This site is a deliberate exercise in owning the whole stack. There is no web framework here, no content management system, and no third-party newsletter service. Posts are Markdown files in a git repository. A small Python script turns them into static HTML. That's the entire machine.
The reasoning is simple: for a personal blog, every one of those "conveniences" is a set of constraints you inherit and then fight the moment you want something slightly different. A static site generator you wrote in an afternoon does exactly what you need and nothing you don't — and when you want to change how a page looks, you change the page.
The one thing worth not building is the Markdown parser. Turning CommonMark into HTML correctly is a genuine project with no payoff, so that's a small, single-purpose library doing a bounded job. Everything else is a few hundred lines of code that fit in one head.
Two kinds of posts live here. Articles like this one get their own page and show up on the front page as an excerpt. Linked-list items are short — a headline that links straight out to something worth reading, plus a sentence or two on why. Both flow together in reverse-chronological order, the way a blog should.