On Thu, 2007-04-26 at 16:32 -0700, Ken Arromdee wrote: > What are you talking about? Unless I'm mistaken, the topic is to consider a request for a GR that would add language to the DFSG saying that licenses need not be modifiable. :) > If by "legal composition of copyright" you mean "license texts are copyrighted, > so they cannot be DFSG-free", that's false. We include plenty of copyrighted > materials which are DFSG-free. > > If by "legal composition of copyright" you mean "license texts are used to > indicate how other things are copyrighted, and cannot do that if they are > modified", that's wrong. You're assuming that "modifying a license" means > "trying to relicense the thing the license is attached to". That's incorrect. > One might want to modify a license in order to reuse the license somewhere > else. Modifying a license in this way has no bearing on the licensing of > the work to which the license was originally attached, and the copyright of > the work does not restrict modifying the license this way. Neither of the two options above are what I'm talking about. Obviously, the first option is totally self-evident. All the copyrighted material in Debian main is DFSG-free. The DFSG deals with the licenses of copyrighted material. That's the whole point of the DFSG. The second option also has nothing to do with this GR request. I'm not assuming that changing the license text with the purpose of reusing it somewhere else will magically allow anyone to relicense the work to which it was originally applied under terms of their own choosing. I'm also not saying that creating a derivative license should necessarily be prohibited. It's not without complication, though. For example, subtly changing a license text (as a work) might lead to a recipient of another work licensed under the original license mistaking the modified license for the real one. It is in the interests of both the copyright holder of the work and all recipients of it to prevent such confusion. (For example, the GPL suggests to state that a work under the GPL is licensed under "The GNU General Public License, as published by the Free Software Foundation", in addition to FSF requiring a name change on modifications.) What I'm saying is that the DFSG can only be applied to a certain point. We can require that license terms applied to works are DFSG-free. We can require that license terms applied to those licenses-as-works are DFSG-free. We can require that the license terms applied to those licenses-as-works are DFSG-free and so on, "moving up the chain", until we hit bare copyright law at the top of the chain (meaning that there are no specified additional terms to apply; the license-as-a-work at that point has no explicit license). We would then need to add an "exception" for copyright law, because what we originally set out to do was to claim consistency under a certain (flawed, IMO) interpretation, because the consistency would stop at that last link in the chain, and because there is no way we can affect the existence or nature of copyright law by simply changing words in the DFSG. We could try to avoid "moving up the chain" by requiring that all licenses for licenses-as-works be recursive, but then we would be requiring a stricter set of requirements for these licenses than those that we require for other works in the DFSG -- and we would be inconsistent again. (Because we wanted the same freedom requirements for everything.) Plus, we are unlikely to achieve this in practise, because this interpretation goes against that of most copyright holders, who get to choose the license for their works. This is what I mean by saying that the "legal composition of copyright" results in the distinction between a license and a work in the language of the SC and DFSG, as in any document dealing with copyright. Do you think this is nitpicking and that the "license chain" line of thought is splitting hairs? I do -- the same kind of nitpicking, hairsplitting and overinterpretation that led us into this discussion in the first place. We can try to shift the problematic step to any place in the "chain", but we can never get away from the fact that it's there, and that the only way to understand its implications is to draw a line beyond which the DFSG doesn't apply. Refusing to do so means creating a confusing grey area, specific to Debian, and prone to misunderstanding because it is unconventional in the domain of copyright. It's usually assumed, as I said in a previous message, that works and licenses belong to different domains and that licenses form the line beyond which the DFSG doesn't apply. The DFSG seems to be drafted on this principle, and there is therefore no contradiction that needs to be clarified or given explicit exception. Not starting to "move up the chain" spares us from approaching the slippery slope of mistakes that others in this thread have brought up. So let's not pursue this GR. Cheers, -- Fabian Fagerholm <fabbe@paniq.net>
Attachment:
signature.asc
Description: This is a digitally signed message part