Do code quality norms built for human maintainers still hold as models approach superhuman coding?

burny_tech · x · 2026-09-07

A discussion sparked by a repost asks whether our understanding of good code and architecture—shaped by the need for long-term human comprehension and maintenance—will remain valid as models reach superhuman capability and optimization pressure. While practices like hand-rolling random()/hash() were never sane, the author argues it's deeply unclear whether battle-tested, human-ergonomic conventions will stay optimal for models.

Related event: Do human-centered coding rules still hold in the age of superintelligent models?(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →