All notes
1 min read
Pragmatic architecture is a sequence of small bets
Big rewrites feel like the senior move. Smaller bets you can undo usually get you further.
The advice that ages worst is the kind that flatters the person giving it. Rewrite it clean. Get the domain model perfect. Build the platform that will finally unlock the roadmap. Those can work. They're also big bets.
A sequence of smaller ones is easier to live with. Pull out the decision, keep a shared boundary stable, add the test for a bug you already paid for. Shared packages only help if someone owns them and the API stays boring.
Sometimes you do need the migration. More often you need the thinner slice that tells you if the big one is even worth it.