Re: Draft permissive ballot option (was: GR: Ban LLM contributions from Debian)

Johannes Schauer Marin Rodrigues <[email protected]> Mon, 27 Jul 2026 17:25:50 +0200

Newsgroups	gmane.linux.debian.devel.vote
Message-ID	<178516595033.4088.12353627106655050780@localhost>

Hi,

Quoting Aigars Mahinovs (2026-07-27 17:04:23)
> Rejecting a class of contributions, without any reference to the *actual*
> quality or merit of the contribution sounds like a classic case of
> prejudice. Consider what your or public opinion would have been if the same
> package-specific policy were applied to all contributions from people from a
> specific continent or of a specific gender?

I'm not sure whether you are trolling but I'll assume you really believe this.

Consider the opposite case: suppose there was a machine creating perfect
software you have to constantly feed it baby kittens (yes, this is
intentionally an unrealistic example). In that case, I'd like Debian to say
that we do not accept contributions from somebody using this machine. And even
though we cannot really 100% check whether somebody is using that machine
behind our backs, I'd find it important for Debian to make a public statement
where we, as a project declare that we find the use of such a machine to
produce software unethical and that these contributions are not welcome.

> LLMs are not people, but the internal logic of the prejudice is exactly the
> same - you are assuming bad quality based on a very broad classificator
> applied on the source of the contribution.

It doesn't matter how good the output of that machine is (because we'd consider
the way the work was created to be wrong). In the end, the output of the
machine is just an interesting sequence of zeroes and ones. It's just software.
Its importance pales in contrast to the life of the baby kittens. Software is
not important at all in comparison no matter its quality.

> And you are also insulting the review capabilities of the actual person
> standing behind that submission.

Instead, I'd find it insulting if somebody would expect me to spend my valuable
free time on reviewing what comes out of such a machine automatically.

> That might sound acceptable to you. I am sure that there are many other
> people in Debian who would think otherwise.
> 
> Additionally it breaks the doocratic principle.

Aha.

I would say it breaks the doocratic principle if my own contributions become
insignificant because those with enough money to buy tokens are able to produce
a thousand times more output in a fraction of the time.

Why would I want to spend my time on a packaging patch if I can assume that
others in the team are willing to automate its creation? Then I'll just spend
my free time doing other things that I'm having fun with.

Thanks!

cheers, josch

signature.asc (application/pgp-signature, 833 B)