Re: GPL, "license upgrades", and the obligation to offer source code
On 5/11/05, Glenn Maynard <glenn@zewt.org> wrote:
> On Tue, May 10, 2005 at 06:52:41PM -0700, Michael K. Edwards wrote:
> > work you wrote?" And by extension, "Why should a court feel obliged
> > to grant any copyright holder's request to enjoin Does and Roes from
> > claiming rights to his work under license terms that he did not have
> > the option of reviewing at the time he last contributed to the work --
> > even if he consented to the license upgrade clause at the time?"
>
> Most simply, I guess I'm of the mindset that I should be allowed to place
> my work under whichever terms I please. For example, I strongly dislike
> the law which allows revocation of all past copyright licenses after ~30?
> years, overriding even written contracts and transfer of copyright. No
> matter how strongly I wish to Free my work and give it to the world, it's
> impossible to truly do so, since my heirs, if I live such a short life,
> can invoke that law and override my wishes completely. I'm not permitted
> to truly Free my work, because the law's condescendence doesn't allow it.
> I don't go so far as to claim that I wish the right to sign myself into
> slavery, but I do wish to be able to license my works as completely freely
> as I wish.
I agree with your wish, with respect to certain of my works.
Unfortunately, under 17 USC, the only way to avoid transfer of my
termination interest in copyright assignments and licenses is for my
work to have been a "work made for hire". That's one reason why
movies are developed under the auspices of a shell company formed at
the outset; I believe that in most cases the executive producers
require that there be a clear "work-made-for-hire" escape from 17 USC
203 for the work of any living creative contributor.
On the other hand, courts are not absolutely mad, and I would not be
surprised if a court would rule that a work which the author clearly
attempted to convey into the public domain at the time of first
publication is in fact public domain, heirs or no heirs. It's not as
if 17 USC says "copyright expiration is the one and only way that a
work can become part of the public domain". That's a term with a long
legislative and judicial history, and there's room for a court (of
law) to correct an unintended consequence of the 1976 Act's removal of
the notice requirement.
But I'm afraid I see no way of disclaiming copyright on a work once it
has been published. This may become interesting with respect to, say,
Phil Zimmerman's work within our lifetimes.
> > I constructed the "DNPLv2/v3" strawman as an example of how a "license
> > steward" can effectively close off people's access to their own work
> > with a little bit of technical maneuvering but without explicitly
> > biased language. It is often said that ignorance of the law is no
> > excuse, but there ought to be a limit to just how big a rights grab
> > you can pre-authorize sight unseen.
>
> Well, it's not quite a "rights grab": they're giving themselves rights,
> but not diminishing those of others. I guess I'm not a big subscriber
> to the GPL these days, so it doesn't particularly bother me that someone
> might be able to "take my work proprietary", since I don't lose anything
> in the process, but I can see a copyleft-minded person taking strong
> objection to it.
My example attempted to exhibit a case in which others' ability to go
on using the extant works collapses down to printing them and using
them for wallpaper. Just because they can still exercise DNPLv2
rights to copy and modify doesn't mean they can do anything useful
with the data next year without accepting DNPLv3.
> > Like the right to renegotiate the terms of a copyright license after a
> > certain number of years, the right to decline terms that you haven't
> > seen ought to be inalienable.
>
> (This, by the way, is the "right" which I objected to above, which I noticed
> on rereading that you referred to.)
It's a classic "widows and orphans" exception to the copyright
monopoly, and exists to put some kind of limit on how much damage a
lack a foresight can do to an author's residual interest in his or her
works. If you had negotiated a fixed royalty per copy for your magnum
opus in 1969, and it still had commercial value in 2004, wouldn't you
want a shot at renegotiating to handle inflation?
> > Oh, a bounded set of options is fine, like "my agent can commit me to
> > pet any fixed number of cats, not to exceed three, per copy I
> > distribute"; but not "my agent can bind me to do, or not do, anything
> > he pleases". And remember that a copyright license is just a
> > contractual promise not to sue. Blanket authorization to alter
> > license terms on a work is not just authority to adjust the scope of
> > the license but authority to add arbitrarily onerous conditions on the
> > exercise of my own copyright. Not to force copyright assignment,
> > though; that's among the things which a US court cannot construe
> > without positive written evidence.
>
> It seems like a lot of your arguments have a feel of "revokation" behind
> them: you speak of these upgrade clauses as if the original terms are no
> longer available. There's nothing like copyright assignment--I can still
> offer the work under whatever terms I want (regardless of additional terms
> having been offered). License terms on the work are not being "altered";
> they're being amended in a way that can only ever result in a net loosening
> of restrictions, since the original terms always remain valid. (We might
> disagree on the significance of that; given that I tend to X11-license my
> work, if you copyleft, that's an understandable disagreement of philosophy.)
Funny you mention the X11 license. Do you think the license change on
XFree86 upstream has disrupted the lives of people who must fork, or
switch forks, or whatever if they want their incremental contributions
to be usable with (the FSF's interpretation of) the GPL?
The "derivative work" approach makes sense when the jumps are big --
first edition to second edition to musical to screenplay to movie. It
doesn't represent the workflow of modern software, which (as my
somewhat off-topic "why copyright?" post hyperbolizes) is why licenses
with a myopic focus on copyright and a highly abusable "upgrade"
mechanism don't cut it. And that's not just GPL "v2 or later"; it's
MIT/X11 and BSD too.
> I guess that's the essense of my confusion, though: upgrade clauses never
> allow additional restrictions to be forced upon a work, since you (the
> licensee) are always free to ignore the option.
>
> (FWIW, I'm not sure that I buy "contractual promise not to sue"; it seems
> like a reasonable analogy, but not really an accurate one. If I incorrectly
> sued you for distributing my GPL-licensed work, you might countersue for
> the cost to you--but I don't think you'd claim that my lawsuit was violation
> of contract. Deja vu--I suspect we've had this exchange before.)
Horse beaten elsewhere; thread at
http://lists.debian.org/debian-legal/2005/01/msg00176.html .
> As an aside: while I'm sure the "similar in spirit" clause in the GPL is
> too vague to satisfy your "bounded set of options" in general, if a GPLv3
> was released, and it actually was, to a typical, reasonable Free Software
> developer, similar in spirit[1], would you object to its use, given that
> the alternative is the existance of some fatal loophole, which otherwise
> (due to the very nature of the GPL) is essentially impossible to fix? It
>
> [1] For example, if the GPL came to court with a licensor claiming the
> "modified files to carry" clause means modifications must be painted on your
> car, and a court agreed, rendering all GPL software practically unmodifiable.
> The FSF fixes this, by correcting the ambiguity. I don't know of any *real*
> loopholes in the GPL to offer, but note that I'm specifically referring to a
> real, clear-cut loophole such as an ambiguity, and not to issues like the
> "ASP loophole" being "fixed" by adding new, controversial restrictions. (I
> suspect that if the FSF releases a GPLv3, it *will* include restrictions that
> people won't like applied to their software, but let's limit ourselves to
> the smallest possible change, "fixing a bug", which is likely to be accepted
> by the biggest group.)
The GPL is an offer of contract, not a statute! In any particular
court case, determinations about what contract terms the parties
believed they were accepting are issues of fact, not of law. They
don't have any value as a precedent, except insofar as the parties in
a later case can be demonstrated to have accepted them as
clarifications of the applicable law. So there's no need for a "bug
fix" for some stupid interpretation that somebody managed to put over
on a district judge.
Did you notice that the GPL not only doesn't contain an integration
clause ("this supersedes any existing agreements, written or oral,
etc."), it explicitly contemplates parallel agreements ("you may offer
a warranty")? This means that any public statement, any evidence of
conduct, any grounds for showing that the person you're in a dispute
with is bullshitting when they say that's what they understood it to
mean is allowable in court.
There are only two things the "v2 or later" clause can be used to do.
One is to allow a subset of copyright holders to fork under the GPLv3
when it comes out, without the consent of the rest. The other is to
allow a licensee to claim rights under the GPLv3 that the language of
the GPLv2 clearly can't be construed to allow. It's bad public policy
to permit either of those consequences, which is why a court is
(IANAL) very likely to agree with anyone who claims (in a court case
where it becomes relevant) that the "v2 or later" clause should be
struck during construction.
Cheers,
- Michael
Reply to: