The choices were:
-
Choice 1: No LLM contributions to Debian via Social Contract
-
Choice 2: Allow AI-Assisted Contributions with conditions
-
Choice 3: Reject LLMs as far as practical, update Code of Conduct
-
Choice 4: Accept AI contributions for Debian specific work
-
Choice 5: Responsible Use of Generative AI
-
Choice 6: A cautious approach to generative AI
-
Choice 7: Debian is created by humans
-
Choice 8: Avoid the use of LLM: climate destruction is a deal breaker
“Responsible use of generative AI” is literally an oxymoron lmao. So fucking stupid.
Responsibly kick the child of omelas, just like, not too hard.
the status quo won, I can’t say I’m surprised.
A handy graph that someone else prepared makes the result a lot more digestible: https://people.debian.org/~lucas/gr-2026-002/results.png
this is just about the least digestible image I have ever clicked on.
Linux people when they say something is simple/trivial:
I cackled
Debian uses condorcet ranked choice voting https://en.wikipedia.org/wiki/Condorcet_method
Each line represents the results of a theoretical competition between the two.
it made sense after i stared at it long enough, but it was very funny to me when I clicked on it initially.
Aaaaaaaaaaaaaaa this bubble can’t pop soon enough I’m so Godamn tired of this bullshit
Damn bro I thought this was one of the safe distros
Edit: what are some more principled ones? I’ve used Gentoo many times but never got any benefit from having to regularly recompile everything. I’ve used guix and it was very close to great but had just enough pain points that I had to drop it to avoid spending too much time on admin. Fedora and derivatives led the way on allowing LLM results in the codebase.
There’s no getting around it at this point. LLMs are extremely useful tools in the hands of a proficient code-understander (some examples from my own experience here), and we’re already at the point where models that run on high-end laptops are useful coding assistants.
Even if openai and anthropic go bust tomorrow - which they might, and they should! - there’s still no putting the toothpaste back in the tube.
This is completely correct. I hate the conflation of openai/anthropic with all ai technology. AI companies will go broke in a few years as better on-prem open weight models will destroy their business case, (although whoever ends up owning all the datacenters will be fine since they are valuable in and of themselves) but that will have no impact on those open weight models continuing to exist and continuing to be useful.
Seems to boil down to “it’s still your shit, so you’re responsible for it”
As a maintainer it’s pretty difficult to reject a contribution on the basis that the contributor appears to be using LLMs irresponsibly without coming across as hostile when the project explicitly allows LLM contributions.
Even just asking whether a contribution was authored by LLMs is perceived as taking part in a witch hunt by some. And debian now doesn’t even require disclosure of AI use.
It’s debi-over
Time to hop to a new distro that isn’t Debian based
We already lost when code generated by LLMs was allowed into the kernel. All Linux distros are affected by upstream code, now.
Install gentoo
