Re: GPL as a license for documentation: What about derived works?
On Fri, 28 Jan 2005 23:56:25 -0500, Raul Miller <email@example.com> wrote:
> On Fri, Jan 28, 2005 at 07:44:38PM -0800, Michael K. Edwards wrote:
> > Any given country's implementation of the Berne Convention may vary
> > somewhat, but the US statute (at least as of 1986) and the case law I
> > have seen are consistent with the interpretation that "compilations"
> > (or the subset "collective works") are a disjoint category from
> > "derivative works". See 17 USC 101, and compare the UK CDPA sections
> > 3 (1) (a) (a "table" or "compilation" is a subtype of literary work)
> > and 21 ("adaptations" are defined to cover essentially the same scope
> > as Berne Convention "derivative works").
> The Berne Convention does not appear to use the term "derivative"
> at all. The only place I can find that uses related worde
> (derived, and collection) is Article 14 and 14ter, in reference
> to ("derived") cinematographic production based on other
> kinds of works and ("collection") the collection of fees.
I was going on the title ("Derivative works") of Article 2, Clause 3
in the edition at
http://www.wipo.int/treaties/en/ip/berne/trtdocs_wo001.html , which
provides titles for each clause. The US and UK definitions of
"derivative work" and "adaptation" respectively match this definition
pretty closely. As I wrote, Article 2, Clause 5 defines "collections"
as a disjoint category, and the rest of the Convention and its US and
UK implementations appear to me to be entirely consistent with this.
I don't have the expertise to say this for certain, though; is there
some piece of statute or case law that you read to contradict this?
> Furthermore, if 17 USC 101 is a basis for "collective works" being
> disjoint from "derivative works" then it's also a basis for "computer
> program" being disjoint from both. Which is clearly not the way a court
> would treat computer programs.
I'm not saying that two categories defined in 17 USC 101 can't
overlap. I'm saying that collections clearly aren't defined to be a
sub-category of derivative works, and that neither the Berne
Convention nor 17 USC 101 gives one any reason to think that putting
two or more works side by side creates a derivative work. As I wrote
elsewhere in my analysis, the hypothetical book without
Chapter/Appendix X may be a derivative work but is a protected "fair
use", and is a separate work of authorship from Work X itself.
I don't think that combining the two creates another, larger
derivative work; I think it creates a compilation, and not necessarily
a separately copyrightable one at that, since there's nothing
particularly creative about binding the two together. If I write a
critical essay on Don Quixote, and you publish it under non-exclusive
license from me together with the text of the novel, you don't hold
copyright in that compilation (it's not sufficiently creative) and
can't bar me from authorizing another publisher to do the same.
> I think that the mathematical concept of a disjoint set misconstrues the
> nature of legal definitions. The difference is more one of focus. There
> is no need to believe that there is some precise legal difference between
> "an anthology" and "a work based on one or more preexisting works".
I think case law is pretty clear that what's independently
copyrightable about an anthology or other collective work is its
selection and arrangement, and that protection of "derivative works"
is intended instead to cover cases like the Berne Convention's
"translations, adaptations, [and] arrangements of music" (and, under
the mise en scene doctrine, sequels and the like) that are themselves
In theory, "other alterations" could cover almost anything including
compilations, in which case the authorization to create a compilation
(like any other kind of derivative work) could not be inferred from
authorization to copy. But I think that would conflict with a great
deal of case law in which license to publish, say, a recorded song,
unless explicitly limited in scope by language such as "in the context
of the entire album called Foo", permits its use in other
compilations. But I don't have time right now to follow up this line
> Finally, I agree that "linking doesn't create a derivative work".
> But I also feel that this statement about linking can be misleading.
> Before a program can be successfully linked with some other body of work,
> it is already a derivative work, or not a derivative work, of this other
> body of work.
Right. As I wrote, the argument that a program written to use a
library isn't a derivative work of that library relies on different