I think the web/system dichotomy is also a major conflating factor for LLM discussions.
A “few hundred lines of code” in Rust or Haskell can be bumping into multiple issues LLM assisted coding struggles with. Moving a few buttons on a website with animations and stuff through multiple front end frameworks may reasonably generate 5-10x that much “code”, but of an entirely different calibre.
3,000 lines a day of well-formatted HTML template edits, paired with a reloadable website for rapid validation, is super digestible, while 300 lines of code per day into curl could be seen as reckless.
Exactly this. At work, I’ve seen front-end people generating probably 80% of their code because when you set aside framework churn, a lot of it is boilerplatey and borderline trivial (sorry). Meanwhile, the programmers working on the EV battery controller that uses proprietary everything and where a bug could cause an actual explosion are using LLMs as advanced linters and that’s it.
A “few hundred lines of code” in Rust or Haskell can be bumping into multiple issues LLM assisted coding struggles with. Moving a few buttons on a website with animations and stuff through multiple front end frameworks may reasonably generate 5-10x that much “code”, but of an entirely different calibre.
3,000 lines a day of well-formatted HTML template edits, paired with a reloadable website for rapid validation, is super digestible, while 300 lines of code per day into curl could be seen as reckless.