Re: Archive section for open source models
A virus detection database is another prominent example where it is
absolutely clear there is no way to get consent from the copyright
holders of the original training data (the authors of millions of
viruses). And it is (I think) perfectly clear that we as a society do
*not* want the copyright of those virus authors to be restricting the
distribution freedom of virus detection databases - either direct ones
that contain snippets of actual virus binary code, nor statistical
model that have extracted clues about code structures that are likely
to appear in virus code and at the same time unlikely to appear in
non-virus code (which include both virus *and* non-virus software in
their training data sets).
On Thu, 8 May 2025 at 21:38, Sam Hartman <hartmans@debian.org> wrote:
>
> >>>>> "Clint" == Clint Adams <clint@debian.org> writes:
>
>
> >> Maybe the answer is that they're just too useful to the
> >> distribution to not package regardless of our opinions about
> >> whether they're free software. User experience and free software
> >> principles *are* often in tension and it's fine for us to shift
> >> that balance, in my opinion. But I guess I would have expected us
> >> to do that via a mechanism similar to non-free-firmware if we
> >> wanted to make it easy for users to use software that is
> >> OSAID-approved but not DFSG-free, at least if we have a lot of
> >> it.
>
> Clint> Maybe that is what we should be doing; I'm not sure.
>
> I'd support this, especially if
>
> 1) the name of the section did not make a negaive judgment about using
> it. Between the text in the social contract and the name non-free, we
> come across as making a judgment
> against using non-free software. People who do that are tolerated; we
> recognize that their needs exist, but we hope for a world where they are
> just able to use free software.
> We have (or had) programs like vrms to encourage people to use only free
> software.
> But under Russ's reasoning at least, I will never get a free spam
> classifier, or a free writing assistant. I'm fine with that but not okay
> making the same judgment against wanting spam classifiers or writing
> assistants that we do make about non-free software.
>
> 2) Providing some mechanism (allowing recommends is the obvious solution
> to me) so that programs in main can get their model data without having
> to download it from non-Debian sources. Being able to have a complete
> system with things like spam classifiers, OCR, text to speech, only
> given a Debian mirror is very important to me.
>
> important
>
--
Best regards,
Aigars Mahinovs
Reply to: