ESR: "We Don't Need the GPL Anymore"

Michael K. Edwards M.K.Edwards@gmail.com
Tue, 12 Jul 2005 21:55:57 -0700


On 7/12/05, Rick Moen <rick@linuxmafia.com> wrote:
> Quoting Michael K. Edwards (m.k.edwards@gmail.com):
> > Whoops.  To be clear: I am of the opinion that the GPL is enforceable
> > (as a contract) in all legal systems about which I know anything at
> > all, and is effective in requiring distributors of binaries made from
> > GPL source to offer corresponding source code, to within a reasonable
> > standard of accuracy.
> 
> As you are perhaps well aware, it seems likely to be adjudicated
> enforceable merely as a copyright licence, irrespective of whether
> it also has all the required elements of a valid contract.  (In the very
> unlikely event of this being a new thought, please see Prof. Moglen's
> "GPL Not a Contract" piece linked from
> http://linuxmafia.com/kb/Licensing_and_Law/ , and also Lawrence Rosen's
> licensing book, available online at http://www.rosenlaw.com/oslbook.htm .

Professor Moglen is not in the habit of citing actual historical law
in support of his position, which I find particularly reprehensible
given that he holds a chair in the history of law.  And though I have
considerable respect for Mr. Rosen, his statements about the GPL in
Chapter 4 of "Open Source Licensing" are best taken as a cautious
hedge rather than a real analysis.  In fact, there is no such thing as
a copyright license that is not a (term in a) contract, and never has
been, from the Statute of Anne onward; and to the extent that there is
a form of "bare license" (a creature of equitable estoppel, not a
copyright license at all) that sometimes quacks like one, the BSD and
MIT X11 "licenses" may fit but the GPL doesn't.  I feel confident in
saying that no US circuit court would seriously consider the GPL in
any posture other than as an offer of contract.

In my (non-lawyer but heavily researched) view, the GPL is a valid
offer of bilateral contract, with ample return consideration, and
there is no particular difficulty in finding that it has been accepted
through conduct by anyone who knowingly exercises a right that would
otherwise be reserved to the copyright holder.  If you were to read
one appellate case touching on these matters, I might suggest Specht
v. Netscape, and note that picking up a free newspaper from an
unattended newsbox doesn't convey the right to modify or republish its
contents.  If you have time for a few more, you might want a copy of
my essay in preparation, such as it is, which cites specifics from a
couple dozen cases.

> > But I don't think a US court would rule that a
> > program that uses a library becomes a "derivative work" of that
> > library....
> 
> Yes, the FSF-promoted notion that linking alone suffices to make a work
> derivative seems highly doubtful and to fly in the face of precedent,
> e.g., the CAI v. Altai decision's test.

The term "derivative work" is quite well defined in both 17 USC and
the Berne Convention, and US case law is also quite clear on the
matter.  I will presume so far as to recommend two precedents: 
Lexmark v. Static Control Components (not so much about the term
"derivative work", but an excellent survey of the limits of copyright
when interoperation is at issue) and Woods v. Bourne (addressing the
usage of "derivative work" that prompted the definition in the 1976
Copyright Act, and the only one that matters unless someone insists on
using the legalism in a contract: the 17 USC 203(b)(1) / 304(c)(6)(A)
exception to license termination).

> Your analysis (judging by current evidence) tempts me to take back
> all... well, some... of the rude things I've said lately about
> debian-legal:  http://linuxgazette.net/116/lg_launderette.html#nottag.22

Don't take them back on my account.  While I have found debating with
debian-legal denizens to be stimulating, and have met a few
individuals there whose opinions are well grounded (even when they
don't always agree with mine :-), most of the dominant contributors
seem to live with Humpty Dumpty in through-the-looking-glass-land
(where the question is who's to be master, that's all).

Cheers,
- Michael