The method aims to avoid technical debt from incremental edits.
what did he know
“oh, every morning at 1:00am our language model regenerates the whole codebase from scratch based on the current requirements document. it’s more reliable than trying to make incremental edits”
Many users reacted positively to language models regenerating full codebases daily from requirements documents, praising it as a prescient and eventually sensible idea, while others criticized related quotes as having aged poorly.
No Digg Deeper questions have been answered for this story yet.
Funny to read this.
I am right now in one window pretty much regenerating a codebase from scratch, in order to get to a new architectural layout while preserving functionality. That AI (codex) is in a race with another AI (claude) in a second window, which is trying to achieve a similar endpoint by doing a sequence of neutral refactors from the current codebase.
I have a third window keeping tabs on them. I'll compare results when they're both done.
it's been a busy few years in software!
@norvid_studies Sales guy at my company unironically said something like this, although it was specifically for a website for tech demos so it does kind of make sense, tailoring the website for the partner he’s targeting.
@norvid_studies
@sharifhsn it's going to make sense at some point! the only question is when
@norvid_studies Prescient. "our language model" sadly aged like milk
@norvid_studies before cgpt was public, even! lesswrongers stay winning
@norvid_studies @moultano Stochastic coding
@norvid_studies WOW
@alexisgallagher @norvid_studies Where will you post the comparison?
@norvid_studies huh
