I keep seeing this claim that software has suddenly become more buggy and unreliable since the advent of agentic coding, and I'm like… have folks just not been paying attention for the last decade?

Software has been getting progressively buggier and more unreliable every year for as long as I can remember

I'm not saying AI isn't increasing the rate of garbage code, but I do wonder how you can tell the difference between vibe coded shit and regular organic shit, because it smells the same to me

@nicklockwood every claim made about AI—that it's reducing employment, that it's making code worse, that it's increasing energy or water consumption—is backed by a trend that began before AI was productised.

Unless Eliza is to blame for all of this, of course.

@nicklockwood Software has been buggier and more unreliable since features features features, no time to look back came in. I think a lot of places do agile very, very badly (in the long term) and businesses reward based on moving metrics, not fixing the millions of papercuts.
@nicklockwood Yeah, surprisingly human coders can produce shit as well, lots of it actually …
@nicklockwood “Move fast and break things” didn’t include “fix the things you break”
@nicklockwood I think this is just a factor of there being more software in the world
@nicklockwood Vibe coding is just the ultimate expression of the move fast and break things culture.
@nicklockwood I guess people just prefer hand written bugs 😅
@jandamm artisanal hand-crafted slop
@nicklockwood software has gotten worse the bigger the push to deliver value to the shareholders has gotten.
@nicklockwood on the other hand, people report that the reduced (!) mental load has enabled them to build higher quality code: https://go.peterfriese.dev/coding-agents-for-production
Coding agents for production iOS: a senior engineer's setup for 2x the output

How I doubled my output as a senior iOS engineer using coding agents daily for 8 months.

Vova Ignatov

@nicklockwood This is why vibe coding is a good way of development for enterprise. The software was always a mess, buggy, and nobody could understand what someone before had implemented.

So actually, GenAI is simply producing legacy code from day one. Overblown with features management so desperately wanted and technical debt developers were never allowed to reduce.