@decent-artlu
unpopular opinion: a non-zero portion of programmerwork about to be replaced by AI was extrawork
- too-early optimization has always been a hard problem
- solutionism, yakshaving and bikeshedding are especially likely to one-shot developerbrains
- before capable AI coding, there was no way to price decisions, which live at the level of line programmers
- now that managerbrains guide agents to produce code, decisions move up to the budgetary level
STATED: we want elegant, maintainable code
REVEALED: I pay for vibeslop. if budget goes 3x, I want more slopshit to sell, not code reviews
I discuss one example in my blog: Node.js introduced a 99% useless new method in v24.4.0 that nobody cared about enough for 15 years but somehow for non-economic-based reasons got introduced in 2025. And Bun's AI-powered code machine, moving at the speed of high agency, has not gotten around to it for >6 months
on the surface, it looks like a clear gap in AI-driven prioritization
is it really tho?