On Tue, 28 Jul 2026 at 11:59, Ian Jackson <[email protected]>
wrote:
> Charles Plessy writes ("Re: Ballot option: Call the Free Software
> community to develop Free AI."):
> > I would like to reach a point where we can agree to disagree without
> > having one side imposing its own terms on the other side.
>
> That's always nice when we can do it, of course. (Historically, we
> have been very bad at that, though.)
>
> My proposal is in fact trying to do that, as far as possible.
> Most of it is simply opinion. It provides for individual projects
> (package maintainer teams) to be able to set their own rules.
As you stated in your other email, your proposal only allows projects
setting other rules if they are *stricter* than your baseline rules. Not
the other way around.
The sticking point comes when we must interact with each other,
> as fellow members of a common project. [1]
>
> Let's take as an example racist material: we must either impose on
> racists a ban on them writing as they wish, or we must impose on
> racialised people that they must read racist material (or leave).
This analogy fails completely. Racist material is racist because of its
**content**. Not just because it was written by racist people. You refuse
to even look at LLM-generated content. There is nothing in good
LLM-generated content that differs from code written by hand. As an extrame
example, an LLM-generated patch removing one line of code freeing a
spinlock in the wrong place is (on binary level) no different from a patch
created by a human in vim. The content is not "racist" or "LLMish". The
creator of it is. We as a community do not tend to summarily reject
software or contributions of people *just* because of their personal or
political views, as long as that is not reflected *in the content* of their
submissions.
It pains me to explicitly spell it out, but a very close analogy is more
apt - some people in the project are refusing to read content (any content)
written by "black people" (LLMs or LLM-users) and want to establish rules
enabling them to have their own, segragated communities where no
"black-created" (LLM-created) content is allowed to enter. Maybe conceding
(as a compromise) that other communities may also exist that permit this
"tainted" content.
That is not a healthy state of mind to hold or to propogate.
And, it could be easily argued, that such approach violates the spirit of
DFSG Article 5 (No Discrimination Against Persons or Groups) and with that
is contrary to our social contract, specifically Article 4 in context of
"We will support the needs of our users for operation in many different
kinds of computing environments." as well as restricting the development of
Debian in ways that are contrary to "We will be guided by the needs of our
users and the free software community. We will place their interests first
in our priorities." by placing the interests of the Anti-LLM community
first instead.
--
Best regards,
Aigars Mahinovs
