Post #2167754

2026-05-07 14:45 UTC

AI slop is good, actually. Slop is what enables fast parallel experimentation. The etiquette and skill is understanding the boundaries of where slop exists and the extent to which it should be cleaned up and how. A few examples: I’m working on the internals of some system right now. The API and GUI of this thing is fully zero shame slop. It’s horrible. But it lets me focus on the core quality while shipping a usable piece of alpha quality software to testers (transparent about the slop frontend). Similarly, this system has plugins. We sent agents in Ralph loops overnight to generate dozens of plugins. The plugins are slop. The quality is bad. The plugin API/SDK is absolutely not done. But we can test a full GUI with a full plugin ecosystem. When we change the API, we can regenerate them all. The cost of change is just tokens, the velocity is incomparable to before. I built Terraform. We tested and shipped TF 0.1 with about 3 very weak providers. Because we ran out of time. Building was slow. And when we changed our SDK the cost was immense. Totally different today, 10 years later. Today, I would’ve slop generated 100 providers (again, with transparency and cleanup later, but just to prove it out). As an anti example, I would not PR this (without prior warning) to another project. I would not throw this onto customers without full review or transparency (as I’m already doing). I would not accept first pass slop. It’s almost never right. Slop is a tool. And like anything else it’s not blanket bad or good. The context is everything.

Replies (8)

@chfkch@ruhr.social 2026-05-07 14:58

@mitchellh@hachyderm.io I agree to a certain extent, but there aren't many people who know or accept those boundaries. In addition, alot of people do not posess the skill to have the skill to improve quality out of slop. Also, the industry marketing does not nourish the way you described, they promote some fire and forget tactics. That all is besides ethical and ecological concerns in regards to training and such.

Open ##2240168
@HugeGameArtGD@mastodon.gamedev.place 2026-05-07 16:48

@mitchellh@hachyderm.io Trying to polish and understand AI slop often takes more time than rewriting the code from scratch. It works in a specific case but might be completely broken otherwise, e. g. when used as a library. I'm not sure whether releases are meaningful at all if the API changes with every release. It looks like something that seems to work when it actually doesn't. It only makes sense when there's no time to do it properly imo.

Open ##2240169
@thejacenallen@mastodon.social 2026-05-07 17:47

@mitchellh@hachyderm.io lol no. Youre ignoring the environmental harms.

Open ##2240175
@_grahamdavison@hachyderm.io 2026-05-08 03:35

@mitchellh@hachyderm.io is that AI slop, though? To me, there’s a difference between “yeah, this is sketchy code that we’re going to fix up later” and “this code is something that I think is ready to go [narrator: it was not ready to go]” My examples come from receiving PRs on the AWS Terraform provider. There have been some surprisingly low quality PRs that waste our team’s time and that of the submitter. I think you addressed this as an etiquette issue. I would definitely call those PRs slop, but i wouldn’t call your experimentation slop

Open ##2240178
@martin@hachyderm.io 2026-05-08 06:35

@mitchellh@hachyderm.io I wouldn't call that slop until someone else has to deal with it. But I really like the use case!

Open ##2240179
@cdlhamma@hachyderm.io 2026-05-08 15:45

@mitchellh@hachyderm.io the terrifying thing is people who don’t understand the code or what it’s doing then releasing it in prod on the internet 😬

Open ##2240180
@Noisecolor@toot.community 2026-05-16 17:30

@mitchellh@hachyderm.io That's not slop, that's just integrating ai into the work process. Slop is low effort stuff.

Open ##2628007
@barubary@infosec.exchange 2026-05-16 18:12

@mitchellh@hachyderm.io The cost of change is just tokens haha, fuck you

Open ##2628008