but ethically I wasn’t so sure. Ideally I wanted to extract the adverts from a given episode, and then preserve these and share them with the user alongside the summary, but in practice this was hard. I found that the LLMs struggled to pick out which material was an advert from a transcript alone. Worse still, more fundamentally, the way that lots of podcasts serve adverts is more personalised than this, with targeted ad content being injected into the media file before being downloaded by the user, meaning that that targeting would be lost.

I was happy testing out the idea on a small scale, but didn’t like the idea of building a business off the back of other people’s hard work, without having a clear way to share that revenue. If I was running this as a full-time startup, I might have run full-speed ahead, and then worked backwards to solve the advertising revenue issue, negotiating a revenue share deal after we’d scaled enough to be relevant. As it was, I took this as a natural limit on the growth of the product, and have now removed its web presence.

What I learnt

• A hands-on understanding of the capabilities and limitations of LLMs for this type of task

• How to ration tokens and get the most out of a context window

• How to choose between different models for different needs and cost-optimisation points

• The speed of change in this area - with model performance and context windows improving each month

• That generative AI products work really well with the product-led growth model, whereby you charge users for the valuable outcome that they receive (rather than for something arbitrary like the number of accounts that an organisation has). This is because with LLMs, assuming you’re calling an API rather than hosting your own, you’re paying based on your volume of use of the API. So you’re paying the LLM to generate useful outcomes for your user. If the per-request cost of providing users with a valuable outcome is less than the price you’re able to charge a user for that outcome, then you have a potentially viable business.

• The rapidly-changing frontier of economic possibility for intelligence-driven businesses. Some of the new models like GPT 4 were prohibitively expensive, and so I avoided using them, as I couldn’t make a profitable business around them. But with each month, intelligence gets better and cheaper, and new types of business harnessing this intelligence become viable as a result. (And, of course, you should plan your business based on where this trajectory is going, rather than the current state of things.)

• Developing a pricing and business model for a B2C startup. I was able to work out how much it would cost me to save a user an hour of their time, and then I could work out how much I’d charge for that. Charging people for the outcome - the time saved - is a key principle of product-led growth, and it fit well with how the economics of per-user LLM API calls work, and aligned nicely with the core value proposition of the product.

• The nuance of working out price, combining a range of considerations: the go-to-market and product-led growth approach, the different LLMs and their associated costs, and different price points to encourage paid subscriptions. It was the most complex part of the whole development. I built a spreadsheet to model these dimensions to help me plan this product as a business, and work out whether I thought I could make a profit. It included foundational factors like:

  • Transcription and summarisation cost for 1 hour (GPT 3.5 16k)
  • Transcription and summarisation cost for 1 hour (Claude 2)
  • % of non-paying users that we expect to convert (from an industry benchmark)
  • Expected card payment fee
  • Number of free hours that non-paying users get per month
  • Number of hours that a paid user gets each month
  • Price charged to paying user per month Some of these factors were fixed, and others I could vary at will. I used the modelling spreadsheet to play around with the variables that I could control (such as price), and updated the other details like cost of transcribing an hour’s material as these factors changed over time.

I explored different levels of generosity in the free offering, and different tiers of payment and the number of hours of time saved that the user would get each month for that price. The spreadsheet calculated a ‘profit per user per month’ value for a given price level. This modelling helped me explore a range of different options and think more widely and creatively than before - whilst being grounded in my current best understanding of the economics of things like API costs, likely conversion rates from freemium use, and payment provider fees.

I created different tabs, exploring freemium and free trial options, testing them against each other, before arriving on the freemium model.

• Calling out and testing risky assumptions and hypotheses in a commercial startup context, as opposed to my day job in government, where this activity is quite different

Overall, my two takeaways were the clear product capabilities opened up by LLMs, and the excitement of the challenge of interacting between a compelling product idea and the rigours of working out how to design a viable software business around it.

What I’d do differently next time

• Have a cleaner approach to the infrastructure. Using Whisper locally was economically fantastic - I could essentially transcribe podcast episodes at zero cost. This made the business viable, as Whisper API pricing was prohibitively expensive. But it also came at a cost: running the product off my regular desktop was messy, with the code running alongside everything else. If I built this again, I would take a container-based approach to orchestrating the infrastructure running locally, so that it was split off. Or I’d just use Whisper via API and incorporate it into my pricing model, doing some optimisations upstream of sending the data to it (e.g. trimming out silence) to reduce costs.

• Bake in a formal evaluation approach to ensure quality. I eyeballed a range of difficult cases early on, and monitored a % of the emails going out by hand. This quickly became unsustainable. If I was doing this again, I’d want to review summary quality in a reproducible way on each significant change (e.g. for each pull request, like you’d do in other software as part of a suite of automated tests).

What’s next?

LLMs are better and cheaper than when I built this product, and they have longer context windows. I want to use these capabilities to help people in the charity sector carry out parliamentary work more effectively.

“We start our journey to our dreams by wanting, but we arrive by focusing, planning, and learning.” Christina Wodtke

“you can release all the features you want, but if it doesn’t solve the underlying business problem, you haven’t really solved anything” Marty Cagan

This post is a summary of Christina Wodtke’s Radical Focus, on the method of using Objectives and Key Results to set goals and measure progress towards them.

Common reasons for things not getting done, and how to do better

The path to success is simple but hard

“To be successful, you have to focus on what matters. You have to say no a lot. You have to check in with your team and hold people accountable to their promises. You have to argue about whether or not your tactics are working and admit when they aren’t. There isn’t much complexity to OKRs, but they take quite a bit of discipline to do well.” Wodtke

What are Objectives and Key Results?

An Objective is what you want to accomplish. It’s inspirational. A Key Result (or set of key results) is how you’ll know if you’ve achieved a given Objective. A Key Result is measurable and grounded. A key result should be a stretch - you should have a 50% chance of achieving a given Key Result.

Neither the Objective nor the Key Result specify how the goal is to be achieved:

“Don’t tell people how to do things, tell them what you need done and let them surprise you with their results.” General George Patton

You’ll also set Health Metrics so that you make sure that you don’t lose sight of underlying performance of your organisation or product - for example, you might track service availability as a health metric, or customer churn, or team health (assuming that these aren’t things you’re trying to change through your Objective and Key Results).

Wodtke suggests setting 1 Objective, with three Key Results, per quarter. In my experience in larger, more complex organisations, you might need more. In the year ahead for my products at GDS, I’m setting out 3 Objectives.”

Objectives and Key Results process

1. Set inspiring and measurable goals
2. Make sure you and your team are always making progress towards that desired end state. No matter how many other things are on your plate.
3. Set a cadence that makes sure the group remembers what they are trying to accomplish and holds each other accountable

Now you can evalute your work against the OKRs, and ditch any work that doesn’t work towards them.

OKRs shouldn’t just be about meeting targets - they should be about learning

“Just setting metric-defined goals - even if you call them OKRs - isn’t enough. OKRs without focus and a cadence of learning become an exercise in making your numbers.”

How OKRs relate to performance management of individuals

Keep OKRs separate from individual performance reviews.

In this context, how do you assess performance and decide who gets pay and promotions?

“Instead of a one-time performance review event, use continuous conversations to coach and calibrate. Have one-on-ones at least twice a month and calibrate on three things: engagement, performance, and alignment. We use five levels of each and recommend both manager and employee share their view so perception gaps can be addressed quickly.”

Bonus insights

“Starbucks had started by bringing high-quality, European-style coffee to one small Pike’s Place cafe and charging three times the going price for a cup. Coffee used to be bottomless and a buck. Now it was a single cup of single origin and cost three. Starbucks had invented a marked and then owned it.”

When faced with an uncomfortable decision as a leader, you can use Intel’s Revolving Door test. “If we got kicked out and the board brought in a new CEO, what do you think he would do?” e.g. in their case, getting out of the memory business.

‘AI for good’ is the use of AI to prevent, mitigate, or resolve problems affecting human life or the environment.

Some things to be wary of from the outset when thinking about harnessing AI:

1. AI doesn’t always add value - in fact, most of the time in his career at Apple, Google and others, Robert Monarch has found that it doesn’t add value where you thought it would. So be wary of falling into an AI-first mindset and seeing every problem as an AI problem. E.g. covid swab tests were quick and cheap. So even though researchers showed that you could use AI and medical imaging to test for covid, that would have been much more expensive, and impractical.2.

Rather, AI is one potential tool in your toolkit:

1. Be careful with personally identifiable information. It can be used by authoritarian regimes to crackdown on dissidents.

2. Be wary of the impacts of potential errors of your application - e.g. mis-identifying illegal deforestation, or giving a wrong cancer diagnosis.

3. Unforeseen consequences can happen if your solution does work as intended. So think about adversarial use cases - ways in which others might use the system or the data to do harm rather than good. E.g. publishing black rhino preservation data, including location, could assist poachers. Speaking to stakeholders is essential to help understand potential impact and do no harm.

The AI for Good Framework

A structure for approaching AI for good work.

Most AI projects, and most for-good projects, fail to achieve a demonstrable positive impact.

“The majority of ideas, however well intentioned, are not going to work out. Being able to recognise that you are not likely to be able to reach a successful outcome is one of the most important leadership and design skills that you can develop”

At the end of the Explore phase, you should be able to answer the following questions:

• What is the specific problem you are addressing?
  • A good problem statement should:
    • Define the problem you are hoping to address
    • Identify key stakeholders
    • Give an idea of what success looks like
    • Not mention the specific technology you aim to deploy
  • You might take 4 months and 50-70 versions of a 1-page problem statement
• Who are the stakeholders?
• Do you have access to, or can you collect, the necessary data?
• Could AI add value? Where and how specifically?
  • Before testing out an AI approach, work out a baseline performance without AI to have something to compare to.
  • You might find that a simple solution works well enough, and that you don’t need to do anything more complicated.
  • Spotting correlations and patterns in the data is a good sign that AI could provide value. e.g. temporal patterns, correlations between different variables, spatial patterns
• How does the “do no harm” principle come into play?

At the end of the Design phase, you should be able to answer the following questions (or you may learn that you need to go back to the Explore phase):

• How will you address issues with imbalances (e.g. over- or under-representation of certain elements in your training set), biases, privacy, or other concerns with our data?
• What kind of model will you implement, and how will you measure its performance?
• How will your design address the problem you set out to work on?
• How will the end user interact with your system?

At the end of the Implementation phase, you should be able to answer the following questions:

• Is your model’s performance acceptable?
• Are end users able to successfully use your system?

Evaluation stage:

• Use the problem statement to guide your evaluation
• The AI is only part of the system that you’re delivering and evaluating.

E.g. Robert Monarch’s company Ibidon was working with healthcare providers in Nigeria to classify and triage incoming text messages. It was performing well technically - classifying well, and getting better as clinicians provided more and more labelled data to help train it. And the volume of responses, and patient response time, was improving. BUT clinicians felt that they were spending too much time annotating data, not caring for patients. (Perhaps they didn’t get a sense first-hand of the benefits that the system was bringing). So the human-computer interaction was a failure, and this led to the ending of the project.

AI and Public Health

Neonatal asphyxia is one of the top causes of sudden death and disability in infants.Every year over 1 million babies die of it, and 1.2 million develop medical conditions from it.

Ubenwa Health created an app to detect signs of asphyxia from a baby’s cry. It uses an algorithm trained on a labelled dataset of labelled cry recordings, to output ‘healthy’ or ‘not healthy’. They used a large and diverse dataset, with input from Brazil, Canada and Nigeria.

-The average number of premature deaths from air pollution globally each year is similar to the number of annual deaths from covid at the peak of the pandemic.

-The death rate due to air pollution is 50% higher today than 20 years ago.

-98% of urban areas in developing countries don’t meet accepted air quality standards

-Every $1 invested in preventing air pollution has an estimated $30 benefit

The type of air pollution that causes the most deaths globally is particulate matter less than 2 microns in diameter (“PM 2.5”).

Robert Monarch recommends Purple Air as a source of crowdsourced air pollution data

AI and Climate Change

Some examples of how AI can help with climate change:

Mitigation - reducing climate change

• Forest protection/development
• Renewable energy - wind and solar power forecasting, thereby making these renewable resources more valuable and viable; planning new facilities to be as impactful as possible
• Energy efficiency
• Carbon capture (perhaps, in the future)

Adaptation - adapting to climate change

• Building more resilient infrastructure
• Proactive disaster management - e.g. translation of messages in a disaster; damage assessment and resource allocation through image analysis
• Biodiversity monitoring - e.g. analysis of images

Mitigation + Adaptation

• Urban greening - cools the air and reduces greenhouse gases
• Water conservation - reduces water scarcity, and reduces carbon impact of water infrastructure
• Sustainable agriculture, which uses less water, land and energy, and also helps adapt to a changing climate

AI can help improve the predictability and usefulness of renewable energy:

Given fluctuations in weather - e.g. wind and cloud cover - it’s hard to predict how much power you’ll generate from solar or wind.

It’s important to be able to do this, because we need to generate enough power on the grid to meet energy needs at a given moment.

DeepMind found that they could increase the value of wind power by about 20% by increasing the reliability of forecasting wind power output up to 36 hours in advance. This allowed them to make advance commitments to the grid that were not previously possible.

To help manage fluctuations in renewable output over time, you have to keep coal power plants running at a low % utilisation (because they take hours to go from 0 to some % output, so you generally run them at a small % output and then ramp up when you need them, as you can do that much more quickly). But running at low % utilisation is inefficient - they are more energy efficient at high utilisation.

Large-scale energy storage doesn’t exist yet, hence the importance of forecasting energy output. (But this might change, with battery technology continuing to improve, and electric cars becoming more widespread)

A comparison of the potential quality of machine learning versus simple linear models:

A neural network, capable of describing non-linear relationships between variables, does a much better job:

Climate change and biodiversity

Biodiversity helps combat climate change, and is a good measure of ecosystem health.

68% of animal population lost since 1970 (WWF, 2023)

As a result of climate change, 50% of all species are on the move towards colder areas: 17km per decade on land, and 72km per decade at sea

1.5 degree increase in temperature will lead to a 70-90% destruction of coral reefs 2 degree increase in temperatures globally will lead to 99% of coral reefs disappearing

How can AI help monitor biodiversity?

• Bioacoustic monitoring - set up microphones in lots of locations, and then use AI to identify the different creatures that you can hear over time.

• Overhead imaging of natural ecosystems. E.g. NASA’s Earth Observatory tracking deforestation, or Sen2Coral project monitoring coral bleaching
• Citizen science: iNaturalist platform allows people to upload pictures. E.g. European Plants project
• Camera traps triggered by motion (Use the Megadetector project, which classifies things as animal, vehicle, human or empty, to filter out empty images, or ones containing humans)

Given the amount of data we’re collecting, manual data processing does not scale. So you need to use AI:

There’s a data bias issue: most global biodiversity is in SouthEast Asia, sub-Saharan Africa, and Southern America. But most data we have is of Europe and North America.

Starting with an existing network, tackling a very similar task to the one you’re interested in, and then fine-tuning it on your field-specific data, can work very well. E.g. taking a network trained on identifying cars and then training it to identify planes. This is transfer learning: you benefit from the earlier layers’ handling of more generic image recognition-type capabilities, and just fine tune the deeper layers.

AI and Disaster Management

A disaster is an event or situation that causes extensive harm to people, property or damage to the environment. The degree to which a disaster-impacted community can recover depends largely on risk management, response and recovery efforts, before, during and after a disaster.

The Disaster Management Cycle: Phases of Disaster Management

Mitigation

• Action to prevent or reduce the risks, impact, and consequences of disasters
• Resilient Infrastructure
  • E.g. dykes and levees in the Netherlands
• Early warning systems
  • E.g. Japan’s advanced earthquake and tsunami warning system

Preparation

• Actions to plan, train, and educate on procedures in the event of a disaster
• Emergency plans
• Training and drills
  • E.g. public education on natural disasters in Japan

Response

• Actions immediately after a disaster that provide aid
• Search and rescue
• Provide essential aid
  • E.g. temporary shelters, relief centres, food and water, medical care
• N.B. immediately after a disaster is often the worst time to try to introduce a new technology or process to a community impacted by a disaster or to disaster response organisations. So most AI assistance is generally in Preparation and Mitigation phases.

Recovery

• Actions to restore regular operations
• Assess damage
• Financial assistance
  • E.g. post-Katrina, financial assistance for businesses and individuals, rebuilding of damaged infrastructure, and a housing assistance programme
• Loop back into mitigation, to reduce the impact of future events

Some Examples of AI Being Used to Help with Disaster Management

• Cyclone Idai in Mozambique in 2019. Extensive manual water sampling and AI modelling (using rainfall patterns, water quality, and water levels) was used to predict where water-borne diseases were most likely to break out, which allowed health resources to be prioritised in the right place.
• Search engines and translation apps can be incredibly helpful in disaster response. Westerners often take the value of these for granted. “Unfortunately, the majority of the world’s languages are not supported in search or translation applications.”
• UNDP has created an AI early warning system using a number of different sources, including meteorological, hydrological, seismic and geophysical, and satellite imagery. Trained on historical data from data-rich locations, and extrapolates to cover areas with little on-site observation data. AI automatically detects and triggers early warning alerts. Deployed in Bangladesh, India, Indonesia, and the Philippines.
• The company Pano AI uses machine learning to do early detection of wildfires, which can have a big impact on being able to stop them getting out of control. This significantly increases the speed of alerting a professional, and the speed of telling the professional the specific location of the fire.

Limitations of Using AI in a Disaster Response Situation

• Need accurate data
• Potential for bias
• May have challenges relating to communication with local communities
• May require significant resources and technical expertise
• Privacy and accountability

… so incorporate AI, where relevant, as part of a broader approach. And always work to do no harm.

Helping Communities Help Themselves

Helping communities help themselves is generally the most impactful intervention you can make. A lot of this is about access to communication. Often communities in impacted areas are not included in tools like machine translation and search engines.

“Any technology that helps get information to linguistically diverse communities will help them in many situations, including disasters.”

“I believe that some of the work that I have done helping large companies deploy technologies in more languages has probably had a bigger impact on disaster response, than some of the time that I’ve spent working in refugee camps for the UN.” - Robert Monarch

Haitian Kreyol was underserved by AI translation after the 2010 earthquake. This meant that the large volume of requests for help couldn’t be understood by outside workers. Project 4636 - text 4636 with requests for help. Diaspora population manually translated messages from Haitian Kreyol. The labelled corpus that they created formed the basis of the first machine translation services from Google to Haitian Kreyol.

Analysing the messages can help build an understanding of needs and patterns changing over time, that can be used in preparation for future disasters.

Hurricane Sandy, 2012, Rob Monarch’s company had ahead of time produced a system for automatically assessing damage from aerial photos. They did this by using multiple humans to label test data, and then combining these assessments into the label for a given image, to address for human error. The map of damaged and undamaged locations was used to plan aid distribution and to prioritise reconstruction efforts. This was combined with other sources of information to inform decision-making.

They boost the number of example labelled images they have with augmentation:

• Rotating
• Enlarging
• Zooming
• Changing brightness
• Flipping

This augmentation effectively creates extra training data, derived from the training data that you already have.

In many real-world damage assessment projects, you would be likely to be looking at many more images containing no damage than those containing damage. In a scenario where you are training an AI system to do damage assessment using overhead imagery with a labeled dataset containing 99% images identified as showing no damage and only 1% images identified as showing damage.

If you trained your model directly on this dataset, the model could learn to simply predict all images as containing “no damage” because in this way it might achieve a higher accuracy (99%) than any attempt to correctly identify those containing damage as well. This is a common outcome when attempting to train a machine learning model with an unbalanced dataset.

Using methods of upsampling and downsampling, as well as data augmentation, could help to balance the dataset slightly, and train a more robust model. Collecting more images showing damage (if possible) would help balance the dataset and allow you to train a more robust model.

Metrics: accuracy, precision and recall:

• accuracy: all correct predictions / all predictions
• precision: true positives / all positive predictions
• recall: true positives / all actual positives

Harm and getting involved

You should not be the only person involved in working out how to do no harm. Local community members and other stakeholders should be involved.

Work on general-purpose technologies to help communities help themselves:

• Support low-resource languages with better technologies like translation and search
• Default to private data practices. Aggregate data and ML models can amplify privacy risks
• Avoid projects that involve analysis of social media data and work funded by oppressive governments
• Engage with impacted communities

Strategic challenges with unrepresentative datasets:

Example of the importance for diversity in the datasets used to train models:

• An image recognition surgery model trained only on white hands
• An algorithm for assessing damage to cars that was only trained on American cars

Examples of the potential problems of lacking positive examples in your dataset:

• Air pollution datasets often have fewer examples of high levels of pollution than low levels. This means that predictions of pollution at high levels will be less accurate than those at lower levels.
• Breast cancer detection is difficult because you have many more images or non-cancerous scans than cancerous scans
• Over-representation of one type of animal in a dataset of different animal photos could lead to an algorithm over-identifying the over-represented animal in images

Problems of insensitivity to local contexts or unintended consequences:

Example of problems caused by insensitivity to language:

• In Sierra Leone, during the Ebola crisis, an international news agency made broadcasts in the Mande language to Temne speaking areas. Mande was associated with the rival political party. This led to distrust, and to Temne speakers being less likely to visit health clinics. As a result, for every person who died from Ebola, 10 people died from preventable diseases because they’d avoided the clinics.
• Arab Spring in Syria. Some people posted information on social media about local conditions and road closures. These posts were often reshared by the UN - which linked these people, without their consent, to the UN, which was often seen as a malign foreign influence or invader. Ask: What is the impact of recontextualizing the data or model by you or an organization republishing it?
• Oppressive governments often use disasters as cover to identify and silence their critics. So be particularly wary of how information relating to personal information could be used to violate human rights. People speaking English on social media can be a targeted group (it’s white- and affluent-coded as a language). After Typhoon Haiyan in the Philippines, a small number of English-speaking responders were in danger of reprisals from the government.

Rob Monarch’s article of advice to people thinking about using AI for disaster response

Source: Research on Machine Learning for Disaster Response

• The majority of disaster response is helping the crisis-affected community help themselves. Therefore, tools that empower disaster-affected communities are most valuable, especially speakers of low resource languages.

• Information management is a much bigger problem than information discovery for professional disaster responders.

• Papers focused on machine learning for disaster response should not cut corners on the science.

• Papers promoting systems built by research labs have no place in disaster response.

• It is not possible to fully evaluate the sensitivity of data during a disaster, so responses to ongoing disasters should default to private data practices.

• Disaster response is often used as a cover for human rights violations, especially under authoritarian regimes, so research from authoritarian regimes that can violate human rights should be rejected.

• Researchers should not partner with non-operational aid organizations and should know how to spot the difference between operational and non-operational organizations.

• English social media processing is not interesting or useful for disaster responders.

• Ignore anything that relies on research published in “ISCRAM”.

• Apply the “Do no harm” principle to evaluating impact.

What is machine learning?

Machine learning is a sub-field of artificial intelligence. It is “the science of getting computers to learn without being explicitly programmed”.

Machine learning is useful because there are lots of things that we want machines to do that we aren’t capable of describing how to accomplish. e.g. identifying an image or getting a car to drive itself. So with machine learning, we design machines to work out how to do it, and then do it.

Some examples of machine learning applications: film recommendations, speech-to-text, spam identification, optimising wind turbines, search results, inspecting factory items for defects

The two types of machine learning

There are two types of machine learning: supervised and unsupervised. Supervised is more commonly used, and more technically mature, than unsupervised.

In both types of machine learning, the inputs are called features - and choosing which features to pay attention to, and which to ignore, and when to derive new features from existing features, is a key task called ‘feature engineering’.

In supervised learning: an algorithm is given a set of ‘right answers’ to learn from, and trains itself to work out more ‘right’ answers when confronted by new input that it hasn’t seen before. So you input a load of training data - input and output pairings, and then the algorithm trains itself to work out the output for a new input. For example:

In unsupervised learning, the algorithm is just provided with input data. The job of the algorithm is to find structure and pattern in this unlabelled data. For example:

1. Clustering algorithms group similar data points together. (E.g. grouping customers into market segments.)
2. Anomaly detection - finding unusual data points. Useful for fraud detection.

Supervised Learning

Linear regression

Linear regression is fitting a straight line to your data, using a machine learning algorithm. It’s probably the most widely used machine learning technique.

You provide the algorithm with a training set of ‘right answers’ - pairings of input and output data. The algorithm then works out how to work out, for a given input value, the appropriate output (this is known as ‘training’). This allows the algorithm to then take new input values and provide predicted output.

When training our model, we’re looking to make the predictions as good as possible. The way we track this is through a ‘cost function’. It tracks how close the algorithm’s predicted output is from the actual output value provided in the training data. So in training the algorithm, the goal is to minimise the cost function. (There are a few different types of cost function, e.g. a squared error cost function.)

Gradient descent

Imagining a straight line graph of type y = mx + c, we want an algorithm that works out the values of m and c, to give us the best predictive result (i.e. minimising the cost function). To do this, we can use an algorithm called gradient descent.

To carry out gradient descent:

• Start with some values of m and c
• Work out the cost function (i.e. how well this line fits your training data)
• Change m and c to try and reduce the cost function. Test values of m and c around the current values,
• Adjust m and c in the direction of the steepest reduction in the cost function (hence ‘gradient descent’)
• Continue until you reach a minimum. (If you’re using a squared error cost function, there will only ever be one minimum, so you don’t need to worry about local minima)

Tips for making gradient descent work better:

• Use vectorisation to allow for parallel execution (if you use a machine learning code library, it’ll be set up for this)
• Use feature scaling, so that all your features are in a more consistent range, making it easier for gradient descent to find the minimum point. Aim for each feature to be between -1 and +1. (This is particularly important if you’re doing polynomial regression)

Classification with logistic regression

Classification is most commonly done with a logistic regression algorithm.

Logistic regression takes in a set of features, and outputs a value between 0 and 1 This output is the probability that the class is equal to 1. E.g. an output of 0.7 means that there’s a 70% chance of the input having a positive output.

It’s similar to linear regression, but uses a sigmoid/logistic function to take the input and map it to the 0 to 1 scale, and uses a logistic loss function rather than a squared error cost function.

The ‘decision boundary’ is the line that divides your 0 classifications from your 1 classifications. Generally people set a threshold, and say that things above it are 1 and below it are 0. 0.5 is a common threshold. But it won’t always be the right value - e.g. if you had a tumour detection algorithm, you’d probably want to set a lower threshold for review by a specialist.

Underfitting and overfitting

You want your algorithm to generalise well - i.e., you want it to fit well to the training data, and to make high-quality predictions for examples that it hasn’t seen before.

If your algorithm doesn’t follow the pattern in the training data well, we’d say that it is biased (in a technical sense) and ‘underfit’.

If your algorithm is over-optimised for fitting the training data, you might find that it is ‘overfit’. This is because it’s so optimised for perfectly fitting to its training data, that it doesn’t do as good a job of predicting more generally. We’d say that this model has ‘high variance’. This can be the case in regression and classification:

Approaches to address overfitting:

• Get more training examples
• Feature selection: use fewer features when training your algorithm. (Lots of features + a small training set generally leads to overfitting)
• Regularisation: reducing the impact of a given feature, without eliminating it outright. Regularisation lets you keep all your features, but it prevents the features from having an overly large effect.

Regularization to address overfitting and underfitting

We can update our cost function so that it now trades off between two goals:

• The mean squared error function encourages the algorithm to fit the training data well.
• The regularization term encourages the algorithm to avoid overfitting, by keeping parameters small.

The value you set to lambda determines the tradeoff between these two goals:

• If lambda is 0, you’ll overfit, because you aren’t regularizing at all.
• If lambda is huge, then you’ll underfit, because you aren’t using the mean squared error value to make sure that the algorithm does a good job of predicting input:output pairings.
• So you want a lambda value somewhere in the middle

Neural networks (aka deep learning)

These algorithms are inspired by the way the human brain works. They are great at tasks like speech and image recognition.

Logistic and linear regression (which we have looked at so far) can only scale in performance to an extent with additional data. Neural networks, on the other hand, are much more effective at taking advantage of additional data.

As we now have more data, faster processors, and new computational capabilities unlocked by using GPUs for machine learning, we have been able to take advantage of neural networks in recent years.

The smallest component of a neural network is a neuron. A neuron takes an input - or set of inputs - does some computation, and then produces an output (called its ‘activation’)

A layer is a group of neurons that take as an input the same or a similar set of features, and that in turn output a few numbers together.

In the demand prediction example below, in addition to the input and output layers, there is a layer of 3 neurons in the middle.

The network will learn, over time, for each neuron, which of the features in the layer to its left are more important in making a prediction.

Neural networks give us the benefits of logistic regression, whilst effectively creating their own features from the input layer, and finding the best way to weight them - e..g affordability, awareness, and perceived quality being more usefully predictive than ‘material’ and ‘shipping cost’ alone. So instead of manually having to do feature engineering, the neural network figures this out for us.

Implementing neural networks

Tensorflow and pytorch are great for implementing neural networks.

They do a lot of the mundane work for us, allowing us to follow these steps:

• define the model (its layers, and their relationships)
• compile the model
• fit the model (running gradient descent and fitting weights to the data)

Then you can use the model to predict the output for new input values.

How to evaluate and improve a machine learning algorithm

Instead of using all your data for training the algorithm, hold back some of it to use for testing:

• Use 70% of your data for training, then keep 30% back to use for testing against the algorithm that you train.
• Compute the training error (how well the algorithm performs on the training data): this is called Jtrain
• Compute the test error amount (how well it performs on the test data): this is called Jtest
• If you have an overfit model, Jtrain will be low (i.e. the model will perform well against the data used to train itself, meaning that this cost function is low), but Jtest will be high, because the model isn’t good at predicting things not in its training set

(The above steps are for regression. For classification algorithms, work out the % of items that have been misclassified in the training set as Jtrain and the % of items that have been misclassified in the test set as Jtest.)

You can take this a step further, and add a cross-validation set (aka validation or development set). This helps you compare between different models, using the cross-validation set to do so, before subsequently using the test set to independently test its quality. The ideal is a model that performs well on the training and cross-validation sets. This implies that it’s able to learn patterns from the training set without overfitting. (The cost for the cross-validation set is JCV.)

Diagnosing bias and variance

Your model generally won’t work well the first time. Looking at bias and variance can help you work out what improvements to make.

You want a model with low Jtrain and low JCV - this shows that it isn’t overfit, and that it isn’t underfit (biased).

(“»” in the above means “much greater than”. I.e. JCV being much greater than Jtrain is a sign of overfit.)

How to establish a baseline level of performance

It’s useful to establish a baseline level of performance, against which to compare JCV and Jtrain. You could use:

• Human level performance
• Performance of similar, competing algorithms
• A guess, based on your prior experience

If you have a big difference between your baseline and the training error, you have a bias problem. If you have a big difference between your training error and JCV, you have a variance problem.

Working out whether it’s useful to add more training data

Graph the learning rate to see how far it is useful to add training data. To do this, plot the size of the training set as the x axis, and the error as the y axis.

If a model has high bias, adding more training data will not, by itself, help much. Resolve the bias problem before adding any more training data.

High bias:

With a high variance model - i.e. one that is overfit to the training set - you may find that your Jtrain is better (i.e. that this error rate is lower) than human performance. But that’s illusory, as it’s overfit - you will see that JCV has a higher error rate.

If you have high variance (as shown by a big gap between JCV and Jtrain), increasing the size of the training set should help improve performance:

If you have high variance:

• Get more training data
• Try a smaller set of features
• Increase the regularisation parameter lambda, so that your algorithm has less flexibility to fit very complex, wiggly curves

If you have high bias:

• Make your model more powerful, and more flexible to fit more complex data:
• Add additional features
• Try adding polynomial features
• Decrease the regularisation parameter lambda, so that your algorithm has more freedom to fit more complex curves.

Bias and variance in neural networks

Large neural networks are generally low bias, so the process for refining them looks a little different:

Machine learning development is an iterative loop

Transfer learning

Transfer learning is using data from one task to accomplish another, similar task.

You can take an existing neural network and apply it to a new task (as long as the inputs are of the same type). e.g. a network trained on identifying different classes of objects in images could be reapplied to identify the digits 1 to 9 in images.

Option 1: Copy over the entire neural network, keeping all the parameters and weights, and just retrain the output layer, as you want it to give a different output. This approach is best for small training sets, and is called ‘supervised pretraining’. There are lots of pre-trained neural networks openly licensed and available online, so you can get started on this quickly.

Option 2: Retrain all the parameters, initialised using the values of the parameters that you copied from the other context. This approach is best for anything other than small training sets, and is called ‘fine tuning’.

Fairness, bias and ethics

Examples of bias (in the ordinary meaning of the term, rather than the more technical use elsewhere in this course):

• Hiring tool discriminating against women
• Facial recognition having an unfair propensity to match dark skinned people to criminal mugshots

Examples of negative use cases:

• Deepfakes
• Spreading inflammatory speech by optimising for engagement
• Generating fake content for commercial or political purposes
• Using machine learning to build harmful products or commit fraud

Some tips for thinking about fairness, bias and ethics:

• Get a diverse team to brainstorm things that might go wrong, with an emphasis on vulnerable groups
• Carry out literature review on guidelines or standards that might exist for your industry
• Audit systems against possible harm prior to deployment
• Set up a mitigation plan ahead of deployment, and then monitor for possible harm after deployment, implementing the mitigation plan as required.

Decision tree algorithms

The job of a decision tree learning algorithm is to create a decision tree that works as well as possible for the training data given, but that also generalises to new input that it hasn’t seen yet.

So it will generate a bunch of possible trees, like below, and then work out which is best:

How to choose which feature to split on at each node, when making a decision tree?

Use an entropy function and maximise ‘information gain’.

We use an entropy function that gives a value of 0 when we hit 0 or 100% cat % in a given grouping. It has a value of 1 (an ‘entropy of 1’) when the group is 50% cats 50% not cats.

When choosing which feature to split on, go for the one that reduces entropy the most. The reduction of entropy is also known as ‘information gain’.

Keep splitting until the stopping criteria is met. This could be:

• When a node is 100% one class: i.e. entropy has been reduced to 0
• When splitting a node will result in the tree exceeding the maximum depth (to minimise complexity of the model and avoid overfitting).
• Information gain from additional splits would be below a defined threshold
• When the number of examples in a node is below a given threshold

Represent non-binary features with ‘one-hot encoding’

You can handle features that could have more than two values using ‘one-hot encoding’ - e.g. instead of having a feature ‘ear shape’ that could have 3 possible values - floppy, pointy or oval, you create 3 features, one for’ floppy ears’, one for ‘pointy ears’, one for ‘round ears’. And you set those as 0 or 1 appropriately.

More generically: “One hot encoding: If a categorical feature can take on k values, create k binary features (0 or 1 values)”. This also works for neural networks, and is a great way to encode categorical information so that it can go into a neural network.

Use tree ensembles to make decision trees more robust

A weakness of decision trees is that they can be quite sensitive to changes in the data provided. One solution is to build multiple decision trees: a tree ensemble.

Run each of them against a new test example, and then get them to vote on the final prediction.

Use ‘sampling with replacement’ to get variants on your training set, to train different trees of your ensemble. You might build something like 64-128 different trees in your ensemble

The XGBoost algorithm is the most common implementation of tree ensembles. It boosts the probability of picking examples that were mis-classified in previously-trained trees.

When to use decision trees and when to use neural networks

Decision trees and tree ensembles:

• Work well with tabular (structured) data. (i.e. the kind of thing you’d put in a spreadsheet)
• Not recommended for unstructured data (images, audio, text)
• Fast
• Small decision trees may be human interpretable (but don’t overstate this, particularly if you’re using a forest!)

Neural networks:

• Work well on all types of data, including tabular (structured) and unstructured
• May be slower than a decision tree
• Works with transfer learning. This is very important.
• When building a system with multiple models, it may be easier to string together several neural networks than several decision trees / ensembles.

Unsupervised Learning

In contrast to supervised learning, in which the algorithm is provided with labelled input and output data to train itself, unsupervised learning is just provided with input data. The algorithm’s job is to find structure and pattern in unlabelled data. For example:

• Clustering - finding data points that are similar to each other. (E.g. Grouping similar news articles, Market segmentation and DNA analysis)
• Anomaly detection - working out if a new example is similar or different to your training set.

Choosing between supervised and unsupervised approaches to spotting anomalies

Choosing between supervised and unsupervised approaches to spotting anomalies can be subtle.

Generally, anomaly detection is best if you only have a very small number of positive examples (e.g. 0-20).

Anomaly detection is better than supervised learning if you think that you might encounter deviations from what you’ve seen in the training set. If you want to be able to identify types of fraud or engine failure that haven’t happened yet, use anomaly detection.

A supervised learning approach learns in a positive way what an anomaly is, and tries to spot it - anomaly detection learns what normal looks like, and flags things that differ from normal.

Reinforcement learning

Reinforcement learning trains an algorithm by giving it rewards to tell it when it’s performing well.

The reward function tells the algorithm when it’s doing well and when it’s doing poorly. So reinforcement learning focuses on describing the outcome that you want, rather than describing the behaviour to get there. So it gives much more freedom for that to emerge organically.

Example applications:

• Controlling robots
• Factory optimisation
• Financial (stock) trading
• Playing games and computer games

Reinforcement learning isn’t used as much commercially yet, but is a bit of a frontier of research:

• Much of it has been done in simulations, which are easier than the real world
• There are far fewer applications of reinforcement learning than supervised or unsupervised learning

At each step, the agent is in a state, and chooses an action, in the pursuit of maximising a reward function. And as a result of these actions it will get to a new state.

Choosing between rewards

The agent needs to be able to work out tradeoffs between lower rewards that are quicker to obtain, and greater rewards that take longer to obtain. To allow it to do this, we specify a ‘discount factor’, by which we exponentially multiply rewards that happen later in time.

You can change the discount factor depending on how much you want to encourage aggressive pursuit of immediate rewards over long-term rewards:

• A discount factor of 0.99 encourages the algorithm to optimise for long-term rewards.
• A discount factor of 0.5 means that the reward is getting much smaller very quickly into the future, so it will likely try and optimise for immediate gains.

Markov Decision Process

The ‘return’ is the sum of the rewards that the agent gets, each of them weighted by the discount factor raised to the appropriate power based on the delay.

The goal of reinforcement learning is to work out what actions, given an input state, maximise the return.

This paradigm is called the Markov Decision Process. One key aspect is that the current state and future states are all that matters - it doesn’t hold context or baggage from the past. The future depends only on where you are now, not on how you got here.

Reinforcement learning and uncertainty

In some situations, the outcome of an action is not completely reliable.

In a stochastic (random) problem, there isn’t one sequence of rewards that we can perceive with confidence. So our focus can’t be on optimising the Return, as that’s a random number, but instead in maximising the average value of the sum of the discounted rewards. This is called the ‘expected return’, and it’s the average reward that the agent obtained after running the full sequence through many times.

Specifying rewards (positive and negative) is central to reinforcement learning

Specifying rewards incentivizes the behaviours you want to see (and not see) your agent perform.

For example, a lunar lander might have the following rewards specified:

• Getting to the landing pad: 100
• Crash: -150
• Soft landing: 100
• Leg grounded: 10
• Fire main engine -0.3
• Fire side thruster: -0.03

The negative reward for crashing is a clear disincentive, as we definitely want to avoid that result. The small negative rewards for using the thrusters are to encourage economical use of fuel.

Sometimes we deliver a feature and it doesn’t have the impact that we’d hoped for. One central challenge of our product work is reducing the likelihood of being in this situation.

As our focus is on delivering value, not just on delivering features, we need to be rigorous in pursuing only the ideas that are most likely to successfully drive our desired outcomes.

The solution

Adopt a ‘continuous discovery’ approach to product work. Continuous discovery is:

• At a minimum, weekly touchpoints with customers

• By the team building the product

• Where they conduct small research activities

• In pursuit of a desired outcome

Teams using this methodology can test 10-20 new ideas or approaches a week. This is only possible if, instead of testing whole ideas, we test the underlying assumptions that need to be true for the ideas to succeed.

The process of continuous discovery

1. Define an outcome

2. Map the opportunity space (Map the customer’s current experience; interview customers to understand their experiences by asking for specific stories, not their general ideas; make an easy-to-share-and-reference snapshot after each interview; automate the interview recruitment process so you always have a customer to talk to each week; map out the opportunity space by using the steps of the experience map or using interview summaries to identify key moments in time (nesting sub-opportunities under their parent opportunities))

3. Compare and contrast the parent-level opportunities (e.g. on opportunity sizing, market factors, company factors, customer factors); take the top one and do the same exercise on its children until you reach a target opportunity with no children. This is your target opportunity. (“Instead of asking “Should we solve this customer need?” ask, “Which of these customer needs is most important for us to address right now?””)

4. Generate ideas for the opportunity You want 15-20 ideas in total.
   • Generate ideas individually, then share, then generate more individual ideas.
   • “research shows that your first ideas are rarely your best ideas. The goal is to push your creative output to find the more diverse and more original ideas”

5. Evaluate the ideas together and make sure that they map up to the target opportunity.

6. Dot vote to pick the best 3 ideas. We’ll use prototyping and assumption testing to go from 3 ideas to 1 idea. We want to set up a compare-and-contrast decision. “Which of these three ideas best delivers on our target opportunity?” This helps us get past falling in love with our first ideas - and makes sure that we’re generating lots of ideas, the latter of which tend to be the best. Don’t test the ideas against each other directly - that would be loads of work. Instead, identify and test the most important assumptions that underlie the ideas.

7. For each of the 3 ideas, to work out the key underlying assumptions, generate lots of assumptions - we won’t need to test them all. But generating lots makes it more likely that we uncover the riskiest ones. Types of assumptions:
   • Desirability: does anyone want it? Will people get value from it?
   • Viability: will it work for our business?
   • Feasibility: can we build it? E.g. is it technically viable?
   • Usability: are people able to use it?
   • Ethical: is there potential harm in building this? How to identify assumptions:
   • Story map your idea. Show explicitly the steps that the user (and anyone else involved) must go through to get value. Review each step for the 5 assumption types.
   • Pre-mortems. “Imagine it’s six months in the future: the product launched and it was a complete failure. What went wrong?”. It’s crucial to frame this as certain failure - i.e. it did fail - otherwise we don’t generate good ideas.
   • Walk the lines of your opportunity solution tree, from your idea up to your opportunity and up to your outcome. Ask:
   • • Why will the idea address the opportunity? There will be assumptions that you’re making here.
   • • Why will the opportunity address the outcome?

8. Decide which assumptions to test: Map your assumptions relative to each other, from least important to most important, and from weak evidence to strong evidence. Test the most important assumptions for which you have the weakest evidence. Important: “We aren’t testing one idea at a time. We are testing assumptions from a set of ideas”

9. Test the key assumptions: We want to get data on people’s actual behaviour, to generate evidence on a key assumption. (We’re seeking to mitigate risk, not establish absolute truth) So we should simulate an experience, and give the participant an opportunity to either behave in the way our assumption expects - or not. Ideas will often share underlying assumptions. So often a single assumption test can help us test multiple ideas.

Methods for testing assumptions:

• Unmoderated user testing (e.g. provide a stimulus, then provide tasks to complete and questions to answer. Do this asynchronously. “These types of [unmoderated user testing] tools are game changers. Instead of having to recruit 10 participants and run the sessions yourself, you can post your task, go home for the night, and come back the next day to a set of videos ready for you to watch.”
• One-question surveys - e.g. if we wanted to test “Our subscribers want to watch comedies” we could create a one-question survey asking “When was the last time you watched a comedy?”. As with everything else, make sure to ask about specific instances of past behaviour, rather than generalisations, or future-facing predictions.
• Using your own data

1. Update your mapping of risky assumptions, and revisit the ones that are in the ‘high importance, low knowledge’ quadrant (assuming they are above your organisation’s risk appetite). If you have the same things at the top, even after a smaller-scale test, then you could justify doing a more expensive test, e.g. a smokescreen test in production. “With assumption testing, most of our learning comes from failed tests. That’s when we learn that something we thought was true might not be. Small tests give us a chance to fail sooner.” Design for the best case scenario. They will still often fail. If you fail in the best-case scenario, then your results are clear. If you test with a less-than-ideal audience, then someone will argue that you need to tweak the audience to test it properly.

2. Measure the influence that this idea has on your outcome. Setting up your infrastructure to measure your outcome might take some work, but it’s essential. e.g. measuring successful job applications is hard. “Just because the hire wasn’t happening on our platform didn’t mean it wasn’t valuable for us to measure it. We knew it was what would create value for students, our employers, and ultimately our own business. So, we chipped away at it. We weren’t afraid to measure hard things”

3. “Track the long-term connection between your product outcome and your business outcome.”

How to get started with continuous discovery

Start talking to customers every week. This is the foundation for everything else.

How to measure adoption of these habits

• How long since you last did a customer interview? (We’d expect this to be 1 week or less)

• How long since you last threw out an idea after an assumption test? (We’d expect this to be 1 week or less)

I start by reviewing where our platforms have come from and the tremendous value they provide. I then talk about our current work - uncovering an extra layer of service transformation by helping teams without access to technical skills to transform their service.

From there I look ahead to the future of our platform offering for 2025-30. I argue that the next step for government as a platform is to directly help services transform. We’ll do this in two ways: first by going much further to help people make better design decisions for their services, and second, by helping services continually optimise themselves. For the first, we can make use of generative artificial intelligence to help people make better decisions when designing and creating services, and for the second, we can make use of platform-scale performance data and machine learning.

Where have we come from?

GDS started by significantly improving the online experience of interacting with the state, at the same time as saving a huge amount of money by working more efficiently. GDS brought together hundreds of disparate government websites into a single domain, with a vastly improved user experience, overhauling content and a number of key transactional services.

To help scale the transformation of government services, and drawing on the insight of all of this work, GDS then created Government as a Platform - a set of platforms and components that teams across government could use to build and run services (e.g. GOV.UK Pay, GOV.UK Notify, GovWifi, Registers, Performance Platform, GOV.UK Verify and Platform as a Service). These platforms are focused on doing one thing, and doing it really well. In addition to this, GDS bundled the design and frontend insights gained from its partnerships with departmental service teams into the GOV.UK Design System, GOV.UK Prototype Kit and GOV.UK Frontend. Taken together, these platform products meant that services across government now had easy access, without procurement or financial difficulty, to most of the building blocks needed to create services.

Services adopting these platforms benefit from consistent branding and quality of experience, allowing them to meet user expectations of how interacting with government looks and feels. This improves ease of use, and builds trust. Services benefit from being able to get up and running more quickly, from ease of procurement, cost efficiency, and the accessibility of these products.

At first these platforms were mostly focused on helping teams who already had access to technical capability, as shown by the excellent APIs and developer documentation of these products. But in recent years, these platforms have been expanding into helping teams with less access to technical skills - for example Pay’s payment links feature.

What’s happening now?

GDS’s Digital Service Platforms are operating at scale: each week Pay processes around £30m and Notify sends around 13 million emails, 14 million text messages, and 200,000 letters; the Design System is used on thousands of services. These platforms are the foundation and engine room for a huge range of services across the public sector, from schools and local councils up to massive central government departments.

The procurement, contractual and financial arrangements are designed to make these platforms as easy as possible to adopt. GDS pays centrally for running the platforms, and outside of passing on any per-transaction costs (e.g. the fee for taking a payment, or sending a text message) the offering is free to use. GDS operates a self-service model that keeps operational costs as low as possible, and makes adoption quick and easy for departments. The tremendous growth that we saw during covid was testament to this institutional self-service model, and to the design of the economic approach to the platforms as a whole.

Building centrally and reusing widely is incredibly economically efficient, with a massive return on investment. Notify is projected to save the taxpayer £63 million this financial year, and the Design System £38 million. As we widen the range of services that we can serve, the potential benefits increase further still.

We’re growing adoption of our existing platform products year-on year. And we’re adding to our set of products by building GOV.UK Forms. Currently in private beta, this product addresses a big need that our platform products weren’t previously serving, by allowing services to collect information from users.

Adding Forms also lets us start to change the nature of our platform offering. Up until now, if a government user has wanted to create a service using our components, they’ve had to stitch them together themselves, and think about hosting, technical integration, and end-to-end service design. Because what we offer is a set of tightly-scoped individual components, adopting one of our products doesn’t help you at all in adopting any of our other products, or in making a whole service that combines more than one product.

Forms gives us a chance to change this. Last summer we carried out research that suggested that there’s value in a more joined-up offering that can be used by teams without access to technical capability - but we want to test demand and value in practice. So in Q3 this year we’ve integrated Forms with Notify in a very basic way, and in Q4 we will be integrating Forms with Pay in a very basic way. We’ll use these integrations to test the demand for - and value of - a more joined-up offering. This quiet expansion of our Forms product amounts to an MVP of a platform that empowers teams across government to design, build, operate and iterate simple transactional services without the need for access to developers.

If this work continues to go well, in 24/5 we’ll build on this, progressing towards a platform offering that can either be consumed as individual components parts, or adopted as a whole, without users having to worry about integration, hosting, or any of what’s going on behind the scenes. Ultimately, we want to make it as easy as possible for government to build and run great services for the public, whatever the service team’s size, shape, budget and capabilities.

Helping teams without access to technical skills to transform their service allows us to uncover an extra layer of service transformation across government. This is the next step in the government as a platform vision, going from providing platform components to providing an easy-to-adopt platform that helps build and run the whole service. This helps boost the quality of these services, makes them easier to adapt over time, and saves the taxpayer lots of money.

With even small, simple services often costing hundreds of thousands of pounds (or more) a year to run, helping the thousands of non-transformed services fully digitise has economic potential in the hundreds of millions of pounds, even with a pessimistic estimate of the percentage of services that would be impacted.

Looking across to the rest of GDS, GOV.UK is soon to start building a native app, with a particular interest in young people, alongside its web estate, and Digital Identity are nearing important adoption milestones with some key departments and services.

Where is GDS going next?

This work across GDS should set the foundations for where we go next, from 2025 to 2030.

The GDS vision is to deliver the public a smoother, more proactive and personalised experience of interacting with government. Working back from this target experience, much of the strategy and work ahead is clear. Data sharing across services, where users consent, to reduce effort and friction, better control of the communication experience, the understanding and anticipation of user circumstances and needs to drive proactive action to help people when they need it, without them having to be experts in government; all underpinned by a smooth, secure and ubiquitous way to prove your identity. This will take a lot of work.

What’s the strategy for delivering this at scale?

To reach this desired future outlined by the emerging GDS strategy, we’ll need to carry out two waves of transformation. One wave is about continuing with the current GDS strategy of digitising the long tail of 9,000 non-transformed document-driven services. The second wave is about building on this and taking digital services to a higher degree of interconnected sophistication.

Hand-crafting all of this transformation would be slow and ruinously expensive, and the ongoing operational costs of a bespoke approach would be prohibitive. Public finances will continue to be under pressure, and government will need and expect rapid deployment of policy into service provision. So the 2025 to 2030 transformation needs to be powered by platforms, to allow it to happen fast, in a cost-effective and high-quality way.

Alongside this, we have an opportunity to be ambitious about our vision for the future of government services. We know that AI will get better over the next 5 years, so we should bake it into the heart of our strategy. This takes us beyond playing catchup to where we should already be, and presents a real vision for transforming services.

The next step for government as a platform is to directly drive service transformation

The next step for government as a platform is to directly help services transform. We’ll do this in two ways: first by going much further to help people make better design decisions for their services, and second, by helping services continually optimise themselves.

We will empower teams to make great design decisions when building digital services

We will use AI, the GDS Design System, and the Government Design Principles.

If our first shift is to go beyond empowering technical specialists to empowering service teams who don’t have access to technical specialists, our second shift is to empower teams who do not have access to service and digital design expertise. This is how we reach the long tail of thousands of government PDF forms.

Artificial intelligence capabilities are improving rapidly. AI can already be used to check content against a style guide, check the reading age of a document, or suggest alternatives to jargon. Soon we will be able to harness AI to help people make better decisions when designing and creating services and their associated content, particularly when we power it with an understanding of our Design System and Government Design Principles.

In this prototype, we use a large language model to suggest improvements to make a form question easier to read.

As these capabilities continue to mature, we’ll be able to help government design and build better services from the start - improving policy formation and service design.

Design decisions over the next 5 years are likely to become harder rather than easier. Service interactions may be increasingly mediated through AI agents and through more personalised or multimodal interaction patterns rather than flat HTML pages. For example, a user might access a GOV.UK benefits checker through an AI agent that operates it on their behalf, rather than by visiting GOV.UK themselves. Or a user might be told the steps needed to renew their driving licence by asking an AI agent that has read lots of pages on GOV.UK.

The experience of interacting with government through AI agents may be disintermediated, with private sector actors being increasingly in control of users’ experiences. Their goals and incentives may not match with user needs. Designing for these interactions, and distilling best practice into readily available components, will be a key role for GDS. Similarly, assuming the trust in government continues to become scarcer, we’ll need to design for the retention and reinforcement of trust in government in ways that go beyond the visual recognition of the GOV.UK Design System and GOV.UK Frontend library.

The Design System will broaden out from being a pattern library for components, and move into setting guidance for apps, operational design, policy design, service design and AI interaction surfaces. Cybersecurity threats will increase - over the last year, NCSC have already seen a nearly 64% increase in cyber attacks. So we’ll need to empower people to build services with great cybersecurity, whatever their level of access to technical and cyber expertise. There’s lots of cross-GDS collaborative work to be done to make this happen, with Digital Service Platforms partnering closely with Digital Identity and GOV.UK on this integrated service platform for government.

Taken together, this work sees us move from empowering technical specialists (the first years of the GaaP programme), through to empowering digital and design specialists without access to technical skills (the current work of the Digital Service Platforms programme), and into empowering policy and operational colleagues, so that we can transform all government services. Better policy and better design to make better services for everyone.

But making better design decisions when creating a service is only the start. We need to help government services improve over time.

We will drive ongoing improvement to individual services, and to the wider service platform

Digital government in the UK has never had good insight into how well services are performing. GOV.UK is working on this, through the Single Data Environment, which will start to build a foundation of data on how users interact with services and where they have problems. We’ll bake joined-up analytics into the Digital Service Platform, along with automatically-summarised insights from user feedback, and an understanding of the Government Design Principles. We’ll use these together to give a clear view of how each service is performing, and how it might be improved to deliver better policy outcomes and user experience.

For example, we might pick out a particular step of a service as having an unusually high dropoff rate, and highlight improvements informed by successful examples that we also see on the platform.

Services will be expected to act on these insights in a swift and timely fashion, and CDDO could be given oversight of these across all services, to make sure that this is happening in practice.

In harnessing data and machine learning, we’ll work carefully to scan for potential unintended consequences, and design deliberately to guard against the dangers of bias and inequity, and to be particularly alert to the needs and experiences of marginalised or vulnerable people.

So every service on the joined-up Digital Service Platform will benefit from continuous data- and insight-driven optimisation. We’ll surface this insight to the public and to the people building and running services (from policy through design, delivery and operations), improving scrutiny and decision-making. This will empower government and enable oversight by the public, with high-quality visibility of detailed service performance out-of-the-box.

In this prototype, we use performance data and machine learning to give visibility of a whole service. Key areas for improvement are picked out using machine-analysed data and user feedback. Suggestions for improvement span all aspects of the service, including content, form steps, and notifications.

These insights won’t just be used to drive improvements to individual services: they’ll feed back in to components and guidance, so that the platform as a whole - and all the services on it - becomes better at meeting user needs over time.

This is where we harness network effects. At the moment we don’t really make use of the benefits of having so many services using our products. In the next 5 years we’ll change this, and use the scale of government digital delivery to help improve the quality of digital services over time. This use of data and machine learning will directly drive the GDS vision of delivering a smoother, more proactive and personalised experience of interacting with government.

The opportunity ahead

The years 2025 to 2030 offer the prospect of all 9,000 government services being truly digitised, through a government-wide Digital Service Platform. More than this, they offer the prospect of these services being better designed, and improving over time - not just individually, but as an interlocking ecosystem that learns and strengthens together, to keep doing an ever-better job of serving user needs across all government services. This strategy builds on what government as a platform has accomplished up to this date, and uses newly-emerging capabilities, whilst harnessing the breadth, diversity and scale of government. This is GDS’s opportunity to drive the next stage of digital government, and for the UK to regain its international leadership in this space.

Disclaimer: This is early thinking, and not an official GDS blog post. Other possible futures are available - if you have one in mind, I’d love to hear about it. What do you imagine for digital government in 2025-30?

Your job is the goal you pursue – not the activity you do

What you do is only ever a means to an end. The end is what matters.

“I’m a taxi driver” vs “I help people get where they want to go”

Perfect organisation design is impossible

“In order to optimise the system you have to sub-optimise the sub-system. That’s a mathematical truth.” In assessing peoples’ performance, and compensating them, you can either measure the system or the sub-system. Both are flawed:

If you provide global incentives, then you’ll create a unified, aligned team, with cooperation and risk-pooling. But it’s harder to manage, you can get free-riders, and great performers will likely feel that their unique contribution is not recognised.

If you provide local incentives (focusing on more narrow performance goals or measures of success) then you get excellence, with focus and accountability, and can attract great people. But you get a silo mentality, with people only thinking and acting to pursue their narrow immediate interests. For example, if waiters keep their tips rather than putting them in a common pool, you’ll get good waiters – but if a waiter sees that another table is having a bad experience, they won’t be incentivized to intervene. Optimising for silos doesn’t make sense – there’s no use having healthy kidneys if the body as a whole is sick.

So incentive systems cannot drive a company properly. You need a mobilising, cohering vision and purpose to transcend this contradiction. This will inspire people to do the right thing, regardless of the incentive system around them.) The good news is that because this is a universal, unsolveable problem, you just need to do a slightly better job at solving it than your competition.

Material compensation is only helpful to a point. People value self-transcendence more highly. So make space for people to be themselves, but bigger.

Leadership is about inviting people to follow a mission

“Leadership is eliciting the internal commitment of others to pursue a mission” – it’s about moral authority rather than formal authority.

Going further: “Don’t work for me, work for the mission.”
“A true leader doesn’t say ‘follow me’, a true leader says ‘join me and we’ll follow the mission’”

You’re a part of everything that happens to you

Everything that happens in life is a result of the challenge and your ability to respond to it.

You can see life in one of two ways, which has a big impact:

1. Your experience of life is determined by what happens to you
2. Life is giving you raw material and you have to choose how to respond. It’s like a game of chess, and you can make good things out of bad raw materials. Take responsibility for your own choices and actions, even if circumstances aren’t great.

Acting with integrity is the only solution to the problems of life.

You must accept that you are part of the problem. You need to take responsibility and work out how you can respond. “If you don’t feel part of the problem, you cannot be part of the solution.” So ask yourself:

• How did you contribute (through action or inaction) to create the problem?
• What can you do now?
• What can you learn from this?

“Die before you die”

We will all die. Start from this inevitable point. What do you want to do with the time that you have left?
Heroism is living by your values, being vulnerable, telling the truth.

(Aside: I disliked how Reid Hoffman acted dismissively towards Rachelle Diamond who introduced the second session. This type of behaviour doesn’t suggest a high level of psychological safety)

Healthy tension and unhealthy tension

Healthy team tension manifests as:

• Constructive conflict encouraged
• Supportive challenge
• People sharing positive feedback with each other, raising each other up
• People offering help to each other.
• Innovation, morale and productivity.

Unhealthy team tension manifests as:

• Infighting and disputes
• Lack of conflict resolution
• Unfair work distribution
• Credit not being given
• Poor performance not being managed
• People ignoring the elephant in the room
• People being confused or lacking direction
• People unable to ask for help or be vulnerable
• Low morale and productivity.

Types of threat that people can feel

Tension often manifests because people’s threat response has been triggered. People can feel threats to:

• Status (importance relative to others)
• Certainty (being able to predict the future. Goals, roles, structure and process help here)
• Autonomy (our sense of control over events. Clear objectives and boundaries help here)
• Relatedness (our sense of safety with other people)
• Fairness (fair exchanges between people)
• Understanding (particularly important when things change)

Trust is foundational – Lencioni’s 5 dysfunctions of a team

Lencioni outlined 5 dysfunctions of a team. They lead into each other in succession:

1. Absence of trust (not trusting colleagues’ intentions or competence. Not able to show weakness, admit mistakes or ask for help)
2. Fear of conflict (no healthy or open debate)
3. Lack of commitment
4. Avoidance of accountability
5. Inattention to results

So the absence of trust is the root cause of all the other problems:

• A lack of trust leads to
• Inability to demonstrate vulnerability to each other, which leads to
• Fear of conflict, false consensus and back-channel conflict, which leads to
• Artificial harmony, which leads to
• Lack of commitment to decisions and plans, which leads to
• Ambiguity, which leads to
• Avoidance of accountability, which leads to
• Low standards, which leads to
• Inattention to results. Individual goals put above team goals.

When you do have trust in a team, you get lots of good behaviours:

• Sharing information
• Accepting feedback
• Accountability – taking responsibility for one’s performance
• Open debates
• Commitment
• Focus on achieving results

How to deal with dysfunction 1. ‘Absence of trust’

The factors that create trust:

Credibility = your expertise and how you appear to others

Reliability = dependability and consistency. Acting in accordance with expectations. Keeping your word.

Intimacy = ability to connect with the other person. Revealing your true self, likes/dislikes, strengths/weaknesses. Being someone that someone can confide in.

Self-interest = how much you put yourself and your needs first

An exercise that can help teams build trust in the ‘forming’ stage – the Trust Shield:

A one-page document with the following:

• Me at my best
• Me at my worst
• What I hope to bring to the team
• What I need from the team

Leaders should share theirs first, to model the vulnerability of sharing this information.

In retrospect, I’d have liked to see more on trust here, and less on the other dysfunctions. The other dysfunctions ultimately stem from a lack of trust, so it seems to me to be by far the most important thing to spend time on. This article on why agile doesn’t work without psychological safety has 5 ideas for bolstering psychological safety.

How to deal with fear of conflict (dysfunction 2)

We want constructive conflict, with dialogue, rather than arguments or debates.

It brings innovation and creativity, better information gathering and sharing, understanding of different perspectives, improved communication and reduced stress.

Techniques to get constructive conflict:

1. Separate interpersonal issues from the problem at hand
2. Turn zero sum negotiation into exploration and collaboration (or compromise, if that doesn’t work)
   1. Focus on interests rather than positions. Ask open questions to understand why someone is behaving a certain way and taking a certain position. Summarise what you hear them say and ask clarifying questions.
   2. Get away from the idea that the situation is antagonistic and zero sum. Invent shared interests and opportunities for mutual gain. Collaborate to create new choices for action.
   3. Make sure that both perspectives get outlined and engaged with. If either party’s views are not engaged with fully, the outcome will be suboptimal.
   4. Sometimes compromise is needed – one side experiences a mini win, and the other a mini loss.
3. Give quieter people space to talk. Sometimes you’ll need to call on people to contribute, and hold space for them to do so safely.

How to deal with lack of commitment (dysfunction 3)

Clear communication on goals, mission and objectives. Revisit these constantly and relate all activity – and individual objectives – back to these.

Make sure that any practical barriers to doing the work are addressed.

Connect work to people’s motivations. The main motivators are:

• Responsibility
• Recognition (inc salary)
• Sense of achievement
• Interest
• Opportunities for advancement
• Opportunities for personal growth

How to deal with avoidance of accountability (dysfunction 5)

Increase ownership:

• Define the outcome and trust people to work out how to get there
• Let team members formulate their own goals
• Use a RACI model to give clarity on roles

Dialogues need four roles to be successful

We need to constructively engage with tension, rather than ignore it. It’s always a sign of something that needs to be worked through. You can set up healthy dialogues by making sure that you have the following 4 roles being taken by participants. (These are fluid, and people can move between them over time. Nothing to do with job title, hierarchy or authority.)

1. Mover – initiates a topic or point of view, and brings focus to it
2. Follower – adds support to what is being said.
3. Opposers – challenges what has been said.
4. Bystander – listens and watches, expanding what is thought about. Doesn’t necessarily take a firm stand in support or opposition of an idea.

You want to have all 4 of these actions present, in balance.

What unhealthy conversation looks like

This happens when roles are missing or are out of balance.

Some common failure states:

• Movers and Opposers creating and destroying ideas endlessly.
• Someone being stuck in a single position. (E.g. someone being stuck in the Mover position, introducing more and more ideas until finally someone takes on the follower role and supports something.) This can cause the conversation to get stuck.
• One or more roles being silenced by group dynamics. This can happen to the bystander role, especially when a group is feeling tired.

How to implement the four roles in dialogue

Introduce everyone to the 4-role model. Having a shared understanding makes it easier to notice when things aren’t going well.

Some interventions:

• Invite a stuck mover to slow down and make space for others.
• Point out a Move/Oppose deadlock and allow the participants to self-remedy
• Invite a Bystander into the conversation to share their perspective
• Split the group into sub-groups, then bring their ideas back to the main group.

To break people out of established roles, use a liberating structure, or give people one of 6 thinking hats to wear. This encourages people to think in a way that doesn’t come naturally to them.

Measuring the value of service transformation – Matthew Lyon, Head of Economics and Analysis, Central Digital and Data Office (Cabinet Office)

Reform of the Animal Licensing service led to a 50% reduction in processing time, and a 30% reduction in time for FOI requests. 6,600 hours saved.

User satisfaction increased 68% to 77% between 2018 and 2020.

Prepare to raise a child service – saved 12 minutes per user. As they have over c650,000 users, that’s about 125,000 hours saved.

Department for Transport research suggests that we value leisure time at about £5-7 per hour, so you can get a £ value for time saved.

The cost of failure demand: measuring the impact of poor user experiences at HMCTS – Aliane Alves, senior service designer, and Sam Brierley, Head of User-centred Design at HMCTS

“A problem for a service almost always becomes a problem for the organisation providing the service”. Stuff like avoidable contacts, unsolicited inbound emails, rejected applications, staff doing data entry, lots of manual checking and cross-referencing.

Easiest way to understand failure demand is to do contextual research with support centre staff as they are usually at the receiving end of it.

The team came up with a snappy meme to help people think about the cost of failure demand: “A typical CTSC caseworker’s time costs 50p a minute” This soundbite was much more effective than showing people a spreadsheet.

41% of calls to the Apply for probate service were from applicants wanting an update on their case. This failure demand cost £40,800 a month.

They wanted to reduce this to 20% of calls. So they started using GOV.UK Notify to tell people about the status of their case. This cost £37,000 over 8 weeks (one content designer, one business analyst, one tech lead, and one developer).

They came within 3% of their goal. (For all phone calls, they record: who, what, why, and did a quant and qual questionnaire with call agents)

Improving the quality of user feedback collection on GOV.UK – Jeremy Yun, Senior interaction designer, GDS

Feedback comes to GOV.UK through several different channels and formats. The team mapped out the different ways that it’s collected and used.

They’re simplifying the frontend survey, helping people classify their feedback (to help downstream use: “can’t find”, “don’t understand”, “doesn’t work”, “other”), simplifying how information is collected and stored, using data science to help automate feedback analysis, and reviewing how we distribute feedback.

Designing and improving the TfL Go app – Hannah Kops, Head of Experience, and Dan Bean, senior product manager, Transport for London

The vision is: “A personal travel assistant for everyone in London, which helps you to make the right choice at the right time, and provides TfL with the insight to keep London moving”.

They’ve been using the pirate metrics for their MVP:

• Awareness – app store impressions
• Acquisition – downloads
• Activation – complete onboarding
• Retention – BAU, opens per day
• Referral – telling friends, (app store ratings?)
• Revenue – money per user

Impressive focus on accessibility from the start.

Solve deep needs, not superficial wants with Top Tasks – Gerry McGovern, author and consultant

Focus on designing for the top tasks that users have- typically 3-5 tasks account for 25% of all user activity.

You can work out these tasks by outlining a long list of about 50, then asking users to choose their top 5 from a randomised list. That gives you a league table.

Then benchmark and work to improve:

1. The success rate
2. The time to complete

Treating people as followers or as cogs in a machine is wasteful

People acting simply to follow orders can lead to big problems:

Marquet shares a story of a time when he ordered the submarine to move forward at ⅔ speed. His officer relayed the order. The helmsman then said that it was impossible, as the setting didn’t exist. Marquet asked the officer why he’d given the order, knowing that it was wrong – the officer said that he passed it on because he was told to.

“What happens in a top-down culture when the leader is wrong? Everyone goes over the cliff.”

Beyond this, if you treat people as followers, as cogs in a machine who are only there to follow orders, you miss lots of potential.

Marquet explains that a 135-person submarine would traditionally only have 5 people at a sufficiently senior position in the hierarchy to be expected to have “fully engaged their capacity to observe, analyse and problem-solve”.

We want to harness the brainpower, creativity and perspective of our teams

[If you’re doing knowledge work, or work that involves uncertainty or change, then you want to make the most of the understanding and experience of your team.]

“With emancipation we are recognizing the inherent genius, energy, and creativity in all people, and allowing those talents to emerge.”

So ask: what authority could we devolve to people to make their jobs easier?

“rejecting the impulse to take control and attract followers will be your greatest challenge and, in time, your most powerful and enduring success.”

Don’t move information to authority, move authority to the information

Instead of channeling information up the chain of command to decision-makers, deconstruct decision authority and push it down to where the information lives.

Focus on achieving excellence, not on avoiding errors

“Focusing on avoiding mistakes takes our focus away from becoming truly exceptional. Once a ship has achieved success merely in the form of preventing major errors and is operating in a competent way, mission accomplished, there is no need to strive further.”

One example in the book is a crew member producing an error-free but utterly useless navigation chart.

“I resolved to change this. Our goal would be excellence instead of error reduction. We would focus on exceptional operational effectiveness for the submarine. We would achieve great things.”

If you connect the work to its larger purpose, and focus on doing a great job of achieving that purpose, everything else will follow:

“Once the crewmen remembered what we were doing and why, they would do anything to support the mission. This was a stark contrast to earlier, when people were coming to work simply with the hope of not screwing up.”

“On [the submarine] Santa Fe, doing well on inspections was going to be the natural outcome of being excellent, not the goal. Operational and tactical excellence and preparedness for service to the country were what mattered. If we were excellent and prepared, the drills and inspections would take care of themselves.”

To divest control whilst retaining responsibility, you need to make sure that technical competence and organisational clarity are strong

If you give people more control to make decisions, but their technical competence isn’t strong, chaos will result. You need to build people’s technical competence to allow them to go from expecting to be told what to do, to being able to take authority and control. As competence increases, you can increase control.

Marquet advises that if you want a training programme that people actually want to go to, focus it on increasing technical competence. And delegate increased decision making as a result of passing the training.

To safely push decision-making down the chain of command, you also need clarity

Alongside competence, people need a clear understanding of what the organisation is trying to accomplish.

Instead of interrogating subordinates about a possible course of action, invite them to lead the questioning process

When deciding what course of action to take, don’t ask people a series of questions that will let you, the leader, determine if the proposed approach is sensible. Instead, ask people what questions they think you will have in your head. They can surface and address these questions.

This approach gets subordinates thinking about the issues that are important to the person the level above them in the hierarchy. “In effect, by articulating their intentions, the officers and crew were acting their way into the next higher level of command. We had no need for leadership development programs; the way we ran the ship was the leadership development program.”

Specify goals, not methods

The fire hose drill used to be very opinionated about who ought to be putting out the fire. The crew decided to change this drill to be less stuffy about who manned the hose, to focus instead on being able to put the fire out. That was the single most important thing. The goal was important, the method was something that could be adapted and improved.

“Provide your people with the objective and let them figure out the method”

Being quiet is important on a submarine. They used to monitor this in a top-down way: observation by the sonar operator, then investigated down the hierarchy. They decided to change this. They started asking people to self-report, and to non-judgementally investigate causes. This led to a reduction in noise in the ship.

The ship used to be slow to submerge. Previously this had been thought of, and optimised, as a series of disparate events. But now, Marquet set an overall goal: a target time for the ship to be submerged and stable. “That forced the crew not to think in terms of disparate events (under way, maneuvering watch, shift the watch below, submerge, trim the ship), with all the discontinuities in personnel and equipment, but to think of sticking all those events together. When challenged like that, they found ingenious ways to trim seconds and minutes from the transitions, which made Santa Fe a much more effective warship.”

Change “I recommend…” to “I intend to…”

With a skilled crew working towards an understood goal, give them space to take ownership by asking them to share what they intend to do, rather than ask what to do, or just make recommendations.

Saying “I intend to…” makes people feel ownership of their actions, making action more deliberate. And by stating an intent out loud, it also gives a chance for others to challenge if appropriate.

Similarly, ahead of setpiece events, don’t brief people about what’s going to happen and what they are supposed to do – instead, invite participants to show that they are ready for the event. This gives them ownership.

“A little rudder far from the rocks is a lot better than a lot of rudder close to the rocks”

Quick, regular sense checks of work in its early stages can help make sure that what’s being done by subordinates matches with the overall intent.

These short, early conversations aren’t about telling people what to do. Instead, they are opportunities for a subordinate to get early feedback on how they are tackling a problem, retaining control of the solution, and to make sure that there’s clarity on the intended goal.

“Many lasted only thirty seconds, but they saved hours of time.”

Don’t try and change people’s behaviour by preaching – instead, make concrete structural changes that give ownership

“Don’t preach and hope for ownership; implement mechanisms that actually give ownership.”

“we didn’t give speeches or discuss a philosophical justification for the changes we were going to make. Rather, we searched for the organizational practices and procedures that would need to be changed in order to bring the change to life with the greatest impact.”

In David Marquet’s case, expanding the power of the chiefs on his submarine started with giving them control over their men’s leave – a decision previously made by the captain.

He chose to change behaviour, hoping that it would lead to new thinking, rather than trying to change thinking, hoping that this would lead to new behaviour.

One concrete structural change that gives ownership is eliminating top-down monitoring systems. (Not data collection and measurement processes, but processes whereby senior staff are checking up on and directing the actions of junior personnel.) These top-down monitoring systems have “a pernicious effect… on initiative, vitality, and passion”.

Guiding principles need to be useful, reinforced and based on the real organisation

Crowd-source strengths and guiding principles and trim them to a page.

Only include a principle if it helps you make a decision between two different courses of action.

Principles need to be reinforced and people evaluated against them.

“Guiding principles have to accurately represent the principles of the real organization, not the imagined organization” – there’s no point including principles that the organisation doesn’t actually hold to.

What’s new in Web Standards? – Dan Appelquist

Dan is Samsung Internet Director of Web Advocacy, co-chair of the W3C Technical Architecture Group and Open Standards Board member.

Connections between the web and your device:

• WebNFC
• Web Bluetooth
• Web USB
• Serial API
• Gamepad API
• SMS Receive API
• Contacts API
• Clipboard Access APIs
• File system APIs

A lot of these need privacy implications thinking about. e.g. clipboard access API could be done without the user knowing what was being done.

New features that make the web able to develop a richer experience:

• Progressive Web Application functionality: Manifest File, Service Worker, Push Notifications, Badging API
• WebXR
• Web Assembly
• WebGL
• Web Payment

New layout capabilities:

• CSS Grid
• “Houdini” APIs (let you script style within the javascript layer)

New communications capabilities:

• WebRTC
• WebSockets
• Streams
• Web Transport

Enhancements to Web Security:

• Feature policy
• Signed Exchanges
• Packaging

What makes the web ‘webby’?

• Open
• Linkable
• Internationalised
• Secure and private
• Multi-stakeholder
• Does not require a single authentication method or identity
• Zero friction

What makes the web open?

• Built on open standards
• Based on user needs
• Transparent and open process
• Fair access
• Royalty free
• Compatible with open source
• Multiple implementations

What makes a standard open?

• Collaboration between all interested parties, not just suppliers
• Transparent and published review and feedback process. Wide review is crucial when thinking about new standards. W3C look at new things and think about them on GitHub

When building new specifications, how do we make sure that they are ethical?

W3C have a privacy and security questionnaire that they encourage everyone to work through when working on a new specification

Some ethical frameworks:

• W3C TAG Ethical framework
• Mozilla Manifesto
• ACM Ethics
• Google AI Ethics

You can get involved in Open Standards by joining a community group or working group.

API for Humans and the Machines – Kin Lane

Kin is General API Evangelist, and Chief Evangelist at Postman

You should follow a design process when creating an API.
That way you can tease out assumptions, and test value before carrying out technical development work. Iteration is quickest and cheapest when you do it before you write any code!

Start off by defining an endpoint for the API, the values that you’d want to send to it, and giving an example of what you expect it to return.
You can use tools like Postman, and publish this test API, giving you an endpoint for testing, and letting people try working with it.

With OpenAPI, Swagger and Postman, you can publish your documentation from the code.

Some things your API should have:

• A choice of response format. Don’t just give JSON – let people receive CSVs if they want them. This makes things more open to non-developers.
• A management layer: access keys, rate limits. Apigee, Tyk,io, Mulesoft and Kong are tools for this.
• Governance
• Monitoring
• Automated testing in your Continuous Integration / Continuous Deployment pipeline
• Security testing
• A clear point of contact for support
• A plan for communications. Announce your API and new versions. Explain the purpose and what’s changing. You should have a comms strategy around every release. Without evangelists and communications, your API won’t last.

The API now exists online.

Here’s an example request for the nearest 5 foodbanks to latitude 52.629958, longitude 1.298408. Results are returned in JSON format, which is machine-readable. I have plans for a human-facing experience too.

The service is running on an entirely renewably-powered hosting provider in Switzerland. The extra time taken to communicate with a server a bit further away isn’t significant.

Here’s what I think the next steps are:

• Create a clear, professional homepage (rather than just a line of text!), so that new users have an idea of the purpose of the API, and so that it looks credible.
• Produce documentation on how to use the API, so that developers understand how to interact with it.
• Create a web page that uses the API, taking a user’s location and showing them the nearest foodbanks and what they need. Having built the API, this feels like a natural next step. Anyone who goes to this page will be able to find out which food banks are near them, and what items they need.
• Tell people about it, so that developers can start using the API, and people can start using the service to find out what items their local foodbanks need them to donate. We’ll have two minimum viable products – one API, and one human-facing service – and it’ll be time to find out if there’s any interest in using them.

I’ll be collaborating to make the above happen, which is exciting! We’ll be doing some user testing as well, to see how people use the API and documentation.

My goal was “Make an API that, for a given geolocation, returns the nearest 3 foodbanks, with a list of the items that they need.”

I’ve achieved this for something running locally (i.e. just on my computer, but not on a web server that anyone could access). You can download the code and follow the instructions to run it yourself, if you have the Python programming language installed on your computer. I actually went slightly further than planned – you can specify the number of foodbanks you want to see, and you can also find out the items needed by a given named foodbank.

The next step is to get it running online so that anyone can use it.

I’ve been testing the risky assumptions

• If I know the URL of a given foodbank’s page on food donations, I can work out what items they need.
  Yes. I’ve written code to do this.
• All Trussell Trust foodbanks follow the same way of organising their websites.
  Mostly. About 9% of them don’t follow the standard format.
• All Trussell Trust foodbanks follow the same way of describing the items they need.
  As above.
• I can access or somehow generate a comprehensive and accurate list of all Trussell Trust foodbanks.
  Yes. I stumbled across this in the HTML on the Trussell Trust’s Find a Food Bank page. I can get this list with a single GET request.
• If I have a list of Trussell Trust foodbanks I can straightforwardly work out the URLs of their pages describing the items they need.
  Mostly, yes. I’ve written code to do this.
• I can scrape the information I need from the relevant server/servers in a courteous way
  Not sure yet. I assume all of the Trussell Trust’s standard sites are hosted on a single web server. I make a single GET request to get the names and URLs of all the foodbanks, but each ‘items needed’ page is a separate request. I’ve included a pause between each request, but I don’t know if it’s too long or too short.
• It won’t be very difficult to build a data representation of food banks and required items, or to store this in an appropriate database.
  This was quite straightforward. And I didn’t even need a database, as I’m going to hold all the information in memory and not manipulate it.
• Building and running the API won’t be too much fuss. (Or, less concisely: It’s possible to build a lightweight, modern infrastructure to host a database for this API and serve requests without too much complexity or cost.)
  I’ve built an API that runs locally. Hosting it online as a real webserver should be reasonably straightforward. That’s the next step.
  I’ve found an entirely-renewably-powered web host, which might help me meet my extra goal of running this API entirely renewably.

Read part 1 of this project. See Part 3 to find out what I did next

An API is a tool that lets you quickly interface between bits of technology. If a tool has an API, it means that web developers can quickly interact with it: either taking information out of it or sending information or instructions to it. Using an API to interact with a bit of software is generally quick and easy, meaning that you can spend your time and energy working on doing something special and interesting with the interaction, rather than spending your effort working out how to get the things to talk to each other in the first place. Twitter has an API which lets you search, view or post tweets; Google Maps has an API that lets you build maps into your website or software. I built a tool around the twitter API a few years ago and found it a real thrill.

The idea for this API came from Steve Messer. I haven’t worked on a creative/web development project for about a year, and I’ve been feeling eager to take one on. I know that I learn a lot working on a personal project. I also experience a fantastic sense of flow.

Inspired by the Weeknotes movement, I’m going to write a series of blog posts about how I get on.

Goal for the project

Make an API that, for a given geolocation, returns the nearest 3 foodbanks, with a list of the items that they need.

How I’m approaching the work

I’m going to focus on the Trussell Trust, as they have a large national network of foodbanks – whose websites seem to work in the same way.

I’m starting by testing some risky assumptions. If these assumptions turn out to be wrong, I might not be able to meet my goal. So I want to test them as soon as I can.

Currently known risky assumptions

• If I know the URL of a given foodbank’s page on food donations, I can work out what items they need.
• All Trussell Trust foodbanks follow the same way of organising their websites
• All Trussell Trust foodbanks follow the same way of describing the items they need.
• I can access or somehow generate a comprehensive and accurate list of all Trussell Trust foodbanks
• If I have a list of Trussell Trust foodbanks I can straightforwardly work out the URLs of their pages describing the items they need
• I can scrape the information I need from the relevant server/servers in a courteous way
• It won’t be very difficult to build a data representation of food banks and required items, or to store this in an appropriate database.
• Building and running the API won’t be too much fuss. (Or, less concisely: It’s possible to build a lightweight, modern infrastructure to host a database for this API and serve requests without too much complexity or cost.)

Side challenge

Can I host this API in a way that is carbon neutral or, even better, renewably-hosted?

If I can’t, can I at least work out how much it’s polluting and offset it somehow?

What next

I’m going to start by working on the first risky assumption – “If I know the URL of a given foodbank’s page on food donations, I can work out what items they need.”

Read part 2 of this project to find out what I did next. See also Part 3

Tim Berners-Lee and Nigel Shadbolt, interviewed by Zoe Kleinman

Tim Berners-Lee described commercial advertising as “win-win”, because targeted advertising is more relevant. But “political advertising is very different… people are being manipulated into voting for things which are not in their best interests.”

Nigel Shadbolt: There’s a risk that people just move on to new shiny things. Creating a common data infrastructure is unfinished business.

Berners-Lee: We should be able to choose where our data is shared, rather than it just being impossible because systems can’t speak to each other. “You can share things with people that you want to share it with to get your life done.”

Shadbolt: Data sharing has to be consensual. Public data shouldn’t be privatised. We need transparency and accountability of algorithms used to make decisions on the basis of data. Platform providers are controlling and tuning the algorithms.

Berners-Lee: How might we train algorithms to feed us news that optimises for ‘aha’ connection moments, rather than feelings of revulsion?

Kriti Sharma – Can AI create a fairer world?

If you’re building tools with data, the biases of that data are perpetuated and potentially amplified, which can worsen existing inequalities. e.g. access to credit or benefits, or deciding who gets job interviews.

• Early on in a design process, think about how things could go wrong.
• Train machine learning or AI on more diverse datasets.

An MIT test of facial recognition found an error rate of 1% with white-skinned men. For darker skinned women, the error rate was 35%.

• Build diverse teams. Only 12% of the workforce on AI and machine learning are women. A more diverse team is more likely to question and correct biases.

Data Pitch accelerator

A EU funded accelerator, connecting public and private sectors to create some new data-driven products and services. A 3-year project.

28 data challenges, 13 countries.

4.6 million euros invested
14.8 million euros “value unlocked” – additional sales, investment and efficiencies. These are actual numbers, not optimistic forecasts.

datapitch.eu/datasharingtoolkit

How do we cultivate open data ecosystems?

Richard Dobson, Energy Systems Catapult
Leigh Dodds, Open Data Institute
Rachel Rank, Chief Exec, 360 Giving
Huw Davies, Ecosystem Development Director, Open Banking

Energy Systems Catapult:
If you want to move to renewable energy, you need to know what’s produced, where, and when.

So BEIS, through a Catapult scheme, set up a challenge on this. Seamless data sharing was crucial.

360 Giving:
Help grant makers open up their grant data in an open format so people can see who is funding what, why, and how much.

Open Banking:
Catalysed by regulation from the Competition and Markets authority. UK required largest banks to fund an implementation entity, to make sure it was effective and standards-driven to set up a thriving ecosystem. So they worked on standards for consent and security. Every 2 months the ecosystem doubles in size.

When encouraging people to contribute to an ecosystem, show value, don’t tell people about it.
Don’t talk to people about organisational identifiers. Show them why you can’t see their grants alongside the other grants because they haven’t been collecting these. People had such low insight into what other people were funding, that this was very compelling. Make people feel left out if they aren’t sharing their data.

Thoughts on making a healthy ecosystem:

• You need standards for an ecosystem to scale
• Accept that even with common standards and APIs you’ll get a few different technical service providers emerge, then people emerge who add value on top of this. (This was the experience in Open Banking)

“You can’t over-emphasise the importance of good facilitation at the heart of the ecosystem”
(I took this as: you need investment from somewhere to make this collaboration happen)
Open Banking did lots of work to collaboratively set up standards

Sophie Adams, Ofgem

Ofgem is working to decarbonise the energy system.

They’ve been working to make their data machine readable. They’ll then publish it on their data hub, through the Energy Data Exchange.

They’re taking in information from the Met Office and matching it up with price changes over time, to see the impact that weather has on energy prices.

17-18 Jan 2020 – Ofgem and valtech will co-host a hackathon on visualising environmental data. It’ll ask questions like “How could we decarbonise the UK in 5 years?”

Jo Judge, National Biodiversity Network

They get data in lots of different formats. Converting this into something consistent and usable is a challenge. Encouraging people to use this biodiversity data also takes work. Their State of Nature report visualises and summarises some of this data.

Philip Taylor, Open Seas

Mapping cod volumes and fishing locations over time, using publicly-available data, provokes conversations about management about this resource. (Of course I disagree with this conception of these creatures as a resource.)

Open Seas tries to take data and turn it into public awareness and better decision-making. They also use data to spot illegal fishing.

Chris Jarvis, Environment Agency

The Environment Agency use data to create UK Climate Projections, looking at the impact that change will have on weather. They’re working on linked data to allow their datasets to be built up in useful ways.

We used to think about flood defence. That’s not viable any more – we now think about resilience. The Environment Agency want to build a “nation of climate change champions” – people who know what’s happening, the risks and impact on them and what they can do.

2/3 of the 5m million people whose homes are at risk of flooding are unaware.

The Environment Agency are great at flood forecasting. Data collected up to every 15 minutes. They collect this over time, and make this over time.

This data is available through API. There’s information on hydrology and flood monitoring, including flood areas.

Aside: The Met Office’s data might also be interesting and useful. Information on their products and hourly site-specific observations.

Ben Rewis, Save the Waves

Dirty Wave challenge – crowdsourced data on dirty beaches, with an incentive to take action.

Users take a photo, tag it to their geolocation, and classify the type of problem that it relates to.

Advice for January hackathon: convene a set of people with shared values. Use technology to add value in some way. Get standards to encourage reuse and interoperability. Connect shared communities to a bigger picture. You might either get people to passively add data, or to interrogate, curate and work with what is already there.

Open Standards are publicly-available agreements on how technology will work to solve a particular problem, or meet a particular need.

The Open Data Institute has a useful definition of standards, open standards and open data.

Open Standards are good for laying the foundations for cooperation in technology, as they allow people to work in a consistent way. e.g. HTML is an open standard, which means that everyone can build and access web pages in the same way.

As technology develops, the standards can be updated, allowing innovation in a way that retains the benefits of interoperability.

How GDS works with Open Standards – Dr Ravinder Singh, Head of Open Standards, Government Digital Service

GDS outlines the Open Standards it supports. You can suggest standards that should exist. You’ll be asked 47 assessment questions. If a proposal comes out of that, GDS will take this to the Open Standards Board, which meets twice a year. The new Open Standard will be published on GOV.UK if it’s adopted. It’ll be incorporated into Service Assessments and the Technology Code of Practice.

PDFs are still the most frequently uploaded filetype on GOV.UK. So there’s a long way to go in making HTML and other open standards the default. (Why content should be published in HTML not PDF)

Supporting the adoption of open standards – Leigh Dodds, Open Data Institute (ODI)

CSVW lets you add metadata describing structure and schema

Open Standards for Data – ODI microsite

“Open standards for data are reusable agreements that make it easier for people and organisations to publish, access, share and use better quality data.”

ODI have produced a canvas to help you think about researching and designing a standard. The technical bit is the easy bit – the hard bit is getting people to agree on things.

Some advice if you’re building a new open standard:

• Don’t just dive in to the technology rather than understanding the problem
• Invest time in getting people to agree
• Invest time in adoption. Don’t just do the specification. You need guidance training, tools, libraries.
• Focus on the value you’re trying to bring – not, just the standard as an end in itself.
• If you think you want a standard, be clear what type of standard you mean. Types of standard include:
  • Definitions
  • Models
  • Identifiers
  • Taxonomies
  • File formats
  • Schemas
  • Data transfer
  • Code of practice
  • Data types
  • Units and measures
  • How we collect data

Opportunities for adopting open standards in government

Some thoughts from my group:

Schemas for consistent transparency publishing on data.gov.uk. Currently lots of datasets are published in a way that doesn’t allow you to compare between them. e.g. if you are comparing ‘spend above £25k’ data between councils, at the moment this isn’t interoperable because it’s structured in different ways. If all this data was published according to a consistent structure, it would be much easier to compare.

Shared standard for technical architecture documentation. This would make it easier for people to understand new things.

Do voice assistants have an associated standard? Rather than publishing different (meta-)data for each service – e.g. having a specific API for Alexa – it would be better for all of these assistants to consume content/data in a consistent way.

The (draft) future strategy for GOV.UK involves getting a better understanding of how services are performing across the whole journey, not just the part that is on GOV.UK. Could standards help here?

Misogyny is a system of hostile forces that polices and enforces patriarchal order.

Sexism: “the branch of patriarchal ideology that justifies and rationalises a patriarchal social order” Belief in men’s superiority and dominance.

Misogyny: “the system that polices and enforces [patriarchy’s] governing norms and expectations” Anxiety and desire to maintain patriarchal order, and commitment to restoring it when disrupted.

A reduction in sexism in a culture might lead to an increase in misogyny, as “women’s capabilities become more salient and hence demoralizing or threatening”

Women are expected to fulfil asymmetrical moral support roles

Women are supposed to provide these to men:

• attention
• affection
• admiration
• sympathy
• sex
• children
• (social, domestic, reproductive and emotional labour
• mixed goods, like safe haven, nurture, security, soothing and comfort

Goods that are seen as men’s prerogative:

• power
• prestige
• public recognition
• rank reputation
• honor
• ‘face’
• respect
• money and other forms of wealth
• hierarchical status
• upward mobility
• the status conferred by having a high-ranking woman’s loyalty, love, devotion etc

If women try to take masculine-coded goods, they can be treated with suspicion and hostility.

There are lots of “social scripts, moral permissions, and material deprivations that work to extract feminine-coded goods from here” – such as:

• anti-choice movement
• cat-calling
• rape culture

There are lots of mechanisms to stop women from taking masculine-coded statuses – such as:

• testimonial injustice
• mansplaining
• victim-blaming

An example of this asymmetric moral economy:

“Imagine a person in a restaurant who expects not only to be treated deferentially – the customer always being right – but also to be served the food he ordered attentively, and with a smile. He expects to be made to feel cared for and special, as well as to have his meal brought to him (a somewhat vulnerable position, as well as a powerful one, for him to be in). Imagine now that this customer comes to be disappointed – his server is not serving him, though she is waiting on other tables. Or perhaps she appears to be lounging around lazily or just doing her own thing, inexplicably ignoring him. Worse, she might appear to be expecting service from him, in a baffling role reversal. Either way, she is not behaving in the manner to which he is accustomed in such settings. It is easy to imagine this person becoming confused, then resentful. It is easy to imagine him banging his spoon on the table. It is easy to imagine him exploding in frustration.”

Praise, as well as hostility, enforces patriarchy

“We should also be concerned with the rewarding and valorizing of women who conform to gendered norms and expectations, in being (e.g.) loving mothers, attentive wives, loyal secretaries, ‘cool’ girlfriends, or good waitresses.”

Misogyny is not psychological

Misogyny isn’t a psychological phenomenon. It’s a “systematic facet of social power relations and a predictable manifestation of the ideology that governs them: patriarchy.”

Misogyny is banal. (“to adapt a famous phrase of Hannah Arendt’s)

This understanding of misogyny is intersectional

Misogyny is mediated through other systems of privilege and vulnerability. Manne does not assume some universal experience of misogyny.

Shout out to “The Master’s Tools Will Never Dismantle the Master’s House” critiquing middle class heterosexual white women over-generalising on the basis of their experience.

A quick note on privilege

Privileged people “tend to be subject to fewer social, moral, and legal constraints on their actions than their less privileged counterparts”

The income of the bottom 90% of earners in the US has stagnated from 1973 onwards.

Pre-1973 growth was more inclusive – the bottom 90%’s income grew faster than the average. Now the reverse is true, with the top 1 and 0.1% growing much faster.

Productivity gains stopped following to workers and started flowing to owners of capital.

A shift from constructive to extractive capitalism.

This has weakened the post-war social contract.

Beinhocker sees this as caused by the rise of neoclassical economics and neoliberal ideology and the associated structural changes driven by this dominant way of thinking about the economy and society.

The breakdown of Western capitalism

The rise of neoclassical economics and neoliberal ideology led to these structural changes:

1. Globalisation of finance
2. Shareholder value revolution (focus on stakeholders as the only legitimate interest in a company)
3. Neoliberal public policy agenda
4. Shift from virtue ethics to selfish utility maximisation

This shifted power in the economy, increased power of capital versus labour, and shifted rents to the top of the system. These rents were used to capture the political system.

Rethinking economics

To change the system, we need to revisit the ideas and assumptions behind it.

Neoclassical economics assumes:

• Micro – individual behaviour
  • People are rational utility maximisers
  • People are self-interested and atomistic
• Meso – social structures
  • Markets are efficient in allocation
  • Markets are self-correcting
  • Firms are optimally run
  • State ‘interference’ causes welfare loss
• Macro – system level behaviour
  • Macro is a linear adding up of macro
  • Natural state is full-employment equilibrium

But the underlying assumptions don’t match with reality:

Hence the following incorrect policy memes:

• Markets are always self-correcting
• Inflation is always and everywhere a monetary phenomenon
• You can either have growth or equality
• Trade is always welfare-increasing
• Raising wages reduces the demand for labour (e.g. minimum wage)
• The goal of business is maximising shareholder value
• Tackling climate change will cost growth and jobs

… and the incorrect belief that equilibrium systems cannot:

• endogenously grow
• create novelty
• generate spontaneous order
• spontaneously crash

New thinking: the economy is a complex adaptive system

• Complex: Many interacting agents and organisations of agents
• Adaptive: Designs and strategies evolve over time
• System: macro patterns emerge non-linearly from micro-behaviour

So in this way of thinking the economy is:

• An unfolding process
• Dynamic
• Non-linear
• Distributed
• Heterogeneous
• Networked
• Computational
• Autopoetic (self-creating)
• Self-organising
• Evolutionary
• Reflexive
• A social phenomena
• A cultural phenomena
• A physical phenomena

Reimagining capitalism

Capitalism isn’t great at efficient allocation, but it’s good at getting people to cooperate at innovate and solve problems. Beinhocker is happy with capitalism in general, just not the current implementation.

Prosperity isn’t money – it’s the accumulation of, and access, to solutions to human problems.

So we can redefine capitalism:

• The purpose of capitalism is to provide solutions to human problems.
• Wealth is the accumulation of solutions to human problems
• Growth is the rate at which new solutions are created and made available
• Prosperity is the set of solutions plus access
• Goal of business and investment is to create new, better solutions and make them more widely available
• Markets and governments together create an interdependent evolutionary ecosystem for solution creation and access.

Traditionally we’ve assumed that price is the measure of value. But that doesn’t help us measure between good and bad economic activity. (e.g. cigarettes or polluting activities. Or creating a fancy derivative product that ultimately makes the economy more unstable.)

The redefinition of capitalism above helps us distinguish between good and bad economic activities. Ask these questions:

• Is my solution creating a problem for someone else?
• Is my solution creating problems for society (e.g. derivatives)
• Is my solution today creating problems for the future?
• Am I solving a real human problem or just rent seeking? (e.g. high-frequency trading isn’t)

Human cooperation is fundamental to problem solving.

To demonstrate this, Thomas Thwaites tried to build a toaster from scratch.

Because cooperation is crucial, and because capitalism exists to solve human problems, then inclusion and a fair social contract are fundamental to capitalist prosperity.

Fixing capitalism

We need to shift our thinking

So we need a broad set of reforms to support:

• Inclusion:
  • Economic
  • Social
  • Political
• A fair, reciprocal social contracts
• Effectiveness in problem-solving innovation
• Demand for, and access to, new solutions
• Problem-solving vs problem creating (e.g. the environment)

• It’s now called the Service Standard, not the Digital Service Standard. This reflects the desire to create end-to-end services. This is better than creating digital services, and then (if you’re lucky) considering assisted digital as an afterthought. People are encouraged to provide a joined up experience across channels. What’s the user experience like if a user phones or emails you?
• Removed the requirement to tell everyone to use the digital service. Because digital isn’t always the right channel. And there’s already a financial imperative encouraging service owners to encourage people to shift to digital. So we didn’t need to push that any more. Instead, we need to encourage people to think more broadly about the service, not just the digital part.
• Focus on solving a whole problem for users, not just a part of it. The Standard encourages people to ask if the service is part of a wider journey. e.g. business tax registration is probably part of a broader journey of starting a business. So you should join up with those services too.
• The team have added more information on why the Service Standard expects certain things, and the benefits of following the Standard. So it’s less doctrinaire and encourages people to do the right thing.
• People are challenged to go beyond just thinking about accessibility, and to think about inclusion more generally: e.g. trans people and same sex relationships.
• The type of approach to meeting user needs is challenged. Is the service the right way to deliver user needs? Or should you publish content or make data available via an API instead?
• The scope of the service is questioned. If it’s too broad or too narrow it’s a problem.
• Removed the requirement to test with the minister.

Individual introductions

Mike Brown, Commissioner, TfL

The high-level vision for accessibility on London’s transport: Everyone needs to be able to travel across the capital easily, safely and spontaneously.

Alan Benson, Chair, Transport for All

Transport for All is a charity that works with TfL as a critical friend. Recent accessibility improvements include:

• new trains, which made many more stations level access
• ‘please offer me a seat’ badge. (The badge is a little controversial, but Alan’s happy with it as long as it’s optional and people don’t feel labelled.)
• disability training programme for TfL managers. Disabled people are teaching people who commission and run the services so that they can better understand the impact of the choices they make. No one else in the British transport sector is doing this. Alan thinks this is the most important improvement.

Gareth Powell, MD, Surface Transport, TfL

Transport exists to get people to a place they want to go. But 84% of disabled Londoners say that transport is negatively affecting their ability to get around and live their lives.

We haven’t designed with everyone in mind. Designing for step-free access is one thing, but what about designing for people with autism? Hence the training of managers, and more involvement of people with disabilities in the design process.

Panel session

Chair: Joanna Wotton, Chair, TfL Independent Disability Advisory Group

Panellists:

• Alan Benson, Chair, Transport for All.
• Michael Hurwitz, Director of Transport Innovation, TfL. (Says that his job “mostly involves worrying and internal procurement processes.”)
• Nick Tyler, Director of University College London’s Centre for Transport Studies.
• Ed Warner, Founder and CEO Motionspot Ltd

What are the biggest challenges facing accessible transport in London, and how might innovation help?

Michael: TfL is more used to working with massive companies than small companies. He’s keen to pave out the route to market for these promising minimum viable products. The commercial and contractual discussion about innovation isn’t trendy, but is super-important.

Nick: We need a better ability to test out new ideas. (His research group’s shiny new lab should help with this.) He wants TfL to be braver in encouraging innovation. Hong Kong and Singapore are more innovative than this country.

Michael: Singapore introduced universal design principles – trying to institutionalise the right types of design considerations. But Hong Kong and Singapore are more top-down, whereas London is much more bottom-up. e.g. there are 34 highways authorities in London. TfL does have the benefit of London’s size – the city is big, so you can test things out in small parts of it.

Alan: Try new things faster. Health and Safety concerns often lead us to hold back from testing things until they’re 100% ready. We could follow the lead of other industries that will launch things that are 80% ready.

Ed: Interesting innovations include the use of colour in wayfinding. e.g. Barajas airport in Spain. Japanese train stations play 7 seconds of melody before announcing a train’s departure platform and it cut accidents by 25%. The music settles everyone down a little bit.

Nick: We should work to make transport more enjoyable. This will make it more accessible. So look at cafes or playgrounds, and see how you might make things better on transport. This changes your way of thinking from “we need to make this system work to the timetable” to focus more on enjoyment (and, implicitly, value to humans).

Michael: Innovation isn’t always about technology. A lot of the most powerful innovations are in behaviour change. e.g. Dementia Friends at TfL. Or more assertive messaging to encourage people in Priority Seats to look up for someone who might need their seat. (They’re trialling this shortly.)

Nick: To get parents to change their behaviour, teach their kids. Parents listen to their children much more than they listen to the government.

How might we have better interfaces between public transport and the rest of the world?

Cities are people and we build the infrastructure around them. If we concentrate on building for people, then everything will get better.

Focus on the emotional side more than the concrete aspects of change. Don’t get stuck arguing about logic.

Anything you can do to reduce the threat that people perceive will help.

For change to happen

The Vision of the desired future state + the Discontent with the present position + the knowledge of the First steps to take MUST EXCEED Cost of change (financial, psychological etc)

People’s reactions to change can be characterised by their level of engagement, and their level of belief.

Why do people resist change?

Control, vulnerability and security.

How to deal with resistance

1. Identify it as resistance. Give people a couple of tries first.
2. Share your perception of resistance: “I notice/sense that…” (don’t make a judgement)
3. Ask a question: “What’s going on here?”, “What are you really thinking?”
4. Let them talk. Stay silent – they talk next.

The aim is to surface what’s really concerning them, rather than the resistance.

Why has government opened up data?

Probably the first motivation for opening up government data was to increase transparency and trust.
The MP expenses scandal led to a political drive to make government and politics more transparent.

Data.gov.uk was commissioned by Gordon Brown and overseen by Tim Berners-Lee, and built in in 2009/10.

In 2010 Prime Minister David Cameron wrote to government departments on plans to open up government data, promising “Greater transparency across government”. He wrote of a desire to
“enable the public to hold politicians and public bodies to account”, “deliver better value for money in public spending” and “realise significant economic benefits”.

What transparency data is published?

Theresa May published a letter in December 2017 clarifying what data departments and the Cabinet Office were expected to publish. This includes things like the pay of senior civil servants, and central government spending over £25,000. (Monitoring and enforcing this is an interesting challenge. Subsequent to this talk, I’ve been having some thoughts and conversations about how we might do this better.)

Transparency data around the world

In the USA you can track spend data back to policy commitments:

It will have taken a lot of political work to have consistent identifiers between different parts of government, so that this type of scrutiny is possible. Not glamorous, but very valuable – a trend you’ll see more of in data work.

My favourite example of transparency work in other countries is DoZorro.org:

Ukraine’s recent reform work is highlighted by its more open online public procurement system. An ecosystem of tools and an engaged community has emerged around this data. Citizen monitoring platform www.DoZorro.org has been used to bring 22 criminal charges and 79 sanctions so far.

This open procurement data has also led to the creation of a tool for identifying corruption risks http://risk.dozorro.org/:

It’s also led to the creation of a business intelligence tool http://bi.prozorro.org:

This takes us to the second big benefit of opening up government data.

Economic value of opening up government data

Open data improves data sharing within government. Previously, having to send a Freedom of Information request to someone else in your own department to access information was a thing that actually happened.

• Over 400 apps have been created using data from data.gov.uk.
• Prescription data was used to identify a £200m saving in switching drug brand.
• UK companies working with open data have total annual turnover of over £92bn.
• TfL’s open data generates up to £130m p.a.

Looking at the datasets on data.gov.uk that are used more, they generally have a clear economic use. These include datasets with information on land and property payments, or information on MOT testing stations. Other popular datasets are more related to understanding society, and are likely used by councils, third sector organisations and other agencies interested in planning service provision – e.g. English Indices of Deprivation 2010 and Statistics on Obesity, Physical Activity and Diet, England.

Measuring value is hard

I don’t think the above section was compelling enough. This is because measuring the value of open data is hard. There are a number of different techniques you can use to measure value. None of them are great – either you have something cheap and broad, which doesn’t give deep insight, or you have to commission a deep and expensive study.

Johanna Walker, University of Southampton, is a great source on this type of thing. She presented on ‘Measuring the Impact of Open Data’, Paris, 14 September 2018. (Aside: Johanna Walker suggested semantically-augmented version control as a way of ensuring quality, consistency and giving a better idea of how a dataset is being used.)

This post has more thoughts about the value of open data.

International Leadership

The UK’s early work on opening up government data has helped set the direction internationally.

International rankings are a relative thing, and the rest of the world has been catching up.
The UK has been first in the Open Data Barometer for 5 years in a row. Now we’re joint first with Canada

Global Open Data Index

Some other countries are doing really good things with procurement data. This work is hard – it took Brazil 5 years to get consistent identifiers so that you can link policy to budget to spend.

Challenges highlighted by user research

International rankings are lovely, but what do we know about the use of open government data and the challenges associated with this?

Government data is hard to find and use. Working with government data is hard – even for users who know government:

• Metadata is inconsistent and incomplete.
• Titles are sometimes obscure.
• Some datasets are published more than once.
• Data users can only understand how useful the data is once they’ve downloaded it and not all data users have the capability to download and view certain types of data.

Competing catalogues, standards, technologies and propositions.

We don’t have consistent, reliable data:

• Basic data skills and literacy aren’t strong enough to consistently produce data that is findable, usable, comparable and reliable.
• This means that many of the solutions designed to make data easier to find are only theoretically possible.
• It means that many services that need consistent, reliable data are also only theoretical

Where there is a relationship between the data publisher/producer and the data user, the quality of the data and metadata is better

Publishing is not enough

Publishing open data is a crucial start. But it isn’t enough.

We need to optimise for use and value.

“It is not enough to have open data; quality, reliability and accessibility are also required.”
Theresa May, December 2017

“Availability is only one aspect of openness, and the government also has a role to play in increasing the value of data, through structuring and linking datasets.”
Treasury Discussion Paper – The Economic Value of Data, 2018

“If no-one can find and understand data it is not open.”
Laura Koesten

How to get more value from data

Some ideas by Elena Simperl and Johanna Walker Analytical Report 8: The Future of Open Data Portals, 2017) on how we might get more value from open data portals. These actually apply more broadly. Some highlights to pick out:

• Linking and interoperability, including consistent schemas. So that you can get network effects from combining datasets.
• Colocation of documentation and tools to reduce barriers to entry.
• Organisation for and promotion of use – thinking about how to get value out of the data rather than seeing the job as finished when the data has been published. So reflecting back the use that has been made of data to teach and inspire others, some level of fostering of a community around the data.

Analytical Report 8: The Future of Open Data Portals, 2017, Elena Simperl and Johanna Walker

Opening up the Geospatial frontier

There’s lots of excitement around the potential of geospatial data. In the UK there’s a newly-created Geospatial Commission looking into how to open up this data. One of its key early tasks is opening up some of the Ordnance Survey’s Master Map data. This looks likely to be on a tiered basis, so free up to a point but paid beyond that.

Some highlights of what this Master Map gives include: Building Height data, Water Network Layer.

Even here we have the question of linked identifiers. The Geo6 are looking at this. This is the kind of graft we need to get the most value from our data.

In summary

Transparency and economic value are the key drivers behind government publishing open data.

But publishing data is not enough – we need to work hard to understand user needs and maximise use and the value derived from data.

Bonus: Extra insights from Steve Messer

TfL

They opened up their data en masse and waited for people to use it before improving it. Here’s the original download page and a post by Chris Applegate on why the first release could be improved. It has since been improved thus proving your point about ‘optimise for use and value’. Optimise after you’ve opened it up. The MVP for opening up data is creating a good standard and putting it out there.

Economic value

I’ve tried to find some £figures to help you back this point up, as below:

• McKinsey report Open data: Unlocking innovation and performance with liquid information
• Market value Open Data to reach 286 billion by 2020

Local government

I once made a timeline of notable events in local gov open data. It’s in this blog post.

He described Martha Lane Fox’s report on digital government as “deliberately a very large stone thrown in a fetid pond”.

What is digital?

“It’s as much about a way of working… the notion of working in small teams, working in multi-disciplinary teams, bringing the technology and design skills back into government to be able to deliver in the same way that Google, Amazon, AirBnB and the best startups deliver.”

“Multi-disciplinary, iterative, incremental, constant improvement, feedback loops, embracing open source as a way forward, communicating openly.”

“Ownership by civil servants of the iterative multi-disciplinary design of services, embracing policy, operations and technology”, focusing on user needs not government convenience.

“We are still in the foothills as a government of adopting internet-era styles of delivery. The humility to start by accepting that you don’t understand what citizens actually need and want and that you iterate your way humbly towards getting the policy outcome you want, rather than pretending you know what the answer is upfront. That is a fundamental change.”

Spending and spend controls

Now that GDS doesn’t have spend control, the GDS service assessment process is like having some “birdwatchers” turn up. (Implicitly toothless and irrelevant.)

“There’s a bit of carrot in the Service Standard. There’s no stick. And you need a big stick”

Treasury doesn’t really understand spending:
“You’ve got a Treasury that likes to spend capex on big projects. It’s easy. This world isn’t about capex it’s about teams. Investing in teams who continually improve and develop.”

What would Tom do if he had the power?

• Change the way that Treasury controls spend. Move from a capex to opex model. “Fund teams to deliver outcomes that ministers want, rather than big capex spend like we’re building a motorway. We’re not building a motorway, we’re building services.”
• Champion non-stop the work of mid-level teams across government doing great things.
• Reinstitute IT spend control

We need horizontal government. Ministers for things like payment or personal data. Platform-based, like internet-era companies, with individual services built on top of that.

Do we just need to get better at writing specifications?

“If you think you can write the answer in a specification and get the right answer, that’s the problem. It’s about my capex to opex point. I still think far too many bits of government think they can design the service upfront in a contract. And then your big suppliers are more than happy to deliver it to you. And when it doesn’t work they’ll charge you change control fees through the nose. That was the business model – you charge for change. You pretended for certainty upfront. I think there are still too many bits of government that would like to think that we can write a specification and get the answer we want, rather than take the harder path which is to start small, iterate, based on reality, with a team and accept the fact that we don’t know the answer upfront, we’re going to iterate our way towards it.”

Jobs to Be Done is about understanding the underlying problem.

What’s the difference between a Job and a Problem that a user has? It adds context: so it includes the fact that someone is driving to work, rather than just being hungry.

This kind of contextual understanding helps you improve your product to better solve some known jobs.

So in the case of selling milkshake to people visiting a fast food restaurant, optimising the product in its own right – making it chunkier or improving the flavour – was missing the point. When the business started solving for the underlying problem – the real ‘job’ the milkshake was being ‘hired’ to carry out – then sales went up.

Jobs to be Done is about user motivations and situations, rather than attributes and characteristics. This is good because those things change, and may not be important. ‘Specialist’ users of GOV.UK are often not very different to ‘Mainstream’ users. I found the same thing at NDCS – the imagined differences between “Professional” and “Parent” users didn’t really hold up to scrutiny.

A Job to be Done is the real world goal that the user hopes to accomplish. It’s not a task – those are the steps that someone takes to accomplish the goal, because they seem like the best way of doing so.

A ‘Job story’ is written in the form:

When [situation]

I want to [motivation]

So I can [expected outcome]

Take a Job and a Job Story, and you have a Job to Be Done. The Job stays the same, as it’s a deep underlying need, but the story and the solution can change over time.

High-level Jobs to be Done on GOV.UK

Do a thing

Life circumstances require you to interact with government

e.g. You want to become a childminder, drive a car in the UK, become self-employed or fish in UK waters.

Trigger: change in circumstances.

Steps:

• Decide where to get guidance
• Discovery loop (find -> read)
• Make choices
• Fulfil legal obligation

Advise on a thing

Help others to understand / interact with / use government structures / policy

e.g. help mum get her pension; help a client understand their visa options

Trigger: client request

Steps:

• Understand client circumstances and request
• Choose the information source (Policy, Data, Regulation, Legislation)
• Research loop (Find / Evaluate / Assess)
• Advise (Explain, report, clarify)

Interrogate / change a thing

Examine, influence, appeal what government is doing and the decisions it has made.

e.g. Appeal my parking fine, raise public awareness about welfare reform, push government to improve policy-making or change policy.

Trigger: Emergency events / data / being told to pay a fine / refusal / ministerial announcement / news / topical events.

Steps:

• Decide information source (generally data and policy papers)
• Research loop (Find -> Evaluate and assess)
• Decide (Support / challenge / prepare for the change)
• Engage (Consultation / campaign / in-person meeting, complaint, appeal)

First of all, tell us a little about your background and how you got to where you are today.

Disclaimer: When you’re applying for a job you spin this stuff into a coherent story.  Certainly there are threads you can follow, but the first part of my career was a series of improvisations and desperate expedients. After a few years I’m now in the fortunate position of being in control of where I want to go, and doing a job I love.

When we were around 9 years old, my mum and dad were dead against me and my twin brother getting any sort of games console. We learnt our times tables and got a family computer. (I later found out that they were planning on getting a family computer anyway…) Me and my brother played lots of computer games together. Starcraft, Little Big Adventure 2, Heroes of Might and Magic, Unreal Tournament. This probably trained my brain to be methodical and be confident in understanding systems, and sold me on the merits of shared exploration and adventure as only one person could play at once.

I went out-of-town to a grammar school in leafy Bucks, and was fortunate to learn HTML. I built my first website on freewebs, about Grand Theft Auto 3. The site had neon text on a black background, character skins that I’d customised myself, car mods, a hit counter and way too many popups. I desperately wish I could share it with you but it seems to have been scrubbed from the face of the internet. Beyond computer games I didn’t really have the confidence, knowledge or interest to dive deeper into what computers could do so this thread lay slightly dormant.

Having originally assumed I’d wanted to do a PPE degree, I decided that history was more my thing. It’s varied, has lots of interesting stories, and is all about people living through whatever situation they find themselves in. I tried my hand at economic history, political history, social history and cultural history. Turns out I really enjoy cultural history and medieval history.

In hindsight this degree was exemplary training to be a Product Manager – you have to rapidly master a new topic and communicate clearly and confidently about it. It trains you to be a bit of a renaissance person, interested in diverse things and with a bit of a sense of adventure. I spent hours each day in studious solitude in various libraries, but also spent a couple of hours each day in excellent conversation with other people in my college over lunch or dinner. That ability to dive deeply into someone’s domain of expertise and get to know how it works, what they care about and what they find difficult, was drilled really well by this.

During this time I assumed that I wanted to be a museum curator. I’d done a bunch of volunteering at the London Canal Museum, working on stuff like an audio tour of the walk from Camden Lock to Islington Tunnel, and a series of audio installations in the museum. In the summer before my final year at uni, I remember a chat with the chairman who advised me that there weren’t really any jobs to be had in the sector. I also reflected on the demographics attending museums and thought that I might be able to have a wider impact doing something else. But I wasn’t sure what that was.

This was a useful if poorly-timed rethink. I left uni without a real plan. I applied for lots of things, including being a radio weather reporter. I came within a hair’s breadth of becoming a trainee television creative, coming up with gameshow concepts. (Apparently some of the concepts I pitched were a bit high-brow. An acceptable rejection.) I nearly ended up working in policy in the charity sector, which is what my twin brother does now.

I’d done a fair amount of digital communications for university societies, just as social media was taking off. A meetup with an old teacher of mine, now working in medical education, pointed me in the direction of an opportunity to use these skills on the creation of an online learning community for training doctors. It felt like a break so I packed my bags and moved to Kentfor the contract.

I didn’t know about agile, or about user needs or iterative design at this point, but I gave it my best shot and started learning more and more about how the different parts of digital fit together. I was working on all aspects of the site, alongside a developer, so built my understanding loads. My manager was excellent at helping me understand the politics as well.

From here I did digital projects and digital communications jobs for a while. I remember one early job, finding out that a ‘web editor’ role in practice needed a firm technical understanding of how website backends worked. I got in touch with my friend Andy Jones, and he talked me through everything in a call that same evening. If ever I’ve been generous with someone else in helping them understand something technical, believe me when I say I’m just paying it forward.

I moved around quickly, much of this as result of the financial crash and funding cuts causing difficulties for the charity sector. Admired GDS since 2013 and was delighted to join in 2017.

Take us through a recent workday.

A standard weekday starts with a half-hour checking up on emails and Slack before a standup with the first of my teams at 09:45. We take ten minutes or so to run through work-in-progress, coordinate activities and discuss any blockers. I repeat the process at 10:00 with my other team.

As a Product Manager my job is about making sure that we’re solving the right problems, that we’re adding as much value as possible, and aligning around the right vision. Day-to-day this means that understanding, assessing and communicating purpose is my role. So I’ll be making value judgements on how to approach certain areas of work – whether to gold-plate a feature or do something basic, for example, or which from a set of options to work on – either with my team or solo. I’ll spend most of the day going from meeting to meeting – which sounds ghastly, but actually is a series of interesting chats with intelligent and motivated people, diving deep into interesting problems. I love working with the different specialists in my teams, and it’s a real brain workout. I do sometimes regret that I don’t carve out more time for solitary intellectual activity, diving deep into a puzzly problem, but if I had to choose, I’d take the group approach every time as it feels more dangerous and collaborative.

When I’m at my desk, I’m usually doing some form of written communication with someone not in my team – a user or stakeholder in another government department. There’s an interesting tradeoff here between ruthless prioritisation of your time, and maintaining relationships and being diplomatic. As a Product Manager, lots of people want your input, or for your to do something for them. So you also spend a lot of time politely saying no to things.

We pause our working day at 3:45 for 15 minutes to have fika – a break for coffee, cake and a chat. (Sometimes just the final one). The key thing is that it’s a non-work environment, so we have an actual break and people can get to know each other.

What apps, gadgets, or tools can’t you live without?

I feel at home wherever my desktop computer is set up.

I put a lot of value on having something that can play music.

What’s your best shortcut or life hack?

Sometimes not doing a thing is the best approach.

Take us through an interesting, unusual, or finicky process you have in place at work.

I insist that every card on my team’s trello board has a ‘What’ and a ‘Why’in its description, so that anyone can understand the purpose of the work. “Can you get a What and a Why on that?” has become a sort of joke catch phrase but I’m okay with that. It makes sure that everyone in a multi-disciplinary team understands the purpose of everything that is being worked on, and it trains people in communicating beyond their own specialism.

How do you keep track of what you have to do?

Google calendar for solo tasks. Trello cards for things I’m working on with my teams.

I’ve tried the Getting Things Done framework, but when things get busy I’ve found it too brittle.

I take notes of things to do on post-its and stick them to my laptop. I triage them to something electronic within the hour.

What’s your favourite side project?

My Social Summary. I built it from the ground up (that’s why the front-end looks rubbish ;)) and I learnt loads. It’s now a paid-for product. For £3 a month you get a daily (or twice daily) summary of the best material shared by your twitter network.

I’m also proud of the technical work I did on a creative project with my ex-girlfriend. I set up most of the standard things that a competent developer would do: version control, automated testing, a decent build pipeline, programmatic creation and destruction of review applications. The product idea was a fun and interesting one, and I got to play around with a few different APIs.

What are you currently reading, or what do you recommend?

For work-type reading, I’d recommend “Turn the Ship around” by David Marquet. It’s about intent-based leadership and how you can give people space to do great things. In knowledge work you need to lead this way, not in a command-and-control fashion.

The novels that have impressed me most as an adult are probably Nabokov’s Pale Fire and Paul Auster’s New York Trilogy.

Who else would you like to see answer these questions?

Some of the Lead Product Managers and Heads of Product at GDS and across government.

What’s the best advice you’ve ever received?

“Time’s short, your life’s your own. And in the end we are just dust and bones.”

“Work smarter not harder”

“Love is letting go of fear”

“Just Fucking Do It”

Empowered teams

Bad approach: Tech teams exist to serve ‘the business’.

Good approach (product culture): We set up teams to serve actual customers, meeting their needs in a way that they love, but in a way that meets the need of the business.

Why don’t most organisations do this?

Trust. Senior management don’t think that they can trust subordinates with decisions. They think the people aren’t capable.

Coach Bill Campbell, who coached Amazon, Apple and Google founders: “Leadership is about recognising that there’s a greatness in everyone, and your job is to create an environment where that greatness can emerge”

The role of leadership

Leadership is about inspiring people to a goal. Leaders need to provide:

• Product vision – a north star aligning every discrete product team in the organisation. Needs to be inspirational. It’s the main recruiting tool.
• Product strategy – the plan for getting from the current state to the desired end state. Not a product roadmap – more a sequence of big milestones.
• Product principles – princples to help you make difficult product decisions. e.g. at eBay resolving the fact that the seller team had a lot of power, and that could lead to buyers being treated worse. So eBay established a principle of always favouring the buyers, because if you take care of the buyers then the sellers will come.
• Product priorities
• Product evangelism. “If you ease up on evangelism there’s always someone that is going to take your resources”

The role of (people) management

“It doesn’t make sense to hire smart people and then tell them what to do; we hire smart people so they can tell us what to do.” – Steve Jobs was negative, but he didn’t tell people how to fix things. He gave them space to do that.

• Staffing. Put the right people in place and don’t set them up to fail.
• Coaching.
• Objectives (Not roadmaps. Give people business problems to solve. Let the team figure out how to solve the problem. Make sure that different teams’ objectives are reconciled with each other.)

The Basis for Trust

You need to get the right competencies in the team.

Trust needs competence and character.

Often managers don’t know or understand the needed competencies.

The All Blacks have a “No Dickheads” rule. They’ve kicked players and coaches over this principle of character.

“Cultural fit” is a bad thing to recruit for because it leads to hiring people like the people you already have.

The best way to recruit is to go for competence and screen out the jerks.

How to tell if a team is empowered

• The team is staffed with competent people with character, covering the necessary range of skills. (e.g. having teams with engineers and product and design)
• The team is assigned problems to solve, and they are able to decide the best way to solve those problems. (Not implementing features on a feature roadmap. 3/4 of those features won’t solve the problem. Do whatever features you need to do to solve the problem.)
• The team is accountable for solving the customer or business problem (i.e. delivering an outcome). Leadership gives the objectives – e.g. make the churn rate better. The team gives the key results – e.g. saying that we think we can do a 10% improvement. Teams can suggest objectives, but the organisation needs to think about coordinating and prioritisation organisation-level objectives, so needs to drive this process.

Bonus: Cagan’s thoughts discovery

3 things you need in a good team:

1. Are they tackling the big risks up front? (value, usability, feasibility, business viability) PM responsible for making sure that we resolve these risks (or at least well enough) before we write production code.
2. Are they collaborating closely around the problem, rather than passing it around between disciplines?
3. Not about implementing features, it’s about getting results. shipping things on time and budget is not important – it’s about getting results. time to money is more important than time to market.

The event was mostly questions and answers, so what follows is a blend of that material, bolstered with content from the Scrum @ Scale website.

The goal of Scrum @ Scale is to allow organisations to scale effectively, rather than finding diminishing returns as growth leads to complexity and confusion. So it’s solving a higher level problem than the Scrum framework, which is focused on individual teams.

The Scrum @ Scale framework is deliberately minimalist. It’s a “minimum viable bureaucracy”. So you can build what you need on top of it to fit your culture or existing practices.

The key point is that you have scaled daily scrums going up the organisational hierarchy. Issues that can’t be resolved at one level are taken up a level.

You have scrums, scrums of scrums and (if you need it) scrums of scrums of scrums. The scrums above the normal scrum are delivery focused – so made up of Scrum Masters or Delivery Managers. Focused on the ‘How’.

The highest scrum is at executive level. This Executive Action Team has the political and financial power to unblock impediments. So if a team needs a policy change it can get senior attention on it that same day. (Of course, that senior level chooses what it wants to do with that escalated issue.)

This Executive Action Team has a transparent backlog and ideally daily standups. It’s responsible for the transformation backlog and for the quality implementation of Scrum in the organisation.

There’s an equivalent Product Owner / Product Manager framework, focused on the ‘What’ and the ‘Why’. It includes key stakeholders, not just Product people. On a sprint by sprint basis you have coordination of backlog prioritization at each of these levels.

The MetaScrum exists to:

• Create and express an overarching product vision.
• Build buy-in to the prioritised backlog.
• Measure value and metrics.
• Create a shared definition of done that applies across scrums of scrums.

The Chief Product Officer sets strategy and vision for the product.

MetaScrums can scale up like Scrums of Scrums can.

You get up to the executive meta scrum level. At this level, organizational vision and strategy is set, aligning all teams around common goals.

Project management

At a high-level in the organisation, an end goal is decided. It’s also determined what work should be done to achieve this goal. Often this type of thinking and decision-making happens in an organisational or directorate strategy.

Project management is focused on delivering the specified output.

A project will usually be a discrete chunk of work, with some concrete deliverables, tied to a fixed pot of funding.

Product management

At a high-level in the organisation, leadership identifies a set of problems to solve, or goals to achieve.

In contrast to the project management approach, this is where senior direction stops. The product team – in particular the product manager – now has to figure out how to solve the problem or meet the goal. The task of figuring out how to reach the goal is devolved, and the product manager takes responsibility for maximising value within constraints. They vary the scope of what they build, and iterate their product over time, starting small and constantly testing hypotheses to test against risky assumptions, and to increase the value of what they’re delivering.

This approach places responsibility for the success or failure of the work on the product manager. Rather than following someone else’s idea, the product manager has to lead the process of working out how to accomplish the goal. They need to deeply understand the value they’re trying to bring, and act as the intersection point between users, analytics, technology and the organisation.

Funding is more usually ongoing. This is because if you have a big problem it probably won’t get solved once-and-for all through a short burst of work. Or if you want to grow something valuable, you’ll get better returns if you invest over time.

Why product management is better

• Understanding is emergent. People closest to a problem have the richest understanding of how best to solve it. We usually don’t know in advance which tactics will work and which won’t. This means that an approach that empowers the team on the ground – who will have far closer understanding of the problems than any senior manager – is better. The product manager, holds a strong understanding of the purpose of the work and of the different domains in their team, and can make very rapidly make holistic management decisions in response to new information.
• Work involving human beings is risky. There are lots of unknowns when building things that interact with humans. Releasing working software regularly, testing assumptions, measuring impact against goals, and iterating to achieve better impact, is better than just delivering something that seemed like a good idea to a senior manager when they were writing a strategy.
• It empowers and energises the team. You let the team own the problem and the solution. Organisational leaders are freed up to set overall vision and goals – a high-value task that better uses senior managers’ broad strategic view than micromanaging teams who have closer situational understanding than they ever will. It’s healthy to give staff interesting problems to solve, and demands more of them.
• Ongoing investment is better than boom-and-bust creation and ending of projects. Once a project finishes, people spend less time thinking about it. Knowledge and context is lost, and money and people are devoted to other things. This makes it more costly to improve this area in the future. It’s more efficient and effective to invest seriously in a problem or an area where value can be added, and build a product that gets better over time, along with a persistent organisational understanding that gets more robust over time, as iteration after iteration builds knowledge.

So if you’re a leader, give people problems not projects. If a problem is worth solving, or you want to create new value in an area of your work, make that an ongoing investment. Make the overall purpose clear and give your product manager the space to work with their team to work out how to deliver the results you need.

Gitlab doesn’t explicitly outline how to setup review apps with Heroku, but the Heroku API is well-documented, and it’s quite straightforward to get this up and running. You’ll need to set up a few environment variables. And make sure that the length of your branch name and your project name is no greater than 29 characters (Heroku has an app name length limit).

To create a review app:

1. The interesting things happen inside (Inside start_review_scripts.bash)
2. Create a new app with a URL derived from the name of the code branch
3. Get the right git branch, then push it to Heroku.

To delete the review app:

1. Delete the app

View this build pipeline in full.

Ending It. Emotionally, Responsibly, With Your Business Intact – Joe Macleod

With death becoming less visible in society, we expect perpetual heavenly consumption.

Consumer narrative structures are unusual in that they don’t have endings.

Consumer packaging often avoids recognising the real end of a product. So people generally aren’t aware of the fact that they can recycle old electronics.

Designing for endings and multiple engagement is a good idea:

• Permitting no fault divorce reduces suicide.
• Even the best gyms have 30% of people leaving per year. So design a good ending experience to get people back later on if they’re a serial gym-leaver.
• 50% of new financial product consumers leave within 90 days.
• 71% of app users stop within 90 days.
• GDPR and the right to be forgotten empower people to initiate endings. How might you make it easier for that person to come back? (GDPR is generally thought of as about privacy and data. But it’s also an opportunity to design for endings, and tidying things up.)

Designing from the perspective of novelty is bad:

• The promise of novelty: New process, materials, products -> an improved life.
• Not sustainable.
• Fails to provide individual reflection about consomption. Doesn’t acknowledge scarcity.

Designing from the perspective of endings is better:

• Design something from the perspective of the end of its use.
• Plan for recycling and for repeat use.
• Reclaim resources
• Reflect on consumption
• Develop and design new things.

What does a good closure experience look like?

Consciously connected to the rest of the experience through emotional triggers that are actionable by the user in a timely manner.

Some well-designed ending experiences:

• Epson’s PaperLab gives a visible ending and rebirth as it devours old paper and makes new paper.
• Marie Kondo’s Tidying Technique includes a thankful, meaningful goodbye to things that we have consumed.
• Fair phone lets users access their phone’s insides and update it.
• Kia cars plan for death in 7 years. This opens up the chance for them to sell more. Increase in market share from 1.3 to 3.5% due to the 7-year warranty. No change in design or price point.

Navigating the ethical minefield of digital design – Per Axbom

We’re hearing a call for ethics because people are being harmed by digital.

Some examples of design hurting people:

One terminally ill person shared their prognosis on Facebook, but Facebook didn’t show this update to their friends. Their friends had no idea they were ill. They probably died thinking that no-one cared.

Grindr exposing HIV status to other companies.

Of people who are getting hurt by design, mostly it’s happening unknowingly.

Can you map out the 1st, 2nd and 3rd order effects of your work? Eating chocolate is good in the moment, but has negative 2nd and 3rd order effects. Exercising is the reverse. How do you prioritise between them?

Good Inclusive Design is changing how we deliver public services – Katy Arnold, Head of User Research, Home Office

How to do it:

• Hire people with access needs Diverse teams are more likely to design well for diversity.
• Dedicated people and time Even if you don’t have budget. You might need to do it on top of the day job. Build networks with different organisations and groups to help you recruit users with different accessibility needs. As soon as you can, dedicate money and people to design.
• Set a high bar e.g. “include 1 person with an access need in each round of user research”. This has helped push contextual research, as you want to test in context with a user’s own devices and setup, not just in a lab environment. It also makes you think about ethics, consent and data protection. So the Home Office team joined the Market Research Society and got their researchers tested.
• Support innovation e.g. Home Office sharing posters with accessibility tips.
• Leadership Provide cover for good work. Set expectations. Reinforce the approach. Give practical support.

Building a new digital culture – Eve Critchley, Gareth John

The team wanted to move “From helpdesk to strategic partner”, and to encourage people away from tactically chasing individual pots of funding.

The team asked “What’s the biggest problem we need to solve?” Mind is good at information provision. But next steps for information-seekers to take aren’t always clear. This is the most important problem to solve.

3 goals:

1. empowering information at all touchpoints
2. marketing and income generation
3. using digital to improve the way the organisation works

Some tips on getting colleagues on board with change:

• 1:1 at the start of the transformation process helped understand pain points and desires. This seeded future conversations making the case for change.
• These champions were cultivated over time.
• Competitor research, including out of sector. But you don’t want to go too far with this because each charity serves a different set of user needs.
• Opportunity cost and ROI are useful frameworks to build shared understanding.
• Some pieces of collateral or public decision-making act as boundary objects. They render the project visible and intelligible to people on the outside and help them feel involved.
• Build buy-in incrementally.
• Storytelling helps.

Tech, polyphony and power – Ella Fitzsimmons

“You can’t control the story anymore, but you can choose what to emphasize”

Stories are political

CIA sponsored abstract expressionism, to boost American soft power, combating USSR’s Soviet Realism. Abstraction was interpreted as about being a space for free, Western individuals. Societ realism showed lots of people working joyfully together, implicitly focusing on social confirmity.

The Iowa Writers’ workshop was also sponsored by the CIA. Individualistic, senses-driven, narrow, quirky, materialist, warm stories. Not about systems, ideas, injustices or castles in the sky, which was more the USSR style.

Most of the stories we tell today are about heroes

In the hero’s journey narrative, tech startup garages are the liminal or otherworldly space.

The heroes journey narrative for tech sector is:

• kind of an outsider (A bit of a lie – rich and well-educated)
• uniquely gifted (See above)
• moments of doubt
• perseveres
• returns with the elixir

What happens when a heroic CEO is the centre of the company’s vision and what it builds?

It encourages certain kinds of personalities and behaviours, and the centralisation of power. It bleeds into the rest of your work – e.g. your software. It also leaves the organisation vulnerable, and shapes unrealistic expectations of a single person.

What can we do to change this?

• Change the stories, bring new people in.
• Amplify different voices.
• Show individuality without setting people up as heroes.
• Start to tell stories about many people.
• Facilitate other people telling stories.
• Shine Theory. Amplify the ideas of other people to start raising expectations.
• Who is quoted matters. Don’t just quote heterosexual cisgendered white men.
• “Attribution is a revolutionary activity”. Show the community of people that you are part of. Disrupts the idea of lone genuises.

Scenius – the genius of the scene. Jobs: “Great things in business are never done by one person. They’re done by a team of people.”

• Show what people do, not just what they look like.
• Ask, ask again, and offer support.
• Talk about money, both with people who are like you, and with people who aren’t
• Mix up who works on important projects
• Ask someone new to do the boring work.
• Give people time to prepare for presentations

Warka Water

A 12-metre tall bamboo tower that gathers water from the air. About 100 litres of clean drinking water per day. Inspired by lotus leaves and spider’s webs.
The project is named after the Warka tree, an Ethiopian fig.

Refugee Text

A text chat-bot to give refugees trustworthy information on their phone.

Refugee Nation Flag

Being stateless makes it hard to access rights, and being away from home makes identity and belonging more difficult. But creating a nation through a flag gives some identity, coherence and hope.

Mrs Fan’s Plug-in House

Designed as an upgrade to the historic hutong districts of Beijing, this pre-fabricated house slots into the existing courtyard structure. It should be affordable so that existing residents can stay living there.

The renovated Sala Beckett Theatre

The building was previously Barcelona’s Peace and Justice Cooperative Building. Some of what was there has been retained – the bare plaster walls, exposed timber and rose windows – but many things have been added.
I enjoyed the playful sense of remixing the past, of function and style shifting over time. In its confusion, it feels alive with possibility.

There were some interesting items here – but I did wonder if the exhibition should shift its focus towards designed things that go beyond the physical. Refugee Text was an example this year (and GOV.UK in the past) – but what about great services, concepts or algorithms?

Print in the Digital age – redesigning the Guardian newspaper

The concise popup on the Guardian’s new tabloid format was worth a quick visit.

I’d assumed that the new tabloid format was primarily about saving money on printing costs. But it was also an opportunity to re-design the approach to presenting printed information. Digital content design has now influenced print design. Writing and formatting are intended to facilitate scanning, with fewer blocks of text. Infographics are used more too.

Compare the two ways of presenting a single article in the images below:

Blogs

• Putting people first
• Inside Intercom
• Nielsen Norman
• Mountain Goat software
• John Cutler
• Melissa Perri
• DEV
• Mind the Product
• Product Hunt (also, the Product Hunt browser extension)
• GoSquared
• Silicon Valley Product Group
• Pivot Product Hits

Books

• About Face – the principles of interaction design
• The Little Black Book
• Step Up Club
• The Design of Everyday Things

Podcasts

• Aurelius Lab
• This is Product Management
• 99 Percent Invisible
• Intercom

Video

• Product School video AMAs

Reading lists

• Simon Cross’s PM essential reading list
• Attack with Numbers
• Ross Ferguson’s PM reading list

Exhibitions

• Design Museum’s Designs of the Year
• Royal College of Arts, Show of the Year

Events

• Product Tank Events
• Lectures, e.g. psychology

Formal guidance on the PM role and product lifecycle

• DDaT skills they need – different levels of PM
• GDS Service Standard
• Service manual
• Braun – 10 principles of good design

People to speak to

• Other Product Managers
• Other teams through internal show and tells
• Other Product Managers on a shared task, e.g. working through 52 weeks of UX together.

Other organisations’ open roadmaps

• Trello
• Monzo

Newsletters

• Exponential View

A service starts with a need and an idea of the outcome, but no clear idea of how this will be achieved.

Service design is the process of designing this set of touchpoints to meet the given goal.

Image credit: Lou Downe. See Lou’s great post on service design

Service design starts with asking the right questions…piloting a new course w/ @Martin_Jordan @gdsteam pic.twitter.com/LCRxx9FZL6

— Fajer (@fajq8) January 19, 2018

Currently there’s a disconnect between a user’s experience of a service and the government’s stated policy intent.

Often senior management will make a pronouncement like “We need a portal so that applicants can upload bank statements”

But your responsibility is to challenge this proscriptive approach, and instead understand the problem and goals before even thinking about building anything.

So ask questions like:

• Who are the users?
• What are they trying to do?
• Why now?
• What is our motivation?
• What outcomes do we want?
• How does it relate to a wider service?
• What are the key metrics?
• How will it help users?

To frame your problem statement, focus on the organisation’s desired outcome, and on what the users are trying to do.

That keeps you focused on what you’re trying to achieve, leaving you free to explore how best to achieve those ends.

As you start building, it’s useful to cycle between optimising the big picture of the service (and, as the policy process becomes more amenable, the policy behind it) and the closer detail of a given task. Oscillate between the meta and the matter.

In the future, services will shape government, not the other way round.

One vehicle for achieving this transition is a service community. Government is made up of disconnected units, but the user shouldn’t need to know how to navigate this complexity. A service community is a group of people whose touchpoints form part of a wider service. Newly-formed service communities include “Start a business”, “Employ someone” and “Import/Export”. They start by mapping out the current service, and then identify opportunities for improvement.

When designing a service, be mindful of: The end-to-end service (from the user’s first step towards meet their goal, through to a successful outcome) The front-to-back service (so make sure to include all back-office and technical functions) Different channels (not just digital!)

The two most important things to do when designing a service: Understand user needs Prototype and iterate

Check out IDEO’s Design Kit for service design techniques you can use. (And read about the design project I worked on if you’re interested.)

To improve an existing service, this flow of activities is useful:

• Service walkthrough
• User journey map + service blueprint
• ‘How might we’ questions
• Prototype and iterate

Great day being a part of the @gdsteam Intro to Service Design pilot course today! Definitely makes me feel that the role of a service designer is challenging, purposeful and important. 🙌 pic.twitter.com/bzDoBhu2q7

— Marie Cheung (@mariecheungsays) January 19, 2018

I’ll assume that you’ve plugged and installed in the Steinberg, installed the Cubase AI Software, and opened up Cubase AI.

Go to the ‘Devices’ menu dropdown at the top of the screen, then select ‘VST Connections’.

Go to the ‘Inputs’ tab and create a mono input using “Yamaha Steinberg USB ASIO” and “UR12 Input 2”.

Keep output as default on “Yamaha Steinberg USB ASIO” – i.e. Stereo speakers and Device Ports “UR12 Output L” and “UR12 Output R”.

Plug your lead into the “Hi Z” input on the UR 12. (This is “Input 2”. The other one is for microphones.) Plug your headphones into the “Phones” socket on the right. (If you’re using 3.5mm headphones you’ll need a 3.5mm to 6.35mm adapter, as pictured above.)

Turn the Input 2 gain knob up to about 1/4 of the max.

Strum a bit, and you should see the Cubase Monitor show the input levels increase as you strum. If it’s flat-lining like in the below picture, check that you’re plugged in and that you’ve set up the input correctly. Play some low, loud palm-muted chords, and turn up the Input 2 gain until you get a red square appear above the blue input monitor. Click on the red square to make it disappear, and turn down the Input 2 gain a little. Your goal is to have it as high as possible without clipping.

Now left click in the audio panel, and select “Add audio track”

A popup with a load of options will appear. Ignore them and add the track.

Click on the speaker symbol next to this track. This will allow you to hear the input as you play. It’ll turn orange to show that the “Monitor” option is enabled.

Make sure to turn the ‘Output’ knob on the UR12 up a bit so you can actually hear it. At this point you should be able to play something on your guitar and hear it through your headphones from the UR-12.

If you’d like to add some amplifiers or other effects, go to “Inserts” on the left-hand “Inspector” and choose one. There are loads of free VST amplifiers and effects available.

Now let’s record something. To record audio from a track (e.g. “Audio 01” in the above image) you need to make sure that the record icon for that track is red. Cubase has a concept of an audio track being enabled or not enabled to record, and it just ignores it if it isn’t enabled.

Before you start recording, turn off the orange “Monitor” option. It’s useful for experimenting with a tone, but adds latency when recording so can mess up your timing. Use the “Direct Monitor” button on the front of the UR-12 box to send a dry (non-processed) signal back to you as you play.

To start recording, click on the recording circle near the top of the screen.

Bonus – fixing audio distortion in Windows 10

There’s a bug with Windows 10 whereby you hear horribly distorted output. The two most useful threads on the topic (first and second) suggest that it’s been an issue for a couple of years. The fix is downgrading from 1.96 to 1.95 of a key driver. To do this, uninstall “Yamaha Steinberg USB Driver” via Add/Remove Programs (not via Device Manager – it won’t fully get removed if you try this route.), then install version 1.95 of the driver.

Bonus – gain screen space by removing VST Instruments and Media Bay

I don’t know what these do, so I’ve hidden them. To do this, click on the button of three rectangles, just under the “File” menu. Then uncheck “racks”.

I’m concerned about the political ramifications of the technical underpinnings of voice interfaces. Voice recognition has advanced through access to lots of data, and machine learning. This means that for people to use this interaction medium, they need to be using tools provided by big providers like Apple, Amazon and Google. How easily will people be able to create their own voice interfaces without relying on a corporate provider? How easily will you be able to identify the thing that you’re interacting with, the logic it’s driven by, and who ultimately owns and controls the means of interaction? What is the product, and what generates value and to/for who?

Voice as an interaction medium has limitations. Voice requires (quiet) private space, and for you to be able to speak in an understood language. And so far voice interactions have focused on information or commerce transactions.

It’s worth reflecting on what problems we’re trying to solve when we pioneer a new interaction paradigm. What value are we trying to realise? Are transactions the big challenge we need to solve? (Would interaction approaches based around critically managing and engaging with information flows, or social connectedness, be different?)

The dominant method of interacting with computers so far has been mechanical, with feedback and state communicated visually by the computer. Thinking more broadly, the interaction channels we use to interact with computers could be mechanical or oral, and each of our senses could be a channel for the computer to feed back information. And we could design for more than one type of input and feedback at one time. This Smashing Magazine piece on multi-modal interfaces outlines interfaces that take visual and audio elements:

“When​ we’re ​designing​ an ​interface,​ ​if​ ​we​ ​know​ ​the​ ​context​, ​we​ ​can​ ​remove​ ​friction.​ Will​ ​the​ ​product be used​ ​in​ ​the​ ​kitchen​ ​when​ ​the​ ​user’s​ ​hands​ ​are​ ​full?​ ​Use​ ​voice​ ​control;​ ​it’s​ ​easier​ ​than​ ​a touch​​screen.​ ​Will ​they​ ​use​ ​it​ ​on​ ​a​ ​crowded​ train?​ Then ​touch​​ing a screen​ would ​feel ​far​ ​less awkward​ ​than​ ​talking​ ​to​ ​a​ ​voice​ ​assistant​.​ ​Will ​they​ ​need​ ​a​ ​simple​ ​answer to​ ​a​ ​simple​ ​question?​ ​Use​ a ​conversational​ ​interface.​ Will​ ​they​ ​have​ ​to​ ​see​ ​images​ ​or understand​ ​complex​ ​data?​ ​Put​ ​it​ ​on​ ​a​ ​screen”

So rather than just focusing on voice input, it might make sense to think about how to structure our data to be agnostically accessible. Agnostic interaction channels don’t privilege one way of perceiving or interacting with the world. Now that more of the population is computer-literate, we might be able to relax skeuomorphic baggage. Skeuomorphic design’s visual language helped communicate but also constrained the behaviours and interactions we could design, because they had to be intelligible as physical metaphor. Old metaphors or expectations, like “saving” files by clicking on a floppy disk icon, could be seen as blockers to thinking more broadly about interaction.

Perhaps the next paradigm of interaction design is to transcend interaction – to read and service our intent directly. This feels a way off, but here’s an early proof of concept.

If we remove friction and translation from our interactions, what are we left with? Our own unmediated desires. Friction in transmitting intent can be a good thing – a chance to reflect and exercise deliberate control, rather than being driven just by desires. The task of cultivating deliberate intention, and making conscious decisions, rather than acting on impulse, is a design problem and a spiritual one.

Introduction – Lou Downe

“Bad service design is one of the biggest costs to government”

In the future, “Services will shape government, not the other way round.” “Transformation will never be done” so focus on helping change to happen sustainably.

Dan Hill, Associate Director, Arup – Design and the public

The contrasting experience of two social housing projects shows the need for service design:

Wohnpark Alt-Erlaa, Viennna; image by Thomas Ledl – Own work, CC BY-SA 3.0, Link

By Thomas Ledl – Own work, CC BY-SA 3.0, Link

The Robin Hood Estate; image by stevecadman – Flickrtik hartua, CC BY-SA 2.0, Link

Hill argues that the key difference between the success of the former, and the demolition of the latter is the ongoing maintenance, service and engagement.

Design needs a service blueprint. You aren’t just designing a one-off object – you’re building for life and for ongoing use. You need to design a service.
This is what was missing in the Robin Hood Estate.

There’s been a steep decline in the percentage of architects employed in the public sector. In 1970, 50% of architects were employed in the public sector; today only 0.7% of architects are.

Hill presents this as an opportunity. Digital has an opportunity to build design intelligence back into government. And we can build in this view of service design, making things better than before.

“Design for government could be more ambitious and innovative than ever” @cityofsound #GovDesign pic.twitter.com/uDbf9NaXgO
— Colin (@htmlandbacon) September 18, 2017

Prepare for the future by preparing to transition smoothly

It’s hard to know what the future will be like. So “The best focus for people is to make the transitions as effective and painless as possible as opposed to worrying about what the end point is.” – Michael Spence

The public sector needs to set out a vision for the future, otherwise the private sector will do it, leeching value out of a city and sending it to California.

Jean-Louis Missika, deputy mayor of Paris:
“We should announce, before 2020, that in Paris, no private owned AV will be allowed it will be only mobility as as service; not mobility as ownership”

Research in Zurich, Singapore and New York suggests that 80% of car use could be replaced by shared autonomous shuttle fleets.

DVLA – John Hewson

90% of DVLA interactions are now digital, up from about 9% in 2007. But that’s still 23 million paper applications per year, 5 million pieces of casework.

DVLA have made big changes to the user experience, but they haven’t yet made improvements to their internal systems.
Redesigning internal systems is another challenge, requiring slightly different expertise to building user-facing websites. These users often value efficiency over user-friendliness. One key piece of software is from 1990 and doesn’t even support a mouse, so this is a quite a challenge.

“The core discipline of prototyping, using best practices from the Service Manual, and testing iteratively will always work.”
— John Hewson

But the core discipline of prototyping, using best practices from the Service Manual, and testing iteratively will always work.

Panel discussion

The discussion reinforced the importance of applying service design to business models and procurement process – not just external user journeys. Take a service design approach to problem solving right through organisations.

Similarly, design for failure. Don’t just design for the perfect digital frontend – plan the whole service, and for all aspects of that service.

Design for the gaps between different nodes in someone’s experiences – e.g. a journey through healthcare. Each node can have its own effective risk register, but things can get lost in the cracks because there’s not always a holistic view.

Introduction

“Critical reflection on my experience as a student in unexciting classrooms enabled me not only to imagine that the classroom could be exciting but that this excitement could co-exist with and even stimulate serious intellectual and/or academic engagement.” [7]

“But excitement about ideas was not sufficient to create an exciting learning process. As a classroom community, our capacity to generate excitement is deeply affected by our interest in one another, in hearing one another’s voices, in recognizing one another’s presence.” [8]

“There must be an ongoing recognition that everyone influences the classroom dynamic, that everyone contributes. These contributions are resources. Used constructively they enhance the capacity of any class to create an open learning community. Often before this process can begin there has to be some deconstruction of the traditional notion that only the professor is responsible for classroom dynamics.” [8]

“Seeing the classroom always as a communal place enhances the likelihood of collective effort in creating and sustaining a learning community.” [8]

“I celebrate teaching that enables transgressions – a movement against and beyond boundaries. It is that movement which makes education the practice of freedom.” [12]

Engaged Pedagogy

“To educate as the practice of freedom is a way of teaching that anyone can learn.” [13]

“many professors have intensely hostile responses to the vision of liberatory education that connects the will to know with the will to become.” [18-19]

“Any classroom that employs a holistic model of learning will also be a place where teachers grow, and are empowered by the process. That empowerment cannot happen if we refuse to be vulnerable while encouraging students to take risks.” [21]

Chandra Mohanty: “resistance lies in self-conscious engagement with dominant, normative discourses and representation and in the active creation of oppostitional analytic and cultural spaces.” [22]

A Revolution of Values: The Promise of Multicultural Change

“In the book Where Do We Go From Here? Chaos or Community, Martin Luther King, Jr. told the citizens of this nation, with prophetic insight, that we would be unable to go forward if we did not experience a ‘true revolution of values.’ He assured us that ‘the stability of the large world house which is ours will involve a revolution of values to accompany the scientific and freedom revolutions engulfing the earth. We must rapidly begin the shift from a ‘thing’“-oriented society to a ‘person’-oriented society. When machines and computers, profit motives and property rights are considered more important than people, the giant triplets or racism, materialism and militarism are incapable of being conquered…’”

“The public figures who speak the most to us about a return to old-fashioned values embody the evils King describes. They are most committed to maintaining systems of domination – racism, sexism, class exploitation, and imperialism. They promote a perverse vision of freedom that makes it synonymous with materialism. They teach us to believe that domination is ‘natural’, that it is right for the strong to rule over the weak, the powerful over the powerless. What amazes me is that so many people claim not to embrace these values and yet our collective rejection of them cannot be complete since they prevail in our daily lives.” [27-8]

Embracing Change: Teaching in a Multicultural World

“We found again and again that almost everyone [in the faculty attending the transformative pedagogy seminars at Oberlin], especially the old guard, were more disturbed by the overt recognition of the role our political perspectives play in shaping pedagogy than by their passive acceptance of ways of teaching and learning that reflect biases, particularly a white supremacist standpoint.” [37]

“… it is difficult for individuals to shift paradigms and … there must be a setting for folks to voice fears, to talk about what they are doing, how they are doing it, and why. One of our most useful meetings was one in which we asked professors from different disciplines [including math and science] to talk informally about how their teaching had been changed by a desire to be more inclusive. Hearing individuals describe concrete strategies was an approach that helped dispel fears. It was crucial that more traditional or conservative professors who had been willing to make changes talk about motivations and strategies.” [38]

“Accepting the decentering of the West globally, embracing multiculturalism, compels educators for focus attention on the issue of voice. Who speaks? Who listens? And why?” [40]

“Often, professors and students have to learn to accept different ways of knowing, new epistemologies, in the multicultural setting.” [41]

Shifting paradigms can be uncomfortable. “White students learning to think more critically about questions of race and racism may go home for the holidays and suddenly see their parents in a different light. They may recognize nonprogressive thinking, racism, and so on, and it may hurt them that new ways of knowing may create estrangement where there was none…” [43] difficult experiences are common, so we need “practice at integrating theory and practice: ways of knowing with habits of being.” [43]

Paulo Freire

Freire: “We cannot enter the struggle as objects in order later to become subjects.” [46]

Theory as Liberatory Practice

Theory helps us “understand both the nature of our contemporary predicament and the means by which we might collectively engage in resistance that would transform our current reality.” [67]

“By reinforcing the idea that there is a split between theory and practice or by creating such a split, both groups [excessively theoretical, anti-intellectuals] deny the power of liberatory education for critical consciousness, thereby perpetuation conditions that reinforce our collective exploitation and repression.” [69]

“I continue to be amazed that there is so much feminist writing produced and yet so little feminist theory that strives to speak to women, men and children about ways we might transform our lives via a conversion to feminist practice. Where can we find a body of feminist theory that is directed toward helping individuals integrate feminist thinking and practice into daily life? What feminist theory, for example, is directed toward assisting women who live in sexist households in their efforts to bring about feminist change?” [70]

“I am often critical of a life-style-based feminism, because I fear that any feminist transformational process that seeks to change society is easily co-opted if it is not rooted in a political commitment to mass-based feminist movement. Within white supremacist capitalist patriarchy, we have already witnessed the commodification of feminist thinking [just as we experience the commodification of blackness] in ways that make it seem as though one can partake of the ‘good’ that these movements produce without any commitment to transformative politics and practice. In this capitalist culture, feminism and feminist theory are fast becoming a commodity that only the privileged can afford.” [70-71]

Essentialism and Experience

“Often it is forgotten that the hope was not simply that feminist scholars and activists would focus on race and gender but that they would do so in a manner that would not reinscribe conventional oppressive hierarchies. Particularly, it was seen as crucial to building mass-based feminist movement that theory would not be written in a manner that would further erase and exclude black women and women of color, or, worse yet, include us in subordinate positions. Unfortunately, much feminist scholarship dashes these hopes, largely because critics fail to interrogate the location from which they speak, often assuming, as it is now fashionable to do, that there is no need to question whether the perspective from which they write is informed by racist and sexist thinking, specifically as feminists perceive black women and women of color.” [77-78]

“I am always amazed by the complete absence of references to work by black women in contemporary critical works claiming to address in an inclusive way issues of gender, race, feminism, postcolonialism, and so on. Confronting colleagues about such absences, I, along with other black women critics, am often told that they were simply unaware that such material exists, that they were often working from their knowledge of available sources.” MDL – I need to broaden my network / inflow of information to broaden my understanding [80]

“…the very discursive practices that allow for the assertion of the ‘authority of experience’ have already been determined by a politics of race, sex, and class domination. Fuss does not aggressively suggest that dominant groups – men, white people, heterosexuals – perpetuate essentialism. In her narrative it is always a marginal ‘other’ who is essentialist. Yet the politics of essentialist exclusion as a means of asserting presence, identity, is a cultural practice that does not emerge solely from marginalized groups.” [81]

“a critique of essentialism that challenges only marginalized groups to interrogate their use of identity politics or an essentialist standpoint as a means of exerting coercive power leaves unquestioned the critical practices of other groups who employ the same strategies in different ways and whose exclusionary behavior may be firmly buttressed by institutionalized structures of domination that do not critique or check it.” [82-3]

Holding My Sister’s Hand

“Although they have written poignant memoirs which describe affectional bonds between themselves and black female servants, white women often failed to acknowledge that intimacy and care can coexist with domination. It has been difficult for white women who perceive black women servants to be ‘like one of the family’ to understand that the servant might have a completely different understanding of their relationship. The servant may be ever mindful that no degree of affection or care altered differences in status – or the reality that white women exercised power, whether benevolently or tyrannically.” [98]

“Until white women can confront their fear and hatred of black women [and vice versa], until we can acknowledge the negative history which shapes and informs our contemporary interaction, there can be no honest, meaningful dialogue between the two groups. The contemporary feminist call for sisterhood, the radical white woman’s appeal to black women and all women of color to join the feminist movement, is seen by many black women as yet another expression of white female denial of the reality of racist domination, of their complicity in the exploitation and oppression of black women and black people.” [102]

“what factors distinguish these relationships we have with white feminists which we do not see as exploitative or oppressive. A common response was that these relationships had two important factors: honest confrontation, and a dialogue about race, and reciprocal interaction.” [106]

Feminist Scholarship

“black gender relations were constructed to maintain black male authority even if they did not mirror white paradigms, … white female identity and status was different from that of a black woman.” [120] “In search of scholarly material to document the evidence of my lived experience, I was stunned by either the complete lack of any focus on gender difference in black life or the tacit assumption that because many black females worked outside the home, gender roles were inverted. Scholars usually talked about black experience when they were really speaking solely about black male experience.” [120]

“the vast majority of white feminists did not welcome our questioning or feminist paradigms that they were seeking to institutionalize; so too, many black people simply saw our involvement with feminist politics as a gesture of betrayal, and dismissed our work.” [122]

Recommendation for Michele Wallace’s “Black Macho and the Myth of the Superwoman” [125]

Building a Teaching Community

“The person who is most powerful has the privilege of denying their body. I remember as an undergraduate I had white male professors who wore the same tweed jacket and rumpled shirt or something, but we all knew that we had to pretend… The point was we should all respect that he’s there to be a mind and not a body… The traditional notion of being in the classroom is a teacher behind a desk or standing at the front, immobilized. In a weird way that recalls the firm, immobilized body of knowledge as part of the immutability of truth itself. So what if one’s clothing is soiled, if one’s pants are not adjusted properly, or your shirt’s sloppy. As long as the mind is still working elegantly and eloquently, that’s what is supposed to be appreciated.” [137]

“The erasure of the body encourages us to think that we are listening to neutral, objective facts, facts that are not particular to who is sharing the information. We are invited to teach information as though it does not emerge from bodies.” [139]

“the ways erasure of the body connects to the erasure of class differences, and more importantly, the erasure of the role of university settings as sites for the reproduction of a privileged class of values, of elitism.” [140]

“professors may attempt to deconstruct traditional biases while sharing that information through body posture, tone, word choice, and so on that perpetuate those very hierarchies and biases they are critiquing.” [141]

“many teachers who do not have difficulty releasing old ideas, embracing new ways of thinking, may still be as resolutely attached to old ways of practicing teaching as their more conservative colleagues.” [142]

“To educate for freedom, then, we have to challenge and change the way everyone thinks about pedagogical process. This is especially true for students. Before we try to engage them in a dialectical discussion of ideas that is mutual, we have to teach about process. I teach many white students and they hold diverse political stances. Yet they come into a class on African American women’s literature expecting to hear no discussion of the politics of race, class, and gender. Often these students will complain, ‘Well I thought this was a literature class.’ What they’re really saying to me is, ‘I thought this class was going to be taught like any other literature class I would take, only we would now substitute black female writers for white male writers.’ They accept the shift in the locus of representation but resist shifting ways they think about ideas.” [144]

“one of the responsibilities of the teacher is to help create an environment where students learn that, in addition to speaking, it is important to listen respectfully to others. This doesn’t mean we listen uncritically or that classrooms can be open so that anything someone else says is taken as true, but it means really taken seriously what someone says.” [Ron Scapp] [150]

Language

“One line of this poem that moved and disturbed something within me: ‘This is the oppressor’s language yet I need it to talk to you.’” [167]

“…these words… make me think of standard English, of learning to speak against black vernacular, against the ruptured and broken speech of a dispossessed and displaced people. Standard English is not the speech of exile. It is the language of conquest and domination…” [168]

“It is not the English language that hurts me, but what the oppressors do with it, how they shape it to become a territory that limits and defines, how they make it a weapon that can shame, humiliate, colonize.” [168]

“How to describe what it must have been like for Africans whose deepest bonds were historically forged in the place of shared speech to be transported abruptly to a world where the very sound of one’s mother tongue had no meaning.” [168]

“I imagine, then, Africans first hearing English as ‘the oppressor’s language’ and then re-hearing it as a potential site of resistance. Learning English, learning to speak the alien tongue, was one way enslaved Africans began to reclaim their personal power within a context of domination. Possessing a shared language, black folks could find again a way to make community, and a means to create the political solidarity necessary to resist.” [170]

“Using English in a way that ruptured standard usage and meaning, so that white folks could often not understand black speech, made English into more than the oppressor’s language.” [170]

“The power of this speech is not simply that it enables resistance to white supremacy, but that it also forges a space for alternative cultural production and alternative epistemologies – different ways of thinking and knowing that were crucial to creating a counter-hegemonic worldview. It is absolutely essential that the revolutionary power of black vernacular speech not be lost in contemporary culture. that power resides in the capacity of black vernacular to intervene on the boundaries and limitations of standard English.” [171]

“That the students in the course on black women writers were repressing all longing to speak in tongues other than standard English without seeing this repression as political was an indication of the way we act unconsciously, in complicity with a culture of domination.” [173]

“To recognise that we touch one another in language seems particularly difficult in a society that would have us believe that there is no dignity in the experience of passion, that to feel deeply is to be inferior, for within the dualism of Western metaphysical thought, ideas are always more important than language. To heal the splitting of mind and body, we marginalized and oppressed people attempt to recover ourselves and our experiences in language. We seek to make a place for intimacy. Unable to find such a place in standard English, we create the ruptured, broken, unruly speech of the vernacular. When I need to say words that do more than simply mirror or address the dominant reality, I speak black vernacular. There, in that location, we make English do what we want it to do. We take the oppressor’s language and turn it against itself. We make our words a counter-hegemonic speech, liberating ourselves in language.” [174-5]

Confronting Class in the Classroom

“During my college years it was tacitly assumed that we all agreed that class should not be talked about, that there would be no critique of the bourgeois class biases shaping and informing pedagogical process [as well as social etiquette] in the classroom. Although no one ever directly stated the rules that would govern our conduct, it was taught by example and reinforced by a system of rewards. As silence and obedience to authority were most rewarded, students learned that this was the appropriate demeanor in the classroom. Loudness, anger, emotional outbursts, and even something as seemingly innocent as unrestrained laughter were deemed unacceptable, vulgar disruptions of classroom social order. these traits were also associated with being a member of the lower classes. If one was not from a privileged class group, adopting a demeanor similar to that of the group could help one to advance. It is still necessary for students to assimilate bourgeois values in order to be deemed acceptable.” [178]

“Silencing enforced by bourgeois values is sanctioned in the classroom by everyone.” [180]

“Even those professors who embrace the tenets of critical pedagogy [many of whom are white and male] still conduct their classrooms in a manner that only reinforces bourgeois models of decorum.” [180]

“Sharing experiences and confessional narratives in the classroom helps establish communal commitment to learning. These narrative moments usually are the space where the assumption that we share a common class background and perspective is disrupted. While students may be open to the idea that they do not all come from a common class background, they may still expect that the values of materially privileges groups will be the class’s norm.” [186]

“I have found that students from upper- and middle-class backgrounds are disturbed if heated exchange takes place in the classroom. Many of them equate loud talk or interruptions with rude and threatening behavior. Yet those of us from working-class backgrounds may feel that discussion is deeper and richer if it arouses intense responses. In class, students are often disturbed if anyone is interrupted while speaking, even though outside class most of them are not threatened. Few of us are taught to facilitate heated discussions that may include useful interruptions and digressions, but it is often the professor who is most invested in maintaining order in the classroom. Professors cannot empower students to embrace diversities of experience, standpoint, behavior, or style if our training has disempowered us socialized us to cope effectively only with a single mode of interaction based on middle-class values.” [187]

Eros, Eroticism, and the Pedagogical Process

“Suggesting that this culture lacks a ‘vision or science of hygeology’ [health and wellbeing] Keen asks ‘What forms of passion might make us whole? To what passions may we surrender with the assurance that we will expand rather than diminish the promise of our lives?’ The quest for knowledge that enables us to unite theory and practice is one such passion. To the extent that professors bring this passion, which has to be fundamentally rooted in a love for ideas we are able to inspire, the classroom becomes a dynamic place where transformations in social relations are concretely actualized and the false dichotomy between the world outside and the inside world of the academy disappears. In many ways this is frightening. Nothing about the way I was trained as a teacher really prepared me to witness my students transforming themselves.” [195]

The Case for Investing in Adwords – Kate Sanger, Head of Communications at Jo’s Cervical Cancer Trust

They worked with a paid search agency, but with the explicit upfront aim of upskilling the in-house team.

The agency audited and restructured their ad copy. Incorporated call out and phone extensions. Refined keyword matching so it wasn’t just ‘broad match’. Set up an information flow between analytics data and PPC performance data. Added ‘do you need adwords?’ question to the comms brief that the digital team receives.

“Do you need AdWords?” – this should be on a brief from teams looking to use digital. Good point. Should definitely do this #nfptweetup — Joe Freeman (@JosephFreeman) May 17, 2017

The My Oxfam app and more – Matt Jerwood, Head of Digital Fundraising at Oxfam UK

The Oxfam App displays content for the period during which you’ve been a regular giver. The idea is to show the impact of your donation. The app displays third party news, to increase credibility.

It shows your gift history, and displays income generated from items you’ve sold in Oxfam charity shops. It lets you manage your direct debit level in-app, moving it up or down.

“The My Oxfam app shows your gift history and displays income generated from items you’ve sold in Oxfam charity shops.” #nfptweetup — Joe Freeman (@JosephFreeman) May 17, 2017

There isn’t currently a designed journey for people who dial down their direct debit – e.g. prompts or encouragement to increase it after a period of time. Again, the app is very much about the soft cell.

I bet the CRM integration was really complicated. But for the user, the experience is simple. That’s the way it should be.

“No firm evidence of success yet, but initial results suggest that it improves retention.” #nfptweetup — Joe Freeman (@JosephFreeman) May 17, 2017

They’ve improved the single donation experience too. They added Apple Pay and PayPal payment options, massively reducing the time needed to make payment.

“We didn’t get stats on the impact on the donation value or reduction in dropoff at the payment page.” #nfptweetup — Joe Freeman (@JosephFreeman) May 17, 2017

From dinosaurs to digital masters: our mission to change our DNA at CRUK – Kate Simmons, Head of Customer Experiences at Cancer Research UK

Cancer Research UK surveyed people’s experiences of working with the digital team.

People felt that digital was something done to them, rather than something they had control over.

They produced a word cloud and the biggest word was ‘patronising’.

“The CRUK team survey relationships with other teams every 2 weeks.” #nfptweetup — Joe Freeman (@JosephFreeman) May 17, 2017

You need to recognise that people go through a change curve. It’ll get emotionally difficult before it gets better. You need to look after your digital team and build their resilience to help them with this element of their work:

“Great to hear @katecsimmons talking about the Change Curve in reference to digital change. Very important to acknowledge impact #NFPtweetup” pic.twitter.com/qvb1sS3IIV — Joe Freeman (@JosephFreeman) May 17, 2017

Interesting lessons from advanced hub-spoke model: people feeling out of place and leaving

”#NFPtweetup Unexpected outcome- after exciting ‘spoke’ projects people tended to leave and take their new skills and approaches elsewhere.” — Jess Day (@day_jess) May 17, 2017

So empower people to make change in their own teams after you’ve upskilled them through cross-team digital working. And if they don’t feel part of the digital team, but don’t feel part of their original team either, help craft a third identity for them.

Cancer Research UK avoid the words: “digital”, “agile” and “ways of working”. They set up a Modern Marketing Academy.

Aside: Cancer Research UK made the case for improving findability on their intranet by working out how much time was being wasted by the poor user experience, and what the resultant cost was.

“Do read this later if you also found @katecsimmons’ #NFPtweetup talk fascinating https://t.co/CVSPghlBwL ‘Building digital confidence’” — Joe Freeman (@JosephFreeman) May 17, 2017

What makes for good user research?

Years ago, the approach to reporting on user testing was different. Jared Spool: “You’d come back and you’d assemble the notes into a giant report. You would write the report in passive voice, and then distribute it. Of course, the thicker you made the report, the more of a thump up it made when you dropped it on the table, which was, of course, the most impressive way to do reporting, and then nobody read it. Then you wondered why it was there.”

Of course, the purpose of user research is to inform high-quality decisions. So research must be read and acted upon. This is why user researchers now produce short, easy-to-digest reports.

Erika Hall: “The value in research is not the answer… It’s creating that culture where you’re constantly asking the same question, and you’re in a position to keep asking the question, finding what today’s answer is, and finding a way to respond to that in your work.”

What is the aim and output of research in a team?

The documentation becomes secondary. Erika Hall: “The goal is not to produce a report. The goal is to create this shared understanding so that everybody in the team knows here’s what our goal is, and we’re very clear about our goal. Here’s what our constraints and requirements are. To really think about the assumptions together and develop the shared vocabulary about here’s what we’re betting on, and here’s our evidence that those are good bets.”

Selling research

Erika Hall: ‘[One thing that] sociologists are studying right now is the fact that data doesn’t change minds… People’s minds are very good at shutting out data that they don’t want to hear…That’s a sales moment for research, when you bring a stakeholder in and you’re like, “Watch this, and see the power, and feel the change in your own mind when you see all of your assumptions blown away.” Those are really, really powerful moments.’

Erika Hall: “Research is challenging given ideas, so it’s naturally anti-authoritarian. If you’re in an authoritarian business culture, you have to work very carefully to change that.”

Asking people what they want leads to unhelpful speculation

Erika Hall: “one of the criticisms of research is that you’re asking people what they want. People will speculate, and this is something you have to be really careful of when you do research about people and their actual behaviors and habits. If you ask the question the wrong way, what you’ll hear is what people are speculating about, which might have no connection to how they actually behave.”

If you talk to people about, “Would you use this feature? What do you like? What do you want?” they’ll imagine these scenarios that may have very little relationship to what they actually do, and what they actually need, and the choices that they make if they’re using something in a real-world scenario.

When building digital products, you should test them early to make sure that what you’re building will work

Jared Spool: ‘if we developed bridges the way we develop online products, the way people seem to want to do it, we would build the bridge and then we would send a car across it. We would watch the car inevitably plummet into the depths below, and then we’d go, “Huh. Maybe there’s something wrong with the bridge.’

Erika Hall: ‘you need to know what your overall goal is, that the research is supposed to be helping you with. you say, “OK, what are our major assumptions that we’re betting on, that carry a lot of risk?”…Then you say, “OK, what questions do we need to ask? What are we trying to find out before we get down to work, or as we continuously work, to help us validate our assumptions?”’

When you work with data, be wary of projecting assumptions and biases onto the data

Jared Spool: In a study, a panel was interviewing for the position of police chief. There were two flavours of resume: academic-oriented, and street-oriented. The male name was attached the to academic resume, the female one the street experience resume. They picked the male candidate. They said “This position requires a real academic approach, so that’s why we chose the male candidate over the female. It wasn’t because they were male. It was because it was academic.” But then the names were switched, so that the female name was on the academic resume, and the male name on the street resume. Then the evaluators would say “This job requires someone with real-street smarts, so we chose…but we’re not biased.”’

But it’s possible to take measures against bias: “they found that if they had the folks rate up front which is more important, street-smarts or academics, and they were to write that down before they looked at the resumes, then they were more likely to choose the person based on the actual criteria they had decided, not based on gender bias.”

There will always be people who won’t or can’t use digital services:

• Access
• Confidence
• Skill level
• Language
• Literacy

When government started creating new online transactions, digital services were created, with offline ‘Assisted Digital’ support, so that in theory no-one was left behind and unable to access services.

Why did I want to learn about Assisted Digital?

Assisted Digital is something that Product Managers are supposed to know about. The DDaT capability framework says that a good Product Manager “Understands the importance of assisted digital” and that a good Senior Product Manager is “Able to identify and implement solutions for assisted digital.” I didn’t know much about it and wanted to learn.

Assisted Digital is the 2nd most searched for content in the service manual. Not having a good Assisted Digital approach is a common reason for services to fail service assessment. Most importantly, it’s my responsibility to make sure that this is taken seriously. This fits with our design principles: “This is for everyone” and “Build digital services, not websites”. I’m not currently working on a service, but it’s likely that I will be at some point in the future so I want to be ready.

So I did some reading and spoke to some knowledgeable and smart people, including Roxanne Asadi and Ben Carpenter at GDS, who’ve both spent a lot of time working on and thinking about Assisted Digital.

How well is government doing with Assisted Digital?

In general we’re not doing a great job. Recent Inclusive Services research showed that Assisted Digital channels “are the most likely to be needed by the most vulnerable users, but are the least likely to be switched on or user-centred”.

The narrow language of “digital by default” and “digital transformation” often led to a narrow view of service design that saw Assisted Digital as of lesser importance.

Tom Loosemore, co-founder of the Government Digital Service, reflected on Assisted Digital recently:

“I could have avoided the need for using the phrase “assisted digital” by making it clear that we should be using Internet-era service design to reach all users, regardless of channel.”

“by separating out “assisted digital” as a different thing, and by handing responsibility for it over to a separate team, we as leaders of GDS created an artificial divide in both our language and our actions.”

“Rather than challenge the “one size fits all” approach to service design, the words “assisted digital” suggested that one size for all was just fine, and that people with additional needs could or should be “assisted” to navigate a single path, common to all.”

And Assisted Digital options are needed by lots of people. For Personal Independence Payments, 50% of users were using Assisted Digital.

How might we be better at Assisted Digital?

• Refer to ‘Service Design’ rather than ‘Assisted Digital’. This makes it clear that this is integral to a holistic service, and not a bolt-on. Focus around the user’s needs and end-to-end problem, rather than the mechanics of the service that you’re offering. This seems like the single highest impact strategic move you can make.
• Include Operations people in designing services – not just Digital people. Otherwise we prejudice ourselves towards digital only, and don’t think about the other channels and implicitly reprioritise them. In government departments, often the Operational people and the Digital people were/are separate teams and directorates. So it wasn’t/isn’t easy to get them to cooperate.
• Include non-government interactions in your service design
• Create and share design patterns for Assisted Digital. It should be as easy to reuse designs for a non-digital interaction as it should be to copy a front-end component like a button on a website.
• Make sure you understand the needs of all your users deeply, including people who don’t have skills and access to digital. If you take this as your foundation you’ll likely do a good job.
• Iterate and test your Assisted Digital offering just as much as you do your website.

Some changes that leaders can make:

• Think more broadly about the goals behind our service design, and the metrics we use to measure the success of services. If we just judge services on how many people they’ve moved online, how much they’ve reduced the headcount in a call centre, and reduced costs, then we aren’t giving teams licence to take Assisted Digital seriously.
• Don’t just build teams of digital specialists to build a digital service. You need operational people as well.
• Accept that Assisted Digital has a cost. Plan to invest some of the money you save from channel shift to digital in Assisted Digital.

Examples of services with good Assisted Digital

Rural Payments. Offered drop-in sessions and home visits. These were cost-effective as they were for a small number of people.

Student Loan applications. They found out that about 5% of users needed support. They mapped out how the call centre would take calls and help people, and tested it. They triaged calls and offered a visiting service. The volume was very small so they could pay for this – they just had to retrain people at their existing contact centre.

A large number of the reported causes for failed agile projects are management-related.

This is down to management’s way of thinking.

Why do managers think differently?

Managers still often follow the scientific management principles of Frederick Winslow Taylor:

1. It is possible to know all you need to know in order to plan what to do.
2. “Planners” and “doers” should be separated.
3. There is only one right way to do things.

The harmful divide between “Planners and “Doers”

This distinction between “planners” and “doers” causes an uneven power dynamic.

The powerful (planners) are focused on ambition, politics, mistrust, greed and fear.

The powerless (doers) are focused on resentment and resignation.

This approach is manifested in management creating plans for resources to follow, in milestones, steering groups (the idea that these people set direction and the doers just follow along), progress reports, measuring individual performance, annual budgeting, organisation silos and timesheets.

The split between “planners” and “doers” may have made sense in the early 20th century factory system, where you didn’t have an educated workforce. But it doesn’t make sense now that university education is widespread – and it particularly doesn’t make sense for knowledge work, where the “planners” probably won’t be expert in the fast-changing specialist domains of their subordinates.

As Steve Jobs observed: “It doesn’t make sense to hire smart people and then tell them what to do; we hire smart people so they can tell us what to do”

Why do we tell people how to do their jobs?

We tell people how to do their jobs if the outcomes we want aren’t materialising.

We set objectives, then make plans to meet these objectives, and then carry out the necessary actions. But we might not get the outcomes we want.

This can happen because of problems in the flow between these steps:

• Knowledge gap: a difference between what we think we know and what we actually know. Assumptions.
• Alignment gap: a difference between what we want people to do and the actions people take.
• Effects gap: a difference between what our effects are and what we want them to achieve.

How does scientific management approach problems?

• Knowledge gap: give me more detailed information
• Alignment gap: I need to give you more detailed instructions
• Effects gap: need more detailed controls

This disempowers people in the organisation.

How else might we approach problems?

Scientific management isn’t the only approach. Prussian General Von Moltke in 1869 advocated a different way.

• Knowledge gap: “Do not command more than is necessary or plan beyond the circumstances you can foresee”
• Alignment gap: “Communicate to every unit as much of the higher intent as is necessary to achieve the purpose” Intent more important than how to achieve it in practice.
• Effects gap: “Everyone retains freedom of decision and action within bounds” everyone decides how to achieve that purpose.

So a superior management approach is to establish alignment on intent and give autonomy on actions.

• Define and communicate intent
• Allow each level to define how they will achieve the intent of the next level up and ‘backbrief’
• Give individuals freedom to adjust their actions in line with intent.

Some examples of this approach

• A BA check-in staff member was presented with a gold customer running late for a flight that was taking off in 20 minutes. All the formal rules suggested that nothing could be done, and she couldn’t get in touch with her manager or with customer services. So she took the initiative to hold up the flight and personally escort a gold customer to the gate, allowing him to make his flight. This was against formal rules, but in alignment with the company’s values “To ensure that BA is the customer’s first choice through the delivery of an unbeatable travel experience.”

• Netflix avoid top-down decision-making. They focus on ‘context, not control’.

• Submarine Captain David Marquet (author of Turn the ship around) divested control to his subordinates. Instead of giving orders, he’d ask people to tell him what they intended to do. He’d then potentially ask a few questions. He’d then give assent. Subordinates would internalise the required diligence, and grow in their own competence and professionalism. His focus was on providing clarity of purpose.
• Buurtzorg – a nursing company predicated on empowered, independent teams of nurses. Teams hire their own people, and decide how to operate. Patients love it because nurses spend more time with them. And yet the need for care is 40% less than it is in other organisations. The US would save $49 billion a year if it had this system.
• Favi – brass foundry. Legendary for on-time delivery, having not missed a deadline in 28 years. Staff empowered to do what it takes to get results, including delivery by helicopter if that’s what’s needed. This builds trust in delivery and the brand, far beyond the costs.

Changing culture results in bigger gains than changing processes

Simply adopting agile practices will generally give about a 20% benefit. Adopting an agile culture gives about a 300% benefit. This is much more powerful.

Simply adopting agile practices will generally give about a 20% benefit. Adopting an agile culture gives about a 300% benefit. #dpml

— Martin Lugton (@martinlugton) March 9, 2017

How people rewarded or punished in an organisation determines your values. Management sets boundaries by how it treats failure. Any cultural change needs to address this.

You need to evolve into theory Y management to realise benefits from agile. Change from theory X management to theory Y management. Put in place supporting structures, processes and practices. Role model these behaviors by people with moral authority in the organisation. Recognise that work is accomplished by teams not individuals. Monitor and value groups. Divest control within teams. Give teams autonomy and boundaries to work unimpeded. Encourage people to explore and challenge their personal beliefs. They’ll leave if they don’t like it.

Command and and control versus team-based approaches to work:

As Product Manager, my role is to make sure that we’re solving the right problems, and that we’re getting the most value from the work we do.

At the start of a mission, I set some clear overall goals for that quarter. Be upfront about what you’re not working on, so that everyone’s clear.

Work back from the end state you want to get to at the end of the quarter – e.g. migrating from one supplier to another, or launching a new feature that successfully meets its goals – and think about the things that will need to happen to get there. This will generate some milestones. You can start to build a high-level plan of what’s coming up in the quarter by mapping when you think you can hit these milestones. Do this with your delivery manager and tech lead, so that it’s sensible. It’ll change as you go, and that’s fine. I like to do this with post-its on a wall in the team’s space.

Within this structure, I like to work in weekly sprints. This means that you plan the detail of your work a week ahead. I like to run sprints from mid-week to mid-week, so that everyone knows what they’re doing on Mondays and Fridays, and those days can be kept free of meetings.

I run planning with the whole team – which is really important to build a shared understanding of the problems everyone is working on, across the disciplines. A day or two before the planning session, I’ll have a series of conversations to decide on the most valuable work to be done in that sprint. Some of these will be with my tech lead and delivery manager, others will be with the different specialisms in the team (e.g. user research, performance analysis or content design). The goal is always to check in with the big picture aims for the quarter, dive into any thorny questions we need to think about, or work to review in detail together, and then agree on what do work on in the upcoming sprint. Often this will be something already mapped out at high level, and now we start to consider the detail, but sometimes it’ll be something new.

These pre-planning conversations are a lot of fun, and a chance to think deeply about each of the disciplines in my team. They align us on goals and scope for the week – so we’re thinking carefully about how much is responsible to try and take on. In most quarterly mission teams I’ve been on, we’ve found estimating doesn’t work too well, as there’s lots of uncertainty inherent in a mission, and with a newly-formed team it takes a while to understand your velocity. So we rely more on gut feel, experience and the ability to iterate week-on-week if we felt a given sprint had too much work or not enough.

Once these conversations have finished, I’ll leave the different team members to write up cards to describe the work. I have a standard format I want cards written to – they need to have a ‘What’ and a ‘Why’ that anyone in the team could understand. But I don’t need to write the cards myself – it’s good practice for people to be able to communicate about their work, and it saves me time. If someone is struggling with writing cards to this standard, I’ll take the time to pair with them to iron this out.

The goal of writing these cards is to inform a robust planning conversation. Going into the team’s weekly planning session, we want clear goals and alignment on the big picture from each of the specialisms. The goal of planning is to think through the complexity of the work as a team, to really flesh out the detail of the cards – adding acceptance criteria, or thinking about how we might tackle a thorny technical challenge. But it’s also to build a shared understanding of what the other disciplines are working on – in a time-efficient way.

Ideally I’ll introduce some goals for the sprint, covering each of the specialisms. Probably tied in to the milestones for the quarter, that get us to our goals.

Once I’ve introduced the overall goals, and we’ve nuanced them if we need to, then we go into talking through the individual cards. I usually invite other people to introduce their own cards, and then I make sure that there’s a good quality conversation on each card – making sure that by the end of the discussion we understand the value we’re trying to deliver, and have thought about the approach we want to take.

So this session is about sense checking, challenging, thinking about approach – but isn’t prescriptive about approach. The person who decides to work on the card takes the lead on that, calling on others for support if needed. Planning is very useful for being prescriptive in defining quality, and in adding acceptance criteria to a card. As with planning overall, making sure we know what we need to do is more important than how – the how is worked out in the ground.

The evolution of (Digital) strategy

Discussions on digital: How strategy is evolving—and staying the same—in the hypergrowth digital age

“My view is that the role of strategist, first and foremost, is the mobilizer. People need to understand where the company is going and take autonomous decisions. “

“the way to manage a company has changed a lot in the last couple of decades, from “top down, the leader knows it all, let’s execute,” to a model that’s more like a federation, where you expect empowered teams to make the right choices and follow in the general direction. So it’s even more important to have a strategy that allows that federation of little teams to do what they have to do and not refer up the chain every time.”

Be wary of taking a narrow conception of what ‘digital transformation’ is

Digital strategy: Understanding the economics of disruption

“what they had wasn’t a digital strategy, it was a list of priorities for digitization. Explicitly, it was how are we going to reduce the cycle time in our end-to-end processes, how are we going to improve the customer experience and build new apps, and so forth. It was about how they digitize. It was not actually the choices they were making about a big disruptive economic force, which is the changes that are made possible by digital technologies.”

“The word ‘strategy’ is used too loosely with digital to mean our priorities for digitization not the choices we’re going to make in terms of where we compete and how we cmopete in the face of a big disruptive force.”

“We found that going back to the fundamentals of economics, trying to understand where there is economic room to be attacked, where your open flank is, or where you could thrust a spearhead, was a much more useful way to approach the problem and simplify the problem and focus on the right priorities.”

“Do not just think digitization. Think digital strategy. How will the economics of my business change in the future? How can I change the economics of other businesses? And, therefore, what should be strategy in the digital age be?”

How companies become digital leaders

“incumbents, particularly laggards and followers, sometimes reflexively try to drive digital marketing and sales. They mistake the world of digital for the world of online-selling success. And that narrow set of levers and aspirations for digital transformation is often unsatisfying and unsuccessful. Because there is a lot of opportunity unlocked by digital: the ability to fundamentally change how productive your assets are, the ability to leverage data and your consumer base in a different way, and the ability to actually develop innovative new propositions that you could never have offered before – the world of universal connectivity and ultracheap and ultrafast processing power. the ability to do these things is often a lot more powerful than trying to get a laser-like focus on pure digital- marketing and -sales performance.”

Thinking digitally can be difficult

Achieving a digital state of mind

“One of the biggest changes digital allows is that what might have been product sales become services. … You may have bought a thermostat, but now you’re buying energy management. You may have bought a TV, but now you’re buying entertainment streaming.”

What ‘digital’ really means “diverse perspectives [on what digital means] often trip up digital teams”

“digital should be seen less as a thing and more a way of doing things.”

1. “creating value at the new frontiers”. This could be about developing new businesses in adjacent categories, or identifying and realising value pools in existing sectors.
2. “creating value in the processes that execute a vision of customer experiences”. Obsessively seek to understand each step of a customer’s purchasing journey, regardless of channel, and think about “how digital capabilities can design and deliver the best possible experience.” e.g. supply chain to give flexibility and deliver the product efficiently and in a way that the customer wants.”
3. “building foundational capabilities that support the entire structure” “Being digital is about using data to make better and faster decisions, devolving decision making to smaller teams, and developing much more iterative and rapid ways of doing things.” [see below on decision making and oversight] I.T. should be delivered in a two-stream model. Legacy systems that support critical functions should work at a slower pace, and a separate stream supports fast-moving, often customer-facing interactions.

Making the cultural case for change

Digital disruption is impacting every sector, even law firms

“We should not try to convince management of the importance of digital. Instead we should frame the conversation around their pain points and struggles. Every senior management team has certain objectives that they need to reach or barriers they need to overcome. The chances are that digital can help with those and that is how we should position it.”

Customer obsession

Achieving a digital state of mind “Everybody will say they’re customer oriented, but, as digital leaders would say, they have to be customer obsessed. And that’s something that a lot of organizations still struggle with. So that’s where you should start. Take the customer perspective.”

How to think about building digital services

Digital disruption is impacting every sector, even law firms Not like creating a building, where you draw up the plans, then build, then do minor maintenance. It’s more like planting a garden. Start small, then add more elements. Keep pruning, trimming and evaluating as you go.

Design sprints

Preventing the Executive Swoop and Poop with Design Sprints

• Takes a week
• Uses a team of participants from different roles and perspectives in the organisation.
• Define and unpack a problem. Built a shared understanding
• Generate ideas and decide which ones to pursue for testing
• Build a prototype of their ideas, and validate assumptions by observing real people using it.

This helps you escape the risk of designing by starting with a list of stakeholder requirements, which may not be correct. “Unfortunately, because there’s usually no validation process built into the conventional design process, it isn’t until late (sometimes as late as when the product ships) that the team learns they went down the wrong rabbit hole.”

“Trying out what seems like a great idea and discovering that you’re wrong is a fantastic way to learn. Doing it quickly and early in the process mitigates the risks associated with heading down the wrong paths, delivering more educational value to your organization at lower cost.”

“When an organization integrates design sprints into projects, they see a dramatic decrease in outside influencer disruptions and an increase in their design quality.”

Continuous optimisation

Optimize Customer Experiences With Online Testing And Continuous Optimization

Improve your online testing through the following continuous optimisation objectiveS:

1. Learn from every possible customer interaction
2. Test customer interactions across the entire lifecycle
3. Align with a customer-centric strategy
4. Deploy testing within every possible digital channel

Failure and experimentation

Amazon’s 2015 letter to shareholders

“failure and invention are inseparable twins. To invent you have to experiment, and if you know in advance that it’s going to work, it’s not an experiment. Most large organizations embrace the idea of invention, but are not willing to suffer the string of failed experiments necessary to get there. Outsized returns often come from betting against conventional wisdom, and conventional wisdom is usually right. Given a ten percent change of a 100 times payoff, you should take that bet every time. But you’re still going to be wrong nine times out of ten.”

Different types of decision need different types of oversight

Amazon’s 2015 letter to shareholders “One common pitfall for large organizations – one that hurts speed and inventiveness – is ‘one-size-fits-all’ decision making.” “tendency to use the heavy-weight Type 1 decision-making process on most decisions, including many Type 2 decisions. the end result of this is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and consequently diminished invention.”

There are actually two types of decision:

• Type 1: consequential and irreversible or nearly irreversible. “one-way doors”. “these decisions must be made methodically, carefully, slowly, with great deliberation and consultation”. A minority of decisions are Type 1.
• Type 2. Changeable, reversible. “Type 2 decisions can and should be made quickly by high judgment individuals or small groups.”

Data-enabled decision-making and the role of top management

An executive’s guide to machine learning “Frontline managers, armed with insights from increasingly powerful computers, must learn to make more decisions on their own, with top management setting the overall direction and zeroing in only when exceptions surface.”

How companies become digital leaders “Not doing anything may be the riskiest move of all.”

Rapid development and measurement

Achieving a digital state of mind “If you’re doing to do a rapid ‘test and learn’ and get those feedback cycles, the whole philosophy has got to be that what we do is measurable.”

The world is bad because we made it this way. “The world is designed to work this way.”

When people talk about changing the world, ask: “How? For who?” For all the excitement about Uber and AirBnB, the service economy is nothing new. “There’s nothing disruptive about rich people getting richer.”

How to change the world

1. Get ignorant.

   Realise that the world as designed works in our favour. What if that wasn’t the case?
   The Veil of Ignorance is “The single most important political and ethical concept in a designer’s toolbox.”

2. Look like the world.

   “Our diversity is our strength, and we’re idiots for not leveraging it.”
   If people have narrow life experience, you just get “white boys solving problems for white boys”.
   “They’ve never been harrassed, so it doesn’t even occur to them that that’s a problem you have to solve for.”
   Similarly with cabs refusing to stop, or being assaulted,

   “Empathy is not enough – we need inclusion.”
   The point isn’t that any particular experience or classification makes you a better designer. People are just better informed about themselves than they are about others.
   Our teams need to reflect the diversity of who we design for. It’s not just about race or gender, but experiences, needs, thinking, solutions.

3. Design the right thing

   The AK47 is easy to use, easy to manufacture. But design is about more than this.
   “Nothing who’s primary purpose is to kill can be said to be designed well.”
   “Attempting to separate an object from its function, in order to appreciate it for purely aesthetic reasons, or to be impressed by its minimal elegance is a coward’s way of justifying the death they have brought into the world and the money with which they’re lining their pockets.”

   “Design is a trade done for money, but we have a choice about how we make that money.”
   “Your role as a designer is to leave the world in a better state than you found it.”
   “You are responsible for what you make.”

There are big design problems for us to solve

• Global warming
• The migrant crisis
• Guns in the US

We’re lucky people – so we’re responsible for helping others who weren’t as lucky.
Change how we design and who designs.
Use your time on this world in the interest of making others free.

Key observations

• The overall design approach is successful. The idea of thinking of content in blocks was easy to grasp, and the context-sensitive popup options were intelligible.
• The function of the up and down arrows to the left of content blocks wasn’t immediately clear.
• Users quickly worked out what the block-positioning arrows did, and were happy with this approach. To my surprise, no one tried to drag and drop blocks after finding out how to reposition blocks.
• In 1.0 people weren’t sure what the S means – is it strikethrough or the option to remove a link? Perhaps we can use proximity to make this clearer – associating this options clearly with presentation controls (underline, italics) or functionality controls (hyperlinks).
• Some users tried to drag and drop images. How will dragging and dropping work – within and between blocks?

Full notes of the testing sessions

First prototype

Tester 1

Prototype: “A beautiful thing about Apple…”

Clicks into a block and sees and instantly understands the alignment options.

Clicks on the ¶ symbol but it doesn’t do anything. Expected it to be interactive as the mouse changed to a hand pointer, suggesting interactivity.

Clicks on the up arrow, but nothing happens. Wasn’t really sure what it would do.

Enters a line return to break up the paragraph of text, but isn’t able to do this. Thought that the arrows up and down might allow the insertion of a line break.

Highlights text to make it bold and expects to have to use a keyboard shortcut as nothing was previously visible on the screen. But then notices the popup. Comfortable with the options presented.

Heading and blockquote options don’t behave as expected. (Dismissed as a limitation of the prototype)

Clicks on the image. The up and down buttons don’t do anything. Tries to click on the image icon but it doesn’t do anything. “I don’t understand what the symbols are. I expected them to be interactive but they don’t do anything.”

Tester 2

Understands the split between code and visual editors.

Likes the context-specific editing as it keeps the screen more focused.

Tester 3

Understands the concept of HTML and editable preview.

Wonders how valuable constant HTML visibility is to the average content editor. Will they break things?

Second prototype

“1.0 Is The Loneliest Number”

Tester 1

Clicks on the heading text block. Then clicks on the up arrow. Then clicks down on it. “Ah, that’s interesting. So these up arrows change the order of the boxes”

User expected to be able to insert line breaks. (How do we want to handle line breaks within paragraph blocks? Do we permit them, meaning that we have big text blocks, or do we take line breaks as denoting the start of a new paragraph block? I’d prefer the latter but haven’t thought about all use cases.)

Clicks on the paragraph of text, and then clicks on the ¶ symbol. Changes to blockquote style from paragraph style. Understands that this allows him to apply style to the block.

Scrolls down the page and clicks on the plus sign at the bottom.

When asked what this is, says that it’s “a shortcut to functions that you use regularly.” When asked: what would you expect to happen if you clicked on one of these items? “It would create a new paragraph, image, heading, quote, etc, which you’d then populate with content.”

Tester 2

Clicks into the block and manipulates alignment options – working as expected.

Clicks down arrow and block responds as expected. Understands that the page is made up of blocks and that these can be repositioned.

Clicks on “+” and explains that this will add a new block. More options than expected: expected just image or text, so the extra options are “a nice surprise”.

Drags and drops an image into the content block. (Is this something we want to design for?)

Clicks on ¶ – and changes the text block type

Tester 3

Clicks into the text block. “I wonder what the arrow does” Clicks on the down arrow and sees the block move down. Understands it fine.

Looks at alignment options. “Does that do the whole block? It does”

In general, understands the popup text formatting and link options. Not sure what the popup “S” with strikethrough icon means.

Clicks on image and manipulates text flow options.

Clicks image icon, but “nothing here”. Expects caption controls.

Clicks on the “+” sign, and understands that this is for adding blocks. Wonders how lists will interact with paragraph blocks, and how we could set levels of header.

Easily understood the blocks concept. Wondered how well this would handle more complex page layouts.

Innovation and failure – Luke Williams, RNLI

Set up organisation-spanning structures to keep an eye on trends:

.@RNLI have a ‘future trends’ group to look 5-10 yrs into the future: digital currency, drones, IoT etc #NFPtweetup — Dan Papworth Smyth (@dansmythphoto) January 24, 2017

Formal approval processes often take longer than tech implementation work:

Half a day to set up being able to accept Bitcoin, 6 months of approvals to actually get live @RNLI @alukeonlife #NFPtweetup — Dan Papworth Smyth (@dansmythphoto) January 24, 2017

How to decide whether to jump a trend:

• Where is this trend in its lifecycle? Has it peaked? Which other organisations have taken part?
• Can we relate it to our organisation?
• What time/resources do we need to make it happen?
• How do we make it authentic to the trend?
• Does taking part present any risks?

RNLI are using chatfuel.com to build a Facebook bot.

Chatfuel is a great tool (ht @ade), we used it to build a chatbot as part of a hackathon https://t.co/dCzYs7h9Pz #NFPTweetup — Jonathan Waddingham (@jon_bedford) January 24, 2017

Tips for getting buy-in for an idea:

Steps to writing a great digital brief. #NFPTweetup pic.twitter.com/RuKEMCivcD — Nathan Murray 👓 (@MrNathanMurray) January 24, 2017

Writing a digital brief – Jonty Sharples, Hactar

Headline structure for a digital brief:

Start with the user need and then think about the product. Not the other way round.

Digital Transformation approaches and language – Joe Freeman, Breast Cancer Now

Some resources on transformation recommended by Joe:

• A digital transformation manifesto
• My love-hate relationship with the term digital transformation
• Eat Sleep Work Repeat podcast
• Transformation but not as you know it…

Digital transformation is mostly about culture and ways of working, not technology

Like @JosephFreeman found at @Sue_Ryder, our events team is our go-to for upskilling/empowering in digital things. #NFPtweetup — David P Hunt (@d_p_hunt) January 24, 2017

(This has been my experience too.)

Two tools for measuring how well your organisation is doing digitally:

• Digital Maturity Matrix

What I built was pretty basic – it was running on an old laptop in the corner of my flat, and it just displayed the results on the laptop screen after a specified number of hours had elapsed. A solid final project, but it only ever felt like a proof of concept. I wanted to build this into something I could use, and something that other people could use too.

So over the next few years I built it from the ground up as a full-blown web application that has now been launched as a real business: MySocialSummary.com.

The CS50 class was great, but once you have the foundations of understanding in place, there’s no substitute for the motivation, exploration, imagination and excitement of working on your own idea. Subject to having a reasonable grounding (and I’ve taken a number of computer science classes, not just CS50), I’d definitely recommend taking on a project that excites you, to take your knowledge to the next level.

Going through this process I’ve learnt a lot. I’ve divided this into two lists: one on specific computer science and business insights, the other on more overarching observations.

Specific computer science and business lessons

• How to embed tweets in emails. (Or, rather, why you can’t embed tweets in emails.)
• Working with APIs. An API is a tool for communicating between two pieces of software. Manipulation of data from the Twitter API is the foundation of My Social Summary.

• How to load test your website. If you’ve built a website, you probably want people to visit. You commission your infrastructure with a certain number of visitors in mind. It’s worth testing how your site would behave if an unexpectedly large number of visitors arrive at once. This is called Load Testing. I used a tool called Load Impact to check out the load speed of My Social Summary when you add more users. As you can see, the homepage performs well with 50 visitors loading it at the same time.

  

• Working with code libraries. When you’re programming, you’ll often want to do something that other people have done before. For example, loading a web page, sending an email, or accessing the Twitter API. If you’re trying to solve a common problem, often there’ll be a code library of pre-written code that will help you with this. By using a code library, rather than producing this code from scratch, you can save a lot of effort, and the quality of the code is probably much higher quality than what you’d produce yourself. This frees you up to focus on the problems that no-one else has worked on yet. Not all code libraries are created equal – some are better documented and more fully-featured than others – so picking the right one is an important decision.

• How to set up login via Twitter. A good example of when a code library can be useful. I’d have definitely struggled to build this authentication myself if I’d had to do so from scratch. Check out this documentation.

  

• How to securely handle user input. If you take any form of user input on your website that ends up interacting with the web server, it’s a security risk. So you need to build your site from the ground up with security in mind. I really enjoyed learning about how you can protect yourself using parameterised database queries and other security best practices.
• How to set up a LAMP server. That’s a common infrastructure stack, using Linux, Apache, MySQL and PHP to run a website. I’ve learnt how to manipulate the different parts of the machine (e.g. updating Apache settings) and use the different management tools.
• The complexities of daylight saving time. When the clocks changed, I noticed that my daily summary emails were arriving an hour early. Fixing this problem for myself was trivial, but solving it for all different users – and possible future users – was harder. I’ve built functionality that checks every evening against the very latest timezone rules, and supports every timezone. (Check out the exhaustive dropdown list on the page where you can sign up for a free trial account)

• The benefits of refactoring code. If code is a set of instructions, refactoring is rewriting those instructions to be more efficient. Refactoring your code isn’t fun – it’s hard work, and involves trawling through your past self’s logic and trying to improve upon it. But it can make massive performance improvements. In one case I was able to decrease the number of database interactions by a factor of a thousand through batch operations.

  

• The importance of good version control and deployment infrastructure. Any halfway competent programmer will tell you that good version control is really important. At the start of this project I was evidently not a halfway competent programmer. If you operate version control you can review the history of your code over time, and quickly track down errors. Your deployment infrastructure is the process from getting your code from a code file on your computer to your web server so that it can start doing real work. Ideally you want this to be as frictionless as possible. In the early days, I had no version control, and my deployment procedure was copying and pasting the contents of one text file into another. I can confirm that this was not a good process.
• How to work with a front-end framework. Specifically Twitter Bootstrap. My interest in this project has been more in concepts and engineering than in presentation. But any website needs to have a front-end of HTML and CSS. Rather than building all of this from scratch, I used the Bootstrap library to help set up the structure and presentation of the site. This made it quite easy to make the site mobile responsive.
• How to integrate with a payment provider. Taking money from people online has a few complexities. The main one is passing data from your payment provider to your web server, so that you know who’s paid what. This entailed using Paddle’s Webhooks. Handling all the different types of event that might happen – and doing so securely – was quite a bit of work, but it was very satisfying to get this set up and working correctly, and see money come in to my account.
• How to use cron jobs. On a Linux machine, you can instruct the computer to automatically carry out tasks at certain times. These are called ‘cron jobs’, and they can be really powerful. I make extensive use of these for My Social Summary, for the management of user accounts, and the sending out of communications to new users.
• How to satisfy the Twitter brand guidelines – and how to update things when they change…. Before coming up with a name, I carefully checked Twitter’s brand guidelines to make sure that the name was consistent. I felt pretty smug about having thought to do this. However, Twitter changed the goalposts, and later the name I’d chosen was no longer acceptable. So I had to come up with a new name for the service – and then roll out this change to all aspects of the service.
• How to set up HTTPS security. Privacy and security are important, so I wanted to make sure that the site I was building used HTTPS encryption. Setting this up was quite easy – much of the process was done by my web hosting provider who installed the certificate. Seeing the little padlock next to the domain name for the first time was very satisfying.

• How to set up email authentication via DKIM. I had no idea what this was. But one day I saw a little question mark when I opened up one of my summary emails, next to the sender name. Gmail said that it wasn’t sure if the sender of the email was actually who they claimed to be. Google had instructions on what to do next if you were the person sending this email, and I didn’t even need to follow all of these – pretty much all I had to do was change a setting with my hosting provider and this was resolved.

  

• How to migrate hosting provider. Part-way in to the project, it became clear that I’d outgrown the infrastructure I’d started on and needed a more robust web server. The process of transitioning from one provider to another is something I’ve overseen before in my day job, but to do all the work myself was useful. Before I started the process, I had to choose a provider with the right infrastructure and an SLA to match my needs. (I chose TSOHost and I’m happy with them.) Then I had to commission a new database within slightly different constraints, migrate the data over, and switch the traffic over to the new site. None of this was massively difficult, but the work took about a day nonetheless.
• How to register a business in the UK.
• How to track business finances. Fortunately the costs of a digital business are easy to track, so a simple Google spreadsheet suffices.
• How to complete the year-end tax process. I haven’t had to do this yet – this is likely to be the real learning curve.
• How to comply with the EU’s new rules on VAT for digital services. The EU changed its laws for how VAT is charged on digital services. Rather than being charged at the rate of the country of the person selling the service, VAT has to be charged at the rate in the country of the person buying the product. You have to collect evidence to prove this, and you can either make VAT payments to each EU country separately, according to its own processes, or you can submit centralised returns through the UK VAT MOSS service. Both of these processes seemed tedious and onerous, so I opted instead to use a payment provider that would handle VAT on my behalf. This meant switching from Stripe or PayPal (I’d been testing out both) to Paddle. Here’s a blog post on Paddle’s site discussing the changes.

• How to set up a Content Delivery Network. Your website lives on a web server – a computer probably sitting in a datacentre somewhere. If someone the other side of the world wants to see one of your web pages, the information has to travel all the way to your web server and back. It would be quicker if they were requesting the web page from somewhere closer to home. A CDN is a network of web servers around the world that does this task. A CDN can also help take some of the load off your web server – if people are getting pages from the CDN rather than your web server, the web server doesn’t have to work as hard. I use the Cloudflare CDN. It has a free tier and is easy to set up. It disrupted the site’s HTTPS setup for a few hours while I turned it on, but other than setting up the CDN was easy.

  

Broader lessons

• Starting with a Minimum Viable Product is the right way to build a digital product. I started off by building the simplest possible tool to meet my needs, and then I successively improved it from there until I had a product that was ready to launch. Had I tried to build everything in one go, I wouldn’t have known how to do it, or what I was trying to build. And I wouldn’t have had the motivation and insights provided by being able to use the tool myself from early on in the process. It would not have even been clear if what I was building was going to useful to anyone until after it had been launched. The Lean Startup methodology, and the Agile Manifesto, tell us that you should start off by building a Minimum Viable Product – the most basic version of your product that is still valuable. I already had an intellectual appreciation of this, but it’s always useful to test theories out in practice.

  

• The power of self-directed learning. This project has been entirely self-directed. This has been thrilling – a chance to follow my interests and try and constantly reassess what I need to do – or learn – next. You learn a lot more, and a lot more broadly, if you’re learning to achieve a personal goal.
• The power of emergent possibilities and understanding. When I started this project, I didn’t know everything upfront. I didn’t even know what I would need to know. The Agile approach to projects – release early, and iterate based on evidence of real user behavior – is thoroughly vindicated by this. Working on this project has strengthened my understanding of, and commitment to, the Agile and Lean principles of releasing early and improving your product from there.
• The importance of knowing how you work – and what you need to work. Working on a side project is different to working your day job. I’ve become more sensitive to when I’m in the right headspace to do some work on My Social Summary, and when I just need to take it easy. I also have a better understanding about the conditions I need to get work done outside of my normal job. I tend to work quite well in uninterrupted stretches of at least an hour and a half – especially if I’m programming.
• The importance of dealing with changing circumstances. The external environment changes. During this project, VAT legislation changed and Twitter’s naming rules changed. You need an approach to building a product that is not only resilient to this, but embraces it. Again, if I hadn’t been following an Agile approach, I’d have been in trouble.
• To trust in your ability to keep learning and know more than you know now. Taking on a self-directed project builds self-confidence and self-reliance.

I’d love for you to check out what I’ve been working on. My Social Summary has a free one month trial, with no card details needed, so see if it can help you get more from Twitter.

What’s next for me?

• Spread the word about My Social Summary.
• Explore how we might overcome the ‘filter bubble’ that seems to be narrowing our political and cultural discourse. This feels like an important design challenge.
• From a computer science perspective, I’d like to build something in a different, and more modern, technology stack. Probably Node.js / Foundation (for front-end) / PostgreSQL / Heroku.

User expectations are set by industry leaders

“You may not be in direct competition with Twitter, Google, or even MailChimp, but users will expect the same ease of use that they have been given on those platforms.”
“You are not just up against your competitors, you are competing with the user experience of every digital player out there.”

If you don’t change to meet evolving user expectations, a competitor will appear and render you irrelevant

Napster changed users’ expectations of how they could consume music. People now expected cheap, fast (digital and on-demand) music, focused on tracks not albums.

“The music industry fought hard to have Napster closed down, but the damage was done. User expectations had changed and there was no going back… Instead of adapting to this change in user expectations, the industry failed to act.”

Between 2000 and 2010, record store sales fell by 76%. HMV and Tower Records “crumbled.”
Apple created iTunes and delivered what people wanted. “The music industry lost an unprecedented opportunity because of its failure to adapt to the changing landscape.”

Similarly, Blockbuster had numerous opportunities to purchase Netflix for as little as $50 million. But Blockbuster didn’t accept that customers didn’t want to visit a physical store. By the time it accepted this and began offering a postal service, Netflix was already transitioning to digital streaming. Once this infrastructure was set up, Netflix’s costs plummeted. Blockbuster, with its physical stores and associated costs, could not compete.

Use digital to help you adapt strategically to a changing environment

“With such a rapid rate of change, creating a three to five year strategy is impossible…
Instead, a digital strategy should help the organization become flexible enough and properly prepared to adapt to new challenges and innovations as they arise. The digital strategy should create a digital team capable of thinking strategically on a daily basis.”

“The web requires fast adaptation and close collaboration of people with very different skills. This means that it is the people working with digital every day who have to make rapid, informed decisions. They can’t wait for senior management’s consent.”

Boag recommends: “Good Strategy/Bad Strategy” by Richard Rumelt. A framework for creating a strategy:

1. A diagnosis
2. Guiding principles
3. Coherent actions

“a responsibility matrix is considerably more effective than a web steering committee”

“Web steering committees are not a bad idea in principle, but in practice they can often significantly slow the agility of an organization in a realm where responsiveness and adaptability are crucial.”

“Important decisions are often delayed until a date can be found for the entire steering committee to meet. The meetings themselves often focus on endless internal discussion, rather than basing decisions on data and user testing. Finally, and probably most significantly, most of the people in the room are not qualified to be making decisions on the subjects being discussed.”

Instead, Boag recommends using a RACI framework, showing who is responsible/accountable/supports/is informed about each area of decision-making:
Responsible: doing the work
Accountable: formally accountable for the success of the work
Consulted: consulted about the work
Informed: told of the outcome

Better collaboration across teams

You don’t need to remove departmental structures – just make the edges of departments “fuzzier”.

“Departments should not be the only structure within an organization; there should also be working groups and other smaller teams that work across these departmental divides. But do not mistake this for more committees. I am not talking about interdepartmental committees. I am talking about real teams made up of people from multiple departments who sit and work together.”

Policies promote good decision-making

Standard sources of conflict experienced by digital teams: homepage space, requests for low-value web content, hostility to removing out-of-date content.

Save time on these arguments by having policies. These are impersonal, and have organisational buy-in.

“Instead of saying no to somebody who wants content on the homepage, you are just implementing a policy. Instead of removing somebody’s content, you are just following the rules. It’s not personal, it’s policy.”

Martha Lane Fox’s strategic review of the UK government’s digital offering, 2012

• Manage digital centrally. Commission content from departmental experts as required. (“complete reversal of the previous policy”)
• Focus on user needs and the delivery of online services – not just communicating information.
• Radically simplify the government’s digital footprint.
• Move from large technical projects to a more agile, iterative approach based on extensive testing.

Predicted that if the government moved 30% of its interactions with citizens online, they could save more than £1.3 billion.

If the UK government can make radical change, so can your organisation

“If an institution with as much inertia and legacy as the UK government is willing to consider such fundamental cultural change, then it demonstrates that this is possible for the vast majority of organizations.”

Common cultural characteristics of effective digital organisations

• Collaboration.
• Agile, iterative development.
• Digital by default.
• Innovation.
• Service-oriented

Generic advantages of digital approaches

• Cheaper
• Faster
• More flexible
• Easier to monitor
• More targeted

Senior management need to completely re-evaluate the business in the light of digital

Boag recommends Osterwalder’s Business Model Canvas as a tool to achieve this. Explore the following areas:

1. Customer segments.

Who are your customers, and can digital help you better understand them?
Could you personalise your offering?
Could you serve niche audiences that were previously not cost-effective?
Could digital help you broaden your customer base?

1. Channels.

Can you deliver your offering through new channels?
Could you move to a purely digital offering?
Can digital enhance existing channels – by making them faster, more efficient or cost-effective?

1. Customer relationships.

What expectations do different customer segments have of you, and could digital help you surpass these?
Could digital help your manage customer relationships more cheaply?
Could you transition from telephone support to self-service?
Could customer service be automated?
Could you build online communities where customers support each other?
Could you use digital to enable customers to take a more active role in evolving your products or services?

1. Value proposition.

What problems does your organisation solve? What needs are you satisfying?
Could digital help you make your offering more valuable?
Could digital allow you to solve new problems for your customers?

1. Revenue streams

Could you collect revenue more efficiently, and more easily for customers, using digital payment?
Could you introduce new revenue streams, e.g. a subscription model?

1. Key resources

What are the foundational resources for delivering your offering?
Could digital be used to replace these or lower their costs?
e.g. do retailers need physical store fronts?

1. Key activities

What are the foundational activities required to deliver your offering?
Could digital streamline these, or even automate them? (e.g. automated dispatch management of ecommerce orders)

1. Key partners

Who are your key partners or suppliers?
Could digital tools help you manage them, or even replace them? e.g. self-publishing.

1. Cost structures.

What are the most important costs associated with your business model?
Could digital lower these costs?
e.g. lowering distribution or customer acquisition/retention or stock management costs.

Looking at competitor organisations during this process can help motivate senior managers.

Embrace failure as integral to improvement

“we need to see failure as a necessary step towards success. Only by failing do we understand what success looks like.”

“In an industry so young and dynamic, the only way to innovate, the only way to progress, is to experiment and that will inevitably mean failure. We need to nurture a culture where failure is acceptable and, in fact, expected.”

“You might think that building something that fails is a waste of time and money. That is true if huge amounts of both have been sunk into its development. However, if you are working within an iterative process, centered around rapid prototyping this will not be the case.”

Test your assumptions by building something and seeing if it works. If it doesn’t, you can learn and improve.
If you have competing ideas, you can test them against each other at prototype stage. This de-politicises decisions, and helps organisations make decisions based on evidence.

Allow staff some time to explore new ideas, and create a culture in which new ideas can be expressed and valued.

Unless you are solely competing on price, you need to embrace customer-centricity

Marketing departments are still focused on broadcasting to mass audiences with mass media, rather than focusing on connections.

“One step in the right direction would be to make user testing a permanent and ongoing feature of your company’s culture.”

Twitter have an ongoing programme of usability testing that is open to anyone.
“user testing is not a periodic event confined to a small team, but an ongoing company-wide policy.”

In “Rocket Surgery Made Easy”, Steve Krug recommends a rolling programme of monthly testing.

How to get the most out of your digital team

“The most important factor is to give your digital team the freedom to do its job and set its own direction. This is a new and very light form of leadership based on respect rather than authority. It means relinquishing control and allowing people to direct their own roles.
I am not suggesting staff should be given the freedom employees of Valve or GitHub have. Although this might ultimately be beneficial to both employee and organization, it requires such profound organizational change that it is beyond the reach of most companies.”

The Harvard Business Review advises that “Senior leaders need to get used to the idea of abandoning absolute control”

“digital workers… need to see the reason for management decisions. It is not enough to tell them how things are going to be; they need to understand why a direction has been chosen.”

Your digital team needs the right tools

“You wouldn’t expect a professional tradesman to work with the same DIY tools we buy from B&Q, so why do so many companies insist that their digital teams use the same technology as the rest of the organization? What some perceive as luxuries such as smartphones, high-end computers, and tablets, are in fact tools of the trade for a digital professional. They shouldn’t have to fight to get these tools, they should just be provided.”

The knowledge of your digital team is their biggest asset. You need to actively invest in growing this

“you need knowledge. You need experts in creating digital solutions. They are your most valuable asset.”

Give staff the time and opportunity to continually strengthen and update their skills.
Digital professionals are motivated to do this – you just need to give them the time to learn and experiment, and to meet with peers.

Traditional project management approaches don’t work well in digital

Digital projects are so complex that traditional project management approaches struggle to scale.

For example, “It can prove nearly impossible to accurately specify large web projects due to the huge number of variables and complexities.”

Incremental change is better than big projects

Monitor user behavior to identify problems, prioritise areas for action (based on value of fixing and ease/cost of fixing), invest in a small incremental improvement, test and iterate, then go live. This reduces the risk of failure.

“This significantly reduces the amount of planning required and acknowledges the uncertainty inherent in running any large website.”

“The idea is to establish a rhythm of building, measuring, learning, and improving so that the site naturally evolves over time.”

Focus development work on user needs, not stakeholder requests

Any development work should be based on a user need, and to serve a defined persona.
“These personas need organization-wide approval and they become the bedrock on which your applications are built. Only tasks that meet the needs of these personas should be considered, and no task should be built that prevents a persona from completing one of their key tasks. They act as a filter for deciding which user stories will be accepted into the backlog of work to be developed.”

Tips for building grassroots change

Support each other, to build a safe and fun environment where you have the support of your team to take some initiative.

Try implementing small improvements to how you work. This will build your confidence and get management used to the idea. Don’t ask for permission – just make the change. Have a clear rationale ready in case you’re challenged. This should be focused on how it benefits the organisation.

Always educate and build bridges. So don’t just say “no” to requests.
Make sure you aren’t seen as a blockage that needs to be worked around.
Help colleagues think through alternatives – focus on their underlying need, rather than on the particular approach they’ve advocated.

Educate colleagues:

1. Highlight best practice. (Competitor examples are particularly effective)
2. Destroy preconceptions, using evidence. (e.g. any weird beliefs that people have)
3. Promote your successes. Explain why things worked – builds the credibility of your team, and the organisation’s understanding.
4. Explain failures. Explain why they happened and discuss how they could be avoided in future. Build a culture that embraces failure.

Senior management

“Getting frustrated with them will not help, but learning more about them will.”

Understand their objectives/targets and needs.

Focus on:

• topics that they are interested in
• their broader objectives and targets
• return on investment
• threats

Julie argues that digital superheroes…:

1. Make products that really help people

   e.g. the Ugly Mugs app, built for sex workers to make sex work safer, or Tony Canning’s use of 3D printing to reduce the cost of prosthetic limbs from £20,000 to £40 – and reducing production time from months to hours – and making the technology available open source so that others can use it.

2. Use any tool or platform that they can get their hands on

   BBC used Whatsapp at the height of Ebola in Sierra Leone in 2014 to provide information. Largely simple infographics. Reached 20,000 people in the first 3 months who were otherwise very hard to reach. The BBC used existing technology as it was cheaper, quicker and more effective.

   Crowdfunding sites – e.g. Kickstarter and IndieGogo – reduce the need for people to go through a middleman. Charities need to think what this means for them.

   Girl Scouts in America used IndieGogo to raise funds after rejecting a $100,000 transphobic donation. The troop of Girl Scouts that turned down the donation started a crowdfunding campaign that raised 4 times as much money – and sent out a powerful message about inclusion. They also changed the policy of how the national organisation takes donations.

3. Aren’t frightened to try new ways to do things

   For example bringing in service design thinking or agile methogologies.

   St Mary’s Hospital in Paddington has created the ‘Helix centre’ innovation lab. It’s focused on lean, iterative solutions and combines online and offline. Projects include increasing rates of bowel cancer screening, offering guidance for clinicians on how to communicate about end of life care, asthma management tools for kids, dealing with storage of IV fluids.

   The Town of Jun use twitter for civic discourse and interactions – e.g. booking an appointment with the doctor or making a complaint. Everyone was trained. Saw a rise in public workers being thanked.

4. Can be found anywhere – not just in tech.

   Google studied a favela in Rio, and didn’t expect to find much technology. Rather, they found a flourishing ecosystem, with radio stations, cyber cafes. With the proliferation of smartphones, people often ‘leapfrog’ the desktop ‘stage’ of development.

5. Can have significant impact on organisations.

   The British Library now conceives of itself as a data institution, rather than a custodian of physical objects.

   The #tweetmythesis movement encourages academics to share their thesis in a tweet.

6. Can have an impact in major commercial brands too.

   Barclays set up Digital Eagles programme, driven by the profit motive of reducing cost by moving more people to online banking. (And thereby reducing interaction costs, e.g. staff and branch costs). So it has trained 20,000 staff to train people across the community.

7. Should work for organisations interested in changing.

   When researching “The New Reality” Julie found some organisations just weren’t interested in digital transformation. She thinks many won’t survive a decade, and wants to spend her energy not fighting those ones to change, but working for the ones that do want to change.

A few miscellaneous recommendations:

• Recommended meetups: Citizen Beta and Tech for Good
• “Apps without marketing are pointless”
• If you do pro bono work with a charity, make clear the equivalent financial value. Otherwise they won’t value it because it’s free.
• “Asking people to experiment is easier than asking them to commit”

• You become confident in dealing with different domains of knowledge and different types of truth, and arbitrating between them.
• You can imagine and weigh up possible futures, and sell and fight for the one that seems best.
• You make decisions with imperfect information.
• You lead from the intersection between optimism and pessimism. Seeing how things could be different and better, and helping others believe, but rooting your thinking in understanding of complexity, risk and assumptions.
• You call out the risky assumptions underlying wishful thinking.
• You’re adventurous and humble and focus on value.

Estimating jobs

Gather the whole team for this exercise. e.g. UX design, developers, testers, product managers.

Discuss each job, and collaboratively rank the jobs in order of complexity.
Think about all the work that will be required to get this job ready to go live.
For each job, some elements might take more time. E.g. UX might take a long time on one job, but the development might be quick. So you need the different perspectives involved in the discussion.

Groups similarly-complex jobs together.

Apply T shirt sizes to these groups: S, M, L, XL, XXL.

If you have anything XXL, break it down into two or more different jobs.

Assign ‘story points’ to each job, based on its size:
S: 1
M: 3
L: 5
XL: 8

(If you’re running an internal team, there’s no need to think in terms of hours.)

Estimating velocity (how much work you can get done in a given time period)

If you’re an established team, you’ll know from experience how much work you can complete in a given period.

If you’re a new team, you’ll need to collaboratively estimate how much work you think you could carry out in a given development sprint (e.g. 2 weeks).

Work through each of the jobs, and combine them into groups showing how much you think you could carry out in a sprint. Then total the number of story points of the jobs in that group.

Once you’ve done this 10 or more times, work out the average number of story points per sprint. (Round down to the nearest whole number if required.)

This is your estimated velocity – the number of story points you think you can complete in a given sprint.

Prioritising jobs to build a plan of work

Prioritise all jobs using force rank: the highest priority job goes at the top, the lowest at the bottom. Nothing has equal priority.

You can also factor in sequence dependencies (and so promote some jobs that are required to allow you to complete other, higher priority jobs) and highlight milestones (points where you’ll have something specific to show off).

Once you have this ordered list of jobs, lay them out into separate sprints, totaling the number of story points you think the team can achieve during that sprint.

N.B. to adjust for team size, the time of year (e.g. holiday season), and to leave a % for emergent scope: between 20 and 40%, depending on how confident you are that the stories are comprehensive.

Managing ongoing development

Scope is variable. The plan isn’t fixed – it’s an overall route map.

Estimates are fixed. Once an estimate has been made, don’t change it. Retroactively changing estimates wastes time, and disrupts your statistics.
If you aren’t able to get through as much work as you expected, you need to have this recorded in the sprint burn-down. Reflect on this in the sprint retrospective. Re-project your velocity as required.

New stories are sized with reference to existing stories. This makes it quicker to estimate new jobs, as you can compare them to other similar jobs already completed. So estimation becomes easier and quicker over time.

Update your burn-up chart each sprint. Track the speed to the target (velocity) and the distance from the target (scope).

Once your team is established, use their measured velocity and use this to re-project timings. This is more useful than the estimated capacity that you use when planning with a new team.

Project variables

In a digital project, you have a set of variables that you can control.

In a Waterfall project, you attempt to deliver a fixed scope. To meet this, the time and cost may have to change – and the quality may also be undermined.

In an Agile project, you fix the time and cost, but vary the scope. You might deliver less, but the quality is high, costs are controlled, you release on time, and what you do deliver is the most valuable work you could have been doing.

(A lean approach is slightly different. You look at flow rather than time, value rather than cost (if you only focus on cost you might not get a good ROI), and quality rather than scope (deliver less, deliver right).)

Agile focuses on the early delivery of value to the customer

A Waterfall project only generates value at the end of the project. You start off by specifying what you want to build, then you carry out design, then development, then testing, before launch. It is only at this last step that any value is experienced, because this is the first time that a working product is used by the customer.

An Agile project goes through all these stages during each sprint. At the end of each time-bound unit of development, something valuable is released to the customer. This means that value is delivered more quickly than in a Waterfall project.

Agile focuses on frequent delivery of value

Working in short sprints allows you to regularly release valuable software.

In 2010 Etsy re-engineered their backend systems to reduce the time taken between software releases. The aim was to be more responsive to customers. In 2011, Etsy released new software 12,000 times – about 30-40 releases per day. They tested code on a small percentage of the audience, and then ramped this up to 100% if it tested well. This new approach reduced defects on go-live by 90%, as each change was so small that it was very difficult to make mistakes.

Think about how you could make a small, safe change to your business process. This would help with risk management, and with agility.

Agile values emergent solutions

It’s important to plan, but plans always have to change. Our competitive advantage is in how well we respond to change.

Management must provide vision and purpose, so that self-organising teams know the overall direction they should be travelling in, and management must clear out blockers for teams. This is different to command and control, with management closely directing the actions of teams.

We must design for uncertainty and change

The ‘Waterfall’ model of developing software was codified by Royce in 1970. He noted that it is “risky and invites failure”. The waterfall approach assumes a simplicity which does not exist in a complex environment like software development.

When applied to software development, a waterfall approach makes three dangerous foundational assumptions:

In an uncertain environment, use an empirical approach

Build something -> Measure how well it works -> Learn

Loop through this cycle as rapidly as you can.

Regular delivery allows for the earlier realisation of value, and for learning and customer feedback that can improve the product. This reduces the risk of building the wrong thing. If you take a single, late delivery approach (as in Waterfall), the risk of not delivering the right product only starts to fall later on when you do UAT or go live. It’s better to deliver regularly, and to focus on high-value and high-risk areas, delivering these early to de-risk the project.

The economic case for Agile vs Waterfall

The Standish Group’s 2002 Chaos Report found that: – Around 1/3 of projects are never completed. – For completed projects, 45% of features were never used. So almost half of all time, effort and cost was for no reason. – 19% of features were rarely used. – 64% of time/effort/cost was wasted. So you need to focus on value, and you need to validate ideas with customers.

The Cutter Consortium found the following improvements due to the adoption of Agile practices: – 61% cheaper. – 24% faster. – 83% fewer defects. – 39% smaller teams. These gains are a result of simplicity – maximising the work not done.

The key learning is: make your products smaller and more focused on the core value.

Report on the delivery of actual value, not on progress through project stages

In traditional project reporting, you monitor progress through different stages of the project. But there’s no value delivered until you go live.

Instead: look at revenue, impact, satisfaction levels, customer calls, organisational KPIs.

As a Digital Product Manager, you own the value and purpose of your team’s work. You are responsible for the vision and the ‘why’. Understanding ‘how’ is the job of your colleagues who are all focused on delivering the vision.

You are never going to be an expert in all the domains of practice in your multi-disciplinary team. You shouldn’t be. You’ve hired designers, user researchers, performance analysts, developers and data scientists to be experts in their craft, so give them space to own their specialism.

So a Digital Product Manager should never write code at work. But I think that you should be able to write some rudimentary code, having had a serious play around, because it will help you think better as a Product Manager.

As Product Manager you should have a sustained critical engagement with the core domains driving value for your product. And you should be interested in the frontiers of value in your team’s work. If you’re a Digital Product Manager that probably means learning about digital technology.

You don’t need a computer science degree or technical background – a humanities degree is a strong fit, given the need to understand what ‘value’ means in a number of different domains, and to synthesize, integrate and communicate this understanding. What matters is curiosity, excitement, optimism and a lot of dour pragmatism.

When working with a user researcher, you don’t tell them how to do their job. But you know the types of question to ask, and how to interrogate and work with the answers. You need to be able to do the same thing with technology – you can’t just leave it as a magic black box. You need to be able to critically engage with software engineering to at least the same level of competence as the other specialisms in your team, and probably more, given its central role in defining what is possible. What’s special is that your engagement is focused on value rather than delivery.

Explore, ask questions, and continually think about how this domain might relate to the problems that your product is trying to solve.

Some practical next steps: If you want to be a Product Manager, learn what a function is, have an appreciation of boolean logic and ‘if’ statements, get your hands dirty writing some poorly laid-out code and deal with the consequences, play with an API or two to see the kind of value they provide, store some material in a database, play around with something that you find fun and interesting. I’ve written some guidance on learning programming here. There might be other specialisms in your team where your knowledge falls a bit short – such as data science. Take the same approach there.

@LucyCaldicott & @Skipinder set up a Facebook group for fundraising discussion

They’ve found it helpful to extend their network.

1. Take yourself out of your own little circle. Go speak to others outside your organisation or outside your area #nfptweetup — Amy J France (@MrsAJFrance) June 9, 2016

Recently I attended an event for digital project managers. Most attendees were definitely not in the charity sector. Hearing quite different perspectives – and different acronyms and terminology – was refreshing and illuminating.

The group set up by @LucyCaldicott & @Skipinder is self-organising and non-hierarchical

What i love is someone who asks about badges gets a reply from a director of fundraising from a huge charity – @LucyCaldicott #nfptweetup — Amy J France (@MrsAJFrance) June 9, 2016

Reminds me of Clay Shirky: organising without organisations has become ridiculously easy #nfptweetup https://t.co/6YVQloqXZb — Karl Wilding (@karlwilding) June 9, 2016

Modbods is another recommended self-organising digital community. It’s on Google+ and is focused on community management:

Modbods? 😉 #NFPtweetup — Kirsty Marrins 📝 (@LondonKirsty) June 9, 2016

@charlotte_cox worked to move somewhereto_ from charity to self-sustaining social enterprise

They had to work on selling their core offer to customers. Potential customers weren’t actually interested in their core vision/mission, so they had to package things differently. They’ve boiled their key message down to: “Rent space from us, support creative young people”. (Or, to actually quote their website: “Find and book the perfect space to pursue your ideas and ambitions while giving a young person the chance to pursue theirs.”)

Biggest take out? Keep it simple. Strip back your message. Key message should be short and sweet. #NFPtweetup — NFPtweetup (@NFPtweetup) June 9, 2016

Tips on event social media from @dansmythphoto of @TeenageCancer

Covering a live event on social media doesn’t need to entail incessant posting of content:

Love the learning of “you are excited, you want to spam everyone but actually that tends to switch people off” #nfptweetup — Amy J France (@MrsAJFrance) June 9, 2016

Facebook will penalise you for sharing too much similar content.

Learnings: ‘Video is awesome, but have to be sensible with it’: FB downgrades similar content. Dedicate resource to find stories #NFPtweetup — Rachel Beer (@rachelbeer) June 9, 2016

It’s not always sufficient to just share supporter videos. You need to invest in editing them:

Dan says, they’d learned it’s not enough just to post supporters’ videos. They invested in editing to create watchable content #NFPtweetup — Rachel Beer (@rachelbeer) June 9, 2016

Teenage Cancer Trust re-cut their longer videos to appropriate length for social. Don’t just cross-post your video on all the different social networks: adapt your content for each. If you want to share a video on social, make it shorter, and make it work with no sound.

Just Giving’s #poweredbypeople campaign at the London Marathon

Just Giving created fundraisers a personalised page that they could share with their friends. It showed how much they’d raised, and pulled in information on their cause.

“People like presents more than asks.”

The videos were ‘a gift’ to fundraisers and didn’t appear like an ask to their friends – which really worked #NFPtweetup — Rachel Beer (@rachelbeer) June 9, 2016

Also, snapchat filters are apparently quite cost-effective at the moment, with a good ROI.

Design is inherently political, and we must not ignore this

Ask yourself: who is the “we” that gets to build the future?

If you don’t understand how something works, you are a consumer, not a citizen. Don’t be fooled by ‘magic’.

Richard Pope – “Software is politics, now” – it shapes power dynamics.

GDS came up with the design principles so that people would have a new language to use to change reality.

The advantages of working in the open

Child benefit tax calculator. They made a mistake, so someone suggested a fix on GitHub which has now been incorporated.

Ministry of Justice – problems with a form used by divorcing couples. Proprietary software. Took months to fix. The change on GitHub took 3 days. Massive difference.

What is digital, and what is our job in a digital world?

Definition of digital: “Applying the culture, practices, processes & technologies of the Internet-era to respond to people’s raised expectations.”

It’s not about technology – it’s about taking advantage of technology to redesign services and organisations to meet changed expectations.

Focus on delivery

Martha Lane Fox’s 4-page report gave just enough cover to start delivering. No need for a big strategy.

“The strategy is delivery” – key phrase at GDS.

Internal metric: write 100x more lines of code than lines of business cases justifying code.

Guy Moorhouse designed icons for GDS. But then he tested and found out that they didn’t help people, so he removed them and blogged about why.

Building the political case for change

GDS alpha was done openly. This was to create buzz outside the system to convince ministers that it was a good idea. This helped overcome reluctance from senior civil servants.

Do something valuable -> build political capital through an early win -> get rid of the ‘no’ people (spending all of the political capital)

Old approaches to service delivery are flawed

When Tom Loosemore started at the DWP in 2013, he asked ‘so, what have you been doing with all this time and money?’ For 3 years of work, they showed a 600 page policy design manual.

The DWP senior leadership thought of Universal Credit as a policy. But they hadn’t designed anything – they’d written a document. It had thousands of untested assumptions about people’s behavior. “a document full of false certainty”

When Tom arrived, the DWP processes were as follows (with each step done by a different team):

1. Invent policy
2. Guess requirements
3. Procure IT system
4. Inflict on users
5. Operate (aka ‘stasis’)

This is the wrong way to deliver services.

You must observe real user behaviour

People don’t know what they need. You have to observe real people in the real world “observe their actual behavior. Surveys are useless. Actually focus groups are useless.”

“Watch what they do, don’t listen to what they say”

“False certainty is our mortal foe”

“Start humble, stay humble”

Start small, build a shared vision and empower the team

Start really small. Iterate based on how people actually use the service.

Craft a vision that everyone can use to steer every decision. Use simple language.

Empower people to make decisions based on this vision without having to run it up the hierarchy. And because you have governance check-ins every 2 weeks through a show-and-tell (demo), things won’t go out of control.

Build an empowered multi-disciplinary team

The multidisciplinary team worked together in a room.

To enter the room, you had to be fully empowered by your bit of DWP or HMRC or LA to make decisions in the room. No one senior. It was surprising how easy it was for the organisation to identify who needed to be in the room.

Video of user testing convinced IDS to make a change to the benefits policy immediately.

Start multi-disciplinary; stay multidisciplinary. Don’t just remove these people once you’ve ‘launched’

Obtain a mix of mindsets: Pioneers, Settlers, Town Planners.

“User research is a team sport”

Continually assess your knowledge and your readiness

Each sprint, they asked themselves: What have we proved? Do we understand user needs better? Have we designed the service to scale massively? Do we know how to operate?

“If you can’t release software every day in an emergency you’ll never be secure, because a new threat will emerge and if you can’t respond like that clicks fingers, your organisation is inherently insecure”

Governance

“Governance was very simple: Ministers come to the show and tell, we’ll show you what we’ve made, we’ll show you what we’ve learnt, and what we’re going to do next, and we’ll talk about risks and issues if you want. But the real governance is seeing the thing being made and seeing the evidence and user research that it’s likely to have the intent that the minister wanted. Every week. And give credit to ministers, they turned up.”

“If your senior management doesn’t show up to show and tells, look them in the eye and tell them that they are failing at governance. Use that word.”

“Show the thing” – a thing you can use, not a thing you can see. If you’re sending screengrabs, you aren’t showing the thing, you’re showing pictures of the thing.

Psychology knows that behaviour is seldom rational. So we need to study behaviour.

Create a common design language with “nonviolent communication”

Marshall Rosenberg argued that there were 3 categories of non-violent communication:

1. 88 human needs
2. 91 positive feelings we wish to experience
3. 153 negative feelings we want to avoid

To be non-judgmental, clear and constructive in our use of language, use a “Rosenberg deck” of feelings cards as a conversation prompt.

Map behaviours with “rational emotive behaviourism”

Albert Ellis, the founder of CBT, argued that Activators trigger Behaviours, which lead to Consequences.

Map out a user journey. Use an Ellis Matrix. Identify the causes of user behaviours. Propose what new Consequences could be, and what new activators and behaviours could be.

Map motivations with “guiding self ideals”

A lot of what we do is a result of feelings of inferiority. (See the work of Alfred Adler.)
We seek a “fictional final goal” – if I do [BLANK] I’ll be finished and happy.

So ask ourselves: what would happen to us as an organisation if we never tried to solve this problem?
What would happen to the user if we never tried to solve this problem?

Then ask: What is an aspirational place for us to be? What if we did do this?
Can you clearly indicate what the result would be? – for us and for users

You need to map this out to get an overview of a potential new area of work.

Good leaders and designers empower the team

Lao Tzu quote on leadership, from Tao Te Ching:
“A leader is best when people barely know he exists, not so good when people obey and acclaim him and worse when they despise him. But of a good leader, when his work is done, his aim is fulfilled, they will all say we did it ourselves.”

1. Identify and challenge assumptions

   Think about what assumptions you’ve built into what you’re designing. What will happen if someone falls outside these?

   Facebook’s Year In Review – a feature designed to help people celebrate and share their great year – wasn’t designed with the experiences of people who’d not had a great year in mind.

   Inappropriate Year In Review images included:

   • a photo of the user’s apartment on fire
   • a photo of an urn containing the user’s father’s ashes
   • a sonogram of a pregnancy that later ended in miscarriage
   • a photo of a friend’s gravestone

   Facebook’s design team had a narrow vision, and so excluded all of these users. Meyer and Wachter-Boettcher challenge us to bring “edge cases” to the centre. “Instead of treating stress situations as fringe concerns, it’s time we move them to the center of our conversations – to start with our most vulnerable, distracted, and stressed-out users, and then work our way outward.” All users will benefit from this more focused, understandable and empathetic approach.

2. Make space for real people

   Give people “enough room within our interfaces to be themselves.” For example, gender is often presented as a binary choice between male and female, which doesn’t fit with our current understanding of gender. Facebook is an example of best practice here, allowing people to choose male, female or custom – which is a free text field with a list of common choices as prompts.

   Other examples of systems not giving people space to be themselves include systems that can’t handle names longer than a certain length (e.g. 15 characters), systems that don’t accept hyphens in names, or ones that don’t accept names that don’t pass culturally-specific test of validity. (e.g. Facebook rejecting Shane Creepingbear’s name as not real.)

   Organisations often make assumptions about what matters to users, or about who they are. The ‘Apple Health’ app didn’t include period tracking on launch, even though it boasted that it tracked ‘all of your metrics that you’re most interested in’. Its implicit focus was on men. And period tracking apps themselves often have a bias towards straight, sexually active, partnered people.

3. Incorporate stress cases

   A DIY and home appliance retailer was looking to improve its product guides. Originally these were written in a chirpy, positive tone, for happy, confident home-improvers. But sometimes users are more stressed when carrying out these tasks. The team found that there were two general categories of use: “urgent” and “upgrade”. They updated their style guide to write for the urgent case. This improved the guides for all users, as the clarity of information increased. Guides now feature installation availability and time-frames, estimated cost ranges, greater user of subheadings to allow for easy skimming, one-sentence summaries, reassuring tone.

   You can incorporate stress or crisis cases in usability testing. And you can test how a product performs in a cognitively-demanding environment by either testing in that environment, or by tiring people out mentally before the testing – e.g. by giving them some maths tasks to carry out.

4. Only ask necessary questions in forms

   Organisations are often pushy to obtain as much information as they can from every web form. Often this is done with a total disregard for the user’s experience. Caroline Jarrett has a protocol for evaluating each question you want to include:

   1. Who in the organisation will use the answer?
   2. What will the answer be used for?
   3. Is the answer required or optional?
   4. If the question is required, what happens if the user enters rubbish data just to get through the form

   This question protocol can help open up a discussion about the true business value of each question.

5. Learn from users

   Work to understand how your users see the world. This goes deeper than just testing top tasks on your website, or discussing product features.

   Steve Portigal recommends three types of question:

   1. Gather context and collect detail. e.g. asking about a sequence (Describe a typical workday) or specific examples (What was the last app you used?)
   2. Probe. e.g. ask for clarification of how a system works.
   3. Draw out contrasts. Useful for uncovering frameworks and mental models. e.g comparing processes or approaches.

   Open-ended research is about opening up questions and ideas, expanding your vision and the types of question you ask. This helps you move towards a design process centred around real people and their needs.

   Customer mapping can help you identify pain points, broken flows, and content gaps, through analysis of lenses, touchpoints, channels, actions, thoughts and feelings. Adaptive Path have produced a guide to customer experience/journey mapping.

6. Making the business case for accessibility

   Karl Groves, an accessibility consultant, argues that there are only three business cases for anything. Here’s how to argue for accessibility for each of these:

   1. It will make money. You can use accessibility to stand out from your competitors. e.g. Slack gaining users through ease-of-use. You can reach new audiences if more people are able to use your product.
   2. It will save money. You can cut customer service costs. The UK government found that as of 2011 it was receiving 150 million avoidable calls a year – calls for which an online service existed. This represented a possible annual saving of around £4 billion a year. Improving accessibility saves you money by increasing user retention – which is between 5 and 25 times more cheaper than acquiring new customers.
   3. It will decrease risk. Accessibility helps you avoid negative experiences and associated backlash – e.g. Facebook’s year in review generated a lot of negative press.

In August 2015, OFCOM announced that smartphones are now the most used internet device
“Mobile is not the future – mobile is happening now”

“Where are they now? Probably not in front of a computer…” – Tim O’Donnell, operations director, Precedent

Mobile isn’t just about access to the internet – it’s about location services, and functionality like sending text messages.

“It’s not your sector that sets the benchmark.”

Rising expectations are set by the best experiences. e.g. people pay for coffee or the tube using their phone, so they expect donating or signing up for a fundraising event to be similarly easy or they’ll just leave. (I think there’s also a potential for competitive advantage if you do a better job than other organisations in adopting.)

Mobile is different as a channel and a technology to desktop. People think about, and use it, differently.
Use is often more casual, time poor, and relaxed. Perhaps copy needs to be custom-written for mobile, to be shorter and snappier.

Thinking with your thumbs – catering for your users on the move – Chloe Bryan-Brown, web manager, Coram

Coram’s mobile strategy covers:

1. Design (have clear call-to-action boxes; make forms as easy as possible to complete)
2. Content (make your content easy to scan. Chunk it.)
3. User journeys

SafetyNets – innovative support for sex workers via smartphones – Matt Haworth, co-founder, Reason Digital and Hannah Shephard-Lewis, safety outreach digital lead, National Ugly Mugs

Reason Digital worked with National Ugly Mugs to create a decentralised, real-time, location-based information sharing system that sex workers can use to share ‘dodgy punter’ information with other app users nearby. This avoids an organisational bottleneck. People can contact the police if they choose, but they are in control.

A good example of the usefulness of user testing in generating insights that a design team would probably never think of: one user remarked that as the background was white, it would illuminate her face. This would mean that she wouldn’t use the app. So they knew that they had to change the background colour to black, so that use was more discreet.

The Big Pathwatch – crowdsourced data gathering – Daniel Brett-Schneider, head of engagement and Eleanor Bullimore, engagement manager, Ramblers

The Ramblers built an app to generate information on the accessibility and state of repair of the country’s path networks.

Lessons learned building the app:

1. Give yourself lots of testing time. The app was set to launch on Monday, but limited testing time meant that it wasn’t ready in the app store when they left the office on Friday night.
2. Check that registration is easy-to-use and that it doesn’t have problems
3. Location services were faulty

The user base was a challenge:

• Fewer than 50% of app users were already members of Ramblers. This wasn’t expected. (I wasn’t sure if they’d planned to convert this new audience into members.)
• Existing members of Ramblers weren’t keen on the app. Most didn’t have smartphones, and often weren’t sold on the idea. I wonder how early on in the project this was known.

Favourite jargon for the day: “onward supporter journey development”.

Attending via Periscope

This was my first periscope event. Here’s what I thought:

• At the start of the event, there were around 15 people. By the end, there were about 10 people.
• We lost connection for a few minutes a few times during the afternoon event. Everyone seemed to be re-invited, though, so most stayed in the broadcast.
• The picture quality was a little grainy, so text on slides had to be large to be legible. Audio quality was quite good.
• You could share messages with other people watching the periscope broadcast, but they disappeared after a second.
• There wasn’t really a back-channel, like you’d have on twitter. I was on twitter at the same time and there was more engagement there.
• Charity Comms did a good job resourcing the periscope feed, reinviting people when there were connection troubles, and scanning for questions to be asked during the in-person question-and-answer sessions. If you are thinking of hosting a periscope session, I’d recommend having someone dedicated to inviting people and handling the logistics.

The usefulness of combining Scrum with DSDM

The Agile Project Framework for Scrum “brings together the strength of DSDM at project level and the streamlined simplicity of Scrum at the delivery team level”
Scrum is focused on the product rather than the project, “so more emphasis is placed on incremental release of a product in the context of a product lifecycle than is placed on formally ending development work after an agreed period of time.”

Documentation

DSDM ‘Foundations’ phase covers the full breadth of the project, but deliberately avoids going into detail. This is very different to the ‘Analysis and Design’ step in a waterfall approach.
The Agile Manifesto values working software above comprehensive documentation.
The DSDM Agile white paper follows this: “break the illusion of security and stability that comes from document-driven, predictive processes. Specification of every detail of requirements, solution design, plans etc. in documents that get ‘signed off’ by stakeholders before work is allowed to progress is now widely accepted to be both wasteful, in terms of time and effort, and ineffective as the basis of governance and control. AgilePF for Scrum embraces the need for high-level versions of requirements, design and planning artefacts in the early phases of the project to frame development and delivery and to support governance.”

Collaboration over contract negotiation

“Typical commercial contracts assume that a traditional Waterfall process under-pins development and, accordingly, ‘a fixed price for a fixed specification’ is the standard for project contracts. Agile projects emphasise collaboration, and therefore contracts need to reflect this.”
Contracts should be “‘light touch’ and ‘guiding’ rather than being ‘detailed and prescriptive’.” “the Product Backlog may represent a contract, effectively defining the scope of a project. But it is cast at a high level and requires customer collaboration with less formality to flesh out the detail of requirements throughout the iterative development of the solution during the project lifecycle.”

Principles

1. Focus on the business need
2. Deliver on time
3. Collaborate
4. Never compromise quality
5. Build incrementally on firm foundations
6. Develop iteratively
7. Communicate continuously and clearly
8. Demonstrate control

Variables

Traditional approach:
Fixed: Features
Variable: Time, Cost, (in practice) Quality

AgilePF Approach:
Fixed: Time, Cost, Quality
Variable: Features

One of the four project variables must be flexible: “With proper planning, any three of the four project variables can be fixed provided one is allowed to vary.”
Agile PF fixed time, cost and quality and varies the scope of the features delivered.
To reach this point in the foundations phase, “an understanding of the high level features is required, sufficient to provide a sensible estimate for those aspects of the project that are fixed. At the same time, it is normal for a subset of the features to be identified as mandatory.”

Feasibility phase

Quickly established whether the project is likely to be technically feasible and cost effective from a business perspective before proceeding.
“The effort associated with Feasibility should be just enough to decide whether further investigation is justified, or whether the project should be stopped now, as it is unlikely to be viable.”

Plan for organisational/business change

“In the context of a project, the solution will include both the Product (often software) and any associated changes within the business wanting to exploit that product.”

Role of the Project Manager

“Managing an empowered team requires a facilitative style rather than a command and control style. It is usual that the Project Manager takes responsibility throughout the duration of the project. This must include both business and technical delivery aspects of the project, from establishing the foundations of the project through to the deployment of the solution.”

• Promote regular giving
• Amount shopping lists and user choice
• Ease of use
• Images of people
• Social proof, trust seals, where your money goes
• Emotional reinforcement
• Use donations as part of a multi-channel relationship

I’ll now elaborate a little on the processes of Macmillan, Cancer Research UK and charity:water

Macmillan’s single donation process

The Macmillan homepage focuses on people and their stories – each of its sections is lead by a photographed person sharing their story.

The donation ask comes either in the rotating carousel or through the site-wide donation button at the top right of each page. This provides a clear, route to making a donation for those who want to, but the site as a whole leads with support.

The donate page:

• Focuses on people and feelings.
• Directly addresses the visitor as essential, and asks how much they’d like to give. This appeal directly builds on the core Macmillan brand message “so no one has to face cancer alone”
• Empowers the user with choice – of donation type and amount.
• Shows, for each of the different donation amounts, how many people are giving the different options. This provides social proof.
• Prioritises regular giving over single donations, because regular giving is more useful to charities.

Two ideas for how this page might be made even more effective:

• Reduce the number of options at the top of the page – they may distract the user and reduce their propensity to give.
• Change some of the accompanying images so that they show people. The image that accompanies the £25 donation features a nurse and a patient or family member smiling at each other. I suspect that this is more compelling than the £50 level which shows a plate of pasta.

Here’s what you see if you choose to make a single donation page:

• From the user’s perspective, the entire process happens on the Macmillan website, and with the Macmillan brand.
• Emotional reinforcement through the form.
• Trust seals increase credibility
• Clear phone number to contact in case of difficulty
• People can see how far through the process they are.

After the 3d secure step, which I won’t show here, you’re taken to the thank you page:

The thank-you page:

• calls you out by name and praises you and your impact.
• Gives additional actions – sharing through different channels, or campaigning.
• Isn’t the end of the journey – the user also receives a thank-you email:

The thank-you email:

• Gives users a chance to find out more about their donation – which is working to build support or implicitly to upsell.
• Outlines Macmillan’s offer of support – building a reciprocal relationship.
• Begins to build a multi-channel digital relationship with the donor.

Cancer Research UK’s single donation process

The Cancer Research UK homepage:

• Use lots of social proof.
• Features a clear box for people who have already decided to donate before coming to the site.
• Incorporates the site-wide donate button.

The Cancer Research UK donate page:

• Uses human pictures
• Is focused and calm, despite including a range of options. Clearly the different options have been prioritised.
• Prioritises regular giving.

The single donation form:

• Subtle attempted upsell to regular donation
• Empowers visitor with choice about where their money goes – but a simple choice.
• Gift aid option explained visually.
• Visitor empowered to choose the easiest payment options that suits them. I chose to pay with paypal.

After taking payment, I then had the opportunity to choose my communications preferences:

I was then taken to the thank-you page:

As with Macmillan, the email thank-you message was used to drive multi-channel digital engagement:

The email shows the difference that your donation makes, but also has secondary actions: taking an exciting quiz on improving lifestyle; an opportunity to get support or to connect; and an invitation to deepen engagement with research.

charity:water’s single donation process

The donate page:

• Uses powerful visuals, communicating the impact of donations.
• Reassures you about where your money goes.
• Is clean and focused.
• Makes paying incredibly easy

Marvel at the ease of use of the Stripe popup. Why aren’t more charities following the lead of ecommerce and using Stripe?

The payment form only asks for the bare minimum it needs to take your payment

The thank-you page:

• is focused, and shows the impact of your donation.
• upsells to a bigger supporting action – using the principle of consistency.

The email thank-you leads with an image, and articulates the impact of the cause.

I’d be interested to see this exercise carried out on mobile. How well do the top charities compare?

What is AI?

AI is not about machines ‘thinking’ like humans. AI is the theory and development of computer systems able to perform tasks that would usually require human intelligence.
e.g. cognitive (planning, reasoning, learning) and perceptive (recognising speech, understanding text, recognising faces)

“As soon as it works no one calls it AI any more”

We expect AI agents to:

• operate autonomously
• perceive their environment
• persist over time
• adapt to change

Drivers of change in AI

• Moore Law – microprocessors are 4 million times more powerful than they were in 1971.
• Big data – low cost sensors, social media, mobiles, the internet gives us more data; combined with better techniques for working with this data.
• The internet and cloud computing
• Improved algorithms

Cognition

Representing knowledge in a computer, using it to reason and plan automatically.

1. Rules-based systems: Rules base, inference engine (to apply rules), working memory (contains all the information it has to assess). Best for situations with a small number of variables.
2. Taxonomy: Helps to organise data into a hierarchy.
3. Bayesian networks (Bayes nets): Useful for situations in which your confidence about a belief may change as your knowledge changes. They can represent assertions, and degrees of certainty. Can help with diagnosis, reasoning from symptom to cause, or for prediction. Less good when you have lots of variables, or when you want to recalculate the entire network.

Some algorithms used in machine learning:

• Neutral networks – Good for pattern recognition. e.g. speech recognition. (segment audio signal onto phonemes, then associate phonemes with words in the dictionary; named entity recognition.)
• Support vector machines – good for classification and regression. Often used for off-the-shelf supervised learning. Straightforward to train and implement, and allow a lot of variables. Helpful for Feature engineering.
• Ensemble learning – using a collection of different models, and combining the output to obtain a stronger result. IMB Watson used this when playing jeopardy. Better than just using any one method.

Planning

Automatically devising a plan of action to achieve goals given a description of the initial state, the desired goal, and the possible actions.
e.g. getting from Times Square to the Bronx Zoo.
Search through possible actions to find a sequence that achieves the goal.
Challenge: managing complexity and computation time: combinatorial explosion.
Replanning is important too, to deal with developing situations.
Applications include: Google navigation, unmanned vehicles, robotics.

Learning

Improving performance automatically. Machine learning is the process whereby machines improve their performance without explicit programming. Machines discover patterns, make predictions, and become better over time with exposure to data. This helps in situations where we can’t anticipate all situations, or when we don’t know how to program the solution (e.g. facial recognition)

Types of machine learning:

1. Supervised learning – learning by example.
   An agent is given pairs of information – input (or a number of inputs) and output.
   This allows the agent to understand how to produce the desired output, even for unknown inputs.
   It’s called supervised learning because we use labelled data to train the model.

   Main tasks: Classification (output is one of a set of discreet values) or Regression (output is a number)

   Applications: Sales forecasting, image recognition, text classification, health.

   Challenges: Acquiring and labelling training data; can be expensive to create data set.

2. Unsupervised learning – discovering patterns in data even though no specific examples are provided.
   e.g. clustering – given a large set of similar items, discover ways to group them into subsets

   Challenges: algorithm has to determine which attributes should be used to group items; sometimes it’s hard to decide where to place an item.

   Applications: Customer segmentation; Social network analysis; Defining product baskets; Topic analysis; Anomaly detection – e.g. looking for outliers in manufacturing.

3. Semi-supervised learning – unsupervised learning with human interaction to fine-tune
   e.g. giving feedback on the number of clusters, or suggesting attributes for matching.

4. Reinforcement learning – learning by trial and error.
   Agent acts in unknown environment, responding to sensory input. Responses shaped using rewards or punishment.
   Agents take into account actions and sequences of actions when associating them with rewards or punishments.
   Works best with closed-loop problems – i.e. ones in which there are no inputs other than those caused by the action of the agent

   Challenges: time consuming with many actions or chains of actions; requires a lot of computing power; trial and error has a cost – e.g. learning how to trade on the stock market, so use it when the costs of trial and error are low.

   Applications: physical control systems e.g. elevators or helicopters, or recovering from damage by learning new ways of walking; in some domains it’s our only option.

Perception

The ability to take in information in a human-like way: through speech, text or vision.

1. Natural language processing (NLP) – software that processes human language.
   e.g. understanding or producing. Break down doc to sentences, then words, which are understood using grammar rules

   Challenges: context is tricky: e.g. “he saw her duck”

   Applications of NLP: summarising documents, translation, extracting info, question answering, writing stories, analysing customer feedback. Medicine and Law

2. Speech recognition – recognising words, tone and emotion of human speech
   Steps: break wave form into phonemes, then match these to words, then put these into an appropriate sequence.

   Challenge: accents, background noise, homophones, need to work quickly. (I wonder how we could add contextual information to understand the set of phonemes)

   Applications: hands-free writing e.g. medical dictation, controlling devices, computer system control, surveillance,

   Future: mine broadcasts and recordings of human speech.

3. Computer vision – the ability to identify objects, scenes and activities in images. e.g. face recognition.
   Has to build up from pixels to coloured areas, and then objects.
   Machine learning can be used to train object recognisers. error rate 2010-14 reduced four-fold

   Applications: Handwriting, medical imaging, autonomous driving, surveillance, gesture detection. One useful current application is recognition of where spare spaces are in a car park.

   Future: recognition in video, and events detection. This is hard because of the complexity: connecting recognition over time

Physical interaction

Types of robot:

1. Manipulators – physically anchored to their workplace
2. Mobile robots – e.g. drones
3. Mobile manipulators – e.g. humanoid robots in films

Elements of robotic systems:

• Mechanical and electrical engineering
• Machine learning
• Computer vision
• Planning
• Speech recognition
• Sensors – e.g. range finders, location sensors, proprioceptive sensors (knowledge of own position), force and torque sensors
• Effectors

Applications of robotics:

• Manufacturing
• Agriculture
• Healthcare
• Hazardous environments
• Personal services
• Entertainment
• Human augmentation

Uncertainty is a challenge for robotics – e.g. needing to take action based on incomplete information, or dealing with an unexpected environment.

Business applications for cognitive technologies

1. Product
2. Process
3. Insight

Product

Embed cognitive technologies in a product or service to help the end user.
e.g Netflix film predictions, which drive 75% of Netflix usage; Google Now / Siri; predictive text.

How cognitive technologies can improve products:

1. Convenience
2. Simplicity
3. Confidence
4. Emotion

Questions to help you decide whether to embed cognitive technologies in your product/service:

• Would people like to use it hands-free?
• Is your product too complex?
• Do customers have to make complicated choices to buy your product
• Would a natural interface help customers bonds with your product?

Process

Embed technology into an organisation’s workflow, to increase speed, efficiency, quality.

Automate internal processes, e.g.:

• The Hong Kong subway system’s preventative maintenance programme. Scheduled by algorithm.
• Georgia’s campaign finance commission. Uses handwriting recognition to handle the volume of work.
• Cincinatti Children’s Hospital. Uses NLP to read freeform clinical notes to find patients who might be eligible for clinical trials. Reduced nurse workload on this area of work by 92%.

Automate expert decisions.
Relieved skilled workers of unskilled tasks.
Automate unskilled work.

Insight

Improve decision making by analysing large amounts of data – including unstructured data – to discern patterns or make predictions.
e.g working out someone’s risk of developing metabolic syndrome, and which medical interventions were most likely to improve patient health.

Benefits: better, faster decisions that can improve operating and strategic performance

How to find opportunities: See where you have large or unstructured datasets that haven’t been fully analysed; look for processes where the value of improved performance is high.

How to decide whether and where to incorporate cognitive technologies in your organisation – use the “Three Vs” framework

1. Viable – e.g. perceptual tasks (involving vision, speech, handwriting, data entry, first tier customer service), analytical classification and predictive (forecasting, document review and summarizing), decision-making tasks (situations where knowledge can be expressed as rules, data-driven decisions), planning and optimisation tasks (e.g. scheduling)
2. Valuable – where it’s worth applying. Involve business processes with costly labour, where expertise is scarce, where there is a high value in improving performance, or where you can deliver features or experiences that your customers care about.
3. Vital – may be required if: industry standard levels of performance demand their use: online product recommendation, spam filtering, fraud detection; scalability – e.g. processing handwritten or printed data, analysing large amounts of social media.

The impact of cognitive technologies on work

There’s a debate – will machines take our jobs, or will they increase productivity and growth – and demand for human skills? Tasks requring adaptability, common sense, human interaction, ambiguity and creativity will be beyond the reach of machines for a long time. AI is most likely to replace highly-structured back-office roles that don’t involve many customer interactions.

Risks of automated systems:

• Not infallible. They may eliminate operational human error, but that doesn’t mean that they’re always right.
• Humans can lose skills if they don’t practice them
• Humans are bad at monitoring information that remains constant for long periods of time, which may lead to errors being undetected.
• Poorly automated systems can undermine worker motivation

Approaches to automation:

1. Replace – completely replace a human performing a job with a machine
2. Atomize and automate – break jobs into narrow tasks, and automate as many of these as possible. Humans are still employed, but in more of an oversight/remedial capacity.
3. Relieve – automate tasks that are dull, dirty or dangerous.
4. Empower or augment – make workers more effective through technology, e.g. by automating brand-new processes.

Strategic choice for approaching automation:

• Cost strategy – use technology to cut costs by reducing the workforce, or through reducing errors and rework.
• Value strategy – use technology to make workers more effective, or reassign workers to higher-value work.

Skills that will probably be desirable in the future:

• The ability to work with cognitive technologies
• Hyper-specialisation of skills or knowledge that are unlikely to be automated by computers
• Empathy, creativity, emotional intelligence

Personalisation is adapting a digital experience to a user’s attributes and presumed interests or needs.

Personalisation isn’t necessarily a good thing – it can come across as good or creepy

A reminder that you should start by understanding your goal, and then work out how to deliver it. Personalisation won’t necessarily be the tool/approach you’re looking for, so don’t ever pursue it for its own sake.

Good personalisation:

• Marie Curie created a mobile-specific text-to-donate ask because they noticed that mobile users rarely converted on their normal responsive donate page.
• L.K. Bennett found that UK users were unusually keen on finding out the returns policy, so they highlighted this for UK users

Creepy personalisation:

• Target predicting when a person becomes pregnant, by looking at their browsing behaviour, and marketing to them on this basis.
• The well-intentioned Samaritans Radar ran into trouble because of concerns about privacy, stigma and consent.

There’s sometimes a question about whether to talk about the fact that you’re doing personalisation. Would explaining it make it seem creepy? e.g. saying “We think we know how many children you have.” Would not explaining it be underhand?

Where can you get the data you need for personalisation?

• 3rd party
• From the user themselves, voluntarily (personal data)
• User behaviour
• Context

Trust is important

When thinking about organisations possessing and using their data, people are concerned about:

• usage creep
• lack of personal benefit
• loss of data

So to some extent trust is derived from user experience. That’s a interesting and unsettling conclusion, but one which explains why we seem so relaxed about giving corporations loads of personal data in exchange for easy-to-use free tools.

Giving people the option to access/update/change their data and communications preferences is a better user experience and also good reputation management, as some users will be concerned about this.

Start small

It’s better to begin experimenting with personalisation on a small scale:

• Easier to get the budget
• Easier to get buy-in and sign-off
• Easier to manage
• Easier to track and measure the results – and therefore to prove success
• Easier to manage data on a smaller scale
• Easier to manage project risks, and to deal with unanticipated ones that emerge

Some things you can do to get started with personalisation

There’s a lot you can do without carrying out heavy engineering on your website:

• Video personalisation
• Email personalisation – and A/B testing to verify the effectiveness of this approach
• Social personalisation. I’d didn’t get much on this in the presentation and would like to explore this a bit more.

A useful morning session – I expect to attend the next Together We’re Better event.

The core points can be summarised quite easily, but I’d recommend reading the full text to enjoy the examples and case studies. My edition was published by Matador, 2011 (2013 reprint).

What are the three values groups, and what distinguishes them?

Which group an individual inhabits is determined by their unmet needs:

• Everyone starts life as a Settler.
• Once their Settler needs are met, they can become a Prospector.
• Once their Prospector needs are met, they can become a Pioneer.
• It is possible to go back from being a Prospector to a Settler, but generally going from a Prospector to a Pioneer is one-way as it is based on self-esteem.

Why values groups are important

“you can get people in any values group to do something if they are allowed or enabled to do it in a way that meets their needs.”

“The kiss of death for communications across values groups is to try and impose the values or ‘reasons’ of one group on another… Despite what some campaigners may believe there is no universal ‘right reason’.”

“Many ‘conventional’ campaigns have centred on universalist-ethical (eg, save-the-planet) and sometimes rationalistic (eg, save-money) propositions. They have lacked fun, fashion, emotion, visible success – in short, the values to appeal to Prospectors.”

Settlers

• Socially conservative
• Belonging
• Tradition
• Focused on the past
• Like predictability
• Preferred charity brands: family, life-saving, simple/small-scale actions e.g. Guide Dogs for the Blind or RNLI. Clear, personal, bounded actions – not social change or global issues
• Discipline
• Following the rules
• Being in control
• Standing up for your family and community
• (National) Security
• Being normal
• Doing your duty
• Being loyal to those who have helped you and your family/friends
• Making sure the basics get done – e.g. food, health, shelter
• Saving lives

Prospectors

• Being successful and being seen to be.
• Material wealth
• Shopping
• Don’t like being told not to do things or to give things up
• Focused on the future
• Avoid political controversy
• Opportunities rather than connections
• Looking better or best
• Soundbites of stories – can be global or local
• Having fun
• Being a winner
• Celebrity

“To engage Prospectors it has to be done on their terms: a better wind turbine than your neighbour, or a donation to Oxfam which will get you talked about in the gym, and preferably make you more attractive to others.”

“Prospectors… tend to be more selective and demanding than those from other Values Worlds. You generally need to create experiences, or desirable things, not just offer ideas or information, and whatever you create has to compete with what social events, brands, commerce and media entertainment has to offer.”

“more fun and outgoing, optimistic and expressive than the Settler World, and much more relationship- and possession-centred than Pioneer World with its emphasis on ideas. Anyone wanting to target Prospectors … must focus on these core Attributes.”

Pioneers

• Love questions, new ideas and the unknown
• Personal ethical responsibility
• Thoughtful about ethics
• Focused on the present
• Benevolence, global justice, openness
• Preferred charity brands: environmentalism, overseas aid.
• Like authenticity
• Like forming connections and networks

New behaviours generally start with Pioneers, then are commodified and consumed as fashionable by Prospectors, then become normal and adopted by Settlers.

“The persistence of individual action because it’s the ‘right’ thing to do, whether or not there seems any prospect of success, can be a characteristic of Pioneers when they act for ethical reasons (this is, internally referenced ideas of right and wrong), and of Settlers when they act for moral reasons (ideas of right and wrong derived from an external, respected Authority).”

What percentage of the UK population is in each of these groups?

• Settler – 31%
• Prospector – 28%
• Pioneer – 41%

UK, 2008. I don’t know what evidence was used to generate these percentages.

How to get different values groups to take action

Communicate in a way that appeals to each group’s unmet needs:

• Enhance the safety, security or belonging of Settlers
• Enhance the esteem of others, or self-esteem, felt by Prospectors
• Engage Pioneers by communicating in ways that involve ideas, innovation, self-direction and ethics.

What actions are appropriate for different values groups?

You need to make sure that your action is something that the values group you are targeting is comfortable with:

Settler actions should be: simple, discrete, achievable and dependable, definitely uncontroversial and ideally officially-sanctioned, familiar, normal, risk-reducing, control-, belonging-, safety- and security-boosting.
Settler actions should not be: innovative, uncertain, controversial, framed as part of a ‘bigger picture’.

Prospector actions should be: visible, immediate, proven, achievable, fun, displayable, fashionable, uncontroversial, socially recognised, celebrity-endorsed if possible, commodified, esteem-boosting.
Prospector actions should not be: innovative, old fashioned or traditional, about following or breaking rules, renunciation (unless it is to gain more), doing things for other people without social reward, framed as ethically- or idea/theory- motivated, controversial or related to an open-ended problem.

Pioneer actions should be: interesting, novel, ethical, complex, change-focused, about ideas not just ‘things’, authentic, an opportunity to connect with new people and ideas, concerned with beauty, nature or justice, about the bigger picture, framed in a way that allows pioneers to make up their own minds and ask questions.
Pioneer actions should not be: doing things because of other people’s – or authorities’ – beliefs or requirements, restricted choices, based on justification by tradition, or justification through a need to be bigger and better.

How to think about communications

Never think of the population as a homogenous mass.

Try to avoid the term ‘message’. Think instead of using the COMPACT list:

• Channel
• Audience
• Messenger
• Programme (why you are doing what you are doing: intention and objective)
• Context (what’s around it in time and space)
• Action (the action you’re asking them to take)
• Trigger (motivation of intended audience)

Just Eat base their social strategy on engagement. Reach and audience size arise out of this. Anthony advises against doing things the other way round: focusing on growing follower numbers can be self-defeating as you under-invest in engagement and lose followers in the long-run.

So how do you ensure engagement? Having a clear tone of voice is crucial, so that you avoid being plain and uninteresting.

Brands invest heavily in a tone of voice, and then lose it on social because they’re playing too safe #nfptweetup @meanwritehook

— Kate Brennan-Rhodes (@kate_brennan) September 17, 2015

Questions to ask to discover who your brand is on social media from @meanwritehook #nfptweetup pic.twitter.com/xfz2OXPgg9

— Adrian Salmon (@adriansalmon) September 17, 2015

Don’t ask ‘What content is right?’ Instead, ask ‘How does my brand behave?’

Imagine your brand as a person, and ask:

• What makes them excited, frustrated?
• How does this person speak – funny authoritative?
• How does this person react to subjects that matter to your brand?
• How does this person handle a bad situation?

Decide what behaviour you want to be known for, and encourage this behaviour in your supporters too.

Bringing marketing and customer/supporter care teams closer together allows you to harness the strengths of both. They’ll be responsive and on-brand, and you’ll all increase your understanding of your audiences.

Alex Goldstein – Senior Social Media Manager at TMW

Alex found that the corporate sector has more visible silos, whereas the often under-resourced charity sector tends to require individuals to take on a wider range of tasks. This can be empowering, and help charity sector workers get things done faster, but it can mean that they don’t always have the required expert support. And having to hold all those disciplines in your mind at once can cause confusion.

#nfptweetup fascinating insight from @mokuska – don’t forget the silos in your brain – in charities we fit several job roles into one

— Mandy Johnson (she/her) (@MsMandyJ) September 17, 2015

Broadly speaking, Alex has found that the corporate sector has more money, and is more courageous with risk. The charity sector has better stories and passion.

Where is the crossover? #nfptweetup pic.twitter.com/oVSs1qKahn

— Rachel Beer (@rachelbeer) September 17, 2015

Endangered Emoji – Adrian Cockle, Digital Innovation Manager at WWF International

The campaign was based on the insight that many of the species included in the emoji range are endangered. So why not highlight #NFPtweetup

— NFPtweetup (@NFPtweetup) September 17, 2015

WWF wouldn’t say how much money the campaign raised, but the primary objective was awareness not fundraising.

I do love this campaign. Tweeting emojis is fun and people did it lots. But we’re all waiting to hear if it raised 💰, right? #NFPtweetup

— Joe Freeman (@JosephFreeman) September 17, 2015

It’s interesting that ‘raising brand awareness’ is still a key aim for someone so recognised and loved as WWF #NFPtweetup

— Joe Freeman (@JosephFreeman) September 17, 2015

Guess that’s the whole ‘reaching younger audience’ thing and ensuring that recognition is maintained #NFPtweetup

— Joe Freeman (@JosephFreeman) September 17, 2015

Adrian summarised some observations on what makes for effective social sharing material:
To be effectively shareable on social media, a campaign/action should follow the NUDES approach:

• Networked. e.g. social nomination mechanic.
• Unexpected.
• Dumb. Be easy to understand.
• Exhibited. Involve a shareable behavior. Make it aspirational.
• Stories. Enable or include stories?

This innovation was something new and untested. Charities tend to want to minimise risk by following the successful actions of other organisations.
But that aversion to risk holds charities back from innovating.

#charities need to stop asking “who else has done it?” & just GO for it to avoid missed opportunities #nfptweetup

— Social Misfits Media (@MisfitsMedia) September 17, 2015

WWF encountered problems while innovating…

Measuring emoji use was a challenge, as no-none of the monitoring systems were capable of this at the time #nfptweetup

— Martin Lugton (@martinlugton) September 17, 2015

… but were able to overcome these because of pro bono support. Had the additional unanticipated costs of innovation not been borne by a third party, how different would this case study have been?

General planning thinks about one future. Futures thinking, in contrast, thinks about multiple possible futures. (The field of futures/foresight used to be about concrete things like predicting future demand, but there’s so much uncertainty that this was a waste of time.)

Thinking about different futures helps us remember that different possibilities exist, and that they are man-made.

We should see the future as an asset – an input into the world we wish for. It’s an input into questioning and changing today. Ivana Milovević said that “The future is not an empty space but, like the past, an active aspect of the present”.

Strategic Foresight is the creation of alternative and preferred futures, with an understanding of the worldviews and metaphors that underlie them, and the working out of what we need to do differently to manifest them. By thinking about different possible futures, we can deliberately work to bring about the one that we want to see. “The grass is greener where you water it.”

The point about worldviews and metaphors is crucial – if we ignore this, then the work is superficial and incremental. In most of our questions there are assumptions about reality, culture, and the right way to do things. If you’re doing foresight, you’re questioning these things. So we need to question the cultural basis of our questions, seeing them not as universal, but as problematic as well. “Our questions are actually congealed knowledge. Thus questioning also has to be questioned.”

Metaphors and frames shape people’s responses. If you think of crime as a ‘beast’ you will be inclined to put money into policing. If you think of it as a ‘virus’ you will be inclined to put money into education and ending poverty. Librarians are moving how they see themselves from ‘keepers of the collections’ to ‘innovators of the gardens’.

So we need to ask: how is the world changing? And do I have the appropriate paradigm or worldview to handle this? What do I need to change about how I see the world to have more choices, and see more possibilities? You need to transform your culture and your metaphors, otherwise you won’t be able to see or realise these desired futures. So you need to find a new story that will help create a new future.

The way you think about the world is a risk. It could be too narrow – so you need to involve different perspectives in thinking about possible futures. Ask: who’s not in the room?

The general approach to Futures is:

• Challenge the Used Future (old practices that persist even though they don’t help us now)
• Scan the environment for emerging issues and weak signals
• Don’t wait until the bad times
• Create alternative futures
• Find the macro-pattern
• Make the vision real
• Link story to strategy
• Focus on change agents
• Make sure everyone is in the room. (When you’re running a workshop creating a vision, you need funders in the room. You need experts with data, citizens with ideas, administrators who can carry things forward, and funding agencies who can provide money. It’s crucial to include people who object to change – put them in charge of the old paradigm that you’re disrupting, as you make things safe in the new paradigm for them.)

The course recommends two different methods for doing this – a shorter method and the longer 6 pillars method if you have more time.

Short method

Outline the broad question that you’re interested in, e.g. ‘the futures of nursing in 2030’.

1. Where from? What is the history of the issue? E.g. write down 3 events that led to the present.
2. Where to? What are your forecasts and fears relating to the issue, if current trends continue? E.g. by year X, Y will be true.
3. Assumptions. What are the critical assumptions you have made about the issue when creating your forecasts?
4. Alternative futures. Develop 3 or 4 alternative futures based on different assumptions/trends.
5. Want. What is your preferred future? Pick characteristics of your desired future. E.g. certain things that you’d like to be true.
6. How. Create the preferred future through backcast: Imagine that we have already got to the desired future. How did we get to the preferred future?
7. Inner story. Transform the old story to the new metaphor. What is a supportive metaphor for your preferred future?

6 Pillars Method

1 Mapping

Tool one: Futures Triangle

Label the sides of your triangle: Pulls, Pushes, Weights.

Pulls: What’s your vision of the future? What does it look like? What are its characteristics? Where do you want the organisation to be in 30 years? Where do you want to be in 30 years?

A good vision enables, ennobles, is rational, emotive, neither too near nor too far.

Drivers / Pushes: What are the trends pushing things into this future? E.g. new technologies, social trends, demographic, policy, legal or funding changes.

This course mentions 2 sources of insight into future trends: Futures Platform and Shaping Tomorrow. There will also be figures in your industry who are doing this type of thinking – e..g Azeem Azhar and Ben Thompson in technology.

Weights: What are the barriers holding back change? Who benefits from the status quote – or loses if it changes? What are the deep structures that resist change? E.g. organisational structures, policies, regulations, procedures, knowledge structures, ways of thinking, historical narratives, existing investments in infrastructure, technology, education, social contacts, demographics. Barriers might also include a paradigm fixation – a focus on what has worked in the past, or what the organisation was established to do.

Once you’ve produced a triangle, explore its implications for your strategy.

Perhaps you want to find a way to lighten the weights, or to ride an inbound wave of pushes. Or you may find that no-one really understands where you’re trying to get to, which suggests that you need to create a compelling vision. You may identify some messages that you want to communicate.

You may end up with competing images of the future.

You can use the triangle to outline desires or preferences, or for more pragmatic thinking.

An example futures triangle on China’s futures (N.B. much of this article seems suspect, and likely orientalist):

Tool two: Futures Landscape

Ask people where they are spending most of their time:

• The jungle (day-to-day survival, competition, scarcity)
• The chess set (strategic thinking dominates)
• The mountaintop (broader strategic futures and the ‘big picture’)
• The emerging future and vision (the transformative, purposeful, directive vision for the future)

Then reflect on whether you might want to change this and expand your thinking.

2 Anticipating

Ask: What might disturb things?

Emerging Issues Analysis (Graham Molitor): Spot weak signals that could be dramatic if they come to pass.

Futures wheel: take an emerging issue. As a group:

• Brainstorm first and second order consequences
• Then think about the wider impact that each of these would have on industry and/or society.

Then think about how we might enhance or suppress each effect. (Disagreements are good – capture divergent understandings in separate bubbles.)

3 Timing

Ask: are there deeper patterns to the future you’re interested in?

There are 4 common patterns: pendulum (e.g. moving between centralisation and decentralisation over time), linear (things getting better over time), cyclical (rise and fall), spiral (going into the future, but with some of the past going with you).

Knowing your context will help you think through how you want to try and change things.

Aside: The Sarkar Game is a group educational roleplaying game that helps people think about group dynamics and power relations over time, through the interplay between Worker, Warrior, Intellectual and Capitalist. Here’s an article with several interesting examples of it being played (see also this overlapping writeup).

4 Deepening

Causal Layered Analysis is a great way to explore what’s blocking the success of a desired outcome – the blocker is usually at a deep level, so needs unpacking and understanding before it can be resolved.

Causal Layered Analysis suggests that there are 4 levels of reality:

1. Litany: The data we see every day. Headlines. The surface problem. The official public description of the issue.
2. Causes / System. Systemic – social/economic/political/technical/environmental. Incentives and structures that bring about the surface litany. How might we incentivize the outcomes we want to see?
3. Worldview / culture/ ways of knowing. The big picture, the paradigm that informs what we think is or is not real. Sits beneath the systemic stuff. The cognitive lenses that we use to understand or shape the world. What’s the worldview that we want people to have? What are alternative discourses?
4. Deep metaphor / the unconscious story. Myths and metaphors – you generally cannot take a rational approach here. Is our metaphor supporting our strategy? Do we need a new metaphor?

We need to work at all 4 levels. And we need to unpack how all stakeholders see the future.

Once you’ve worked out the myths, metaphors and worldview, you need to map that to systems and incentives/measurements.

You can do Causal Layered Analysis for a current situation, or for a desired future situation.

5. Creating alternatives

Create scenarios to think about risks, find new possibilities, clarify hidden assumptions, understand different people’s perspectives, or to help resolve a conflict by getting people to imagine different futures.

Some standard scenarios to think through: best case, worst case, what happens if you don’t change (BAU), outlier.

There are 5 methods for coming up with scenarios:

Once you have your scenarios, apply Causal Layered Analysis to each of them.

(Some words of warning/advice: If you don’t think in terms of CLA, then people won’t buy in to things – people can’t buy into the story, and they can’t see themselves in that future. If you don’t follow the previous steps in the Pillars (Mapping, Emerging Issues Analysis, etc), then you just get safe scenarios that are rooted in the present.)

Ask: What’s a day in the life like in this world? This makes it real for people. What does your day look like?

This creates space for pillar 6, renegotiating with stakeholders: what is the ultimate preferred or desired future?

6 Transformation

This step entails visioning and then backcasting.

3 methods for visioning:

1. Analytic: write down the Causal Layered Analysis for the preferred future. How is reality measured, what is the prevailing system, the preferred worldview or culture, and the underlying narrative?
2. Interview: get quick-fire responses to get at peoples’ real thoughts and feelings. In each different future, what does your daily life look like? (Work, technology, family, nature, challenges.) What do you wish the future to be like? What is your desired future?
3. Creative visualisation (by Elise Boulding and others). “Step along a path, one step per year into the future. Arrive at a hedge separating the present from the future. Travel through a door in the hedge to your preferred future. See it. Then see an ancient tree. Walk into it, and walk up the spiral stairs to the sixth floor where you meet your future self (or a mentor, teacher or loved one). This person gives you life advice, and an artefact to help on the journey. Then open your eyes and write down aspects of the preferred future.” Audio version of the visualisation narrative

Once you have visions and preferred futures, you can negotiate between them.

Then you can back-cast:

• Start at the future point of the goal having been achieved.
• Ask: what needed to be true to get there.

This helps avoid people feeling downbeat about how far away the future state is. It can also be useful for negatively-inclined people: instead of just saying “but we won’t get funding”, invite them to work out what will positively need to be true to get funding. Then work out how to make those things true.

• Facebook and twitter are seen as the most important sharing networks.
• Google+ is reluctantly included. Given recent announcements about Google deprioritising Google+, it seems likely to be even less of a focus in future designs.
• Most publishers aren’t using whatsapp sharing yet. It seems to me like a good opportunity, and Buzzfeed and The Atlantic are leading the way. Some info on setting up whatsapp sharing. Note that Whatsapp desktop isn’t currently widely used.

• Few publishers are using share by text. Again, Buzzfeed and The Atlantic lead the way. With mobile such an important part of the web traffic mix, we should plan for mobile-specific sharing options.
• If comments are permitted, sometimes the option to comment, and the number of comments, are shown alongside social sharing options. Other times these are separate.
• Few publishers are using social proof – showing the numbers of people sharing to encourage more people to share – perhaps because this would increase the page load time.

Some recommendations for best practice

• Know what the most important actions are.
  The best buttons for your site depend on your users’ needs, and your objectives. In most cases, I’d start with: Facebook share, Twitter share, whatsapp share, phone share, pinterest share, email share. If you’re interested in information provision, printing / PDF download options will be important too. You should do user research to find out how your audience would like to store your content before committing to a design that doesn’t help people. e.g. if you’re designing a site for an older audience, a pocket sharing option is probably less useful than a print option.
• Prioritise your top actions by making them larger.
  The Daily Mail has a Facebook share as its largest button, and iMore’s Facebook and twitter buttons are the largest by far.

The Washington Posts’s social sharing options are quite cluttered. Perhaps some of these (e.g. tumblr and Google+) could be deprioritised or removed entirely.

• Design with different sharing contexts in mind – e.g. mobile as well as desktop.
  Try out whatsapp and text sharing for mobile users only. Most people will be printing from a Desktop.
• Think about how best to present your content when someone shares it socially.
  e.g. making sure that you have appealing sharing copy and images for pinterest. Customise your social sharing metadata for each page, so that you don’t rely on the social network trying to work out some copy for itself. There’s a great Moz post on configuring your social sharing markup.
• Social sharing isn’t just about blocks at the top or bottom of an article. Calls-to-action weaved in to the page are very important. E.g. encouraging people to share a blockquote through twitter.
• Think about how to encourage people to share.
  People are used to sharing content from publishers, but what about other content that might be valuable to share, but that people aren’t as used to sharing? Just Giving’s case study of increasing social shares of donations is a useful case study here.

I’ve surveyed the websites of the top 10 UK charity brands, as identified by the 2014 Charity Brand Index. From a performance perspective, the ideal is to have a site that loads quickly, and for the amount of downloading required to be as small as possible. So you want a low ‘load time’ and a low ‘page weight’.

Here’s how the sites compare:

• BBC Children in Need and Comic Relief had the fastest home pages.
• If you’re commissioning a new website, or want to improve the performance of your existing site, aim to beat the average performance of these sites. Aim for pages that:
  • load in 3.5 seconds or less
  • have a page weight of 1.9 MB or less
• If your pages load in 2 seconds or under, and if your page weight is 1MB or less, you’re doing very well.

This data was obtained by running 4 unthrottled desktop speed tests using GTMetrix between 16 August and 22 August 2015 and averaging the results. An interesting follow-up would be to look at mobile performance: set a mobile browser and throttle the connection to a common mobile download speed.

Female infanticide kills at least 2 million girls per year.

“Investment in girls’ education may well be the highest return investment available in the developing world.”

Major General Patrick Cammaert (former UN force commander) on use of rape in war: “It has probably become more dangerous to be a woman than a soldier in an armed conflict.”

Numbers and statistics are much less compelling than stories in motivating people to act.
e.g. it’s more effective to ask for money to help one named girl than it is to ask for money to help 21 million people. And even mentioning the context alongside the named girl makes people less likely to give.
e.g. in another study, people gave twice as much to save one child from cancer than to save eight.

We need to be empirical in our approach, and not defined by conservative/liberal ideology. The best AIDs prevention strategy in one study was neither abstinence-only education nor condom distribution, but education on the dangers of sugar daddies.

The Oportunidades program of financial incentives for education/public health outcomes achieves strong results, increasing school attendance by 10% for boys and 20% for girls. Children grow 1cm taller per year than those in the control group. The scheme encourages poor families to invest in their children, helping to break down the generational transmission of poverty.

Kiva is a microfinance organisation, allowing donors/financiers to loan to organisations vetted by local on-the-ground microfinance organisations.

Male-controlled family budgets in the poorest families in the world spend about ten times more on alcohol, prostitutes, sweets, drinks and feasting than on their children’s education.
Putting money into women’s hands improves children’s experience, with studies in Ivory coast, South Africa and Indonesia showing an increased spend on nutrition, medicine and housing.

After a 1993 Indian stipulation that 1/3 of village chiefs had to be women, bribery was reduced and water infrastructure improved, but satisfaction in the leadership fell. However, once a village had had a female leader, this bias against women chiefs disappeared.

The authors tackle the question of whether cultures can change, and the issue of cultural imperialism:
“We sometimes hear people voice doubts about opposition to sex trafficking, genital cutting, or honor killings because of their supposed inevitability. What can our good intentions achieve against thousands of years of tradition?
“One response is China. A century ago, China was arguably the worst place in the world to be born female. Foot-binding, child marriage, concubinage, and female infanticide were embedded in traditional Chinese culture. Rural Chinese girls in the early twentieth century sometimes didn’t even get real names, just the equivalent of ‘No. 2 sister’ or ‘No. 4 sister.’ Or, perhaps even less dignified, girls might be named Laidi or Yindi or Zhaodi, all variations of ‘Bring a younger brother’. Girls were rarely educated, often sold, and vast numbers ended up in the brothels of Shanghai.
“So was it cultural imperialism for Westerners to criticize foot-biding and female infanticide? Perhaps. But it was also the right thing to do. If we believe firmly in certain values, such as the equality of all human beings regardless of color or gender, then we should not be afraid to stand up for them; it would be feckless to defer to slavery, torture, foot-binding, honor killings, or genital cutting just because we believe in respecting other faiths or cultures. One lesson of China is that we need not accept that discrimination is an intractable element of any society.”

Cultural change has to be driven locally. It cannot be imposed: e.g. 1970s and 1980 efforts against FGM, or efforts to empower Afghan women.
The exception is public health measures that depend on research, materials and knowledge that don’t exist at grassroots.

“the sex slave trade in the twenty-first century… is bigger than the transatlantic slave trade was in the nineteenth.”

Happiness levels seem to be largely innate, and not greatly affected by external forces. But feeling connected to something larger can help us feel better.

Some name-checked organisations I wanted to investigate further: Camfed, Plan, Women for Women international, Tostan.

Evidence for the impact of page load speed on business performance

• Walmart: reducing page load time by 1 second causes an up to 2% increase in conversions. (Study slides, analysis)
• Yahoo: reducing page load time by 0.4 seconds increased traffic by 9%. (Reference)
• Google: increasing page load time from 0.4 seconds to 0.9 seconds decreased traffic and ad revenues by 20%. (Reference)
• Amazon: every 0.1 second increase in load time decreases sales by 1%. (Reference)
• A 1 second delay in page response can result in a 7% reduction in conversions. (Reference)
• Radware: “one-second HTML delay resulted in… 3.5% decrease in conversion rate…” (Reference 1, reference 2)
• Shopzilla: reducing average load time from 6 to 1.2 seconds increased revenue by up to 12%. (Reference)
• Mozilla: reducing average load time by 2.2 seconds increased conversions by 15.4%. (Reference)
• AutoAnything.com: 50% reduction in page load times increased sales by 13%. (Reference)

See Pedal to the Metal: Speed up your load times for more conversions from Tammy Everts

• Obama campaign: reducing load time by 60% increased donation conversions by 14%.

How to test and improve page speed

To see how fast your pages load, go to gtmetrix.com or webpagetest.org and plug in your URL (or a competitor’s ;)). GTMetrix uses insights from Yahoo and Google’s separate page speed testing tools, and draws from their banks of improvement recommendations.

These free services will provide recommendations for how a developer could improve your site’s performance. Things like compressing images, making effective use of browser caching of static assets, and optimising the critical rendering path.

To learn more about the practical measures you can take to improve page load speed, this free online course on page load speed is useful.

The steps of the Human-Centered Design process

“Human-centered design is all about building a deep empathy with the people you’re design for; generating tons of ideas; building a bunch of prototypes; sharing what you’ve made with the people you’re designing for; and eventually putting your innovative new solution out in the world.”

Human-centered design is based on creating opportunities for “high-impact solutions to bubble up from below rather than being imposed from the top.”

You can use this approach for:

• Products
• Services
• Spaces
• Systems

Start by choosing a design challenge. Collect your thoughts; review what you already know; define what you don’t know; review constraints or barriers.

You move through three spaces (not always sequentially):

• Inspiration – exploring the design challenge. “Too abstract and the brief risks leaving the project team wandering; too narrow a set of constraints almost guarantees that the outcome will be incremental and, likely, mediocre.” “learn directly from the people you’re designing for as you immerse yourself in their lives and come to deeply understand their needs and aspirations.” “focus groups and surveys, rarely yield important insights. In most cases, these techniques simply ask people what they want..” But people aren’t good at expressing their needs or imagining possibilities. As Henry Ford said – people would have wanted a faster horse. (Case Study: Clean Team toilets in Ghana. When interviewed, people said they’d prefer to dispose of their own waste if it could save them money, and were reluctant to allow service people into their homes. However, when they prototyped: people quickly realised the value of someone else handling waste disposal. A good reminder that self-report is highly flawed. Similarly, Ideo: Rockefeller Foundation health project in Bangkok: a Burmese immigrant self-reported as having no network, but actually it became clear in interview that she had a strong network.) Research methods: Learn from people, learn from experts, immerse yourself in context, seek analogous inspiration After each research item: Regroup, pull out sound-bites, interesting or surprising stories, interesting interactions and remaining questions. Don’t try to interpret yet – that comes later.
• Ideation – Exploring your design opportunities. Generating, developing and testing ideas. Distil insights from research. Brainstorm to generate ideas – and withhold judgement. Linus Pauling: “To have a good idea you must first have lots of ideas” And you need multidisciplinary teams with multidisciplinary people to make interesting connections.
• Implementation – Making your concept real and a sustainable success. “the path that leads from the project stage into people’s lives”. Prototyping is at the core of this – turning ideas into products and services that you test, iterate and refine. Prototyping – “cheap, quick, and dirty” helps de-risk the process. Have a main idea for the prototype to convey, and test that idea.

Mindsets of a Human-Centered Designer

• Learn from Failure: “Don’t think of it as failure, think of it as designing experiments through which you’re going to learn.” Through prototyping, you’re de-risking by making something simple first, and checking it early in the process.
• Creative Confidence “the notion that you have big ideas, and that you have the ability to act on them.”
• Empathy – you can’t come up with any new ideas if you don’t go beyond your own life.
• Embrace Ambiguity – Let multiple ideas exist simultaneously. You don’t know upfront what’s going to work out.
• Be Optimistic
• Iterate, Iterate, Iterate “we gain validation along the way… because we’re hearing from the people we’re actually designing for.”

Featuring “The power of supporter-focused video to drive fundraising” – Francis Mason and Rebecca Highfield, Anthony Nolan; “Delivering for public health and fundraising? How Breakthrough Breast Cancer did it with TLC” – Matthew Jupe, Breakthrough Breast Cancer; “MyStory: Lessons learnt from delighting users through video content” – Jamie Parkins, JustGiving; and a panel discussion.

Anthony Nolan – supporter-created video and a brand storytelling video

Great examples of a supporter-created video and a slick brand video:

It’s worth reflecting on what distinguishes these two videos, and what unites them.
The formal quality of the supporter-created is lower, but it’s clearly good enough.
The second video is slicker – better film quality, better sound, transitions, and more clearly on-brand.

But the main feature of both videos is strong storytelling. The supporter-created video has an off-the-cuff charm, but it’s also clearly been carefully put together, by a funny and sweet supporter.

Much of the evening was focused on the second type of video – material produced by people working for charities – but I’m very interested in how we can help supporters to create their own content.

Touch Look Check – Direct Marketing on TV and trains

A promotional campaign asking people to request a guide for the signs to spot breast cancer, with a follow-up direct debit donation ask.

Oh, sorry, it was about £8 per lead from train panels and £3-3.50 from DRTV #NFPtweetup
— Rachel Beer (@rachelbeer) May 21, 2015

Intriguing that television was the dominant channel in this TLC campaign and that PPC and display weren’t used so much.

TV might sound like an expensive option but it actually is cheaper per lead than train advertising #nfptweetup
— Lesley Pinder (@Skipinder) May 21, 2015

MyStory

A JustGiving project that generated a personalised thank-you video for each London Marathon runner using their platform, and sent it out the day after the event.

I think we’re gonna start seeing a big trend in personalised video. A great way to thank supporters #nfptweetup
— Nathan Murray 👓 (@MrNathanMurray) May 21, 2015

45 seconds – optimum personalised video length. Viewer decides whether to watch within 10 secs so needs front-loading #NFPtweetup
— Karen JK Hart (@KarenJKHart) May 21, 2015

The traffic to MyStory was overwhelmingly from the Facebook mobile app.
90% of MyStory referral traffic was from Facebook mobile. Only 1% from Facebook desktop. 5% from Twitter mobile.

Sounds like we need to forget all social media but Facebook on mobile. #nfptweetup pic.twitter.com/jY0wubTXl3
— Matt Collins (@charitychap) May 21, 2015

Interesting to contrast the automated, perhaps slightly sterile, MyStory video generator with the rough-and-ready supporter stories. I guess the personalisation gives these videos credibility.

Panel discussion

Here’s a great example of a video that is designed for this short, silent, autoplaying context, by my colleague Eleanor Bowes:
— Martin Lugton (@martinlugton) May 21, 2015

One good point that came up was that it’s not enough to just create some high-quality video: you need to plan how to promote it. And, of course, you need to start with a clear purpose for any video.

A good video is not enough. You need to plan how to promote it. #NFPtweetup
— Martin Lugton (@martinlugton) May 21, 2015

The Importance of Research and Partnership in Tackling Digital Exclusion – Sarah Bridges, Go ON UK

This session was a useful reminder of the importance of accessibility and inclusion. This session focused on the needs of older users, but made it clear that unequal access to digital is not just about a user’s age: nearly half of people who lack basic digital skills are of working age:

.@SarahBridges “It’s not just older people that lack #BasicDigitalSkills -over 40% are of working age” #CampDigital pic.twitter.com/sWw1nwMMHe — Go ON UK (@Go_ON_UK) April 23, 2015

Sarah shared some useful information on the specific needs of older users:

Challenges for older users: memory, aversion to novelty, vision + hearing difficulties, slower movement, dexterity #campdigital — Martin Lugton (@martinlugton) April 23, 2015

Older people are twice as likely to abandon a task online, and tend to blame themselves, not the design #campdigital — Martin Lugton (@martinlugton) April 23, 2015

The move away from skeuomorphic design towards flat design might have a negative impact on older people, who have a more mechanistic mental model of digital.

“Older ppl tend to have mechanical/physical mental model of how digital works” (Flat design = fail? Affordance, metaphors, etc) #CampDigital — Barry Briggs 🐝🏴 (@quiffboy) April 23, 2015

Lean UX and making sure that everyone in your team is on the same page – Imran Younis, Laterooms.com

My highlight of the day. An excellent talk, brimming with examples.

Ask “why?” five times to understand the problem properly before you start thinking about the solution. Otherwise you risk rushing to implement the wrong solution to a poorly-understood problem.

We need to focus our thinking around the problem/desired outcome because this is ultimately the only thing that matters. People want solutions not tools:

People don’t want to buy a quarter inch drill, they want a quarter inch hole! Brilliant @ImranYounis #CampDigital — Danielle Lyon (@danielleclyon) April 23, 2015

The lean startup workflow for ongoing improvement is simple and easy to communicate:

#leanstartup #CampDigital pic.twitter.com/S24KxHG5nR — Dr Aleksej HEINZE 🇪🇺 (@AleksejHeinze) April 23, 2015

The lean startup model stresses that you need to keep learning after you’ve launched.
Constantly validate with users. You’ll never be perfect, but keep perfecting. (I like the optimistic yet humble humanism that seems to exist at the heart of this method.)

Close cross-team working helps retain/share knowledge and gives shared ownership of solutions:

Keeping the research, ux, design & dev team together helps create sense of ownership of a solution. Measure & learn as a team. #CampDigital — Barry Briggs 🐝🏴 (@quiffboy) April 23, 2015

Play back your results to your organisation. Show the value of what you’ve done after each sprint.

Understanding the context of use is really important:

Imran used an example of the HSE, who wanted tradespeople to be more aware of the dangers of working with asbestos. They researched the context of use before designing their asbestos-education solution.

Users didn’t actually find an alarming health app to be helpful or relevant to them.

But they did find the app helpful to communicate with their customers, and to charge them for the asbestos-related work.

Understanding this motivation helped Imran design the app so as to appeal to both sets of objectives.

That moment when you realise a practical benefit is a stronger driver for app adoption than fear of a health risk: @ImranYounis #campdigital — Piers French (@PiersFrench) April 23, 2015

Prototyping content to drive app uptake: making money more persuasive than not getting asbestos (!) #CampDigital — Dafydd Singleton (@dafsingleton) April 23, 2015

Similarly, a classic case study shows that you sell more milkshakes not by focusing on improving the milkshakes themselves, but by understanding why people buy milkshakes. (See Clay Christensen’s Milkshake Marketing)

Not ‘What’s your favourite flavour?’ but ‘Why do you always buy a milkshake on way to work?’Uncovered the insight @ImranYounis #CampDigital — Jane Neale (@Jane2N) April 23, 2015

Evidence from user testing is hard to argue with:

#CampDigital — Martin Lugton (@martinlugton) April 23, 2015

Jargon interlude

The day’s key jargon takeaway was definitely ‘solutionise’. I think it’s analogous to ‘design’, but with the implication of ‘designing/developing/manifesting a solution’.

#CampDigital — Martin Lugton (@martinlugton) April 23, 2015

Garnering positive engagement from stakeholders who don’t understand UX – Fritz von Runte, NICE

A collection of tips for getting non-UX people usefully engaged with a UX process:

• Always have a clear shared definition of success before you start
• Simplify stakeholder requirements in their presence.
• Ask why 5 times.
• Turn opinions into questions.
• Make it clear that business goals are not the same thing as user needs.
• Bring stakeholders along to user testing, but don’t let them moderate it, as that’s a skilled task.
• When stakeholders share a blog post, or journal article, or similar, be sure to follow that source in future. You’ll get an insight into their context.
• Be experts on every metric that your colleagues use. They help you understand what is important to them, and how they frame their decisions.
• Present UX learnings in a concise format. Use bullet points, and prioritise the list.
• After doing UX research, define actions (not necessarily solutions) so that you’re proactively leading the decision-making process.
• “Learnings from tests shouldn’t define new solutions, they define new problems.” (Can we do better than this? Expressing the value of testing like this might make it hard to get buy-in from internal stakeholders. Can we articulate the value of testing while also explaining that it won’t tell you what to do – it will help you understand the problem space better. Perhaps “Learning from tests won’t define solutions. It will hopefully help you better understand the problem, and hopefully will make you aware of further problems.”)
• Do visual design last. This helps prevent stakeholders from focusing on little details of visual design, and keeps the focus on UX.
• Even when you’re working in a team, you’re still the designer – it’s not design by committee.
• Define success before you test.

Why design matters? How a design-led process delivers better digital services – Ben Holliday, Department for Work and Pensions (DWP)

Some evidence in favour of user testing: Jared Spool found a correlation between number of hours spent with users and the quality of the design outcome. (Of course, correlation is not causation, but the kind of team that does good design is the kind of team that takes user testing seriously.)

Very cool article by @jmspool mentioned by @BenHolliday #CampDigital http://t.co/dPcPKPqBAt #CampDigital — Rick Threlfall (@rick_threlfall) April 23, 2015

Ben reiterated Imran’s earlier point that design must be a process, not a one-off event.

Turn research into tangible insights, setting boundaries for design challenges.

Ben’s core message was that we need to achieve shared clarity of purpose:

Develop a shared understanding of why we’re doing what we’re doing. Get these hypotheses written down and shared #CampDigital — Martin Lugton (@martinlugton) April 23, 2015

Example format for recording assumptions/understanding:

We believe: X, we will know we’re right/wrong when we see: qual evidence Y and quant evidence X #CampDigital — Martin Lugton (@martinlugton) April 23, 2015

How’s the Mobile Apocalypse treating you?

A brief interlude because the Mobile Apocalypse has just hit. (Google has begun penalising non-mobile optimised sites in search returns.) A few websites have been caught out – I wonder how quickly each of the below organisations will be revealing a responsive redesign? (I wasn’t in this presentation, but wanted to share the slide anyway.)

#CampDigital — Martin Lugton (@martinlugton) April 23, 2015

There’s gold in them there Hills! Creating ‘Hills’ to frame your releases around user-centric market outcomes, not feature requests – Daryl Walker-Smith & Richard Halford, IBM

I liked the idea of a ‘hill’ as a place you want to get to – a bit like a military goal.

You don’t know how exactly you’ll take the hill, but you know the purpose behind your work.
Once again, understanding purpose is at the heart of our work.

I thought this session struggled a bit under the weight of IBM’s history, and the terminology used. Apparently people usually have a one-week induction into this method, which makes it harder to unpack this methodology and get stakeholders using it.

IBM uses sponsor users, not personas. Personas aren’t real users, of course, they’re just fabrications. Sponsor users are real users. But it’s difficult (and presumably expensive) to have users visit you multiple times a month for testing.

Good observation:

An occurring theme here at #CampDigital is war strategies and the man on the moon chuckle — Jamica (@jamicaortizbr) April 23, 2015

How does this belligerent (masculine?) language reflect or shape our discipline? What other metaphors could we employ to describe our processes?

Digital disruption and the challenges faced by organisations with an focus on in-person, ‘real world’ transactions – Bea Karol Burks, Citizens Advice and Matt Lindop, Premier Inn

The idea of “government as a platform” – putting service delivery before policy – is disruptive to service delivery organisations like CAB, who exist to provide services because government isn’t sufficiently service-focused.

Echoing the core message of the conference, Bea advised us to “focus on problems not products.” Start with understanding user needs. For an organisation like CAB, user needs include ‘changing the world’.

CAB have an alpha blog showing what they’ve done and learnt. It’s open to the public, and to comments. Critical feedback is really valuable.

Great quote from Bea, against commonplace sloppy use of the term ‘agile’: “Agile is a discipline; it doesn’t mean not being disciplined.”

Bea has an interesting vision for CAB’s future:

• Use data on people’s problems to understand the performance of policies/services before the govt does.
• Use data to connect and enable local campaigners to fight battles that CAB cannot.

Splitting the Atom – Nick Wiles & Stewart Bromley, Atom Bank

A talk about the Atom bank – a soon-to-launch bank that will be entirely online. This apparently makes it “a digital pure play.” (Second key point of jargon for the day.)

Interesting contrast to the earlier talk:

Comparing this talk about being online-only vs the digital inclusivity talk this morning, Atom seem happy being non-inclusive? #CampDigital — Chris Northwood (@cnorthwood) April 23, 2015

The talk started to develop some interesting thoughts on information architecture and personalisation…

#CampDigital — Martin Lugton (@martinlugton) April 23, 2015

… but mainly this session felt like a product launch for a product that doesn’t yet exist.

This was a strong conference with good inspiration and practical points. I’ll be sharing them with my team.

Here’s what I’ll be doing next:

• Learn more about the metrics that my colleagues use. (Information, Fundraising, Campaigns)
• Read the milkshake marketing case study.
• When developing new products, or trying to understand users and products, focus more on understanding the context of use.
• Don’t ask “How can we make our events pages more attractive?” Ask “When and why do people sign up for the marathon?” “What problem are they solving by signing up for the marathon?” “Where and why do people access our information products?” “What problems do they want to solve?”
• Once I’ve got some development momentum, I’m going to resume playing back the results of sprints to the organisation, but focusing more on what we’ve learnt.
• I’m going to think about how to incorporate the lean UX focus on continual learning into ongoing development sprints.
• Introduce my team to the lean startup model and talk about how we might use this to help us with continual improvement from a content / user journey perspective.
• Start using the word ‘solutionise’ and see if I get away with it.
• Think about whether we could consciously map out our hypotheses and assumptions with different teams.
• There was lots of buzz about The Ethical Designer – Cennydd Bowles, Twitter, so I’m going to watch the recording when it goes live in a couple of weeks.

The division of work between digital and IT is often unclear

In most organisations, the division of responsibility between IT and digital is:

• IT support internal infrastructure.
• Digital support externally-facing initiatives. eg social media, web and CMS, SEO and PPC, online giving.

But often the scope of digital has not been clearly defined.

As digital has grown beyond marketing and communications and started to manage emerging projects (eg hosting, mobile, other digital services (eg Elefriends), database systems, and (for some) service transformation and data-driven initiatives), this can cause confusion of roles.

Digital teams should remember that IT teams are useful to them

IT teams often hold useful resources – eg developers and expertise with procurement (and processes? – eg agile).

How to make the case for digital in your organisation: think operationally

You need to make the case for change, and you need to explain what digital can do for the organisation.

Take up the discussion at senior level.
Focus on user needs.

(Don’t focus on faults with underlying legacy systems or structures, and don’t just take up the discussion with IT)

If a lack of formal structure or governance is an obstacle to change, set up that structure.

Digital is often thought of as just concerned with fundraising and campaigns. But by thinking operationally you can go beyond income generation and move into efficiency gains and potentially cost savings. eg use of data can refine service delivery; digital can join up disconnected or outdated processes and services.

So identify what issues are stopping your organisation from operating effectively, and identify where digital can play a role in joining up processes and reducing reliance on manual activities.

Look beyond your organisation for inspiration, and articulate your results in the operational language of the rest of the organisation. Think about proving your claims, and about how you will provide evidence of success.

Appropriate revolutionary structures and methods

Who is a radical?

“What distinguishes revolutionary leaders from the dominant elite is not only their objectives, but their procedures.” (148)

“trusting the people is the indispensable precondition for revolutionary change.” (42)

“The radical… does not consider himself or herself the proprietor of history or of all people, or the liberator of the oppressed; but he her she does commit himself or herself, within history, to fight at their side.” (21)

“The man or woman who proclaims devotion to the cause of liberation yet is unable to enter communion with the people, whom he or she continues to regard as totally ignorant, is grievously self-deceived.” (43) (What happened in early c20th Russia, with ‘going to the people’?)

“it is necessary to trust in the oppressed and in their ability to reason. Whoever lacks this trust will fail to initiate (or will abandon) dialogue, reflection, and communication, and will fall into using slogans, communiqués, monologues, and instructions.” (48)

“constant, humble, and courageous witness emerging from cooperation in a shared effort – the liberation of women and men – avoids the danger of antidialogical control.” (157)

“The essential elements of witness which do not vary historically include: consistency between words and actions; boldness which urges the witnesses to confront existence as a permanent risk; radicalization (not sectarianism) leading both the witnesses and the ones receiving that witness to increasing action; courage to love (which, far from being accommodation to an unjust world, is rather the transformation of that world in behalf of the increasing liberation of humankind); and faith in the people,…” (157)

“Instead of following predetermined plans, leaders and people, mutually identified, together create the guidelines of their action.” (162)

“if at a given historical moment the basic aspiration of the people goes no further than a demand for salary increases, the leaders can commit one of two errors. They can limit their action to stimulating this one demand or they can overrule this popular aspiration and substitute something more far-reaching – but something which has not yet come to the forefront of the people’s attention. In the first case, the revolutionary leaders follow a line of adaptation to the people’s demands. In the second case, by disrespecting the aspirations of the people, they fall into cultural invasion. the solution lies in synthesis: the leaders must on the one hand identify with the people’s demand for higher salaries, while on the other they must pose the meaning of that very demand as a problem. By doing this, the leaders pose a as a problem a real, concrete, historical situation of which the salary demand is one dimension. It will thereby become clear that salary demands alone cannot comprise a definitive solution.” (163-4)

Emancipation cannot be imposed

“the great humanistic and historical task of the oppressed: to liberate themselves and their oppressors as well.” (26)

“Freedom is acquired by conquest, not by gift.” (29)

“The correct method for a revolutionary leadership to employ in the task of liberation is, therefore, not ‘libertarian propaganda.’ … The correct method lies in dialogue. The conviction of the oppressed that they must fight for their liberation is not a gift bestowed by the revolutionary leadership, but the result of their own conscientização.” (49)

“The struggle begins with men’s recognition that they have been destroyed. Propaganda, management, manipulation – all arms of domination cannot be the instruments of their rehumanization. The only effective instrument is a humanizing pedagogy in which the revolutionary leadership establishes a permanent relationship of dialogue with the oppressed.” (50)

“The revolutionary’s role is to liberate, and be liberated, with the people – not to win them over.” (76)

“Revolutionary leaders cannot think without the people, nor for the people, but only with the people.” (112)

On charity and deference

“Any attempt to ‘soften’ the power of the oppressor in deference to the weakness of the oppressed almost always manifests itself in the form of false generosity; indeed, the attempt never goes beyond this. In order to have the continued opportunity to express their ‘generosity,’ the oppressors must perpetuate injustice as well. An unjust social order is the permanent fount of this ‘generosity,’ which is nourished by death, despair, and poverty. That is why the dispensers of false generosity become desperate at the slightest threat to its source.”(26)

“True generosity consists precisely in fighting to destroy the causes which nourish false charity. False charity constrains the fearful and subdued, the ‘rejects of life,’ to extend their trembling hands. True generosity lies in striving so that these hands – whether of individuals or entire peoples – need to be extended less and less in supplication, so that more and more they become human hands which work and, working, transform the world.” (27)

I think this is a useful observation for the UK Labour Party

“In a situation of manipulation, the Left is almost always tempted by a ‘quick return to power,’ forgets the necessity of joining with the oppressed to forge an organization, and strays into an impossible ‘dialogue’ with the dominant elites. It ends up being manipulated by these elites, and not infrequently itself falls into an elitist game, which it calls ‘realism’.” (130)

Upskilling everyone equally might not be as useful as significantly upskilling some key individuals.
Build a community of product owners, train them and empower them through management of their own backlog and budget.

Digital transformation sees digital outgrow its regular home in communications and move into:

• Service delivery
• Business transformation
• Making sure that the organization meets user needs.
• Foundational business processes and infrastructure.

Don’t be proscriptive with change. Instead, invite teams to let you know where you might be able to help them achieve their goals more effectively with digital tools.

If you start by optimising a small number of key user journeys, this can give you a clear way in.
Follow this thread towards transformation. The necessary changes emerge organically, and it’s easier than getting buy-in upfront.

The point of education and human action is “the individual’s ontological and historical vocation to be more fully human.” (37)

Two models of education

The banking model of education is about depositing information into passive students

“an act of depositing, in which the students are the depositories and the teacher is the depositor. Instead of communicating, the teacher issues communiqués and makes deposits which the students patiently receive, memorize, and repeat. This is the “banking” concept of education,…” (53)

The banking model requires students to adapt to the world, and encourages servility

“the banking concept of education regards men as adaptable, manageable beings.” (54)

“The more completely they accept the passive role imposed on them, the more they tend simply to adapt to the world as it is…” (54)

“Implicit in the banking concept is the assumption of a dichotomy between human beings and the world: a person is merely in the world, not with the world or with others; the individual is spectator, not re-creator.” (56)

Libertarian education

Education is not about integrating people into an oppressive society, but about understanding and transforming the world

“Authentic liberation – the process of humanization – is not another deposit to be made in men. Liberation is a praxis: the action and reflection of men and women upon their world in order to transform it.” (60)

“Problem-posing education affirms men and women as beings in the process of becoming – as unfinished, uncompleted beings in and with a likewise unfinished reality.”(65)

“Whereas banking education anesthetizes and inhibits creative power, problem-posting education involves a constant unveiling of reality.” (62)

“Education as the practice of freedom – as opposed to education as the practice of domination – denies that man is abstract, isolated, independent, and unattached to the world; it also denies that the world exists as a reality apart from people. Authentic reflection considers neither abstract man nor the world without people, but people in their relations with the world.” (62)

What does libertarian education look like in practice?

“Through dialogue, the teacher is no longer merely the-one-who-teaches, but one who is himself taught in dialogue with the students, who in turn while being taught also teach. They become jointly responsible for a process in which all grow… Here, no one teaches another, nor is anyone self-taught. People teach each other, mediated by the world, by the cognizable objects…” (61)

How to create a libertarian program of education

“The starting point for organizing the program content of education or political action must be the present, existential, concrete situation, reflecting the aspirations of the people.” (76)

“education… cannot present its own program but must search for this program dialogically with the people,” (105)

“the investigation of thematics involves the investigation of the people’s thinking – thinking which occurs only in and among people together seeking out reality… Even if people’s thinking is superstitious or naive, it is only as they rethink their assumptions in action that they can change. Producing and acting upon their own ideas – not consuming those of others – must constitute that process.” (89)

“the team of educators is ready to represent to the people their own thematics, in a systematized and amplified form. The thematics which have come from the people return to them – not as contents to be deposited, but as problems to be solved.” (104)

“after several days of dialogue with the culture circle participants, the educators can ask the participants directly: ‘What other themes or subjects could we discuss besides these?’ As each person replies, the answer is noted down and is immediately proposed to the group as a problem.” (104-5)

The costs and risks of security are increasing

• 90% of large businesses had a security breach last year.
• Security breaches are becoming more expensive: £1.15 million in 2014; £3.14 million in 2015
• 1,000,000 new pieces of malware are created each day.
• 14x increase in infections year on year.
• When Target was hacked, its CEO and CIO resigned. So directors’ jobs are more vulnerable to security risk.

Basic framework for cybersecurity

• Assess
• Detect
• Protect
• Respond

The future of cybersecurity? Inc. better use of data, more collaboration, security as a designed-in service #sf2015 pic.twitter.com/kCkllUTUdP

— Martin Lugton (@martinlugton) November 12, 2015

Reactive approaches to security aren’t good enough. We need a proactive approach, with a broader vision

Talk Talk only noticed they were hacked because their website slowed down. So they probably aren’t an exemplar of best practice.

Threat Intelligence allows us to:

• identify and resolve internal and external threats.
• understand how attackers think and behave, and select appropriate countermeasures
• focus our security resources in the most important areas

We need a broader approach to cybersecurity:

• Holistic view of systems needed, not a silo approach (e.g. via end-to-end cloud systems, rather than having server security separate to firewall)
• of external threat environment (e.g. via a threat exchange)

Analytical detection is more sophisticated than signature detection

A signature approach is the most basic; rules are a bit more sophisticated, correlations are better still, but an analytical approach is strongest for identifying security issues.

Increasing levels of abstraction for identifying attacks #sf2015 pic.twitter.com/C4ZTbpG5E3

— Martin Lugton (@martinlugton) November 12, 2015

Footprinting

Your internal understanding of your network estate is different to how external attackers will see it. So it’s useful to carry out external footprinting.

Hackers can use less well-defended elements of an organisation’s web presence to successively break in to other areas.

e.g. lots of the 2011 Sony hacks were on smaller, national sites. The hackers had a better understanding of Sony’s footprint and vulnerabilities than Sony did.

The commoditization of exploits

Charl van der Walt’s excellent presentation was my highlight of the day.

Zerodium recently paid a bounty for jailbreaking iOS9:

Our iOS 9.1 and 9.2b #jailbreak is one of the most sophisticated exploits ever developed! #Zerodium pic.twitter.com/0NBRnq4I4Z

— Zerodium (@Zerodium) November 2, 2015

Zerodium can now sell knowledge of this exploit to clients.

Because Apple doesn’t know what the exploit is, these clients can reliably use this exploit to attack people.

Government security services want reliable exploits like these, which leads to this commoditization of the exploit market. It also leads to the industrial use of these purchased exploits as they are employed at scale. This is in contrast to the more patient, research- and expertise-led footprinting carried out by other attackers, which presumably requires much more resource for a given output.

I found the day an interesting overview of the cybersecurity space. The biggest single improvement I’d recommend for next year is including female speakers – the keynotes were entirely presented by men.

People use their smartphones a lot

• 79% of smartphone owners check their device within 15 minutes of waking up every day.
• People use their smartphones around 150 times a day, according to industry professionals

What is a habit?

• A habit is an automatic behavior triggered by situational cues.
• Habits require little or no conscious thought.
• A product has high habit-forming potential if it is seen as useful and is used frequently. Fundamentally your product must solve users’ problems.

The four stages of the Hook Model

1. Trigger

A trigger instructs the user to take an action. Triggers can be external or internal.

Habit-forming products start by using external triggers like email, app icons or notifications, but over repeated hook cycles users form associations with internal triggers. Internal triggers are based on existing behaviours or emotions.

• External – tells the user what to do next by placing information in the user’s environment. e.g paid (unsustainable, earned (media and PR), relationship (peer recommendations), owned (“Owned triggers consume a piece of real estate in the user’s environment. They consistently show up in daily life and it is ultimately up to the user to opt in to allowing these triggers to appear.” e.g. an app icon, email newsletter). “Without owned triggers and users’ tacit permission to enter their attentional space, it is difficult to cue users frequently enough to change their behavior.”
• Internal – associations in the user’s memory tell them what to do next. Achieved by tightly coupling a product with a thought, emotion or existing routine. e.g. “A need is triggered in Yin’s mind every time a moment is worth holding on to, and for her, the immediate solution is Instagram. Yin no longer requires an external stimulus to prompt her to use the app – the internal trigger happens on its own.” So you need to understand a user’s internal triggers – the pains they seek to solve. Focus on these emotions rather than product features. (“Only an accurate understanding of our user’s underlying needs can inform the product requirements.”) And don’t just ask people what they want “talking to users to reveal these wants will likely prove ineffective because they themselves don’t know which emotions motivate them… You’ll often find that people’s declared preferences – what they say they want – are far different from their revealed preferences – what they actually do.” Ask “why” 5 times to arrive at an emotion.

2. Action

Activity undertaken by a user in anticipation of a reward. To increase the likelihood of an action being taken:

1. Make it easy
2. Maximise the motivation

Core motivations:

• Seek pleasure and avoid pain
• Seek hope and avoid fear
• Seek social acceptance, avoid rejection

The 6 elements of simplicity:

• Time
• Money
• Physical effort
• Brain cycles
• Social acceptance/deviance
• Whether an action is routine/disruptive of routine

To make action more likely, simplify it with regard to the user’s scarcest resource at that moment.

Utilise heuristics to encourage people to take action:

• the scarcity effect: items that appear scarce are valued more highly
• the framing effect: people assess information in context – e.g. ignoring Joshua Bell when he performed in the subway, or enjoying identical wine more if told it cost $90 rather than $5.
• the anchoring effect: people often fixate on one piece of information when making a decision, e.g. buying something because it’s on sale even though another item is actually better value.
• the endowed progress effect: people want to continue with progress towards a goal. So make them seem like they are already making good progress – e.g. giving loyalty cards starting part-way through rather than at 0%. In a study, both groups had to purchase 8 further car washes to gain a free one, but one group started with 2/10 completion rather than 0/8. This group had an 82% higher completion rate. Linkedin Profile strength uses this heuristic too.

Respecting people’s autonomy makes them more likely to take the action you want. Telling people “But you are free to accept or refuse” makes them more likely to comply. So when you make a request, affirm their right to choose. Leverage “familiar behaviors users want to do, instead of have to do.” “Companies that successfully change behaviors present users with an implicit choice between their old way of doing things and a new, more convenient way to fulfil existing needs.”

3. Variable Reward

Predictable rewards don’t create desire. But variable rewards are compelling. This isn’t because of the sensation from the reward itself, but the need to alleviate the craving for the reward.

e.g. looking through social media and scanning through to find material that might be relevant to you. “The exciting juxtaposition of relevant and irrelevant, tantalizing and plain, beautiful and common, sets her brain’s dopamine system aflutter with the promise of reward.”

Types of variable reward:

1. The Tribe: feeling “accepted, attractive, important and included”. e.g. people liking your Facebook post or upvoting your Stack Overflow answer. These social rewards, of connectedness to other people, are all variable.
2. The Hunt: seeking resources to aid survival e.g. material resources or information. e.g. searching through Twitter to find something interesting, or carrying out a Google search to answer a question.
3. The Self: acquiring a sense of competency. e.g. getting better at a computer game, or improving at codeacademy. In both cases you have feedback on your performance, are improving your self and your skills, and experience variable rewards.

On gamification (“the user of gamelike elements in nongame environments”)

Points, badges and leaderboards only prove effective if there is a fundamental match between the customer’s problem and the company’s solution. Otherwise no amount of gamification will help. “Likewise, if the user has no ongoing itch at all – say, no need to return repeatedly to a site that lacks any value beyond the initial visit – gamification will fail because of a lack of inherent interest in the product or service offered.”

“Variable rewards are not magic fairy dust that a product designer can sprinkle onto a product to make it instantly more attractive. Rewards must fit

There was an initial presentation aimed at a broad group, presumably ranging from people who don’t use twitter through to charities making heavy use of the platform. This meant that there was a good range of material explaining and selling the platform as a whole. There were also some good statistics on the behaviour (and, implicitly, value) of twitter users to charities:

Propensity of twitter users to give

96% of UK twitter users have donated to charity this year. 50% follow charities. Of these, 53% have donated after a tweet #twitterforgood
— Martin Lugton (@martinlugton) November 7, 2014

32% of UK twitter charity followers report an interest in fundraising content #twitterforgood
— Martin Lugton (@martinlugton) November 7, 2014

Taken together, these figures suggest that the interest in fundraising content is higher if the ask or content is right.

Importance of mobile

For anyone making the case for a responsive website, there was a great stat on mobile use of twitter:

80% of UK twitter users access the platform via mobile #twitterforgood
— Martin Lugton (@martinlugton) November 7, 2014

Twitter cards

Twitter cards can be used for signups too:

Twitter cards – Eg images, video, sound, lead generation. Can include sign ups for further contact #twitterforgood cc @Sedom19
— Martin Lugton (@martinlugton) November 7, 2014

We’re currently thinking about Data Protection, so I wonder how easy it would be to integrate twitter cards into our overall data capture setup.

This event signals the intent of twitter to deepen engagement with charities

Twitter’s interest is to deepen and extend use of its platform – and, implicitly, to monetise this use. To my surprise, adverts weren’t mentioned at all. I guess growing an engaged user base is the key challenge; advertising comes second.

I’m looking forward to what may happen next. Could twitter operate as a hub for charities sharing best practice, and for condensing and disseminating these practices back to the twitter-using charity community? There’s clear mutual benefit there. Hopefully we’ll see some developments in the new year.

GamerGate began on unsound foundations

GamerGate started off with Eron Gjoni posting an angry rant about his ex-girlfriend Zoe Quinn. Zoe Quinn was accused of using a sexual relationship with journalist Nathan Grayson to obtain favourable coverage for her work.

This claim isn’t convincing: Grayson never wrote a review of Quinn’s work, and the article that he did write which mentioned her work was published before they were in a relationship.

But even if these claims were convincing, GamerGaters targeted the wrong person. Grayson was the journalist, yet Quinn was the target of vitriol. If GamerGate is a movement about ethics in games journalism, I would have expected it to focus on the journalist.

I don’t think that policing the sexual conduct of one female indie game developer is the best way to make the case for ethical improvements to games journalism. If you are concerned with unethical actions of games producers or publishers, I think it’s much more useful and courageous to focus not on minor indie developers, but on big AAA publishers. These are the companies who have the power to undermine ethical games journalism, and who actually have a track record in doing so.

There are real, known ethical issues in games journalism, but GamerGate doesn’t seem to be focusing on them

GamerGate did not arise in response to the following ethical breaches:

• A retail chain owning a gaming publication. (The same company owns Game Informer magazine and Game Stop games stores.)
• Journalists being fired for standing up to the demands of a publisher. (Jeff Gertsmann being sacked from GameSpot after posting a negative review of Kane & Lynch)
• Product placement in Youtube reviews. (Geoff Keighley being interviewed by Pixel Perfect with product placements for Halo 4, Doritos and Mountain Dew)
• PR companies controlling the editorial output of reviews. (Plaid Social requiring Youtube reviewers of Middle Earth Shadows of Mordor to submit their reviews for approval before publication.)
• Certain publications being granted exclusive early preview access, incentivizing them to produce positive copy. (Eg IGN First.)
• (See also Leigh Alexander’s list of ethical concerns that have not led to furious activity.)

Out of concern for journalistic ethics, GamerGate has challenged journalists using Patreon, which has led to Kotaku banning its journalists from using Patreon, and to Polygon requiring staff to disclose Patreon use.

But this relatively small area of attack doesn’t address the above concerns. Nor does it identify or seek to challenge the key conflict of ethics at the heart of games journalism. As this article published on Deadspin puts it: “From the top down, publishers ranging from AAA behemoths like Electronic Arts to the IndieCade crowd do in fact enjoy symbiotic relationships with gaming media outlets, and if it came down to nothing more than sex and petty corruption, that would be nice, because the problem would certainly be a lot more easily solved.” In short, games companies have a vested interest in supporting journalists to promote their games. Changing Patreon policy in a few places doesn’t challenge this core problem.

So far I haven’t seen GamerGate seriously engage with ethical questions in games journalism. But I have seen GamerGate actively working to attack journalistic ethics.

GamerGate’s main campaigning effort directly opposes journalistic ethics

Gamergaters have attempted to punish sites that they disagree with, by acting to remove their advertising revenue. As a result of Operation Disrespectful Nod, Gamasutra lost advertising from Intel; Gawker lost advertising BMW, Mercedes (later reinstated, I believe) and Adobe.

If GamerGaters are unhappy that the press is not sufficiently independent and principled, attempting to influence publications by lobbying advertisers to withdraw their adverts seems like an odd move. The aim of Operation Disrespectful Nod appears to be to make the press less independent and to intimidate journalists who hold opinions that GamerGate supporters disagree with. In other words, this course of action suggests that GamerGate is opposed to ethics in games journalism.

I want to look at sexism now, as it’s a dominant theme of GamerGate.

GamerGate is an opportunity to attack women

GamerGate coincides with a wave of threats of violence against women. Zoe Quinn, Anita Sarkeesian and Brianna Wu have recently left their homes in response to threats of violence.

Some people have dismissed these threats as publicity stunts: e.g. “Let’s be honest. We’re all used to feeling a niggling suspicion that “death threats” sent to female agitators aren’t all they’re cracked up to be. And indeed there is no evidence that any violent threat against a prominent female figure in the media or technology industry has ever been credible.”

I strongly disagree with this. If someone is making something up they probably wouldn’t go to the FBI with evidence. And violence against women is a big problem, and not just something that people make up on the internet to get attention. In Britain, more than a million women experience domestic violence each year. The UK government reports that approximately 85,000 women are raped on average in England and Wales every year, that over 400,000 women are sexually assaulted each year, and that 1 in 5 women (aged 16 – 59) has experienced some form of sexual violence since the age of 16. We need to take threats of rape and violence seriously, not as the currency or cost of online discourse.

Some people have said that the women being attacked online have brought things upon themselves. e.g. “goading people into making unpleasant remarks and then using those statements to publicly beg for sympathy and cash,” or “They’ve become professional victims.” I think this is incorrect. I haven’t seen any deliberately inflammatory material put out by Sarkeesian or Quinn prior to their being attacked. Rather, I think that their main ‘crime’ has been speaking about and analysing games. Regardless, nothing justifies rape or death threats. To say that Quinn, Sarkeesian etc have brought things on themselves is to harness a common misdirection in our culture: blaming the victim of a crime rather than the perpetrator. In GamerGate as with society more generally, it’s wrong.

Anita Sarkeesian should have nothing to do with GamerGate. Her Kickstarter-funded project Tropes vs Women in Video Games is an example of open, clean funding for independent journalism. But Sarkeesian has been under attack since before GamerGate, and things have intensified:

Women engaging with GamerGate in a cautious, neutral fashion are being attacked. Felicia Day aired her concerns about being attacked if she shared her views on GamerGate: “I realized my silence on the issue was not motivated by some grand strategy, but out of fear that the issue has created about speaking out… I am terrified to be doxxed [having your public contact details / address shared] for even typing the words ‘Gamer Gate’. I have had stalkers and restraining orders issued in the past, I have had people show up on my doorstep when my personal information was HARD to get.” Day’s article included passages like “Games are beautiful, they are creative, they are worlds to immerse yourself in. They are art. And they are worth fighting for, even if the atmosphere is ugly right now.” She was doxxed shortly after publishing this post.

Jennifer Allaway was working on a study to understand “the importance of diversity in game content to game players, and whether or not the game industry is able to predict this desire.” Gamergaters discovered her survey and spammed it with loads of sexist entries. “if you’re even asking about equality or diversity in games, being shouted down in a traumatizing manner is now a mandatory step that you have to sit back and endure.” “Had #Gamergate participated in my survey honestly, as a researcher, I would gladly have taken their data… But instead, #Gamergate left me with hundreds of replies consisting of bald-faced mockery and threats.”

But men are safe to participate in online discussions about GamerGate, even if totally inflammatory.
Chris Kluwe, former American Football player, wrote an article on Why #Gamergaters Piss Me The F*** Off, in which he called GamerGaters “Basement-dwelling, cheetos-huffing, poopsock-sniffing douchepistols”, and a load of other things, and did not get doxxed.

And for the record, none of you fucking #Gamergate tools tried to dox me, even after I tore you a new one. I’m not even a tough target.
— Cassandra, Pudgy Nobody (@ChrisWarcraft) October 23, 2014

Instead, you go after a woman who wrote why your movement concerns her. Fuck #Gamergate and anyone aligned with it.
— Cassandra, Pudgy Nobody (@ChrisWarcraft) October 23, 2014

I haven’t seen sexualised, violent abuse and doxxing levelled against JonTron and Adam Baldwin, prominent figures in the pro-GamerGate movement. This suggests that there is a problem with the treatment of women in GamerGate. As such, I was not convinced when Huffington Post Live asked two women who support GamerGate whether equality is important and needs talking about, and the response was that “This has nothing to do with women. It is about journalistic integrity,” and that sexism in gaming “isn’t really an issue”. From the evidence I’ve seen, I disagree. Of course, not all GamerGate supporters are sending rape and death threats. But GamerGate supporters do seem to have a women problem. Gender seems to be more of an issue than ethics here. Where’s all this gender stuff coming from?

GamerGaters feel that (feminist) critics are vilifying gamers

Lots of the complaints against Sarkeesian or ‘Social Justice Warriors’ are based on the idea that gamers are being vilified and victimized because some material in games is being called out and analysed as sexist:

• e.g. “culture warriors, who thrive in an atmosphere of fear and moral condemnation … the entire gaming community is attacked as a pack of bigoted savages corrupted by gaming tropes”
• e.g. “#GamerGate supporters are constantly being told they’re horrible, misogynistic, gross nerds who just want to harass women, so they’re perpetually on the defensive.”
• (Around 22:40 in this Huffington Post Live video) “We just want to enjoy games without being told we’re horrendous people for doing so.”

I don’t think that this stance stands up to scrutiny. Anita Sarkeesian is not calling gamers terrible people in her criticism. (NB that criticism in this context means “intellectual analysis” not “saying something is rubbish”.) At the start of each of her Tropes vs Women in Games videos, she says that “It’s both possible, and even necessary, to simultaneously enjoy media, while also being critical of its more problematic or pernicious aspects.” Sarkeesian’s criticism is calm, evidenced, and highlights some games as positive counter-examples:

Similarly, Rock, Paper, Shotgun, a news site that has been attacked by GamerGate, responded with an assertion of values saying that “It is possible to criticise games and gamers, while at the same time being a gamer, loving games, loving gamers.” This is a really important point.

There is a real sense of gendered grievance in GamerGate, so let’s explore it a little more.

GamerGate is a reaction against feminism, ‘social justice warriors’ and the broadening of what is meant by games and gamers

GamerGate opposes a perceived intrusion of a liberal/left-wing social agenda into gaming and games criticism. In the words of some male developers surveyed by The Escapist:

• Daniel Vávra: “The root cause of Gamer Gate [is that] people had enough of those hypocrites that started to inject their ideology everywhere… accusing millions of people of misogyny…”
• Xbro: “I’ve personally been against the phenomena of ‘social justice warriors’, as well as the path certain gaming publications have taken in recent years (more and more discussions on morals, feminism, misogyny and other non-game related issues, and less talk about actual games and the industry).”

In the Huffington Post Live interview (about 9:40) one participants argues that it isn’t appropriate for a game to be criticised for being sexist in a review, that reviews should be objective and about mechanics.

I disagree here. Criticism should be, and always has been, about more than mechanics. Criticism is also about theme, dialogue, writing, art direction and pace. And if a game is sexist or racist or homophobic, I’d want to know about it because that’s a factor in how much I’d enjoy the game – and in some cases whether I’d even play the game at all.

I support criticism of games, and greater diversity and more inclusiveness in games, because it’s good and because it makes games better

I’m pleased to see games increasingly subject to intellectual and cultural analysis – it’s a sign of the medium becoming more respected. I’m also pleased to see games becoming more diverse in their subject matter, themes and mechanics, and in games being produced by a wider range of people for a wider range of purposes. The demographic of gamers has shifted so that it’s no longer male-dominated. The majority of gamers are now women (see research by the Internet Advertising Bureau and by SuperData Research).

No one’s going to stop people playing Call of Duty – particularly not as it remains massively profitable – but maybe we can consume other games that are totally different. As games become more diverse, we can play through a broader range of exciting, interesting, challenging stories and challenges. I’m pleased to see journalists respond to this change.

Rock, Paper, Shotgun stated that “We’d love gamers to mean ‘everybody’. It can mean everybody, if we let it.” I hope and expect to see games continue to develop and grow as a unique and awesome artistic medium. But to fully realise this, we need to let everyone speak, think and create. If we want gaming to flourish, let’s take ethics seriously and build an open, critical, loving community.

Chemnet gains – online support for the next generation of membership

Planning and designing Chemnet focused on understanding user needs, difficulties and what they could gain #HTWEB15 pic.twitter.com/F9FlKk18xT
— Martin Lugton (@martinlugton) November 4, 2015

Users expect:

• Easy-to-find content
• Easy-to-access content
• A focused, uncluttered experience
• Consumable, interactive content
• Trusted quality

User participation encouraged by giving them points and badges. Probably socially powerful signals in their own right, but additionally powerful motivators for an audience thinking about UCAS applications.

Forums not dead, if you use them well. Avoid ending up with just handful pple talking, need recognition for users & good content #HTWEB15
— Dominique Lowenthal (@DomLowenthal) November 4, 2015

Great start #HTWEB15 @MemberWise Gareth Davies from RSC ‘making science cool’ & increasing member engagement from 0.5% to 33% in 12 months
— Jonathan Ashby (@JonathanAshby7) November 4, 2015

A membership community requires ongoing investment:

• Content planning is very important to sustaining engagement
• Chemnet is planning to make improvements over time.

Creating a thriving online membership community

User needs must be the driver for creating a community space. If it doesn’t meet a user need, you’ll struggle to build engagement.

“Focus on user behaviours, not features.” Focus on understanding your users’ needs, not the specific mechanisms you’ll use to satisfy these needs. That’s your designer’s job.

Know the purpose of your community, and have a plan in place for community management and content. And resource it.

Know what will make your space better than anywhere else. The importance of quality, easy-to-find content was mentioned again.

The GpmFirst platform is focused on social learning – a community of practice, participating, sharing, and creating knowledge. It seeks to empower users.

When starting a new community, take the ‘lean community’ approach:

• Start very small, with a minimal investment, with a basic product.
• Observe how this product performs and how people use it.
• If it fails, you can close the project having invested only a small amount.
• If it goes well, you can make targeted improvements.
• This approach reduces risk.

Summarised in three steps:

1. Start by finding an overlap between user needs and organisational goals
2. Plan for active, high-quality community management
3. Test, measure, learn, repeat

Web chat – making life easy for your members – Caravan Club

I eavesdropped on this presentation via twitter.

@thecaravanclub – webchats can help you work out where the ‘pain points’ are on your website by member usage. #HTWEB15
— The MemberWise Network (@MemberWise) November 4, 2015

https://twitter.com/MemberWise/status/661862787526864896

A web chat is a text based conversation that knows where someone is stuck on the site, better than generic ‘contact us’ form #HTWEB15
— Dominique Lowenthal (@DomLowenthal) November 4, 2015

Web chat demand @thecaravanclub is highest at 10pm at night. #htweb15
— Chris Hoyle (@HoyleTime) November 4, 2015

55% said they would have called without webchat. 86% said they received an excellent/great service #HTWEB15
— Dominique Lowenthal (@DomLowenthal) November 4, 2015

@thecaravanclub: 5K webchats in 2010 to over 40,000 in 2015! 13.7% reduction in email queries. #HTWEB15
— The MemberWise Network (@MemberWise) November 4, 2015

So web chat can help you understand where people are having problems on your website, and obtain a stronger understanding than a Contact Us form.

I’d like to know how the costs and staffing challenges compare to phone calls.

How digital acted as a catalyst to transform a traditional business (YHA)

Challenges faced by YHA:

• cost per booking was very high.
• Membership in decline.
• Tactic of forcing membership on all visitors unsuccessful

There was sufficient senior buy-in for investment in digital, and some acceptance of risk.

They used data to inform marketing and development work, which led to increased revenue, which led to increased confidence.
This mechanism powers digital transformation at YHA.

Digital transformation means a cultural shift within the organization. From confidence, to data, to promotion & finally to results #htweb15
— Patrick De Sutter (@prdsutte) November 4, 2015

Taking an iterative approach is critical to success in digital transformation. Take quick wins & experiments to show that it works #htweb15
— Patrick De Sutter (@prdsutte) November 4, 2015

Some specific actions taken:

• Embedded Trip Advisor reviews on site to show quality of experience. A tough sell internally, and took conviction in the product, but greatly increased conversions.
• Google 360 tours are very effective marketing tools
• Users can now search the YHA site based on their interest/activity, then the YHA website suggests hostels nearby. More focused on user needs than asking users to pick a region first, particularly if regions are not intuitive.
• YHA has ended the tactic of “tricking” people into becoming members. For a number of years, if you stayed at a YHA hostel, you would be signed up for membership automatically. Ending this tactic has reduced the number of members, but increased their engagement.

YHA’s digital transformation has seen good results:

@YHAOfficial. Changes have resulted in 11% increase in Net Promoter score to 72% + 30% rise in online sales. #HTWEB15
— The MemberWise Network (@MemberWise) November 4, 2015

Using user experience to improve online member journeys, optimise new member and student acquisition

The problems that CIPFA user research uncovered are pretty common:

• Registration was confusing
• Labelling was based on internal structures
• Content was overwhelming

I was pleased to see the information architecture techniques of card sorting and tree testing advocated.
Card sorting groups your content into categories, and tree testing checks whether your categories make sense to users.

Perform tree testing & card sorting to inform information architecture (basically see how easy members can find what they need) #HTWEB15
— Dominique Lowenthal (@DomLowenthal) November 4, 2015

CIPFA has improved its structuring of content, reduced the volume of text on pages, and moved away from internal language towards language that makes sense to users.

I’ve heard stories of organisations thinking about their website and content at five year intervals so many times. Why is everyone still getting this wrong?
The ever-changing digital landscape – and user expectations – mean that we have to keep adapting. Even if neither of these factors existed, ongoing investment is important to refine and improve your digital offer.

Organisations reviewing website + content every five years is still embarrassingly common. Embrace + invest in ongoing improvement #htweb15
— Martin Lugton (@martinlugton) November 4, 2015

Using digital to streamline the member acquisition process

Round Table had a complicated membership sign up flow. Not so much because they had massive sign up forms, but because of the different internal steps involved in the process, and the amount of administrative overhead associated with these. Processing each new member enquiry took about an hour of admin time.

Digital agency IE carried out four phases of improvement to the process. This was more manageable for everyone. (I lost track of the boundary between iterations 3 and 4 – sorry for any inaccuracies.)

Iteration 1:

• Remove non-essential form fields from initial form. Ask the bare minimum number of questions to increase conversions. Ask the other questions in a follow-up survey.
• Validate user details automatically.
• Improved dashboard for recruitment team.
• Process for recruitment team to transmit info to colleagues streamlined via email templates.

Iterations 2 and 3:

• Mobile web forms for recruiters.
• Contact information sent to recruiter via text, with click-to-phone link allowing instant and easy follow-up of leads.

Iteration 4(?):

• Recruiter replies to the text message telling them about a prospect now populate the CRM log. This has massively increased compliance with the CRM’s data needs.
• Website allows users to text to register interest in membership.
• New website focused on location.

Why was this approach a good idea?

make quick decisions, get feedback and iterate. don’t try to be perfect first time. make impact measurable #htweb15
— Patrick De Sutter (@prdsutte) November 4, 2015

For @RoundTableBI we did something quickly, and then iterated based on what we learned. #HTWeb15
— IE (@iedesign) November 4, 2015

If you want to improve a process, set out objectives & success criteria, try something that you can do quickly & cheaply. @iedesign #HTWEB15
— Sarah Alder (@sarah_alder) November 4, 2015

What improvement did Round Table see as a result of this work?

Round Table removed 4 fields from their registration form – registrations increased 50% #htweb15
— Martin Lugton (@martinlugton) November 4, 2015

Round Table increased acquisition by 100% as a result of this work. Doing hard work to make things easier + achieve objectives #htweb15
— Martin Lugton (@martinlugton) November 4, 2015

Holding the line: the long and short of a successful CRM integration

“This wasn’t an IT project, it was a businesses transformation project.” Digital and IT projects so often involve business change – I wonder how many people plan for this from the start?

The two hardest parts of the project were:

1. Data integration.
2. Process and culture change. You need to run a parallel cultural change project. But how many people plan or put resource in place for this?

At the start of the project, they spent a week or two on “benefits dependency mapping”. This produced an intricate diagram, but, more importantly, a shared definition of project drivers, objectives, benefits, outcomes, and necessary organisational and IT changes. Although I’m not sure that all projects can predict what their impact will be at the start.

One other factor behind the project’s success was setting clear governance from the start, so that decisions could be made with authority.

Before they got a supplier on board, they held workshops to “drain swamps” in advance – explore and investigate contentious areas so that the organisation isn’t considering them for the first time when the agency arrives.

Once the development agency had been chosen, they spent six weeks building a technical proof of concept, to check the technical feasibility of the project, and to check the cultural fit with the agency. Its success built organisational trust.

One quite sad statistic on CRM projects circulated during the day, which came from a MemberWise survey:

40% of organisations haven’t integrated their website and CRM. Of those who have, 50% aren’t satisfied with the results #htweb15
— Martin Lugton (@martinlugton) November 4, 2015

What is Orientalism?

Orientalism is a body and tradition of Western representations of the Orient, created in the context of Western political dominance over the Orient, which understand and master the inferior, inherently opposed Orient, and which bear more relationship to each other as a discourse than to the real, diverse, experiences of people who live in the Middle East.

“from 1815 to 1914 European direct colonial dominion expanded from about 35 percent of the Earth’s surface to about 85 percent of it.” [41]

“Orientalism is fundamentally a political doctrine willed over the Orient because the Orient was weaker than the West, which elided the Orient’s difference with its weakness.” [204]

“It is rather a distribution of geopolitical awareness into aesthetic, scholarly, economic, sociological, historical and philological texts;” [12]

“so far as the West was concerned during the nineteenth and twentieth centuries, an assumption had been made that the Orient and everything in it was, if not patently inferior to, then in need of corrective study by the West. The Orient was viewed as if framed by the classroom, the criminal court, the prison, the illustrated manual. Orientalism, then, is knowledge of the Orient that places things Oriental in class, court, prison, or manual for scrutiny, judgment, discipline, or governing.”[40-41]

“Orientalism as a Western style for dominating, restructuring, and having authority over the Orient.” [3]

The division between monolithic West and Orient is man-made

“such … geographical sectors as “Orient” and “Occident” are man-made. Therefore as much as the West itself, the Orient is an idea that has a history and a tradition of thought, imagery and vocabulary that have given it reality and presence in and for the West.” [5]

“neither the term Orient nor the concept of the West has any ontological stability; each is made up of human effort, partly affirmation, partly identification of the Other.” [xii]

“The geographic boundaries accompany the social, ethnic, and cultural ones in expected ways. Yet often the sense in which someone feels himself to be not-foreign is based on a very unrigorous idea of what is “out there,” beyond one’s own territory. All kinds of suppositions, associations, and fictions appear to crowd the unfamiliar space outside one’s own.” [54] “…We need not decide here whether this kind of imaginative knowledge infuses history and geography, or whether in some way it overrides them. Let us just say for the time being that it is there as something more than what appears to be merely positive knowledge.”[55]

Debates about identity are important. Identities create outsiders and enemies

“Debates today about “Frenchness” and “Englishness” in France and Britain respectively, or about Islam in countries such as Egypt and Pakistan, are part of that same interpretive process which involves the identities of different “others,” whether they be outsiders and refugees, or apostates and infidels. It should be obvious in all cases that these processes are not mental exercises but urgent social contests involving such concrete political issues as immigration laws, the legislation of personal conduct, the constitution of orthodoxy, the legitimization of violence and/or insurrection, the character and content of education, and the direction of foreign policy, which very often has to do with the designation of official enemies. In short, the construction of identity if bound up with the disposition of power and powerlessness in each society, and is therefore anything but mere academic wool-gathering.” [332]

“A fourth dogma is that the Orient is at bottom something either to be feared (the Yellow Peril, the Mongol hordes, the brown dominions) or to be controlled (by pacification, research and development, outright occupation whenever possible.” [301]

People are more diverse than this binary

“the terrible reductive conflicts that herd people under falsely unifying rubrics like “America,” “The West” or “Islam” and invent collective identities for large numbers of individuals who are actually quite diverse, cannot remain as potent as they are, and must be opposed,” [xxii]

Antidotes and alternatives to division and Orientalism

“critical thought does not submit to state power or to commands to join in the ranks marching against one or another approved enemy. Rather than the manufactured clash of civilizations, we need to concentrate on the slow working together of cultures that overlap, borrow from each other, and live together in far more interesting ways than any abridged or inauthentic mode of understanding can allow.” [xxii]

“Since an Arab poet or novelist – and there are many – writes of his experiences, of his values, of his humanity (however strange that may be), he effectively disrupts the various patterns (images, clichés, abstractions) by which the Orient is represented.”[291]

Be aware of differences; avoid sweeping groupings; look at mingling and exchange between groups. Look at individuals and their expressions of their own feelings and thoughts. Look at self-expression, art and literature.

Recommended works: Amiel Alcalay’s Beyond Arabs and Jews: Remaking Levantine Culture Paul Gilroy’s The Black Atlantic: Modernity and Double Donsciousness Moira Ferguson’s Subject to Others: British Women Writers and Colonial Slavery, 1670-1834. [353] “In these works, domains once believed to have been exclusive to one people, gender, race or class are re-examined and shown to have involved others. Long represented as a battleground for Arabs and Jews, the Levant emerges in Alcalay’s book as a Mediterranean culture common to both peoples; according to Gilroy a similar process alters, indeed doubles, our perception of the Atlantic Ocean, previously thought of as principally a European passage. And in re-examining the adversarial relationship between English slave-owners and African slaves, Ferguson allows a more complex dividing white female from white male to stand our, with new demotions and dislocations appearing as a result in Africa.” [353-4]

“My aim … was not so much to dissipate difference itself … but to challenge the notion that difference implies hostility, a frozen reified set of opposed essences, and a whole adversarial knowledge built out of those things.” [352]

Course outline

1. Process analysis
   Measuring the flow of units through a production process; Little’s Law; inventory turns; inventory buffering: make to stock (McDonald’s) or make to order (Subway); working out bottlenecks when there are different types of flow units, processes with attrition loss; reasons for inventory.
2. Productivity
   Lean operations and waste reduction; the seven sources of waste; KPI trees and sensitivity analysis; overall equipment effectiveness framework (OEE); reducing idle time through line balancing and standardising processes; labour, material and capital productivity; return on invested capital (ROIC) trees;
3. Variety
   Motives for variety; batch processes and setup time; working out a good batch size; Single-Minute Exchange of Die (SMED); benefits of partial flexibility; delayed differentiation (via product design) to reduce costs of variety;
4. Responsiveness
   Reasons for waiting: insufficient capacity and variability of arrival times and/or processing times; coefficient of variation of demand and processing time; how to compute the average waiting time; measuring inventory over the course of a day; usefulness of pooling; strategies for prioritising work: first come first served aka first in first out, versus sequencing, shortest processing time rule; problems with appointment systems; efficiency gains are often about process redesign rather than just optimising/balancing: value stream mapping aka process mapping aka service blueprints: Yves Pigneur’s 7 ideas for redesigning processes; waiting time and attrition loss (using Erlang Loss table);
5. Quality
   Reasons for defects – performance and conformance quality; redundancy; impact of scrapping and rework on flow; buffers reduce risk of resources being starved or blocked, to keep flow rate up; in contrast: Toyota production system: reduce inventory to expose problems; Kanban – demand pull: work is authorised by demand, so you reduce the number of Kanban cards over time; six sigma: checking units produced against a specification; control charts: normal and abnormal variation; Jidoka system sacrifices flow for quality; Kaizen and Ishikawa diagram for root cause problem solving.

Short of summarising the outline above, I won’t attempt to share everything I learnt. Instead, I’ll share what was most relevant to my own practice.

The distinction between project management and process management

This course was about process management – about doing the same thing over and over. My job is incorporating more process management elements, so I took this course to improve my understanding, and so that I could begin to make more effective process improvements.

Some elements of my work are process management – communicating, planning, and running an agile sprint production cycle. I run through this process every couple of weeks.

Other elements of my work are project management – leading a large web development project; providing consultancy on a project; overseeing the scoping, planning and creation of a new element of functionality or a user experience improvement: each discrete piece of development work is unique. So sometimes it’s useful for me to think in terms of projects, and other times it helps to think more generically and look at underlying processes.

Little’s Law

In any process, the average inventory (number of units in the process) = the average flow rate x the average flow time (the time it takes a flow unit to go from the start to the end of the process)

Key implication: if the flow rate is constant, reducing inventory will reduce flow time, allowing work to be completed more quickly.

See more about Little’s Law.

The seven sources of waste (Taiichi Ohno)

1. Overproduction – to produce sooner, or in greater capacities than demanded. These goods need to be stored; their production slows the rate with which you turn your inventory; they could become obsolete or be stolen.
   The solution: match supply with demand.
2. Transportation – unnecessary movement of people or parts between processes.
   The solution: relocate processes, then introduce standard sequences for transportation.
3. Rework – repetition or correction of a process.
   The solution: do it right the first time. Find out the reason for the quality problem and put a stop to it.
4. Overprocessing – processing beyond what the customer requires.
   The solution: make sure you have guidance for what your standards are.
5. Motion – unnecessary movement of parts or people within a process.
   The solution: create and use standard workspaces that have been created to minimise movement.
6. Inventory – the number of flow units in the system. The biggest source of waste. Bad for inventory turns, increases customer wait time and flow time. Inventory needs to be stored, which is costly.
   The solution: improve production control system and reduce unnecessary “comfort stocks”.
7. Waiting – underutilising people or parts while a process completes a cycle. i.e. a flow unit waiting for a resource. Often a direct result of inventory. Waiting can happen at the resource: this is idle time.
8. Intellect – an eighth source of waste. Don’t waste workers’ abilities to help solve problems and improve processes.

Increasing profitability is easier if you’re constrained by capacity than if you’re constrained by demand

If you’re constrained by capacity, increasing the productivity of the bottleneck can help you significantly increase profits. (This is particularly the case for businesses with low variable costs and high fixed costs)

If you’re constrained by demand, you’ll only be able to significantly increase profitability if you’re able to lay off workers.

Variability increases wait times, even if resource utilisation is less than 100%

If people arrive at regular intervals, and take a fixed length of time to process, then you can plan your processes to avoid waiting time. But real life is less predictable.
Variability of arrival times and processing times can lead to inventory in a process, even if utilisation is 80%.
So variability means that even if you aren’t utilising all your resources all the time, you’ll still have people or products waiting in the process.

Two reasons for waiting: insufficient capacity; variability of arrival times and/or processing times.

If you’re constrained by capacity, you don’t need to worry about demand variability as you already know there will be bottlenecks. If demand is the constraint, and it’s variable, then you need to think about it, as it will cause waiting times.

Partial flexibility is usually the best way to deal with variety

If there is variability in demand, you need to accommodate it. Total flexibility is expensive, and usually not needed.

Eg. It’s sensible to hire developers who have skills in two areas, so you have flexibility, but don’t have to pay the costs of a developer skilled at everything.
For each area of your work, hire at least two people with those skills.

“The way we frame a problem determines the types of solutions we come up with.”

A surprisingly philosophical insight. Often it pays to be more creative than just doing queuing analysis and line balancing.
Question your processes at a strategic level – don’t just think tactically and inside the box.

Value stream mapping is a tool to help you focus your process on valuable activities

Value stream mapping, aka process mapping, aka service blueprints – map out the steps the customer has to go through, then divide them into ones that add value and ones that don’t, or which are waiting time.

Yves Pigneur has a framework for this: Customer actions; onstage actions, backstage actions; support processes.

7 ideas for redesigning processes

1. Move work off the stage.
   E.g. online airport check-in.
2. Reduce customer actions / rely on support processes.
   E.g. rather than requiring customers to fill in all their medical details each time they come to visit, you could have a database to store them.
3. Instead of optimising the capacity of a step, see if you can remove it altogether if it isn’t really needed.
   E.g. Hertz Gold removed the airport check-in step as it provided no value.
4. Avoid fragmentation of work due to specialization / fragmentation of roles.
   E.g. in a bank, it’s annoying to have to fill in different forms for different people, rather than just doing everything at once.
5. If customers are likely to leave the process due to long wait times, move the waiting time to later in the process if you can.
   E.g. Starbucks making you pay first, then wait for your coffee.
6. Have the waiting occur out of a line.
   E.g. restaurants in malls using buzzers to let people know that their food or table is ready, rather than having them wait in a line.
   E.g. appointments to see a doctor.
7. Communicate the wait time to the customer – set expectations.
   E.g. theme parks.

How much do defects cost? It depends on where they are detected.

If defects are detected before the bottleneck, the cost is driven by the input prices.
If defects are detected after the bottleneck, the cost is the opportunity cost of the lost sale.

The step at which the defect happens isn’t important – what’s important is the step at which it is detected.

Therefore it’s very important to test flow units as much as you can before you put them into the bottleneck.

For me, this means that we want to catch problems in the specification stage if possible. Minimise potential problems by making sure that the work is sensible, and that the requirements are clearly articulated.

Kaizen and Ishikawa are tools for root cause problem solving

Kaizen equips front line workers to identify and solve problems.

Ishikawa diagram – structured brainstorm. Shaped like a fish bone. Try to identify root causes. Asking ‘why’ 5 times helps.

Once you’ve done this exercise, go out and measure instances of the identified defects.
Plot these on a pareto chart.
See which defect is most frequent and focus on that first. Generally the pareto principle applies: 80% of the defects are caused by 20% of the root causes.

These methods are recommended because they oscillate between thought and reality, gaining the benefits of both:

Reality: Jidoka – the process is triggered by real-world defects.
Thinking: Ishikawa diagram, to think about what might be causing the problem.
Reality: Pareto chart – collects data to see which causes are most frequent.
Thinking: Think up alternative solutions.
Reality: Experiment with the solution you choose.

Actions I will take as a result of this course

• Map out the production process so everyone knows what steps are required. If people don’t understand all the steps in a process, they might have unrealistic expectations of waiting times.
• Create formal processes for new jobs – development/UX work and bugs. This will reduce variability of inputs and reduce risk of defects by improving the quality of inputs/briefs.
• Formalise waiting time processes. Draw up (collaboratively?) and obtain organisational agreement for a set of organisational priorities for bugs and for new work. This will mean that all prioritisation decisions are made according to a clear set of standards. E.g. number of users affected, financial implications, strategic priorities. Currently I don’t do first-come-first-served, but rather prioritise according to business need and urgency, but this does require me to be wise like Solomon. Better to have some commandments to live by, and a supreme court to interpret them.
• Reduce the number of units being processed at any one time. Little’s Law states that this will result in reduced waiting times. This might be a hard sell, but the truth is that we are already constrained by how many hours of development we have in each two week cycle.
• Identify and reduce the sources of waste in my work. eg transportation – reduce transportation costs in communication with agency and stakeholders (make it clearer), reduce movement of work around internal stakeholders (currently I report to internal stakeholders outside of the tools I use for day-to-day project management, which adds costs and the risk of misinformation.)
• Harness worker intellect through more regular review cycles. Constitute regular reviews of processes with all people involved in them, to see how they feel they are going, and what could be improved.
• See if processes could be redesigned to be more efficient. Could steps be removed or automated?
• Conduct value stream mapping – figure out which steps add value and which ones don’t. Use Pigneur’s framework for process redesign to improve these processes.

And here’s my remix:

I started by listening through the original mix and picking out some samples I liked.
Because we have access to the master mix, it’s possible to isolate a single voice or instrument.
I’d be interested to learn techniques for doing this sort of sampling where you don’t have the luxury of access to the original mix – ie where you can’t just isolate the instrument you want.

Once I’d identified the samples I wanted to use, I cut them up in Soundation. I made sure that all of my samples were 4 bars long – this made it easier to work with them later.
If they’d been of different lengths, it would have been much harder to coordinate them and use them concurrently.

My next move was to stitch these into a single song. One way of doing this would have been to record the music “live”, and simply turn on and off the different channels over the course of the song.
I played around with doing this, but couldn’t find a way in Soundation to record “live”.

But doing things more laboriously meant that I had more control – I could take my time applying effects to the mix, in a way that I wouldn’t be able to do live.

So I copy-pasted the 4 bar samples I’d created, and stitched them together into a full song. There was lots of copy-pasting, and it was a bit fiddly to get things to line up exactly. I found that zooming in very closely, and clicking on exactly the start of the bar, before pasting, helped.
Something I didn’t realise at first was that clicking and dragging a single clip from its top right hand corner loops it over and over. So I should have combined all my cut up 4 bar samples, looped them, and then cut down from there, rather than doing lots of cutting and pasting.

To build up the track, I started with the bass, then added in some piano and then drums.
I wanted the track to start with the rhythm section first, to depart from the original. The intro might be my favourite part of this remix.

Shortly after I’d started adding tracks, I began working on the effects and the positioning in the mix.
I wanted the bass to be prominent but not too loud, and I wanted the backing vocals to feel drifty and ethereal, but also still very present.

I used Distortion, then Equalizer then Compressor on the vocals. The distortion had very high gain and low volume, using distortion type WS1 (no idea what that means). The low end of the equalizer was maxed out; Mid was set to around a third, and High was set to nothing. The Compressor was pretty gentle. Very low attack, low release, high ratio and threshold, and medium gain.

I gave the backing vocals some reverb and delay. I used a cutoff filter on the drums towards the end of the track, to give a descending feel. I then removed the drums completely, before bringing them back at the same volume, and then fading out the volume, allowing me to close the remix with the bass alone, after the piano and then the drums have faded out.

I enjoyed picking out different parts of the original track to use in the remix. I might have liked a bit more raw material to work with – probably I should have used some additional instrumentation from elsewhere – but I quite liked the challenge of working with just the original track. I found it a bit tricky to line up some of the timing – and I’m still not quite sure if the piano is right. I think I’d enjoy working on a remix of a different track – or set of tracks – but I’d want to think it through very carefully. I’d also need to think about how to best take samples from tracks when you don’t have access to the masters. So I may give the final module of Play With Your Music a miss for now, but I’ve definitely enjoyed the course, and found that it’s improved my critical listening ability and given me a taste of using a digital audio workstation and mixing and remixing audio.

My plan for this course is to keep half an eye out for things that might interest me. I don’t expect to be closely involved in conversations, or in knowledge creation, as I’m not sure how much time I’ll have to commit.

I’m interested in people using WordPress to run open online courses, so a hangout last week caught my eye. I didn’t attend, but I’ve watched the video of the event, and have produced a rough record of the 1 hour 25 minute discussion. Hopefully this will help people access some of the insights within.

This was quite a technical and practical discussion. Certainly much of this conversation was beyond my knowledge of WordPress. The discussion was very much about the logistics of actually setting up open online courses in WordPress.

Logistics of registration and implications on aggregation

ocTEL is using AMU plugin (Add Multiple Users) to import users to its multi-site setup. It doesn’t work with existing users, though – they need to be added manually.

FeedWordpress pulls in tweets for comments in ocTEL. Potentially this could cause fragmentation of conversation, as people comment on an imported tweet inside the WordPress environment. Outside the course, the comment would not be associated with the original tweet. Some people have seen the pulling in of tweets as a bug rather than a feature.

Martin Hawksey has edited Feedwordpress registration to also check on a user on other networks, so these interactions aren’t lost even if they happen under a different username. (See below for further discussion of this code.)

Delicious‘s feed isn’t very useful any more.

Team blogs are tricky for registration – the system will just think of all posts as relating to a single author.

ds106 and ocTEL keep stealing/sharing code to improve their respective registration systems.
(I found it funny/strange how polite everyone is about this in this call. Everyone is working hard and improving things for the community. There’s no shame in using other’s code as part of this work.)

If a user changes their feed, ocTEL updates that user’s details. In ds106 this has to be done manually.

ocTEL includes a BB press forum as a space for people who want to participate that way.

(Open) Badges added to ocTEL – 36:00

Lots of people were sceptical about this at first, but soon eagerly asked how to obtain the badges.

ocTEL is using the BadgeOS plugin.

While it is possible to link up with Credly and Mozilla’s Open Badges, Martin decided against this as being too much hassle for the user.

The Badge OS plugin actually integrates with the BuddyPress plugin, and allows you to award badges for activities within BuddyPress.

Martin wants ocTEL to issue Open Badges directly. This should be feasible, as the JSON that Mozilla needs isn’t too tricky.

Badges are useful in the context of showing people what activity is happening. Leaderboards of activity could be demotivating, but maybe you could be shown the activity levels around you.

Ultimately we want to use data to help people navigate the course.

Running an open course across two different installs – 52:50

Will this be adversarial, or will the passion of outsiders help people taking the formal course? “We can use the outside people as agitators and provocateurs”

If you have open participants and paying participants in the same space, could that cause problems?

Paid students get a grade, and more feedback and attention, even if the content is the same.

Universities want to capitalise on the energies of outsiders (using them as an element of the course – “Human as OER”), but they don’t want to give them resources. That’s the origin of having an open course across two spaces. Is it really a big deal to just host outsiders? If you talk to lawyers about this, the conversation becomes trickier.

Returning to the problem of bulk-importing existing users with multi-site – 58:44

Boone’s simple import users plugin might help. Imports via an email address paste into a single site of your multi-site. Doesn’t have many options. Auto-generates user logins based on email addresses. Does some basic BuddyPress stuff – modifies the registration email to include buddy press info.

Martin Hawksey thinks this will be a good basis for what he wants to do. He expects to modify the code so that it pulls in an associated twitter username as the account name.

iframe oembed issues on multi-site – 1:04:00

Boone usually creates a shortcode, with 10 lines of code. Create an MU plugins file with shortcodes.

Potential changes to Martin Hawksey’s FeedWordpress member lookup fork, to save effort in maintaining – 1:07:28

Martin uses some code that is from FeedWordpress, but has to modify it every time FeedWordpress updates (thankfully this isn’t too frequent).

Boone suggests using filters, to save effort maintaining. He proposes adding in a line that will do nothing in most cases, but will do the extra cool thing you want to do in your specific case.

There is a debate over putting in loads of filters VS having every special use case having a custom fork. In this case, it’s probably a common enough use case to warrant being a filter.

You want to know which bands clash with each other; you want to be able to work out how long you can spend checking out a new band before leaving to get to the front in time for a band you know you’re going to enjoy.

I love planning activities within constraints to maximise my enjoyment, so I was excited to see the Graspop Metal Meeting 2014 running order announced.

Graspop Metal Meeting 2014 running order is out! https://t.co/8k5z5z5z5z

— Martin Lugton (@martinlugton) April 22, 2014

But I was soon disappointed, as I downloaded the document and tried to start planning which bands to see.

In looking at this document I made two basic visual assumptions:

1. As we go down the page, the hour of day increases.
2. Bands on the same horizontal level are playing at the same time.

This guide went against both of these assumptions:

1. As we go further down the page, as a general rule the hour does increase. But each stage goes at its own pace, so it’s hard to compare between stages.

   

   Meshuggah are further down than Megadeth, but go on stage nearly 3 hours earlier.

2. Bands on the same horizontal level are often playing at very different times.

   

   On first glance it looks like Unida and Sabaton are playing at the same time. But if you look at the times, Sabaton actually starts 2.5 hours after Unida has finished. Plenty of time to see both.

Okay, so things weren’t looking great. I started thinking about making my own spreadsheet, but I thought I’d give the festival the benefit of the doubt and get in touch:

Hey @GraspopMetal, your running order is really hard to use. Any chance of a version where the times are aligned across stages?

— Martin Lugton (@martinlugton) April 29, 2014

Thanks for the feedback, Martin. We’ll look into it and see what we can do.

— Graspop Metal Meeting (@GraspopMetal) April 29, 2014

And in response they made a new version of the running order, and have updated their website with this new version:

We’ve updated the running order. Thanks for the feedback! https://t.co/8k5z5z5z5z

— Graspop Metal Meeting (@GraspopMetal) April 30, 2014

The extent to which the document has changed underlines just how hard it was to use before.

We can now compare Meshuggah and Megadeth much more easily:

And we can see that Unida play their set a long time before Sabaton:

The new running order is so much easier to use:

• Easier to understand which bands are playing at a given time: all stages follow the same time axis.
• Easier to understand the overall shape of the day across the different stages.
• Colour coding helps distinguish between different stages when scanning down the page.
• Easier to read because the background is white.

I think there are a few lessons here:

• Usability is important. Some documents exist almost exclusively to be used. When designing them, we should try to understand how people will make use of them. User testing before launching a product is the safest way to do this, otherwise you risk the product failing to serve its purpose because you miss something important. If everyone attending Graspop in a couple of months was trying to use the old running order, I think there’d be a lot of confusion and frustration.
• Listening is important. Graspop could easily have ignored my tweet and my point, but instead they chose to engage with it. Credit to them for being open to that.
• Responsiveness is important. The festival was ready to act on what I said, and pushed out a new version of the running order in less than 4 working hours after I posted my message. If you don’t get it right the first time, being ready to respond and improve things is a pretty close second. Well done Graspop!

* Other important decisions include where to camp, who to befriend, where to obtain provisions, and whether to risk crowdsurfing or a wall of death. But I’m not going to talk about these things here.

This video (no longer available) features some useful examples of multivariate tests done during the Obama campaign. Here’s my summary:

1. Nothing is sacred. Question assumptions
2. Explore before you refine
3. Less is more
   Some content experiments:
   • One Kings Lane: Removing 2 fields led to an 8.4% increase in signups
   • Clinton Bush Haiti Fund: Removing title field and phone number field saw a 10.9% increase in the $value of each visitor’s donation
   • AVG Australia: Removing links from the checkout process increased the $ value of each visitor by 16.8%
4. Words matter. Focus on your call to action.
   • Donate Now, Please Donate, Why Donate?, Donate and Get a Gift, Contribute: which is better? The status of the individual makes a difference to which of these options is most effective: new people respond best to “Donate and Get a Gift” (+15.2% $ per pageview compared to Donate Now); people who have signed up but never donated responded best to “Please Donate” (+27.8%), and people who have previously donated responded best to “Contribute” (+18.4%)
   • Clinton Bush Haiti Fund’s donate form – changing the confirmation button from “Submit” to “Support Haiti” increased $ per visitor by 15.75%.
5. Fail Fast
   You can get statistically significant results very quickly.
6. Start Today

• Q: How many variations should you test at once?
  Tradeoff between breadth of test and speed of reaching statistical significance.
  A good starting point is to test with 4 variations, then take the top 2 and test them again.
• Q: What percentage of website traffic should you allocate to a test?
  You can control this per experiment.
  They recommend allocating as much as you can, to reach statistical significance sooner.
• Q: Can you target specific types of visitors?
  You can choose to target all, new, or returning visitors.
• Q: How do you target paid traffic?
  Do this with the targeting interface: trigger by origin location, or match certain query parameters (eg utm_source)
• Q: How do you test dynamic pages?
  Go to interactive mode to make the dynamic stuff happen.
  OR
  Tell us when you want to activate optimizely. By default it loads on page load; but you can activate it manually through the API.
• Q: Does Optimizely work for pages behind logins or paywalls?
  Yes, as long as you’ve added the snippet to the page.
• Q: How do you test big changes to a page, or test a new page against an old one?
  If you want to test a big change to a page, or test against an old page, you can make it so that visitors selected for a specific variation are redirected to another URL. (Make sure that the snippet is on both pages)
• Q: How do you test server-side changes? i.e. where the variations are reflected on the backend and not just on the front end (eg graphics). Eg price variations.
  A: Optimizely wants to improve at this. Most people doing this currently set up two different pages, and set up a javascript gate on the front end to pass 50% of visitors to one page, the other 50% go to the other.
  See Optimizely Tutorials and Optimizely Developers
• Q: How soon to pull the plug on a variation when it looks like it’s not performing well?
  A: Wait until you get statistical significance. The early days of a test may not be truly representative.

Here are some things that I wanted to do with my mix:

• Keep the piano as the dominant backing for the first verse, as it’s really the only frame for the vocals at this point in the song. I like how the faster, plonky texture goes with the soft flowing vocals. But the for the rest of the song, I wanted to make the soundscape a little less cluttered, and decided that the piano could be removed.
• Highlight some of the lovely drumming and bass work. I mixed the volume of the bass quite high, and increased it a little for one key passage. For the drums, I used EQ so that the drumming is initially very muted and bassy – a bit like it’s being played next door. I then change the EQ part way through to bring out the rimshot section. (I’m not a drummer, so I could be incorrect here, but it’s the bit that sounds more clickety-clackety). In the final section of the song I set the EQ to gradually change, giving a strange, mechanical swooshy descent.
• Make the baritone guitar a little smoother. To do this, I applied a low pass filter. I made sure not to put the cutoff of the low pass filter too low – eg 308hz, as it makes the sound too bassy and it loses too much of the attack at the edge of the note. 1477 seemed like a good value, but upon listening to the track in the context of the mix it lacked definition and became a bit insignificant now that I’d dialed back the twang.
• Strip the drums down for all but the section where I want to highlight them. So i applied a low pass filter with a very low cutoff.
• Make the steel guitar more shimmery. So I put some reverb on it, with a a fair amount of wet signal.
• Change the lead vocals somehow. I love the solitary vocals and piano in the first verse, so I didn’t really feel any need to change anything here. But for the sake of experimentation, I wanted to see if I could make the vocals more ethereal, or give them more presence somehow. I started off by applying a very small delay. When set to under 30ms, this just made the sound feel a bit wider, rather than adding distinct or overlapping voices. I settled on a slightly longer delay time, keeping it about the same for both channels. I wanted the final line of the song to be free from reverb, so I used the automation setting in Soundation to drop the wet signal for the very last line. Having played around with this for a while, on listening through I decided that actually the extra lead vocals didn’t work so well and I got rid of this effect.
• Give the backing vocals a bit more presence. I added some compressor to do this.

Along the way I found a few artifacts or clicks in the original wav files. It didn’t seem like I could directly edit the audio files that made up each track, so instead I went in and dropped the volume on the relevant track to nothing for the split second where the distracting noise appeared. (I used Soundation’s automation setting here, as shown below). This took a bit of work soloing each instrument to check which track needed fixing.

For anyone else working on this mix, I found artifacts on the following tracks and locations: bar 27 on the bass track (2 clicks), bar 28 in EFX (I’d muted this track anyway), bar 34 and 39 bass, bar 40 backing vocals, bar 42 in the steel guitar, bar 46 lead vocals, bar 56 on the steel guitar, bar 68 bass, bar 73 bass.

What was my overall experience of adding effects to this song? Overall I found that effects were a bit unnecessary, to be honest. I enjoyed being able to highlight the bass and drums, but that was mainly down to volume and EQ rather than anything else. I also enjoyed using compression to bring a bit more prominence to the backing vocals. But I probably enjoyed using the pan and volume settings in last week’s static mix more, as it felt like I was honing in on the best elements of an existing expression, rather than trying to put Christmas lights onto an already beautiful tree. I think that in some cases effects can sound really cool – I’m particularly looking forward to playing around with an electric guitar in a digital audio workstation – but in other cases they might be an unhelpful veneer. In this track it felt that a lot of the time effects were cluttering or obscuring. Maybe that’s more to do with my skill in employing them than their utility in general. From this week’s module I’ve gained initial basic understanding of effects, and I’m looking forward to being able to learn when and how best to use effects in the future.

For example, at Mind, we split our information on depression into subsections, so that if you want to quickly navigate to “Symptoms”, “Causes”, “Self-help, treatment and support”, “How can friends and family help” (etc) you can do so. It also helps people understand the content that the article will cover much more easily than requiring them to navigate through a single long page of text.

But there are some drawbacks of splitting your information into different sections – sometimes people don’t know that they’re reading one part of a larger topic. So the way that you show that there are other sections, and allow people to navigate through them, is important.

Reviews on technology sites are similarly complex, and these sites face the same navigational challenge. How do different technology sites approach multi-section articles, and which approach is best? I’ve assembled a few screenshots here (with added highlighting and some analysis). I’ve also tried to summarise best practice as I see it. I’d be interested to see some user testing on all this, to see if my review is on the right track. This material is all for desktop – there’s another discussion to be had for mobile.

Best practice

• Prominently number sub-sections.
• Include sectional navigation at the top and bottom of each page.
• Include numbers in your section headings if your content is sequential.
• Use the overall article heading and the heading of the sub-page to describe structure. You need to be attentive both to positioning and hierarchy, and to labelling your content clearly.
• Include “next” and “previous” section navigation. Be aware that it needs context to make sense. This can be provided either by the button itself, or by proximity to a summary of the sections of the article.
• List all the sections of the article – don’t hide them in a dropdown – and show the user where they are currently in relation to other sections.
• Make your inter-section navigation tactile and make sure that it is consistent and not obscured.

Tom’s Hardware

View an example article. Click on the screenshot below to see a full-size version.

I think Tom’s hardware is very effective at helping people understand and navigate its multi-section articles:

• There’s navigation at the top and the bottom of each section.
• Sections are numbered – which, particularly when looking at the top section – helps convey the fact that this article is made up of multiple sections.
• The proximity of the navigation elements at the top of the page is useful. The section number, the dropdown, and the next and previous buttons (one of which is greyed out in this case, which suggests that it relates to the number in front of the heading) are all displayed close to the heading.
• The interactive elements are nice and chunky, and are highlighted using a dominant colour. This suggests that they are interactive.
• The bottom navigation section gives more space to the next and previous buttons.
• In the bottom section, under the heading “Summary”, the different sections are numbered and their names are written out. The current section is highlighted in bold, with a chevron next to it.
• The adverts at the bottom of the page clearly come after the body of the article, and so don’t add confusion.

There are a couple of small points for improvement:

• The URL could do a better job of making it clear that you’re in the first section of a multi-section article. http://www.tomshardware.co.uk/gaming-cpu-review-overclock,review-32885.html looks like the URL of a one-off page sitting directly under the home page. It’s only when you get to the second section of the article – http://www.tomshardware.co.uk/gaming-cpu-review-overclock,review-32885-2.html – that you can infer any structure.
• The next and previous buttons themselves don’t make it clear what they relate to. In isolation they could be confusing: “Next” what? Next article in chronological order? Next page in this category? But in context these buttons work well, as the summary section makes them much clearer.

Tech Power Up

View an example article.

I think this design is less effective:

• There’s less consistency between interactive elements, which it makes their purpose less clear. (Compare the “Packaging & Contents” buttons at top and bottom) So while these buttons feature at top and bottom of the page, they don’t have as much impact as they could do.
• Of the two buttons, the one at the bottom is much better. It explains itself as a section navigation by saying “next”, and also explains what section you’ll be taken to. The top button doesn’t make it clear that it relates to another subsection of the current article. It’s also hidden among the social links.
• A dropdown is used to allow navigation between sections. This takes up less space than the summary solution used by Tom’s Hardware, which shows all the heading names, but might confuse some users if they don’t know that they’re currently in the listed section of a multi-section article.
• The use of numbers helps to show that this section sits in a sequence. But the numbers are shown at the bottom and not the top of the page. Tom’s Hardware, on the other hand, brings these section numbers to the fore at the top of the article.
• There’s no back button, unlike Tom’s Hardware. Rather, the dropdown navigation sits in-line with the next button. Again, this saves space but may introduce more confusion.

Overclock 3d.net

View an example article.

• The page title, and the next and previous section buttons aren’t very clear. The “previous” button on the first page is ‘greyed-out’ (lightened in colour to show that it’s not functional in this context), but the difference in colour is so slight as to be potentially missed.
• At the bottom of the page, sections are numbered, and previous and next navigation is used as well (it’s ‘greyed-out’ when not relevant) on the left hand side, and a dropdown showing section contents is displayed on the right hand side.

Hexus

View an example article.

• Advertising distracts attention from the top navigation. Confusingly, it’s in the same colour as the bottom navigation.
• The hierarchy of overall article heading, and the heading of the individual pages, is nice and clear. The same is the case with Tweak Town, below.

Tweak Town

View an example article.

• The use of colour around the top navigation is quite distracting, which is a shame, as otherwise this design is good.

Hard OCP

View an example article.

• The navigation might need to be more prominent to be visible to new visitors.

Anand Tech

View an example article.

• The top doesn’t feature any navigation options.
• The top doesn’t make it clear that you’re in a section of a larger article through use of headings.
• The bottom next navigation says the name of the section, but it doesn’t make it clear that it is the next section in a sequence of sections. It could potentially be anything. The title of the second section in this case is confusing – “Mantle: a low-level graphics API for CGN”. This could quite easily be referring to a completely separate article on Mantle.
• The bottom navigation doesn’t have arrows or next buttons. This can be confusing if you arrive part way through an article from organic search.
• The bottom dropdown shows section titles, but no section numbers, even though these are in URLs after the first page.

Vortez

View an example article.

• There’s a breadcrumb at the bottom of the page, which makes it clear that you’re in page one of a larger article, even if there isn’t an easily scannable summary of the other sections and their contents.

The first part of this week’s assignment was to reproduce Clara Berry & Wooldog’s “Air traffic”, by setting the volume and pan (left-right) for each track. This reproduction is to create a “convergent” mix. The task was to make a static mix, so we’d just have to find a single setting for each track, rather than adjusting things over the course of the track.

There was a video on multi-track mixing and a useful interview with the musicians, producer and engineer involved in the creation of “Air traffic”. Alex Ruthmann has a good interview technique, and challenged the interviewees to unpack things that were obvious to them but not to mixing novices.

We were doing our mixing in-browser in a tool made for the occasion. It basically loads an 128kbps mp3 file for each track, and then uses javascript to handle the volume and panning of each track. The mix information gets stored as a string at the end of the URL generated by the web page. This is how we can share our mixes with each other. (Unfortunately the URLs are no longer accessible.) The string at the end of the URL is pretty neat: a dictionary with global information identifying the song and its length, followed by a list of dictionaries for each track. I actually just used the soundcloud version of the track, rather than the left and right channels in the mix deck, as I found this a bit easier to navigate. Hopefully the versions aren’t too different!

The most fun part of this week was being given the chance to come up with a different version of the mix.

This is the type of “playing with my music” that I’d been waiting for, and having listened to this track a couple of dozen times I had an idea of how I wanted to refashion it: I wanted to focus the track around the lovely vocals and driving bass. While I love multi-layering complexity, I also love uncluttered, focused soundscapes. That’s what I’m aiming for with my mix. Here’s what I did:

• I removed the piano
• I moved background vocals to the left channel, and lead vocals to the right channel, so that they could have more of a duet feel.
• Initially I made the background vocals louder than the lead vocals, to see what that would sound like. But this sounded a bit too strong, so I turned the volume of these down, so that left and right vocals were now equal.
• I love the drum entrance, but keeping the drums at the same level for the whole mix, as we’re doing in this static mix exercise, made things feel too abrupt when certain elements appeared. And I didn’t want them vaguely-present, but not loud enough to listen to closely. So I stripped the drums out entirely.
• Without the piano there’s actually nothing to listen to for the first few bars, so I changed the min time in the URL to 6, to trim the silence
• I then tried to focus in on the timpani, which I’d set to about half of max volume and forgotten about while I worked on the more important elements. I couldn’t hear it very clearly, so I maxed out its volume and listened to its contribution to the mix. It only makes a few appearances – in a way, it’s a frame for the drums – and sounded a bit disjointed without some of the other embellishments that I’d removed, so I got rid of it too.
• I gave the mix another few spins, and decided that the bass was set a bit loud. So I turned the bass down a bit.

But the way that twitter have constructed this feature means that you can’t embed tweets in emails.

Why is this? You can’t display embedded tweets in emails because embedded tweets require:

1. JavaScript
2. iframes

Both of these features are blocked by most email clients for very sensible security reasons.

So an embedded tweet … ends up looking like this instead:

#ed473 this is really thought-provoking. @anildash on gender privilege & social networks: The Year I Didn’t RT Men. https://t.co/qZXGCdv545

— Bonnie Stewart (@bonstewart) February 13, 2014

This isn’t nearly as useful or pretty.

So what can I do instead if I want to embed a tweet in an email?

Lots of people simply share tweets by taking a screenshot of the message, and then pasting it to wherever is needed – eg into a powerpoint presentation. This solution could work in the context of sending an email: take a screenshot of the tweet you’re interested in, trim to size, upload somewhere, and add in to your email. For extra credit, make the image a hyperlink to the original tweet.

This solution isn’t ideal – it’s manual, laborious, and cannot be easily automated. The information about the tweet will become outdated as more people share or retweet it, or reply to it. Worse, it’s not possible to actually interact with the tweet in this setting, nor is this option at all accessible (unless you are very diligent with your alt text).

Okay then, come up with something smarter

I haven’t created anything better, I’m afraid, but here’s one idea to investigate:

If our aim is to take the useful features of embedded tweets – aesthetics and functionality – and to get them into email, I think that Twitter’s “Do you know…” emails could help:

You could take the inline HTML from these emails (I’ve copied the HTML code here) and use them to style up embedded tweets in an HTML email. You’d need to check in with twitter’s brand policies though, as this may not be within their terms of use.

Why I’ve chosen this track: Alcest is one of my favourite bands, if not my favourite. Alcest’s music sounds graceful, elegiac and transcendent. It feels like it’s coming from a context beyond every-day worries, concerns and feelings.

I’ve chosen a track from Alcest’s second album, Écailles de lune, called Sur l’océan couleur de fer. I particularly like its sorrowful and slightly mysterious lyrics (they work better in French, but I’ve pasted a translation below). I think the track is about 30% longer than it needs to be, but I still really like it.

Lyrics:

On The Iron-Coloured Ocean

On the iron-coloured ocean
Cried an immense choir
And those long screams whose insanity
Seem to pierce through Hell

And then death and silence,
Rising just like a black wall
…Sometimes, in the distance, could be seen
A swaying fire

Translation from lyrics translate.

The instruments and their location in the mix:

The main instrument is a slightly shimmering, mellow clean, electric guitar, played gently.

Just behind it in the mix sit the distant, tranquil, male vocals; wavering slightly but with some lovely sustain and slight vibrato on these longer notes.

There are other instruments too – a smooth, firm and unobtrusive bass guitar part in the centre of the mix, additional guitar and vocal parts in left and right channels, vocal harmonies, an acoustic guitar, backing vocals later and a symbol – but the track is defined by this interplay of vocal and guitar over the bass. The parts where the guitar climbs up one channel, and the vocals climb up another, and the two intertwine, are particularly pretty.

I enjoyed this exercise, and it’s helped me to be more attentive to what I’m listening to. I hadn’t noticed the acoustic guitar that is used towards the end of the track, for example.

1. A “bounce” is a session with only one interaction. Bounce visits are assigned a time on page and a visit duration of zero.
2. Goal conversions can only be counted once per visit. Ecommerce transactions can be counted multiple times per visit.
3. Visitor flow behavior report can show you events as well as page views.

So ideas are worthless if they aren’t shared.

I agree with Jesse Stommel when he says that:

Ideas need to connect with other ideas to grow and develop. Ideas can collide unexpectedly and lead to exciting and transformative results.

— Jesse Stommel (@Jessifer) February 2, 2014

Ideas need to connect with other ideas to grow and develop. Ideas can collide unexpectedly and lead to exciting and transformative results.

I don’t think that we should have to mine and labour to unearth a good idea, or to tease it out of some difficult academic prose.

The initial understanding of an idea should be simple. The challenge should be in exploring an idea’s implications, its limitations, its potential, and in applying it.

So what do I think are the attributes of a shareable idea?

A good idea is easy to understand, and portable.

share this quote on twitter

Present your ideas in a concise and intelligible fashion, and in a way that can be easily shared. This will make them more powerful and useful.

Most people going to a train station are interested in leaving, and aren’t there to meet someone who’s arriving. So we should focus on making “Finding out when my train leaves, and from which platform” as easy as possible; if necessary at the expense of “finding out when and where a train will arrive at this station”.

Hence the relative size of the departure and arrival boards.

Design is a decision about priorities, and as such is generally political. But having the courage to upset some internal stakeholders, you can focus on what’s most important to your customers, and therefore to your organisation and its bottom line.

I’ve added the following feeds to my RSS subscriptions, and will be reviewing their usefulness over the next couple of months:

• Smashing Magazine UX design category
  Feed address: http://rss1.smashingmagazine.com/feed/?f=uxdesign-std

• UX Magazine
  Feed address: feed://feeds.uxmag.com/uxm
  I’m slightly wary of this one – will the content just be stripped down teasers for paid material? Will the focus be too long-form and specific for my needs?

• Gizmodo’s user experience tag
  Feed address: http://gizmodo.com/tag/user-experience/rss
  This one was a little harder to find the RSS feed for. I didn’t have to do any filtering like I’ve done with other RSS feeds, but I did need to manually work out the URL. I followed this guidance to generate the URL I needed – you just need to put /rss at the end of the tag’s URL.

• UI Patterns
  Feed address: http://feeds.feedburner.com/UI-patterns-com
  UI design patterns are standard solutions to common design problems. This site discusses them and also collates examples.

• Elements of design
  Feed address: http://feeds.feedburner.com/ElementsOfDesign
  A showcase of different design patterns. The RSS feed itself shows latest additions to the example gallery.

Lots of people told us that the old website was hard to use, and sometimes overwhelming. We wanted to make our website easier to use, so that people can seek our help sooner, get behind our campaigns, and support us more generously.

We’ve tried to base our new site around people’s needs, so the design process was shaped by over 50 users at in-person workshops and user testing, and by over 100 people who participated online. We held workshops, surveys/tests, and user testing.

Workshops

We held workshops so that we could understand what people need from the Mind website. We wanted to find out what their top tasks were, and asked people to devise a site structure and labelling system that made sense to them, using card sorting. The aim of this was to make a site structure that isn’t based on organisational language or the way that we structure our work or teams internally.

We also invited people to discuss their feelings on the Mind brand, imagery, and use of colour online, and their feelings on how Mind’s values apply to the website.

We held the workshops in Bristol, London, and online.

People’s top tasks were:

1. Finding information about mental health
2. Knowing that I’m not alone
3. Finding support/services near me
4. Getting involved (mostly with fundraising)

People told us that they want a greater focus on their most important tasks. They wanted less clutter, a calmer and more sparing visual design, with a greater focus on people. They really resonated with Mind’s values: real, personal, compassionate, courageous.

Using this information alongside other information about our web visitors – eg from Google Analytics – we were able to draw up a clear, shared understanding of what our project was trying to achieve, and what the most important tasks were. This was crucial, as design is about prioritisation, and that often means telling people that their area is less important. Being upfront about your top tasks – and getting these tasks from users – helps with these discussions.

The in-person workshops took about 4 hours; the online workshop was condensed into an hour.

Our understanding of top tasks also drew from our Analytics data, which tells us things like over 80% of our traffic is to our information pages.

Online Survey

At the workshops, participants devised a site structure that made sense for them. We wanted to test if it made sense to other people too, so we tested it.

We carried out an online reverse card sort exercise to do this, using a tool called Treejack.

There were some areas of disagreement between the workshops (eg “get involved” vs “support us”, and “info & support” vs “help & advice”) and this exercise helped us decide between the competing options.

The exercise took each user about 10 minutes.

User Testing

Before we started building the final website, we built a sort of draft version: a prototype. It was just made with HTML and CSS. We did 1:1 user testing and online user testing of this.

By asking people to carry out a number of top tasks, and seeing what they found easy or hard, we were able to improve the prototype before building the final product.

We did this as part of following an agile methodology.

We encouraged participants to think aloud as they went so that we could better understand their thought processes. We took notes and for one of the sessions participants agreed to have their screen and faces recorded, using a tool called Silverback.

Some key learning points

• I’d start the recruitment process earlier, and plan a whole journey of engagement, rather than just inviting individuals to one-off events.
• We had more people sign up than we expected or had capacity for. (This did help us select a more diverse group for our London workshops.) I’d have liked to plan even more opportunities for wider engagement.

How did all this user involvement help us and help the people who took part?

By better understanding our visitors and their needs, we built a better website for them. By testing as we went, we could make sure that our designs were on track.

Participants gave very positive feedback, saying that they felt listened to, valued, and that their feelings and ideas were taken seriously.

What support did you offer?

• We tried to make sure everyone felt comfortable at workshops. We made sure that we had refreshments, a quiet space for people to go if they wanted a break, and we also asked people to fill in a wellbeing plan so that we could know how best to help them if they were distressed. (Most people didn’t fill in a wellbeing plan, but some did and it was very helpful in these cases.)
• We used supportive language and set an expectation of non-judgmental listening and sharing. We made sure not to ask leading questions, so that our results were useful.
• We tried to be clear in advance about what people were signing up for, and we told people that they could withdraw from activities at any time.

Tips for other people doing user involvement activities

• Be clear in your aims. For us, the value was clear: good agile digital project work relies on user engagement, so doing lots of user engagement was always a priority for achieving a good result. Mind takes user involvement seriously, and this made working in an agile, user-centred way very easy to achieve.
• Budget the time and money to carry out engagement activities. If you want the benefits, you need to plan the investment.
• Negotiate administrative support early, to help with booking venues, paying people for participation, etc.
• I pre-scripted everything non-reactionary in the online chat. i.e. all the welcome messages and pre-planned questions were written in advance so I could copy-paste. That left me free to focus on responding to the contributions of participants.

In an earlier post I looked at how charities can use digital to improve their provision of static information, and to provide proactive advice and support. But that is still quite a one-sided interaction, with organisations being in a position of power and enlightening their beneficiaries. What could happen if we go beyond the idea of service provision?

Organisations with social, political, cultural or economic aims want to change people and the world. We know our objectives, and so do our supporters. We could use digital to work together more closely to get things done. This is where digital gets really exciting, and it’s the part we haven’t figured out yet.

I propose that we take our implementation of the social web a step further. By enabling people to connect and plan and create ideas and drive change, digital can change the world. As charities, our job is to help facilitate this but not to lead it – at least, not to lead it in the way that we tend to think of leadership.

Rather, we’re thinking about nurturing spaces for self-organisation, collaboration, network formation, knowledge and content creation. Not telling people what to do, or what’s best for them, but letting them decide for themselves, and supporting without stifling. This will feel creative, exploratory, emergent, chaotic, and probably a bit risky and uncertain.

Facilitating this in a digital space is going to take something better than a walled platform owned by an organisation with corporate interests and an ability to control what people see. So Facebook probably isn’t the ideal space for this. But there may be a problem if any organisation – even a charity – is perceived to own a platform or network, more than the people that make up that network.

What does national campaigning for a charity look like now?

The experience of being involved in national campaigning for most charities is something like this:

1. Sign up for a mailing list.
2. Receive emails from time to time. These emails update you on what’s happening and ask you to take a range of different actions to help the organisation campaign.
3. When your interests and the request from the organisation intersect, you take action.

What do charity networks look like now?

Charity networks seem to largely exist offline, or at least in silos. Initiative seems to come from the centre – the national organisation – rather than arising out of the network. Opportunities for communication and agency between nodes in the network are limited.

UNICEF’s network seems to exist mainly in specific discrete geographical locations – see its Schools Campaign Network and its regional fundraising groups. Involvement in the international cause is channelled through a national organisation, which directs the activity of local groups.

Oxfam lets you choose to take part in an action, but doesn’t seem to be focused on letting you set direction. Just like with UNICEF, Oxfam’s campaign groups seem to mainly exist as an email address on a map, rather than a live conduit into a trouble-making world transforming network of energetic activists. I wonder if these groups could be in touch with each other more powerfully, and work together more ambitiously?

The Oxfam supporter panel seems like an opportunity to contribute to the organisation’s direction, but it looks quite one-sided. The organisation asks you about what you think about what it’s doing, so that it can decide on improvements to make to what it does:

“The Supporter Panel was set up to help us understand more about how supporters view our communications, fundraising ideas and campaign work, and enable us to make improvements. As a member of the Oxfam Supporter Panel, we’ll send you a short survey once a month, which will never take more than 15 minutes to complete. We’ll then use your opinions and insights to help steer the way we fundraise and campaign for a fairer world.”

What about Oxfam’s communities platform? It seems to be based around fundraising groups and get togethers. I wonder if this space could also be used as a hub to plan projects?

Amnesty UK has specialist networks but these seem to be quite one-directional: taking organisation-led actions, reading a blog, and receiving a newsletter. “Create a profile on our website to get involved with any of our networks and sign up to receive regular emails, actions and news on the issues that matter most to you.”

Again, like UNICEF and Oxfam, their local groups live in a directory, separate from each other and on separate websites. (To start a group, you have to fill in a word doc, pay a membership fee, and send it back to the central organisation.) Local organisations being separate makes logistical sense, but in some contexts we want to aggregate them and combine them so that they become something greater. We want the living network to be more than a list of names in a PDF.

What about Avaaz? That’s an international grassroots campaigning network, isn’t it?

My main experiences of Avaaz have been receiving an email telling you to fill in a petition. Upon filling it in, you’re taken to a page placing your action in exciting real-time global, and quite personal, context. This is cool, but this action was still directed by the organisation.

Avaaz has made moves towards empowering its network. I’ve been impressed to see the annual objective setting questionnaire, through which members contribute to the prioritisation of campaigning activity.

I’m also intrigued by the newer Start your Own Petition functionality. I don’t know how much organisational intervention or vetting there is on these petitions.

Metaphors of control and power; insiders and outsiders; sailing ships and motor boats.

Alice Jay, a campaigns director for Avaaz, explained that “We’re like a sailing boat, not a motor boat.” I think the metaphor here is that the organisation’s supporters are like the wind behind a sailing ship. The staff of the organisation have to work out the nuances of the route, they have to sort out the logistics and worry about avoiding risks, but the general direction of travel is powered by the supporters. Supporters don’t need to worry about finesse, or to have too much knowledge or skill – they just need to keep the wind blowing. A motor boat, by contrast, would be an organisation doing things under its own power – not dependent upon outside support.

But what if the direction of supporter enthusiasm is at variance to the presumably high-quality expert insights of people inside an organisation? What if the direction of popular enthusiasm seems like it’s heading for the rocks?

The situation can become tricky when people inside an organisation believe that they can achieve better results through more private negotiation, and perhaps giving some ground, as opposed to a more belligerent and uncompromising public stance. At the height of opposition to Andrew Lansley’s reform plans, campaigners sought signatures to force a parliamentary debate. But campaigners at 38 Degrees thought that this might not be the best way forward. “that ended up as a real shit storm. People were covering their Facebook wall asking why they weren’t supporting the existing petitions. It just doesn’t look very good.”

I prefer metaphors of sailing boats to metaphors of galleys. A sailing boat organisation is powered by supporters, moving towards a shared objective, with the details of the journey worked out by those on the ship. A galley organisation, by contrast, would have its supporters on board, but they wouldn’t set the direction, and would be told when to row.

Can we go beyond metaphors of insiders and outsiders?

The sailing boat metaphor still relies on a distinction between those inside the organisation and those outside it. Why is that necessary?

I think we can nuance this a bit. Within an organisation, we already divide responsibilities. We leave finance to the people who understand and are interested in it. The specialist digital team handles digital work, working with those inside the organisation who are also interested and skilled. So perhaps we can think of charities working with supporters in a similar way. Break down the charity’s work into different areas, and let everyone who is interested be a part of decision making and implementation. (Do we need to set a barrier for competence?)

So I don’t necessarily think that the outsider vs insider dichotomy is something we need to trip up on.

There must already be examples of charities moving away from central control towards a network-led approach

Movements like Anonymous, the SOPA blackout, the arab spring and Occupy suggest that people are able to come together, making use of digital channels, to achieve change. Which charities are doing this sort of thing already?

The draft of this blog was written in January 2013, but never brought to completion. I decided to finish it off and post it today.

Earlier in my career, when starting a new project I’d wonder: Where will I begin? How will I set up the right structures, ways of thinking?

In practice I’ve found that generally you inherit structure, process, documentation. And even if you don’t, you will come in to an established culture with a set of expectations and values (which you likely won’t fully understand straight away).

So actually the challenge is generally more about understanding what’s been done already and why, and working out what needs changing.

I’ve come in to organizations and been initially relieved at the structure and scaffolding in place, only later to realise that there might actually be some quite big problems with those cultural norms or established processes.

Did they start with user needs?
Is their conceptual approach to procurement the right one?
Are the governance ventricles predicated on the right foundations?
Are there cultural warning flags associated with some of the ways of working?

Worry less about what is missing and more about what is present but wrong. Your toolkit of techniques will help you build anything that is missing – the bigger challenge is assessing, challenging and changing what’s already there.

The digital landscape keeps changing. Google alters its search algorithm; new services and platforms are released; user devices and expectations evolve; web standards develop; a new version of HTML or CSS is released; the legislative context shifts as the EU’s cookie laws are implemented. This means that our practices – and our instincts – need to be forever questioned and improved.

The toolkit of digital competencies isn’t a static box of best practice to absorb once and for all. Certainly there are some good foundations to acquire – an inquisitive, analytical mind; instincts for storytelling and visual design; a conceptual understanding of how the internet and websites work; an enjoyment of play, discovery and exploration; a logical approach; knowledge of HTML, CSS and any of the other web languages.

But the core digital competencies are inquisitiveness and an enthusiasm to engage in discourse with other people trying to understand and shape the digital landscape. A restlessness and a desire to figure out how things can be done – or could be done – better. A desire to understand what works and what doesn’t – particularly when evidence is hard to come by, and an open evangelism to share the results.

I find that making sense of the world around me by producing openly shared media is the best way for me to learn. Publicly sharing my thoughts as I try to work through them helps me formulate them more clearly. Working like this is a vulnerable position to be in; and sharing living work-in-progress rather than thoughts that are ‘finished’ means that I won’t always be ‘right’. Rather, I hope to be stimulating and encouraging.

In my posts here, I hope, by learning aloud, and in conversation with others, to help develop our knowledge of digital communications.

This tweetup was actually focused more on ways of working than on income generation, and agile methodology was the big discussion point.

The PlumpyNut Challenge: making earned media work for fundraising – Deniz Hassan, Digital Marketing Manager at Merlin

We wanted to have fun with brand. We’ve spent 10 years sending photos of sad children to people and it’s not working anymore #nfptweetup

Merlin took a strategic move – no ‘sad Christmas campaign’. Went for a positive fr campaign in the new year #nfptweetup

Took a more low-fi, playful approach to content-creation. Shamelessly reached out to celebrities on twitter to encourage them to share the message.

£6k cost for Merlin’s Twitter campaign earned £40k in donations, 500 new supporters and millions of impressions: not too shabby #nfptweetup

Task Squad: Developing a self-sustaining commercial enterprise, by Sam Sparrow, Knowledge and Innovation Manager, and Damien Austin-Walker, Head of Digital, at Vinspired

Most recruitment consultants aren’t doing a great job of finding work for young people without experience, so it’s interesting to see Vinspired take the initiative here.

Task Squad “Post paid tasks, corporate errands and temporary jobs to a community of over 100,000 young people eager to get into the job market.”

This session was an interesting summary of Vinspired’s creation of a spin off micro-recruitment business. Good discussion around agile ways of working and how this can minimise risks, improve focus on objectives, and enhance organisational ways of working through better collaboration.

Lean, agile development process – start small and test! Less risk and easier for SMT to swallow if it fails! #nfptweetup

Test the concept before you build anything #nfptweetup

You learn as you implement. Very good point, you never stop learning when developing in start up mode. #NFPTweetup

Just Giving’s ways of working, by @annanthem

This slot continued the discussion of agile methodology, by looking at JustGiving’s processes.

I’m wary of using terms without pointing people to their origin, so I tweeted out a link to the agile manifesto.

Wondering what ‘agile’ is all about? Start with the v short agile manifesto ow.ly/1Z4oG2 #nfptweetup

“We don’t care who reports to who – we can’t afford to.” Nice. #NFPtweetup

Systematically taking down any barriers to co-creation that we find – @justgiving #nfptweetup

“We work out the ‘why’ of everything. Then, obsessively measure, and measure again” says @annanthem #NFPTweetup

I’m interested in the task of focusing on relevant, helpful material and reducing the mental overhead of filtering out irrelevant material. In this post I look at one way to bring a bit of order and clarity to your RSS feed subscriptions, so that they are more focused and useful.

To do this, we use Yahoo Pipes – an online tool that can manipulate information from around the web, and serve it to you in a range of formats.

You can make your own pipes, and so create your own processes for gathering and manipulating information and content.

Each pipe takes an input – or a range of inputs – does stuff with it – in some sort of sequence – and then presents an output.

Data flows through pipes, gets processed appropriately at different points, and goes down to the output. You can then do stuff with this output – in our case we want to subscribe to it as an RSS feed.

Let’s look at a couple of examples of how we can use yahoo pipes to filter content:

Filtering the SEOmoz feed, so that I only see Whiteboard Fridays material

One very important area of digital change is SEO.

Watching SEOmoz’s weekly Whiteboard Friday video is a great way to keep up-to-date with best practice, and to see experts figure out the implications of any changes in the search landscape.

I’m not a dedicated SEO professional, though, so I don’t really have time to read all the posts on SEOmoz. But I do want to keep up with Whiteboard Fridays. The SEOmoz RSS feed doesn’t know this, of course, so it just serves me all the content produced on the site.

Ideally I’d be able to filter this RSS feed, so that I just get my weekly delivery of Whiteboard Friday and nothing else. This is super-easy to do with Pipes:

In the Pipe editor, click on the ‘Sources’ heading, then drag ‘Fetch Feed’ into the editing area.

This is where we specify the RSS feed we want to filter. In this case we paste in http://feeds.feedburner.com/seomoz

Now it’s time to add in a filter, and set it to only let through Whiteboard Fridays posts. Under the ‘Operators’ heading, grab the ‘Filter’ module.

Connect the output of ‘Fetch Feed’ to this module. (Connect modules by dragging from the circle at the bottom of one (its output) to the circle at the top of another (its input). This creates a pipe. Make sure that the output from your final process is linked up to the output!)

Now it’s time to specify the filter we want. This looks a little intimidating at first, but if you click on the left-hand column Yahoo Pipes makes some suggestions. In this case we want the filter to only allow through content whose title contains ‘whiteboard friday’.

Have a look at this pipe and make a copy of your own – or remix it!

You can add multiple rules for each filter, which can give you lots of power. You could, for example, filter the NHS Choices Behind the Headlines feed so that you see material about ‘cancer’ or ‘dementia’ but nothing else.

You can filter by more than item.title. You can also filter by item.description, which seems to be the item’s content, or at least an excerpt.

Filtering the Telegraph Comment feed so that it just displays editorial content

I’m a leftie Guardian reader, but I want to make sure that I’m presented with a range of different views so that my ideas aren’t just informed by one way of thinking. It’s good practice to see different sides of an argument, to judge for oneself which is better, lest we fall into sloppy thinking. (I used to be an Economist subscriber, but keeping up with the volume of content it produced was punishing.)

I don’t want to read all the Comment pieces in the Telegraph, though. I’d rather just read the editorial. The naming convention on the Telegraph’s Comment feed seems to be that editorials are written by the author ‘Telegraph View’. So I set up a filter to only let through this content:

Head over to yahoo pipes to see this pipe in action.

Don’t kill the noise

A word of caution: you need noise in your life. Noise is new, unknown, challenging, thought-expanding. Processing it takes effort, and can be uncomfortable, but it has rewards. So don’t filter it out entirely. (Indeed, the whole point of the Telegraph filter is to increase the ideological noise in my life)

But sometimes you know what you want from a given RSS feed, and you don’t want to see anything beyond that. Be in control of the noise – this is the time to filter. Save your brainpower for filtering through the noise in areas where you’d like to expand your knowledge and understanding beyond what you currently think you need to know.

Add an Instrument track, and select HALion Sonic SE.

Choose a drum kit using the dropdown in the first channel.

Close this window.

If you’re using a MIDI controller, make sure that it’s plugged in and that the input is set to your MIDI input, if you’re using one. (e.g. an AKAI LDP8 in my case)

(If you’re not using a MIDI input you can still record drums, either with the on-screen keyboard or by drawing/writing the notes: create a blank track using the pencil, and then click on it to open up the note editor. But do that after you’ve set up a drum map, to make your life easier.)

Now set a drum map on the left-hand panel – GM Map.

This replaces the generic MIDI piano roll with named parts of the drum kit – much more useful. It also shows which MIDI note they correspond to.

Before setting a drum map:

After:

Open the HALion Sonic window up again.

Add the same drum instrument in channel 10 of the HALion Sonic. If you don’t do this you won’t get any sound after adding the drum map. (I have no idea why.)

Now you can either record live through your MIDI input, or you can program some percussion via your mouse.

Bonus – change which notes your external MIDI device sends to Cubase

The following instructions are for my AKAI LPD8 but are probably quite generic.

Open the LPD8 Editor (a separate programme).

Create a new preset with the notes you’re interested in from the drum map.

Then save it and press ‘Commit – Upload’.

It’ll now be on your LPD8, ready to use in Cubase.

Gamification is making an activity more like a game.

The definition of ‘game’ used here encompasses rewards, external markers of success, and (often) competition; but it also incorporates play, fun and collaboration.

Gamification defined – taking something that isn’t usually a game, but applying game principles #nfptweetup
— NFPtweetup (@NFPtweetup) March 13, 2013

Don’t just ‘gamify’ – you need a goal and to tailor towards it. Reward the behaviours you want to encourage; don’t go half-baked #nfptweetup
— Lisa Clavering (@LisaClavering) March 13, 2013

#nfptweetup Why do gamification? For fun. To solve problems. To reward people & take advantage of competitive behaviour
— Sara Wilcox (@S_Wilcox) March 13, 2013

A recommended book on gamification

@david_whitney, on the panel, made the original recommendation:

Yes a damn fine book RT @aarora17: This is the book #nfptweetup > Gamestorming: Playbook for Innovators, Rulebreakers + Changemakers
— Daniel Melbye (@DanielMelbye) March 13, 2013

Some examples of gamification

Cancer Research UK’s Dryathalon:

Dryathalon – Not drinking for a month, to raise money for Cancer Research UK.

Based on 12 months of research and preparation.

“Men are willing to support a charity as long as it is low involvement, derives personal benefit, and facilitates banter with friends.”

41% more was raised by people who signed up for the gamification, as opposed to those who only signed up for JustGiving accounts. (Aside: how much of this difference would we expect to see through normal stewardship?)

They used the JustGiving API to grab data each day for the overall leader board.

@CR_UK very cool use of @JustGiving API for #dryathlon to update leaderboard #everyonelovesaleaderboard #nfptweetup
— yasmin (@yasnav_LDN) March 13, 2013

They also used data from the API to send each dryathlete an email when they hit certain fundraising targets. Participants were sent a badge signifying the milestone. Messages thanked them for their support, and told them about the cause they were contributing to. (We don’t yet know the extent to which the cause message appealed to people, or if they were just interested in the personal/competitive milestone.)

Inspiration

Interviews

• We spoke to people with a range of political views, and carried out participant observation of their news-consumption activity.
• We learnt from experts: Framing and messaging from a third sector expert; Rules of social media and political discourse from a digital communications director.

Key quotes

“I don’t need stirring up. I want information and to make rational decisions. It doesn’t need to be heated.”

“It sometimes worries me that I only read news from a source that shares my views. Yet clearly there’s a whole other perspective, because the majority of people voted for Brexit. I don’t understand that, and I never will if I only ever consume news written by people like me. It would be helpful to see other perspectives, even if I don’t agree with them.”

“If I can understand what their argument is, I’m better placed to counter it. But that would need to be from a reliable source, even if I don’t agree with it.”

“Reading contrasting opinion pieces is helpful in forming ideas, and in developing new ways of thinking.”

Ideation

Key themes

• Desire for reasoned argument from both sides
• Time and efficiency
• Payment and paywalls

Ideas from Brainstorming

1. A twitter account – or accounts – sharing ideologically diverse content. Users can subscribe to high-quality content outside of their ideological viewpoint.
2. A news site where diverse content is submitted by users, grouped by theme and ranked for quality. This means that the most high-quality articles to represent a given ideological position on a given topic bubble to the top.

Storyboard overview

The person
Name: Brian Simmonds
Age: 29
Profession: Administrator for a small healthcare company

Prototyping

What did you prototype? What question(s) were you trying to answer?

We ​tested a clickable digital prototype​ made using the Balsamiq mockup tool.

The most important question to research was:

• “Will people actually choose to read an article of an ideological perspective that they disagree with?”

The most important secondary questions were:

• Will people understand our way of representing ideologically diverse content? What interface design approaches might be best?
• Do perceptions of quality actually cross ideological boundaries as our research suggested they could? In practical terms, will people positively rate an article that they ideologically disagree with?

What did you learn from testing your prototype?

• People looked at ideologically diverse content But this may have been ‘under duress’ as they knew they were being observed. To have confidence in this result, we would need to test in a more anonymous fashion, and in a more real-life context.
• Users were unclear what the ‘thumbs up’ and ‘thumbs down’ meant. Does a thumbs up signify endorsement, or liking, or was it a dispassionate quality measure saying that an article makes valid points and makes you think, even you don’t actually agree with it?
• Some users found the left-right split a bit binary. Could this be improved, to better show nuances of different positions? Is it actually useful to divide content up into different ideological sections of the page?

What might you consider for future iterations?

• Change the ‘thumbs up’ and ‘thumbs down’ to something more emotionally neutral. It will need to signify that this piece is well-constructed and reasoned, but must not imply liking.
• Explore whether community submissions are the best model for discovering new content.
• Explore pulling in content automatically from high-quality sources with different editorial perspectives.
• Explore visual and layout approaches to presenting diverse content relating to a given topic that will scale well to mobile devices. (The approach tested was desktop only, as it required more horizontal space than is available in a smartphone or tablet)
• Explore how to categorize content beyond a left-right binary. Consider in relation to the above point about visual design and layout.
• Allow people to add tweets, and embed these directly on the page.