It’s been a year, huh? My experience of being alive in 2025 was: Go to bed way too late, wake up, pick up my phone to check email and news, get slammed with an enormous frying pan straight to the forehead, and then drink coffee and empty the dishwasher and get ready for my commute. You try to anticipate the frying pan, but it’s a different one every day, coming from a different angle. Meanwhile, we’re all still kind of awkwardly building our things, going to work, and watching Pluribus. Which means there’s a lot getting published, and a lot of good stuff to read. Let’s focus on that.

Rather than do my own roundup, I’ll just point to my favorite: The Bloomberg Businessweek jealousy list, which assembles articles the staff wishes they’d published themselves. They’re all great pieces, and many are not about business. They tend to be things I missed, not greatest hits; sometimes they point to pieces I originally hated, and force me to re-evaluate. You’ll find something good in there.

The industry’s book of the year was obviously Karen Hao’s exploration of OpenAI. It’s a pessimistic but valuable framing—and I learned a ton from it. Two further things helped me a lot this year: for obvious reasons, Simon Willison’s weblog about his ongoing journey to understand AI coding, and, maybe less obviously,  Laura Bullard’s long, well-researched article in WIRED about Peter Thiel’s obsession with the Antichrist. 

I keep thinking about this article, because I love tech and tend to have a naïve concept of what motivates people in our industry. When we see tech people doing things we don’t like, we tend to dismiss them as greedy and go back to our laptops. This was a good reminder that other people’s worldviews are just as complex as our own, and that it’s always worth finding out what someone else believes.

I spent a lot of time using AI tools to generate text, and most of it was predictably forgettable, but I did get ChatGPT to help me finally understand what Palantir does. Basically, they strap a graph database on a CRM and some dashboards, solve deploy for police departments, throw consultants in front of that, and use a government-friendly version of agile to actually ship product in months instead of years. In the end, it’s all React components talking to APIs and migrating databases for cops, but they gave things really alpha names like “Foundry” and “Gotham” and are really good at sales. For example, they sue customers who won’t let them pitch. I’ve probably written 400 proposals in my life, and that one never occurred to me.

Let’s end on the cheeriest note possible: Tom Whitwell’s list of what he learned this year is always spectacular, and you can share bits of it to annoy your family over the holidays. My favorite is that “a gram of silica gel has almost the same surface area as two basketball courts.” Two! Work that one out. One fun fact left off the list: Whitwell is a colossal, legendary synth nerd, and built one of the year’s most interesting and well-regarded experimental boopers.

We’re grateful to our readers, our correspondents, the AI haters keeping us honest, and most of all, the clients who trust us to build their software and indirectly fund our newsletter hobby. See you next year! Hopefully in person. We like that.