What am I doing here?
For me, writing online is fundamentally unpredictable, the way so many other creative projects are. I have guesses and intentions, but they don't always mean much. Given that today is my 161st consecutive day of blogging, however, I have more data (both empirical and introspective) about what this blog actually is.
So! As far as I can tell, here's what I post:
- Notes about programming and software more generally, especially from the perspective of someone who (i) is programming every day, (ii) uses AI heavily, and (iii) has embarrassingly diverse professional experience, including at big companies.
- Notes intended both to help readers navigate the AI revolution and to chronicle it,1 especially...
- ...Humanistic reactions either to the experience of agentic software development or to commonly-held perspectives on it.
- "Rule of three" (in the Gwern sense) material that is more efficient to write publicly once than privately many times.2
- Notes on what I read, whether or not I expect those notes to be of wide interest. I like referring back to them, and I want to live in a world where people are writing about what they read.
- What I'm learning about spaced repetition, memorization, and trivia.
Note: that's what I'm doing, not why. (I partially address the latter here.) It makes more sense if I contrast it with what I'm not trying to do:
- Predict the future. I might make occasional guesses, but I am not an economist and don't know, for example, whether there will be a lot more Rust in production or whether the traditional project-manager role will go away. I'm here as a day-in-day-out programmer, and even though I keep track of the industry more broadly, the business of prediction requires a very different kind of sustained intellectual work.
- Comment on every major event and trend. Some people are great at flagging every significant thing that happens, but that's not what I'm doing.
- Make quantitative claims about which new tools are best. I'm more interested in other aspects of the craft, and anyway I'm quite skeptical specifically of the project of benchmarking AI tools.
And, as always, I'm speaking only for myself.