Re: Questions about legal theory behind (L)GPL
On Sat, 8 Jan 2005 06:04:36 -0500, Nathanael Nerode
> Sorry this is so long and meandering...
By comparison with some of the things I write, it's a model of
linearity. :) But my response is necessarily long as well.
> I wrote:
> > There's a reason I used the analogy of "You may walk on my property,
> > provided you walk barefoot". It's different from "You may walk on my
> > property, provided you give me five dollars". Despite the formulation,
> > it actually amounts to "You may walk barefoot on my property."
> Michael Edwards wrote:
> >That's a poor analogy. It's more like "drink all the water you can
> >hold, wash your face, cool your feet; but leave a bottleful for
> >others, thank you kindly, Desert Pete"
> Well, I don't quite understand the details of your analogy (dangers of using
> song lyrics), but it appears flawed. It looks like you would have to get a
> bottle, fill it, and leave it, in order to satisfy the license. If it
> doesn't mean that :-), and instead means "you may take water, any amount such
> that there is less than 1 liter left", then it is the same as my analogy.
> (Well, for purposes of analysis.)
Sorry, I'll try to be clearer. Even if the return performance is
impossible without exercising rights only available under the license,
it's still performance. It's not a limitation on the scope of the
grant, like "you are permitted to copy verbatim, but not to take
excerpts" or "you are permitted to copy in writing, but not to perform
> There's absolutely nothing wrong with my analogy. :-) Please look at the
> GPL's restrictions carefully; essentially every one is a restriction on
> something which you are simply not permitted to do without the GPL's license
> grant. (For purposes of argument, ignore 3b and 3c, which are clearly
> contract-forming; 3a is considered to be the DFSG-free option.)
For what it's worth, the case law I've read (I don't have Nimmer or
the like handy) points out that a "copyright license" is really just
an enforceable promise not to pursue an infringement claim under
certain circumstances. In most causes of action involving a claim of
license violation, a court has to evaluate the facts under contract
standards first (was there a breach on the licensee's part that
triggered rescission of the grant of license? did the licensee's
conduct exceed the scope of the grant?) before any standard
appropriate to copyright law becomes relevant. Sun v. Microsoft is a
good example -- the appeals court vacated the injunction against
Microsoft, and sent the case back to the district court, precisely
because the district court had failed to follow this procedure.
The fact that the GPL licensee's obligations are, in some sense,
impossible to perform without use of the granted license doesn't
change this logic. It just contributes to the evidence that the
licensee knowingly accepted the terms of the GPL.
> My analogy is perfectly correct. However, I really don't know whether "You
> may walk barefoot on my property" forms a contract. It might, under certain
Not as far as I know (IANAL). It's not enforceable on you in any
ongoing way; you can come up, rip up the sign, and say "get lost".
The sign's having been present previously would be evidence of
innocent intent in defending against a trespassing charge, but that's
> >No. I'm claiming that both licensor and licensee are bound by the
> >terms of the contract once it is accepted. I cited the obligations to
> >distribute source code to recipients of binaries and to offer
> >copyright license in derivative works if they are distributed as
> >examples of conditional promises made by the licensee, amply
> >sufficient to form consideration in that direction according to the
> >cases I have cited.
> This is interesting. However, the cases you have cited do not appear to
> address situations like the GPL directly.
> (1) How does this analysis apply to people who do *not* distribute? They have
> few to no conditions on their behavior.
That's what Specht v. Netscape seems to say; anything you could do to
your copy of a free newspaper without violating the copyright holder's
rights, you can do to freely distributed software. Whether that
extends to modifying it and using it locally, perhaps integrated with
non-free code, I don't know; the case law I have found on "misuse" of
licensed software didn't suffer from lack of a valid contract.
> (2) People are only granted permission to distribute source code by the
> license. Given that, how do restrictions on the nature of the distribution
> allowed (source code or binary + source only; legal notices included; copy of
> GPL included; etc.) constitute consideration?
The precedent isn't perfect on consideration, as I said, but the
Planetary Motion opinion was eloquent on the subject of the advantages
of having one's code distributed under the GPL, even without
considering the two obligations of specific performance I named
(making and fulfilling offers to distribute the source code from which
any distributed binary version was built, and offering one's own
copyright in any modified and distributed version under the terms of
> (3) People are only granted permission to create derivative works by the
> license. Given that, how do restrictions on the nature of those derivative
> works constitute consideration?
Not restrictions on their nature. A quid pro quo when distributing them.
> (4) People are only granted permission to distribute derivative works by the
> license. Given that, how do restrictions on the nature of the distribution
> allowed constitute consideration?
We are not talking "you must keep every other line" here, we are
talking "you must offer to license, under the GPL, your copyright in
any changes you make and distribute". That's a valid conditional
> (Perhaps the consideration is the agreement on the part of the licensee not to
> break the law? Can that really constitute consideration?)
Not usually, I think.
> >I have cited cases elsewhere which
> >demonstrate, at least to my satisfaction with regard to US precedent,
> >that the GPL is an ordinary bilateral contract, not some sort of
> >unilateral gift of gerrymandered copyright territory.
> Unfortunately, none of the cases you've cited appear to address these points
> directly, or indeed at all. :-( Perhaps you could point to specific sections
> if you think they really do prove that there is consideration in such
GPL release isn't release into the public domain and is "use in
commerce", even without considering the licensee's obligation to grant
copyright license in modified and distributed versions (Planetary
Motion v. Techplosion); this isn't decisive but it's suggestive.
Conditional promises are good consideration (Fosson v. Palace
Waterland, entire Section 2, citing the California case law). I ought
to visit the law library anyway to get the classic cases out of
Restatements. If you can find it, check out Mattei v. Hopper 1958
(California Supreme Court, referenced in Fosson and in many law course
syllabi, such as
The Progress v. MySQL district court didn't seem to find it hard to
read a binding contract into the GPL, although the judge seems to have
ignored its attempts to render breach incurable, applying a "balance
of harms" standard typical of evaluations of breach of contract
situations. MySQL wasn't permitted to obtain injunctive relief by
yanking the copyright license, and the court declined to apply the
"automatic presumption of irreparable harm" standard of copyright law.
Does anyone have access to the full court transcript?
> >The offer of license is, prima facie, adequate
> >consideration from licensor to licensee.
> Yes, clearly.
> It seems pretty clear that the licensee is bound no matter what. Either the
> licensee is in a contract, or the licensee is the recipient of a limited gift
> of permissions; either way, the licensee can't exceed those permissions.
> The question of whether the licens*or* can unilaterally cancel permission is
> the interesting one. If it's a contract, the licensor can't. If it's a
> promise of a gift, it's a contract (note what was said on Groklaw about
> promising money to charities).
It isn't a promise of a gift, because it's not an offer to transfer
ownership. I know it seems like a small point, but a lot of law
hinges on the distinction between a non-exclusive copyright license (a
promise not to claim infringement by some person under some
circumstances) and a copyright assignment (a true transfer of
ownership). When Richard Brautigan gave his juvenilia to Edna Webster
with a written notice of copyright assignment, that was an enforceable
contract of gift.
> >To the extent that it purports to restrict the behavior of the
> >offeree, it can be another thing: an "illusory contract" and hence
> >unenforceable on the offeree. That's the conclusion that courts
> >usually reach when consideration is not found. In any case, a gift is
> >a transfer of ownership, and a non-exclusive copyright license is not;
> Isn't it? :-/ A non-exclusive license is still an item of value. Is there
> some precedent that such an issuance is not a gift?
Copyright ownership can be gifted. A non-exclusive license, or even
an exclusive license with limited scope, can't, in the sense that it's
not a form of ownership. For instance, it can't be further
transferred (plenty of case law on this; see Gardner v. Nike 2002 and
Morris v. Business Concepts 2001, both cited earlier). Basically, a
license isn't a thing that you can receive and then own, it's a
promise. It's only binding in the context of a valid contract.
> >courts in the US have consistently declined to find implicit transfers
> >of ownership
> ...of the copyright, which is not the same thing...
> >or of the right to sub-license,
> Some licenses explicitly grant the right to sub-license. Would that make a
The context of these remarks was to show that the things about
copyright that are transferable aren't transferable without a valid
written contract. Kind of a "belt and suspenders" argument;
non-exclusive license isn't a gift, and even if it were, that kind of
gift can only be done by contract.
> >and only a valid contract
> >can bind a copyright holder to issue a license.
> The question of unilaterally terminable licenses is an interesting and
> potentially worrisome one.
> (It's worth noting that in civil-law countries, contracts apparently don't
> require consideration.)
Standards of acceptance differ as well.
> What, however, if the license explicitly specifies "perpetual and
> irrevocable"? I don't think the state law affects that. The GPL probably
> should specify, as many licenses do.
Again, the point was that a license isn't a thing, it's part of the
terms of a contract and governed by applicable contract law (mostly
state law in the US). Under some circumstances, the duration of a
copyright license is limited to 35 years (in the sense that the author
retains a statutory right to terminate during the window of the next
five years) by 17 USC; but that's an example of a statutory override
of the ostensible terms of a contract, often addressed in so-called
"separability" clauses ("If any portion of this contract is deemed
unenforceable, the rest stands", etc.).
> ...and this...
> >he misses, however, the fact that copyright
> >licenses in the US can't be made perpetual. 17 USC 203 says that the
> >copyright holder can terminate a non-exclusive license after 35 years,
> (Actually, this is incorrect; the interest is held by the author or his/her
> personal heirs, *not* the copyright holder, and applies only to works not
> made for hire).
You are quite right; I read that section too quickly. But again, a
license is a creature of statute, and 17 USC is just setting out some
statutory termination provisions that override those in the written
contract, leaving the rest of the interpretation up to contract law.
> >even if the ostensible term of the license was longer (or perpetual);
> (This particular law is also not a problem because the notice requirements
> would be prohibitive for a general public license; notices must be served
> individually on each grantee.)
I suspect that a termination notice to a distro would be enforceable
even if that distro got a new copy from someone else. But you're
right, there's a limit to how many parties will receive notice in that
> Furthermore, the unilateral revocability question weighs much heavier on
> MIT/X11 and BSD licenses, with their weak and negative conditions, than it
> does on the GPL -- and code under those licenses is all over the free
> software world, including many GPL'ed works. So, frankly, the revocability
> of the GPL by the licensor is not worth worrying about until the question of
> the revocability of BSD and X11 licenses is dealt with!
I agree that the situation is in some ways worse with the MIT and BSD
licenses, since one would have to rely on the enforceability of a
public statement relied on in good faith instead of a real contract.
But the GPL issues are real and worrisome to me as well, especially
given "version 2 or later" type clauses which could trigger a fork if
version 3 is unpalatable.
> >With all due respect to you and to the FSF, trying to situate the
> >(L)GPL outside the realm of ordinary bilateral contract doesn't
> >simplify our lives.
> Unfortunately, this is a real question of fact and law. We don't get to
> decide the nature of the beast, judges do.
This comment was in response to:
>>> Based on the bit I quoted above, consideration may actually turn out to
>>> be irrelevant for most purposes. Which would simplify our lives. :-)
which was in turn commenting on the Groklaw article. I only meant to
disagree with you to the extent that the "promise to make a gift"
argument is a lot less sustainable than the "valid bilateral contract"
argument. I certainly agree that it's a question of fact and law, and
as such the only guide to what a judge will do is relevant case law
and statutes -- not the FSF's opinion, or yours, or mine.
> >Where precedents are thin on the ground, a court
> >has more scope to create novel law by reaching past the letter of a
> >statute or an agreement to find legislative intent or a principle of
> I would really expect there to be a lot of precedents regarding "gratuitous
> licenses" under common law; such licenses have existed as long as the law
> has, in both written and oral forms, and are a common source of disputes.
> I'm quite sure I remember something about them normally being revocable, but
> not necessarily revocable when people reasonably relied on them.
> Unfortunately, I simply don't know these precedents offhand.
Not much on "gratuitous copyright licenses", though, as far as I can
tell; and the use of "license" as a synonym for "permission" in less
relevant contexts (e. g., defense against trespassing charges)
obscures its meaning with regard to property rights.
Honest to Murgatroyd, I'd cite case law to the contrary if I could find it.
> I think the principles of equity are on the side of enforcing the license
> against the licensor if necessary: because a very large number of people will
> have come to depend on their licenses -- and legitimately believed them to be
> licenses they could rely on -- and a great deal of hurt would be caused to
> them; while the copyright holder is attempting to back out on a promise which
> he made with full understanding of the consequences, or in the case of
> successors, is attempting to collect a windfall based on no work.
Reliance as a substitute for consideration is said to be discussed in
the neighborhood of Restatement (Second) of Contracts sections
347-349, and is related to promissory estoppel; I'll check it out when
I have a chance. Does anyone have a URL for an unabridged version of
D&G Stout v. Bacardi 1991 (7th Circuit, 923 F.2d 566)?