← cd ~/writing
amine@web ~/writing $ cat wrong-optimization.md

why i stopped optimizing for the wrong thing

2026-02-28 reflection 4 min read

for a long time i measured my output in commits, PRs merged, features shipped. the more the better. velocity was the metric and i was going to win at it.

turns out velocity without direction is just vibration.

the realization

it hit me during a production incident at 3am. we'd shipped twelve features that month. we'd also shipped twelve new ways for the system to break. every feature was a surface area expansion, and we were expanding into the void without a map.

the most productive thing i did that quarter was delete 4,000 lines of code and replace them with 300 lines that actually solved the problem. the system got faster. the on-call got quieter. the team got happier.

what i optimize for now

none of this is revolutionary. but knowing it and doing it are different things.

← prev notes on distributed consensus
next → on building systems that don't fall over