Subject: Re: termless copyright and patents
From: simo <s@ssimo.org>
Date: Sat, 07 Oct 2006 01:17:57 -0400

On Fri, 2006-10-06 at 21:17 -0700, Ben Tilly wrote:
> > They have permission under copyright law, that does not say anything
> > about permission to infringe patents. Of course to be able to
> > redistribute a GPL program you must at least grant permission to
> > exercise the claims embedded into the program you are distributing, but
> > I don't see any reasonable interpretation that forces the patent holder
> > to give up other claims as well. It is a problem (ie he will be denied
> > the right to distribute said modification as per GPLv2 section 7) of a
> > the downstream distributor if he, by changing the program, infringes in
> > these other patent claims.
> 
> Actually the license just says that they have permission.  It doesn't
> clarify under what law they have permission.

Sorry, but you are wrong here, in section 0, the GPLv2 clearly refers to
copyright law, and this is clear in many other references. The GPL is a
copyright license.

Furthermore, I think that permission to modify does not necessarily
imply other permissions.

> > > So when are you bound by the terms of the GPL?  Certainly if you
> > > release software you've agreed to it.  (But I'm unclear whether you're
> > > bound by that agreement, see my previous comment.)  But if you
> > > redistribute it, then under item 6 you "may not impose any further
> > > restrictions".
> >
> > Oh but you are not, you give full permission on the embodied claims
> > (this is exactly the implicit patent covenant).
> > But you are not required to give additional permissions (the other
> > patent claims).
> 
> Well the license says that anyone who follows the procedure has your
> permission to make modifications.  So if you say, "No, you can't make
> THAT modification because of my patent" then you're imposing a further
> restriction that is not in the license.  Which section 6 says you
> can't do.

I see your point and I would agree with you logically.
But law is not necessarily logical. I am not saying that you can argue
about this, only that a copyright license is not a contract or a
covenant.

To make it clear I don't think the text of the license itself implies
much about the patents claims (and I think this is the reason why Larry
say he can argue but the text doesn't say anything), I think the
implicit permission to use patent claims has more to do with the fact
that you distribute the program. Unfortunately I don't know the correct
expression in Legalese to express this kind of implicit agreement (I
know it in Italian not in English), I think Larry mentioned the first
sale doctrine (exhaustion doctrine ?).

> > > Again it does not say whether you'd be imposing those
> > > restrictions through contract law or patent law, so you're not allowed
> > > to use either to impose those restrictions.
> >
> > Since no restriction is imposed on the distributed software, as is, I
> > don't see the point.
> 
> If you say that certain modifications are not allowed, and the license
> says that those modifications are permitted, then you are imposing an
> additional restriction on the distributed software.

I know it seems contradictory, but I think we have to separate copyright
law from patent law. Again it is arguable, so I would not dismiss this
point completely, but I am not sure permissions given in the context of
copyright law are necessarily applicable to other laws.

Trying with an extreme example: there are laws that restrict what you
can do. For example you cannot modify and distribute software with the
intent to cause harm. This may be seen as a restriction on modification,
but that doesn't mean that nobody can distribute GPLed software because
of section 7. It is obvious that certain laws take precedence and there
is no point in arguing on what this license says in those cases.

> > > Since the license points out that running the program is outside the
> > > scope of the license, no permission has been granted to run a program
> > > that infringes on a patent of yours, and you're free to sue for that.
> >
> > I think it would be rather difficult to hold such a position in court,
> > as the act of distributing the software is a clear implicit grant of
> > usage of your patent (limited to the specific piece of software
> > distributed).
> 
> I know nothing about implicit grants of anything.  I'm just reading
> the text of the license in the most literal way I can.

Which is not sufficient, some provisions make sense in light of what is
admissible by law (either statutory or common law).
The context is often as important as the license itself.

Another extreme example: it is a criminal offense to kill someone.
But if you do it for personal defense than you can be completely
discharged. Context matters.

> Oh, let's say the ones that believe that they can distribute anything
> they want in binary form as long as they link against a specific
> interface in the Linux kernel.  And let's just say that I've known
> lawyers like that as well.  Most corporate lawyers that I know have
> more experience with, say, contract law than copyright law.  Not to
> say that they couldn't understand it, but copyrights aren't their
> bread and butter and it shows.

This is reasonable and true for my experience as well.

> It is often the case that competence in one area results in
> overconfidence in another.

Oh, this is soo true.

> I am not a lawyer either, but I've spent some of that same time.  And
> I'm not saying that a bunch of hackers can really understand it.  In
> fact I agree that they *don't* really understand it.  I'm just saying
> that their level of misunderstanding is closer to being on par with
> the misunderstandings that I've seen from companies than what you're
> likely to find from the lawyers in committee B.

Ok, this is clear now, and I can agree with this.

> I'm not saying that lawyers don't understand it.  But a lawyer who
> doesn't specialize in copyright law who is asked to review it can't be
> expected to get all of the subtle points.  For instance the presence
> and scope of a non-obvious implicit patent license would be easy to
> overlook.  And the general company may or may not understand that even
> if the lawyer does.

ok, point taken.

> Which leads into my point that there could be more controversy over
> the explicit patent grant in the GPL v3 than there was about the
> implicit grant in the GPL v2 because a lot of companies simply didn't
> realize that there was such a grant in the GPL v2.  And your comments
> about the competence of the lawyers in committee B is irrelevant to
> this point.

Oh I understand your point now. I am not sure I agree with your
conclusion, but I can see that it make sense.

Thanks!
Simo.