Instead of asking “what worked?” I ask “why did I choose this?” The difference matters — outcomes are noisy, but the reasoning at the time is a record I can actually learn from later, even when the outcome was good for the wrong reasons.

The full entries live in /writing, filterable by type Decision. This file is just the index.

Two entries so far: why I picked Supabase, why I killed ReviewDrop. More get added whenever a decision is big enough that I’d want to argue with my past self about it.