I just posted the following to the Wikipedia-L mailing list, on the topic of the GNU Free Documentation License. I am posting it here, so that people who don't subscribe to Wikipedia-L, or would rather discuss it in Wiki format, can do so. -- Simon J Kissane
AxelBoldt raises a number of requirements under the GNU Free Documentation License, such as the requirement to preserve lists of five most significant contributors (how could we ever judge that for Wikipedia? we wouldn't even want to), change the title of a new document, etc. Many of these things don't fit too well with Wikipedia.
The FDL was not really designed for a purpose such as Wikipedia. It was designed for programming manual documentation. It was designed for an environment with a small number of individual authors, dealing with a relatively stable work, where different versions are clearly distinguishable. The FDL was designed for an environment were acknowledgement of individual authorship and the identity of the work is essential, but it was felt necessary to permit later authors to modify the work in order to ensure it does not fall out of date. Wikipedia, by contrast, has no clear individual authorship, lacks a clear identity as a work, and 'Modified Versions' are being created constantly. Its a different paradigm, a different method of production, and a license designed for one paradigm doesn't fit a significantly different one that well.
In some ways I think that the GNU GPL would be a much better license for Wikipedia. Many of the additional requirements that the FDL has over the GPL aren't really relevant to Wikipedia. The GPL doesn't permit requiring even collective acknowledgement, but as I have argued I don't think the machinery of the FDL license can handle the kind of acknowledgement requirement we probably want.
The FDL allows the author to require the inclusion of specific invariant sections. But when we want to require acknowledgement, what is important is the substance of the acknowledgement and its prominence, not its exact reproduction (though as I have said, exactly how exact a reproduction is required is unclear.)
(On a sidepoint: Jimbo says he can't see how putting the invariant section on a different page could comply with the FDL. The FSF, which produced the FDL, creates its manuals using texinfo. Texinfo documents, when converted into HTML format, generally have a separate page for each section or subsection. So, if we can take the authors of the license as a guide, the invariant section can be on a separate page, so long as it is part of the same document, say by being linked from a Table of Contents.)
Which is why I would creating a separate license for Wikipedia, based on the FDL but with modifications to fit Wikipedia's special conditions. Among other things, I would delete the requirement for invariant sections, and the stuff to do with "Endorsements", "Acknowledgements", "Dedications", lists of authors, renaming modified versions, etc. But I would add a specific requirement to provide acknowledgement, with some broad guidelines.
To get this new license we could simply modify the GNU FDL. While the license for the license doesn't permit modifications, I'm sure RMS would give us permission if we explained why we wanted to do this, and called it something else.
Two other issues remain. One is compatibility with FDL documents being used in Wikipedia. Since the terms of the license I am proposing are incompatible with the FDL, since they don't require things that the FDL does, and do require things that the FDL doesn't. Of course, if the author gave permission this wouldn't be a problem. But still, that would be different from the current (informal) policy, which seems to permit any FDL or other "Open Content" material to be added.
But I am unsure how much such material has been added, and I question the legal adivisability of permitting such material to be added, even if we continued to distribute under the FDL. The FDL provides several specific requirements for reproduction, which I doubt anyone who has added such content has paid specific attention to; and even if they did, as I have pointed out many of these requirements (e.g. invariant sections -- how possibly could anything in a Wikipedia article be invariant?) don't really fit with Wikipedia. And other "Open Content" licenses add their own requirements as well. Since licenses can be quite complex and can often impose conflicting terms, I don't think it is advisable to let people add copyrighted material of any sort, without specific permission from the copyright holder.
The other issue is that people who have already submitted their content have done so on the understanding that it will be redistributed under the FDL, and might not agree to having it redistributed under another license. This raises quite a legal conundrum. The only thing I can suggest is, that if sufficent consensus arises in the Wikipedia community on a license change, we widely publicise the license change, allow anyone who does not want their material to be distributed under that license object, and consider those who do not object as consenting to having their material distributed under the new license.
I think that, rather than just creating a notice below the submit button, we should provide a proper terms and conditions notice, which should include that the author consents to redistribution under the terms of the license, or any later modifications of the license or replacement license which may be agreed upon by the Wikipedia community. Of course if we had a "any later modifications or replacement license" clause, we'd need to find some way for the will of the Wikipedia community to be expressed (how about a Wikipedia council? Wikipedia referenda? Wikipedia elections?).
And of course, any such new Wikipedia license or Wikipedia terms and conditions of use should be discussed and achieve wide consensus in the Wikipedia community, before being put into practice. (I suppose that Jimbo, since he owns the Wikipedia website, could just produce his own terms and conditions of use without consultation and impose us on it all. But I trust he won't.)
Anyway, that is my latest thinking on the Wikipedia licensing issue. I'm also going to add this on Wikipedia commentary under Wikipedia, so people who don't subscribe to Wikipedia-L can read/comment on it.
Simon J Kissane
Do we really need to not use the GFDL or the GPL? The GFDL is definitely primarily for stuff that's printed. I actually think it's not such a bad idea that we use the license. The problems with the GFDL:
a) Authorship issue. GFDL assumes authors, Wikipedia essentially doesn't. As long as Wikipedia gives some definition of "principal authors" we'll be fine. I see two options:
- Preserve all diff records so that at least for individual articles, the principal authors can be defined in some way
- State that the author of Wikipedia content shall be "The Wikipedians" or "The Internet community" or "The World" or simply "collective/no author/unknown/various". I prefer the non-branded versions, but it's not a strong preference. In fact, it would be cool if 2-5-25 years from now there are all these reference works that have a "by the Wikipedians" byline.
b) Unclear about internet reuse. Possible resolutions:
- Don't use the optional features.  That makes life easy.
- Work with the FSF to figure out what the internet equivalents of Cover Texts and Invariant Sections are.
My thinking is that if we do this right, the FSF might want to release an updated version of the license to make these issues more clear. Since it is reasonable to assume that Wikipedia (or at least parts of Wikipedia) will be published in book form at some point, it does make sense to use the GFDL instead of the GPL. But unless we go with "Don't use the optional features" (and I say, why not?) we should work with the FSF to figure out how to interpret the license, and either
- Explain our interpretation of the license, but leave it as an interpretation, or
- Modify the license, hopefully as a new official version of the GFDL