The target, of course, is writing correct code.

I find recursive functions over immutable values easier to prove to be correct than iterative loops over mutable storage.

The target is frankly to get paid.

Sometimes you're making a prototype. Sometimes you're making one off get some info out of data and never use it again.

Provably correct app that now costs 10x and is essentially yet another CRM or CRUD app is a failure because competition already delivered theirs 2 years ago and ate your market.

Go too far in another direction and you're living in house of cards with tech debt killing all velocity you had and users leaving over app having bugs or just being slow, while you're drowning in rotten spaghetti.

I do think that "doing the minimum to do the job at hand" is wrong approach for near everything that will live more than two or three months, but there is definitely threshold on other side of that where you're essentially just gilding the lilly and wasting time.

The key is to pick your battles, time is resource that thing that every other component of your infrastructure is going to use probably could use more of that time than random feature someone up top invented on a whim.

I find that I can prototype most quickly when I'm not wasting time debugging, and I tend to have the fewest bugs when writing recursive functions over immutable data and the result is either going to be correct or it's not going to pass the type checker.