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

Re: GPL and linking



Fine.  I have been goaded into rebutting this specimen.

On 5/11/05, Raul Miller <moth.debian@gmail.com> wrote:
> I'm disputing an argument which seems to require a number of such fine points.
> It is difficult for me to raise such disputes without mentioning the the points
> themselves.
> 
> However, I can present my point of view without resorting to this argument:
> 
> Let's say that we have a court case which involves some contested GPLed work.
> How should we proceed?
> 
> First, let's consider a work which doesn't have any binaries.  This would be no
> different from any other copyright case -- you have to show that the work in
> question is copyrighted under the GPL, and you'd have to show that the terms
> of the GPL are being violated.  This should be relatively simple, and we can
> neglect sections 2 and 3 (which are clearly being complied with if the rest of
> the license is being followed).

Nope.  Under US law at least (IANALIAJ), you'd have to show:

1.  that you, yourself, hold a valid registered copyright to a
specific portion of the copyrightable expression in a particular work
A; and

2.  that a portion of your contribution to A has been copied to work
B, using the Computer Associates v. Altai
abstraction-filtration-comparison standard, and that the amount of
_copyrightable_ material that has been copied exceeds "de minimis";
and

3.  that the distributor of B does not have license from you to copy
that material from A to B, or that the distributor's conduct exceeds
the scope of the license (e. g. creation of a derivative work when the
license extends only to verbatim copies), or that the license has been
terminated for material breach not otherwise reparable under the
applicable contract law standard;

After which, the distributor of B has an opportunity to demonstrate:

4.  that some statutory or judicially created affirmative defense,
such as "fair use", justifies the distributor's conduct; or

5.  that public policy or a principle of equity demands that the
distributor's conduct be sanctioned despite the unavailability of any
defense under current law.

Then, and only then, you may be entitled to some relief under
copyright law.  That relief may be as little as one dollar of damages.

> Now let's imagine that we've got a case which involves binaries.  What do we
> have to do?
> 
> First, we need exhibits: the sources, and the binaries.  Out of
> consideration for
> the court, we want to pick examples which are as simple as possible while
> representing all of the important contested issues.  So let's imagine we have
> Exhibit A (the sources) and Exhibit B (the binary).  [We need to also show that
> this binary is representative of something which is being distributed,
> but that's
> not really different from what you have to do in other copyright cases, so I'll
> ignore that part.]
> 
> Second, we need to show that Exhibit B is derived from Exhibit A.  Again, we
> want to present this in a simple and easily understandable form, and we
> want to also present complete information.
> 
> Once we've shown that B is derived from A, we can start examining the terms
> of the GPL to make sure that they are being followed.
> 
> For example, let's say now that we're the defending party, and we want to show
> that the mere aggregation clause applies.  To do this, we would show that
> the disputed work could be replaced by something trivial, and that having done
> so, the program is still the same program -- we might do this by showing that
> it still has the same behavior.

This has no bearing on the definition of "work based on the Program"
or of "mere aggregation" or on any other relevant ambiguity in the
construction of the contract.  The only sense in which I can see it
having any relevance is if the only theory under which B is derived
from A is "characters and mise en scene", as in Micro Star v. FormGen;
in which case the existence of a reasonable alternative to A, under
which B does something similarly useful, may be a successful defense.

> Switching sides again, if someone asserted that the mere aggregation clause
> applied, and used program behavior to make that assertion, and I believed that
> mere aggregation did not apply, I would show how the program failed to
> operate in some independent context, with the disputed section removed.
> 
> Is that clear enough?

Clear as mud.  What do you mean, "used program behavior to make that
assertion"?  Even though this is an offer of contract, its drafter
harps on one copyright note.  "Mere aggregation" is a phrase with no
legal meaning (there is a single usage of this phrase in all of the
appellate law accessible to FindLaw, and it refers to members of a
school prayer club).  According to FindLaw, Merriam-Webster's
Dictionary of Law defines "aggregation" as:

1: the collecting of individual units (as damages) into a whole

2: a collection of separate parts that is unpatentable because no
integrated mechanism or new and useful result is produced

I think it is vanishingly improbable, even if this were a statute or a
US Supreme Court decision instead of a legal dictionary, that the
existence of a specialized meaning in patent law would obstruct the
use of reading 1.  And even if it did, you'd have to prove that it's a
better reading -- which I would expect would involve proving that
combining A with the rest of B is by itself such a novel idea that it
meets the patentability threshold of originality, which is vastly
higher (in the eyes of a judge, anyway; the USPTO is 0wn3d) than that
for copyrightability.

Did I mention that trying to twist copyright into a poor man's patent
is almost as inane as twisting it into a poor man's trademark?

> Now, back to the argument: an argument has been raised that the GPL is flawed
> because a "work based on the Program" defined in two parts, where the first
> part asserts that "work based on the Program" is a derivative work.  The
> assertion has been made that the second part of that definition is meaningless.

No, the assertion is made that the "either/or" clause IS the
definition.  The rest of the sentence is ancillary at best, and
represents an erroneous paraphrase of the "either/or" clause. 
Grammar, common sense, and the principle of construction against the
offeror demand that it be ignored -- in the absence of clear evidence
that the licensee believed at the time, not only that that's what
"derivative work" meant under the law, but that the paraphrase
_replaced_ the actual definition.  What the drafter or offeror
believed doesn't make a damn bit of difference.

> Let's assume that this assertion is true, that the second part of that
> definition
> is meaningless.  Let's further assume that I can construct an example case
> where a work isn't covered by the GPL because the second part of that
> definition is meaningless.  What would that mean?
> 
> Since Section 0 says that the GPL grants you license to distribute this work,
> and since there's no part of the GPL that grants you license where Section 0
> does not apply, in our hypothetical case we would have shown that the GPL
> does not grant you license to distribute this work.

Wrongo.  The GPL grants you license to copy, modify, and distribute A
under the applicable terms.  Whether by "mere aggregation" or by
reductio ad absurdum, you may distribute some collections containing
A; and there is no basis in the text of the GPL for enforcing on the
licensee any division into some permitted collections and some
forbidden collections.  So B may be distributed so long as the
applicable covenants of specific performance with respect to A are
honored.

> At this point, either:
> 
> A) Copyright law doesn't apply, so it doesn't matter that you don't
> have license, or
> 
> B) The GPL doesn't apply, so it doesn't matter that the GPL doesn't grant you
> license, of
> 
> C) Distributing the work is prohibited by law.
> 
> My argument is that if you reach C) by ignoring the second half of the
> definition
> of "work based on the Program", that you're doing something wrong.
> 
> Does that make sense?

No.  Copyright law applies to the copying of A.  The distributor of B
claims license under the GPL to copy A.  The court construes the terms
of that license, settles all other relevant questions of fact, and
either decides that the plaintiff is entitled to some relief or that
he is not.  It is then so ordered, and there's a path for appeals on
points of law.  "Prohibited by law" doesn't mean jack.

Cheers,
- Michael



Reply to: