[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Questions about legal theory behind (L)GPL

On Tue, 18 Jan 2005 17:04:09 -0500, Raul Miller <moth@debian.org> wrote:
> I think you're confusing EULA with Copyright License.
> With copyright, the copyright holder grants license to the publisher
> to make copies and that's usually the end of the story.  There are
> exceptions, of course (developer tools being a fairly classic one --
> developer tools usually grant unlimited redistribution rights to some
> of the contained content).
> End User License agreements are something different and seem to be based
> more on contract law than on copyright law.

All copyright licenses are provisions in contracts, according to the
US case law that I have read.  My reading of Canada and UK case law,
and comments and citations from Batist in Belgium and Henning in
Germany (both civil law countries), seem to agree.  The existence of a
valid license is an affirmative defense against a claim of copyright
infringement, which is a statutory tort.  But the standards applied in
evaluating the validity of the license are those of applicable
contract law.  Fancy dancing in the preamble and citation of the
Statute of Anne don't get around this fact.

> Anyways, in this context it does make sense to consider the distributor
> as an agent of the publisher -- because the distributor has no license
> from the copyright holder, while the publisher does.  [In the classic
> commercial model, this is also the case for developer tools (while some of
> the content has an unlmited redistribution license, most of the content
> typically does not).]
> But under the GPL, the distributor gets a license from the copyright
> holder, so the distributor does not have to act as an agent for the
> publisher.
> In any event, the way I see it you're talking about traditions which were
> developed to deal with an issue which is not present in the context of
> the GPL.  And, furthermore, the GPL seems to contain explicit language
> conflicting with this application of that tradition.

I think there's a case that software ought to be protected by a
different kind of statutory intangible property right from those now
in existence (copyright, patent, and trademark).  (Trade secret law is
often spoken of in the same breath, but it doesn't create an
intangible property right.)  There are some jurisdictions in which
such statutes have been passed (Virginia and Maryland have enacted
UCITA).  So far, I would not say that they improve the situation.

Elsewhere, one is obliged to apply the traditional legal mechanisms,
and that means finding a legal theory which fits one's intent.  It's
normal for contract language to do an imperfect job and for a court to
have to construe something legally possible that fits the parties'
intent, modifying and striking passages that can't be conformed to
statute and precedent.  This has been used to convert oral exclusive
licenses (prohibited by statute) to non-exclusive licenses and to
detect "implied agency" (a separate theory from "apparent agency"; see
http://www.ihatoday.org/public/mangcare/petrovich.htm ) in
HMO-physician relationships, real estate agencies, and partnership

But I seem to have been wrong that implied agency to sublicense is
construable, at least under Ninth Circuit authority.  Gardner v. Nike
2002 declined to read an authority to sublicense into the part of the
1976 Act governing exclusive licenses, and concluded that "explicit
contractual language" was required to create authority to sublicense
without the copyright holder's consent.  On review, I don't think the
passive "automatically receives" language is strong enough to pass
this threshold.

> > In order for C to "automatically receive" a license under GPL from B
> > along with the physical (electronic) copy of the subject matter, B has
> > to have the authority to transfer license along with it.
> Ok... but let's first try to establish why C would need to receive it
> from B rather than from the original licensor (which is what the license
> says happens).
> So far you've only indicated that that's what happens with other licenses
> (which don't have this "receives a license from the original licensor"
> language).
> > Given that B is doing the copying, it seems natural to me to put B in
> > the position of the software publisher and to construe agency terms
> > from A to B.  Grammatically, this ties "from the original licensor"
> > to the noun "license" (C has A's permission, which B has agency to
> > grant) rather than the verb "receives".
> I'm not following you here at all.  A licensor is someone.  But your
> grammatical argument seems to argue that the license is that someone.
> Is the license document (or legal abstraction) the agent now?

No, I'm just distinguishing "license from the licensor" (a permission
which an agent might have an authority to grant) from "receives from
the licensor" (implying more strongly that the licensor is doing the
giving directly).  But if Gardner controls then it doesn't matter
because, while agency in general can be construed, authority to
sublicense copyright can't.

> You could argue that the recipient recieves the license from the original
> licensor via "B", which makes "B" be an agent for the purpose of passing
> the license on.  But the end result of that argument is still a license
> granted directly from "A" to "C".

ACK, except insofar as the grounds for breach of contract, vicarious
liability, and so forth could be radically affected by a finding of
implied contract between A and C versus implied agency between A and

> One thing I think you need to keep in mind: precedent is narrowly
> focussed on the issues of the case at hand.  A judge might suggest some
> general principles, but different licenses are going to be interpreted
> differently.
> A judge isn't going to override the terms of the GPL just because some
> different license was handled differently.  There has to be good reason
> for it.

Of course.  But to decide who has a contract with whom under the GPL,
a court is going to have to construe either contracts between
arbitrary parties without a vestige of meeting of the minds (I have a
separate contract with every contributor to the kernel?) or the
immediate licensor's agency to sublicense, under the GPL, other
copyright holders' rights in modified works.  Denied the latter
recourse (and the simpler conclusion that it's all in the public
domain and no contract need be found) by precedent, I guess a judge
would be obliged to construe the former in order to reach the parties'
intent, but I don't see her liking it.

I was going to add a rant about "version 2 or later" in connection
with the "contracts, contracts everywhere" interpretation, but I think
I'll skip it.

> > Realistically, I think this is the sort of thing where an appellate
> > judge might argue it through both ways, reach the same conclusion on
> > the facts of the case under discussion, and decline to establish a
> > precedent on which is the correct construction.  Sneaky, those judges.
> >  :)
> Yeah -- as a general rule, they like to keep their rulings very narrow.
> > [snip]
> > > You're right -- this does not seem to be written as an exposition of
> > > the legal foundations for the GPL.
> >
> > If my correspondence with licensing@fsf.org is any indication, there
> > aren't any.  No smiley on this one -- this is the crux of my
> > irritation with the FSF.  It's all very well for them to wave the 1709
> > Statute of Anne around when licensing code on which they hold
> > copyright, but they have encouraged the adoption of the GPL by many
> > other projects.  It's past time they showed their cards, either with
> > arguments based on contemporary law or an acknowledgment that it was
> > crafted in ignorance.
> If by "legal foundation" you mean "court precedent", I don't expect there
> to have been any.  Remember that only a few years earlier there wasn't
> any such legal foundation for copyrighting computer programs at all.
> Many legal experts felt that computer programs were uncopyrightable, some
> resorted to rather draconian contract arrangements, and some companies
> (such as IBM) gave out source code for free.
> If by "legal foundation" you mean "reasoning about legal issues which lead
> up to the GPL", I think that any attempt there would be a form of second
> guessing the GPL itself (and would also be marred by imperfect memories).

In this context, I mean "credible analysis of the legal issues".  Eben
Moglen and Bruce Perens were both publicly quoted in the lead-in to
the MySQL trial as being confident that MySQL would win a preliminary
injuction on the GPL issues.  They didn't.  There were several reasons
for this, which mostly add up to "the judge followed precedent in
applying copyright law standards where they were appropriate and
contract law standards where they were appropriate".  Neither the
FSF's subsequent public comments nor the correspondence I have had
with licensing@fsf.org addresses this point, nor do they seem to be
willing to adduce any modern legal precedent in any jurisdiction.

I am far from being the first to make this criticism.  See, for
instance, the comments in
http://www.oslawblog.com/2005/01/static-linking-gpl-and-lgpl.html by
people who cite actual legal precedents.  The emperor is a decent guy
and usually on the side of the angels, but I'm sorry to say that he
has no clothes.

- Michael

Reply to: