Yep, rarely do devs add more code just for fun. They’ll do it when the incentives are misaligned (e.g. the new modern trend of LOC changed), or doing stuff like “speed loops” to appease really shit management.
Usually when there’s some obscure extra rule in the code it’s because there was some hardware fault, an undocumented use case, a complaint from a key client, or an intermittent failure.
Anyways, I’ll be sitting over here with some popcorn.
Hit the nail on the head there.
Every bit of weird code the LLM wants to simplify is a workaround for some old bug.
See also that old Joel Spolsky essay about how rewrites reimplement the bugs the legacy system fixed long ago.
Yep, rarely do devs add more code just for fun. They’ll do it when the incentives are misaligned (e.g. the new modern trend of LOC changed), or doing stuff like “speed loops” to appease really shit management.
Usually when there’s some obscure extra rule in the code it’s because there was some hardware fault, an undocumented use case, a complaint from a key client, or an intermittent failure.
Anyways, I’ll be sitting over here with some popcorn.