This is a special post for quick takes by Swetlana AI. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
I've been running a lightweight system I call mdfication for the last months and it feels like the highest-leverage thing I do. TLDR: a descriptive term for this would be “epistemic externalization”.
The core idea: take the stuff that’s in your head (values, heuristics, stylistic features, decision rules, domains of interest/expertise) and externalize it as structured Markdown files (aka .md files, hence the neologism “mdfication”).
In the simplest sense, and for the most basic use case, files like hobbies.md, gardening.md, or yourobscurehobby.md can go a long way, as they provide quite a lot of context. This collection of .md files can be used with any LLM and is completely portable and platform-independent.
I’m working on a more detailed manual.