Re: LLM usage in Debian -- ballot options comparison

--0000000000001d87c10657d273b1
Content-Type: text/plain; charset="UTF-8"

On Thu, 30 Jul 2026 at 12:57, Johannes Schauer Marin Rodrigues <
[email protected]> wrote:

> Going back to being more realistic:
> if I would now try to weigh costs versus benefits of use of genAI then on
> the
> one hand we have for the costs a long list containing items like: <snip>
>
> And on the list of benefits we have, very cynically speaking: <snip>
>
> Now you can go and you can challenge any of my points above.


I would love to go on for a long time on how many of those concerns only
really apply to a few USA-based companies, but the real AI landscape is far
wider and far more diverse and how there are already LLMs out there that do
not cause most of the problems above. But I do not believe that will be all
that constructive here. And the situation will evolve over time anyway
either affirming or rebuffing these concerns in ways that we can only guess
at for now.

But what I wanted to pick up was the analogy with the free software
movement. Free software did *not* succeed by absolutely rejecting all
commercial software and non-free hardware from the onset. Sure there was
the long running GNU and Hurd effort, but it did not really gain much
traction until more practically-focused projects appeared, like DJGPP for
running the GNU stack on proprietary DOS and Windows systems or Linux
itself, working on closed hardware including on hardware that requires
closed-source components  to work correctly. And Linux also allowing
closed-source software to work on it without much issue. GIMP and
OpenOffice would have been much harder to gain aceptance if they were
unable to open/save Photoshop and MS Office files. Even Debian includes
contrib, non-free and closed-source firmware and kernel modules because the
needs of our users are a priority for our project.

Debian has both the history of raising and enforcing ideological precedents
(see QT issue with KDE) and choosing the needs of its users over pure
ideology (non-free kernel modules). The AI decision in the end is going to
land somewhere on that spectrum. Where that will be, is still an open
question. How that will influence other projects and the IT community as a
whole is another. I don't think this aspect has been mentioned or
considered much in these discussions.

IMHO both full ban and full permission would have very little impact
outside Debian. But a more nuanced, policy-forward position in the middle
might have a discussion-shaping impact beyond the Debian project itself,
just like DFSG did back in its time. Failing to act will leave the
decisions in the hands of the very same corporations and oligarhs. Us
refusing to use their services will not really change their thining.
-- 
Best regards,
    Aigars Mahinovs

--0000000000001d87c10657d273b1
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr">On Thu, 30 Jul 2026 at 12:57, Johannes Sc=
hauer Marin Rodrigues &lt;<a href=3D"mailto:[email protected]">josch@debian.=
org</a>&gt; wrote:</div><div class=3D"gmail_quote gmail_quote_container"><b=
lockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-le=
ft:1px solid rgb(204,204,204);padding-left:1ex">Going back to being more re=
alistic:<br>
if I would now try to weigh costs versus benefits of use of genAI then on t=
he<br>
one hand we have for the costs a long list containing items like: &lt;snip&=
gt;<br>
<br>
And on the list of benefits we have, very cynically speaking: &lt;snip&gt;<=
br>
<br>
Now you can go and you can challenge any of my points above.</blockquote><d=
iv><br></div><div>I would love to go on for a long time on how many of thos=
e concerns only really apply to a few USA-based companies, but the real AI =
landscape is far wider and far more diverse and how there are already LLMs =
out there that do not cause=C2=A0most of the problems above. But I do not b=
elieve that will be all that constructive here. And the situation will evol=
ve over time anyway either affirming or rebuffing these concerns in ways th=
at we can only guess at for now.</div><div><br></div><div>But what I wanted=
 to pick up was the analogy with the free software movement. Free software =
did *not* succeed by absolutely rejecting all commercial=C2=A0software and =
non-free hardware from the onset. Sure there was the long running GNU and H=
urd effort, but it did not really gain much traction until more practically=
-focused projects appeared, like DJGPP for running the GNU stack on proprie=
tary DOS and Windows systems or Linux itself, working on closed hardware in=
cluding on hardware that requires=C2=A0 closed-source components=C2=A0 to w=
ork correctly. And Linux also allowing closed-source software to work on it=
 without much issue. GIMP and OpenOffice would have been much harder to gai=
n aceptance if they were unable to open/save Photoshop and MS Office files.=
 Even Debian includes contrib, non-free and closed-source firmware and kern=
el modules because the needs of our users are a priority for our project.</=
div><div><br></div><div>Debian has both the history of raising and enforcin=
g ideological precedents (see QT issue with KDE) and choosing the needs of =
its users over pure ideology (non-free kernel modules). The AI decision in =
the end is going to land somewhere on that spectrum. Where that will be, is=
 still an open question. How that will influence other projects and the IT =
community as a whole is another. I don&#39;t think this aspect has been men=
tioned or considered much in these discussions.</div></div><div><br></div><=
div>IMHO both full ban and full permission would have very little impact ou=
tside Debian. But a more nuanced, policy-forward position in the middle mig=
ht have a discussion-shaping impact beyond the Debian project itself, just =
like DFSG did back in its time. Failing to act will leave the decisions in =
the hands of the very same corporations and oligarhs. Us refusing to use th=
eir services will not really change their thining.</div><span class=3D"gmai=
l_signature_prefix">-- </span><br><div dir=3D"ltr" class=3D"gmail_signature=
"><div dir=3D"ltr">Best regards,<br>=C2=A0 =C2=A0 Aigars Mahinovs</div></di=
v></div>

--0000000000001d87c10657d273b1--