[R-wiki] License for R Wiki contents

[Philippe Grosjean]

Paul Johnson wrote:
 > [...]
 > And people who add
 > things should have some control over who can edit their input.  If
 > Prof Harrell were to put in something, and he really does not want
 > other people to fiddle it, he should be able to protect.  TWiki allows
 > that kind of control.

Here, I say "no, no way!". This is against the Wiki phylosophy, and 
also, against the license I propose for the R Wiki, which is "Creative 
Commons Attribution-NonCommercial-ShareAlike 2.5" (but see: 
http://creativecommons.org/licenses/by-nc-sa/2.5/).

OK, a couple of explanations are required:
1) The Wiki is for collaborative work. By the way, I believe it is the 
missing block in the R world: to offer a way for *collaborative* 
writting of documentation. Currently:

- The R core team is responsible for edition of .Rd files of base 
packages and for the R manuals. Limited input from users through "bug" 
reports.

- Package authors/maintainers have control over their .Rd files and 
vignettes. Users feedback on the documentation is also limited.

- R News allows for writting papers on R topics. Nice for reviews, or 
presentation of particular features, but static once it is published.

- Other documentation is either in the form of html/pdf documents, or in 
the form of published books. Both need effort from their authors to keep 
them up-to-date with a R software that changes rapidly with two versions 
each year! A lot of these web/pdf documents are not updated, because 
their authors are not "rewarded" enough from such a painful work. Some 
books are updated (MASS, with its 4th edition, comes to my mind 
immediately), but it is even harder for the users (who owns the four 
editions of MASS?).

Collaborative work may be a solution to this problem: if someone notice 
an error in a page, he can correct it himself... even if the author has 
no time, or no interest on updating its own documents. Now, does it 
means that all pages you put on the Wiki are left for free changes by 
anybody? No, because there are several ways to cope with that:

1) The Wiki stores all versions of each page. If you, as the author, do 
not like changes made by others, you are free to revert to a previous 
version. This can lead to a dispute if the other person revert again and 
  again. This sometimes happens on Wikipedia. In this case, the dispute 
is solved by a discussion on a dedicated forum. Here, we have R-sig-wiki 
which could play the same role, if needed. Ultimately, the original 
author could ask for a locking of his page, if needed, which leads to 
the second tool:

2) It is possible to give write access to selected pages to only one 
user, or a specific group of users. However, this should be used only in 
extreme cases, since it plays against the Wiki phylosophy and strength: 
*collaboration on the same documents*.

3) Bad users can be blacklisted. So, they are bannished from write 
access to the whole Wiki, temporarily or definitely.

This mechanism clearly requires a team of maintainers for the Wiki. 
Given the high investment of many people on R-Help, I anticipe we could 
find enough volunteers in the R users community for doing this job.

Now, back to the license. I think that authors on the Wiki should keep 
rights on their outputs... but not  with an usual copyright approach, 
because it is against collaborative and incremental building of 
documents, by definition!

GPL suffers from one big problem regarding the Wiki: it does not 
inhibits a commercial use of the material. This is not much a problem 
for a software (think about Linux distributions: some are sold by 
companies and this is just fine), but in the case of documentation, I 
would not like that someone decide to publish a part or the whole R Wiki 
in a book, claiming to be the editor because he just compiles the 
documents and put them in a format suitable for Springer, Chapmann & 
Hall, or whatever editor... This is the same problem for the GNU Free 
Documentation License.

After a search for a better license term, it appears that the Creative 
Commons license named "Attribution-NonCommercial-ShareAlike" is a good 
one in this context because:

1) Attribution -- (Original) authors must be always explicitly cited 
(except if they dislike changes or use made of their material, in which 
case they could ask for not being cited),

2) NonCommercial -- It is explicitly stated that commercial use is 
prohibited. Thus, nobody can publish a book using material in the Wiki, 
never! But anybody can compile PDFs or other formats that are freely 
distributed on the Internet.

3) ShareAlike -- This is exactly the same idea as GPL: you can use, 
copy, distribute, modify and redistribute your modifications, but at the 
condition that you keep the same license for the redistributed material. 
It is what makes the strength of R (and Linux, and Apache, and...). So, 
the idea is to use the same phylosophy for the documentation on the R Wiki.

By the way, CC Attribution-NonCommercial-ShareAlike is a license rather 
common for public Wikis.

Now the last point that is, I think, important: would it be possible to 
compile part of the Wiki and to publish it in another form? Indeed, I 
think at a publication that is better valuable in a C.V. Yes, I think 
about somethink like JSS. It should be possible to do so, with all 
authors cited, of course, in an order that reflects the amount of work 
done be each people. That is the theory, but I am not sure it would be 
that simple in practice!

Best,

Philippe Grosjean