2026-10-17

At the end of August, I did something I've dreamed of and went on safari in Tanzania! It was amazing, spectacular. I did not want to come home!

I'm still thinking about what I'm going to do with all the photos I took (so many photos!) but for now, here's a few to share. Not the best, just some that I liked flipping through the pile today.

2026-10-17

I've let my Mastodon hosting subscription lapse, and I'm not sure I'll ever bother signing up for an account on someone else's server. I decided I'd just rather follow people's blogs and posts via RSS. I don't really need to reply, and if I'm going to post stuff that never gets read, it might as well be here! :)

2026-10-16

This article is a really good summary of one of the things that has me most worried about AI-fueled development going into the long term.

https://blog.redezria.net/blog/the-gardener-vs-the-winnower/

I am more or less required to use AI in my current role, so I've been experimenting with it more, and this framing of the Winnower resonates.

For example, I find it genuinely useful to talk through an implementation plan for a feature before I start working on it, because I'm not really inventing anything new (building a business app - just data in and data out in various formats), so AI is good at surfacing questions that I may not have thought about properly, and would otherwise probably end up rethinking part way through the project.

But I have a lot of experience to judge the plan, ask questions of my own, navigate around gotchas that the AI just doesn't surface. I learned from errors and mistakes and rework over years. If I were picking up AI-assisted coding as a junior dev, all this would do is help me build a bad solution faster. I think we'll need a very strong culture of design thinking and devs QA testing their own work with a critical eye to edge cases and real-world data. It's something I'm still pondering.