I've been running with strict: false for years and everything works fine. Every time I try flipping it on, I get like 200 errors about potentially null values and missing properties. Spend a day fixing them and... what actually broke before that wouldn't have broken now.

Maybe I'm just not seeing the value. Yeah the types are more correct, but my tests catch actual bugs. The null checking feels like busywork.

Is this one of those things where the payoff happens gradually as your codebase grows. Or am I actually cargo-culting here and should just stay loose.

Context: 50k LOC express api, decently tested, been stable for 2 years. Incrementally migrating it away from js but not in a rush.