AI slop infiltrating #FOSS is so devastating.
Legendary FOSS projects are being ruined by their maintainers merging Claude PRs with HUGE LOC changes.
The number of projects untainted by this shit is getting vanishingly tiny.
I don't want slop in my software, thanks. PS: I'm a FOSS maintainer. Yes I know what it's like to maintain projects. Imo there's no excuse for huge structural LLM driven changes to your projects.
In 2026 and onward, where do we go for artisanal software?
How do we find software that's not been infiltrated by the inhuman scourge of AI?
I'm sick of this shit. I miss the days where FOSS was cool because you _knew_ someone had intentionally coded their shit, _they made it_ because they were passionate.
Now? We're left with software that someone made by proompting: "hi claude, make me software that will get me attention on github"
The level of pure, vitriolic hatered I have for this technology is unreal.
The costs are immense:
- Ecologically, it's poisoning our air and ruining our limited water supplies.
- Societally, we're being conditioned OUT of creating and the artistic process
- Economically it's proliferating inequality at epic scales, people are losing their jobs en masse
- Practically, the results aren't even good. Software feels worse now b/c companies are shitting out changes that outstrip their human capacity
@QueerMatters codeberg and sourcehut, I guess. Right there with ya
Nov 12, 2026, 10:28 PM@aburka @QueerMatters I've seen a lot of slop projects breaking the Codeberg ToS though.
@serak @aburka @QueerMatters At least those can be reported and the users banned. The TOS is not exactly unknown or unannounced. It's pretty clear what kind of person tries to sneak slop past the TOS.
Nov 12, 2026, 10:32 PM@MaddieM4 @aburka @QueerMatters Were there cases where projects reported as slop have been banned? I've reported a few but not seen any banned yet.
@QueerMatters This is why I wont be publishing any of the code for my bigger projects. As soon as someone comes up with a decent wall against the slop machine let me know.
What's more upsetting is that this isn't even an "AI issue". It's an issue with the projects. Why the fuck is it now suddenly acceptable to have such shitty PRs?!?
Any human trying to push something like that in would have gotten ridiculed to hell and back.
Also you can definitely make reviewable and ok-isch PRs using AI. I've experimented with that myself the last few days. It writes too verbose inline documentation though. But besides that? Quite happy so far.
Like look at my testing project here: https://github.com/agowa/nix-main-client/commits/master/
everything marked as "mcp-shellbox" is Qwens doing.
Besides too verbose docs the quality is not bad at all. And all of the PRs are atomic and reviewable.