Re: Plugins, libraries, licenses and Debian

[Jeremy Hankins <nowan@nowan.org>]

Anthony DeRobertis <asd@suespammers.org> writes:
> On Dec 16, 2003, at 11:28, Jeremy Hankins wrote:

>> If I understand him, he's saying that the author of the plugin is
>> doing the work of pairing his code with the host (even if, in fact,
>> it will be paired many times and by many people) and that that's
>> where copyright subsists, and where a derivative work is created.
>> Arguably, the plugin itself (sans host) is a derivative of the
>> plugin+host which the author created first.
>
> Keeping in mind the original situation, host = X11 license, plugin A =
> GPL, plugin B = OpenSSL, we were wondering "can Debian ship host + A +
> B?" I think we agreed that if they were in separate packages, the
> answer would be 'yes'. See my position summary from December 9th, and
> the responses to it, here:
> <http://lists.debian.org/debian-legal/2003/debian-legal-200312/
> msg00121.html>

Yup, I saw that, and with a minor quible (which I posted about at the
time), I agree.

The quible is based on the fact that we're talking in the abstract here,
but I can imagine particular cases where that would not be true.  It's
*probably* true, for most (if not all) reasonable cases, but not all
cases are as reasonable as we'd like.  The prototypical example might be
a case where MS writes an (X11 licensed) program to load proprietary
Word modules and GPL modules so that the GPL code can be used with word
documents.  Especially if they then advertise this as a way to use this
GPL code with word docs, that'd be infringing.

In the end it comes down to psychology, not technology.  What was the
person who did it thinking?  Since we can't get at that directly we try
to infer it from the technology and the circumstances.  I.e., if it's
unreasonable for someone to think that the two pieces of code (SSL &
GPL, say) were part of the same tool then there's no conflict, but if
it's unreasonable to expect that they *aren't* part of the same tool
then there is.  That leaves a fair amount of grey area to be decided by
the courts, but I'd expect that in most cases a generic module interface
wouldn't run into problems.

> The point I am trying to make here is that Debian is fine if we happen
> to put all the plugins in the same package, along with the host. I
> base this on the GPL allowing 'mere aggregations'.

Whether they're in the same package or not is only relevant so far as it
gives us insight into the thinking of the original author.  That is to
say, it's not very relevant at all.

> From the definitions in US law, it's fairly clear, I think, that the
> package, if it is a copyrightable work at all, would be a compilation,
> not a derivative work. It was then argued that it can't be a
> compilation, because when the Debian maintainer put host + A together,
> there was no original work of authorship.

I don't think this is clear at all -- this is the question under
dispute.  I think it's reasonable to argue that:

- Access to the host program is required to write the plugin,
- Details of the host program determine aspects of the plugin,
- The goal of the entire procedure is to produce a work that is
  host+plugin.

And that it therefore qualifies as a derived work.  Under this scenario
one might argue that the author of the plugin and the end user both
worked together to create it, but as you said no copyright subsists in
the actions of the end user.  Compare this to shipping someone all the
pieces of a table with instructions to put it together, and claiming not
to have shipped a table.  Such a claim would (rightfully) get laughed
at.

I certainly accept that this argument isn't necessarily any more
convincing than yours to the contrary, so all I'm trying to say by this
is that the issue isn't clear.  In other words, without cases or
specific law to the contrary, one must assume that this is a possible
interpretation.

>> But even if the courts don't take such a literalist view, I have
>> trouble accepting your claim that since the creative work and the
>> pairing that results from it are separated in space & time that
>> there's no derivative work.
>
> The important part is that since _I_ did the pairing, I'm the only one
> who could have the copyright over the pairing itself. (A's author
> naturally retains his copyright over A, which may be a derivative work
> of host; and host's author retains his copyright over host)

This is precisely what I think needs to be backed up, as it's certainly
not obvious that you were the only one involved in the pairing.  The
original author helped considerably.

>> If I'm a radical artist and fire a cannon from miles away to land on
>> a sculpture, is the resulting "art" not a derivative of the original
>> sculpture because I wasn't there when it hit?
>
> No, because you can mutilate your copy of a work, and that's mostly
> outside the scope of copyright law. Though, before you fire, you may
> want to check in to the sculptor's moral rights to prevent damage to
> his work....
>
>>  How about if it's a copy
>> rather than the original?
>
> Much more likely, but I'd question whether shooting a cannon at
> something meets the standard of an original work of authorship ;-)

So I fire nanobots designed to modify my copy (put mustachios on it,
say).

-- 
Jeremy Hankins <nowan@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333  9E10 D212 B5ED 37D0 0A03