Re: Termination clauses, was: Choice of venue

[Brian Thomas Sniffen <bts@alum.mit.edu>]

Sam Hartman <hartmans@debian.org> writes:

>>>>>> "Brian" == Brian Thomas Sniffen <bts@alum.mit.edu> writes:
>
>     Brian> Josh Triplett <josh.trip@verizon.net> writes:
>     >> I would agree entirely with that assessment.  I personally only
>     >> have a problem with the forced distribution clause, and not the
>     >> all-permissive license to the original developer.  I think the
>     >> requirement for an all-permissive license is obnoxious, but
>     >> still Free.
>
>     Brian> If it were only an all-permissive license contingent on
>     Brian> distribution *to that person*, that would be fine.  The
>     Brian> compulsive license even if that person doesn't have a copy
>     Brian> is not Free.
>
>     Brian> For example, let's say I give some software under the QPL
>     Brian> to Alice.  I also give it under the GPL to Bob.  Alice
>     Brian> doesn't propagate hers, and tells me this.  Bob does
>     Brian> propagate his.  It gets back to the initial developer,
>     Brian> INRIA.  Now INRIA has my code, with a permissive license I
>     Brian> didn't want to give them!
>
> I'm not sure there's anything wrong with this.  I'm certainly having a
> hard time finding a basis in the DFSG to justify this being non-free.
>
> I'm also having a hard time convincing myself that we desire this
> outcome to be non-free from some moral standpoint.
>
> Intuitively I believe that granting additional permissions in a
> license should not make an otherwise free license non-free.  

That seems like a reasonable intuition.

> Given the GPL we seem to have accepted the premise that a license
> may require all modifications to be distributed under the same
> license as the original work itself.

That also seems like a reasonable conclusion.  An interesting
corollary is that you can compel distribution to be under a more or
less restrictive license -- imagine an interactive program which
derives from a GPL'd library, for example, and see that it has a
notice-preservation clause which was null in the original.

So, for example, a license which was a copyleft, but granted a more
liberal license for educational purposes would be Free.  The copyleft
could even carry the requirement of the liberal educational-use license
along.  That would still be free, I think.  I have some doubt about
the requirement, but I *think* it's free.

The doubt comes from the idea that the license document grants me some
particular license.  It's free to compel me to pass along the freedoms
I had; it's not free to compel me to do other things.  I didn't have
the freedom of the liberal, educational-use-only license, so I
shouldn't have to pass that along.

On the other hand, we accept that in the GPL: I may have received an
interactive program and not been free to strip out the notice at
startup, but if I pass along just a library, then some later person
can make an interactive program which doesn't have a notice.

I think those two cases have to be in the same class, and so with some
slight reservation would call them both Free.

> The combined effect of these two statements seems to be that you can
> create a license that grants extra permissions to some class of
> people, even for all modifications that are distributed.  (I don't
> think the QPL is such a license, but the reason it is not such a
> license seems relatively easy to overcome.  It's still non-free for
> other reasons.)

To certain classes of people, yes.  I think that there are probably some
exceptions, though.  I think the original author is probably one of
these.  Hm.  Maybe if not combined with forced distribution at all,
it's not that bad.  It's still not something I'm thrilled about
releasing software under, but if I can make changes, give them to my
friends, and ignore the rest of the world, I suppose I don't have to
care.

I still don't think these should be thought of as extra permissions,
but rather as extra restrictions -- after all, that's why they're
GPL-incompatible.  They're also introducing new incompatibility
classes, which is of course a pain.

> Perhaps we should give up the intuition that additional rights granted
> should not make a license less free.  If I'm going to support doing
> that, I'd like to know why these outcomes are bad.  So far all I know
> is that there are outcomes that seem fine to me that you consider
> unacceptable.  

I don't think that's necessary.

> Note that even if we end up disagreeing on this issue, I'm still
> interested in helping draft GRs to address conclusions of the QPL
> discussion.  I think some of these issues are fairly important to
> actually bring to the project; they keep coming up again in multiple
> contexts and I'd like to know how the project as a whole feels because
> it would make evaluating licenses easier.  However, I think we have a bit of
> work understanding the arguments on both sides of the issue before we
> could have an informed vote.

I agree with everything you said here , but particularly the last.

-- 
Brian Sniffen                                       bts@alum.mit.edu