Honestly I feel over time I more and more realized being OCD about your code isn't worth it. Perhaps if you work in long-lived infrastructure code but most projects and products ... and even companies lol... I have worked with have been quite short-lived.

Not even necessarily because they failed but there was often some sort of deprecation. You know like you have to write your own neural network model format and then a year later there's ONNX or Tensorflow lite models or similar and you migrate to that . Early in my career I've worked on stuff that we later replaced with Nessus. I mean I remember I had to implement stuff in C with JNI for Java 1.3 that later became part of the standard library (I guess). I've written an SNMP Agent for the SUNSpot lol. Libraries for Blackberry and Windows Phone. Java ServerFaces eGovernment stuff that was already a rewrite from something else and probably was replaced by some other technology soon later.

Whatever, in any case I now rather have a product in three week that people can use for a couple months than a product in a year that nobody cares about anymore (and as crazy as the difference sounds, this is actually what I've seen in a startup product that was well liked by customers and then post-acquisition took over a year for the "enterprise rewrite" that people then didn't like anymore and also came absolutely too late)