Ciao. I’ve not written personal weeknotes in a while because I’ve been churning out weeknotes for Extract every Friday afternoon, and that gets out of my system most of what I need to say about work.

The weeknote for this week’s work on Extract is in draft, and I’ll probably finish it tomorrow, but I spent all day Friday writing up our findings and insights for the senior leadership team. A whole day of thinking hard, typing, and suffering with the onset of the lurgy.

Reframe the problem

The beautiful thing about discovery and alpha is learning you had the wrong idea and that you need to change your mind.

There’s some interesting signals emerging. The main one is that the problem we thought was the problem possibly isn’t the real problem. We’d assumed the main problem was the time-consuming nature of drawing geospatial boundaries, but really it’s the lack of capacity for geospatial tasks in local planning authorities. They want to use an automated tool to ‘scale’ the GIS officers, which completely changes the proposition of the product.

Now we have to think about onboarding non-specialist users, teaching them ‘just enough’ about GIS, giving them ways to collaborate, creating governance processes and ways for people to sign-off outputs, and loads more. And we’re learning that the most valuable thing the AI does it not draw shapes, but look at old maps and find the same place on new maps – also known as georeferencing.

So maybe we don’t need to build all that stuff anyway. Maybe the tool could just output georeferenced GeoTIFF files that people can use to draw shapes in their own GIS tools.

Who knows. We’re lining up some discovery work for the new year to dive into that space.

My personal opinion is that it would be better for government to fund local government to hire more GIS officers. You know, create jobs. The demand is there, after all. If we’re being mission-driven about building 1.5 million homes and it turns out a lack of GIS officers is holding things back, surely you’d help the system hire more GIS officers?

The bosses need to learn more about digital

The other thing that’s been taking up my time recently is writing papers, submissions, briefs, etc. – notes for important people to read and make decisions on, or to read out to other important people in a negotiation. These either revolve around the specialist skills needed for a digital venture, the non-linear progression of a digital venture, or draft plans for launching a digital service while accepting handover from our delivery partner.

It has been a lesson in how to use plain language to describe the intricacies of digital ways of working and software development. The problem is, some times the steps involved only make sense to someone who understands how software works. For example, we can own and maintain a codebase, but if the application is deployed on someone else’s computers, we don’t yet have full ownership. In that case, a handover has been part-done but isn’t complete.

Then people start saying ‘What is a codebase? Can we use layman’s terms?’ and you start thinking No, actually, we have a need to be specific.

Anyway, it has highlighted a massive gap in the understanding of digital and its ways of working (in my opinion) at the most senior levels. It’s not a bad thing to not know these things in 2025, it’s normal. Huge swathes of the Civil Service are undergoing digital transformation and it’s a marathon, not a sprint. But the time to skill up is now: the future isn’t coming, it’s already here.

The good news is that government created a Digital Excellent programme for senior and executive leaders, and the content looks decent. Plus there’s the Public Digital book.

Working in the open

Last year, Emily asked me to join in writing this guide to working in the open, and I finally got it published.

More vegetables more of the time

Before the pandemic, we were in the habit of eating vegetarian meals in the week, only eating meat or fish on the weekends. We also ate more seasonally, as we bought our veg from Brockley Market each week. But as we’ve used supermarkets more over recent years, those habits disappeared.

We’ve decided to go back to it, getting a weekly dose of British organic veg from Abel & Cole. It’s been a joy biting into vegetables that taste of something again. Supermarket veg really is bad, the flavours are muted. Big yield in grams, low yield in flavour.

Bookmarks

• On A New Ecology of Tools, 5 mins. Rather enjoyed this, especially line about the ‘means of co-ordination’. A piece encouraging that instead of chasing a solitary AGI, we should design distributed, accountable systems that keep power and co-ordination plural.
• Overfitting and the problem with use cases, 7 mins. Wonderful destiny as this showed up in my feeds right as we were talking about designing wizards versus studios. Designing creative tools often suffers from ‘overfitting’, where solutions become too specialised and fail to adapt to diverse user needs. Instead of creating complex machines for specific tasks, it’s better to offer simple, versatile tools that users can apply in various ways.
• Is the pace of change really such a shock?, 6 mins. Something Michael shared from 2006 that still feels relevant (see my comments above on people needing to learn about digital). Tom Coates argues that the idea of sudden, dizzying change in media is overhyped. The shift from broadcast to digital has been slow and visible for years.
• Your job is to deliver code you have proven to work, 4 mins
• Why and how to retrospect?, 5 mins
• The holy framework war, 10 mins
• APIs all the way down, 7 mins
• I was wrong about Broken Britain, 5 mins
• The power crunch threatening America’s AI ambitions, 10 mins
• The AI boom is based on a fundamental mistake, 10 mins