Slashdot Mirror
Open Source Licenses For Academic Work?
An anonymous reader writes "We're in the process of submitting a scientific paper describing some techniques for data analysis. We'll be releasing the associated code, so we're faced with choosing an appropriate license. My supervisor insists there should be a citation clause, requiring any published article that uses results of the software to cite our paper. Of course, ideally, free software shouldn't have such encumbrances, and I initially tried to talk him out of it. However, in academia, the issue of attribution and citation is very important. Also, it is not a restriction on use of the software per se, only on publication of results. Unfortunately, I couldn't find any such license. So I wondered: what do other academic Slashdotters do?"
Not one of 'em crazy academics but wouldn't this do?
http://creativecommons.org/licenses/by/2.5/
So, if you were to get such a license and then somebody published a result without citing your software (as opposed to mentioning that they used the software), how would you (or your boss) enforce it?
Would your boss really sue another academic for not citing the software?
Of course, as an academic myself, not citing the paper for some software that I used, is sloppy anyway.
The NAMD license has a similar clause. It might be worth looking into.
The Attribution-Share Alike 2.0 Generic license looks
like what you need
http://creativecommons.org/licenses/by-sa/2.0/
It allows others to modify and adapt your work as long
as they attribute the original in the manner specified by you.
Your academic papers don't have such a licensce. They are cited because it's considered unethical not to do so. The same would apply to using your source code.
Also, your license can't actually enforce the citation clause. I mean whoever uses the code won't necessarily be the same person who writes the paper. Additionally I have some doubts that the kind of clause you are interested in would be legally enforceable.
Science works because we trust other scientists to cite our work if they use it. If we kept our work secret unless other scientists signed agreements to do so nothing would get done.
If you liked this thought maybe you would find my blog nice too:
It's not like they can use your work without attribution in academic papers, since that would be plagiarism, right? So any specific clause that required citation would be unnecessary.
Use a BSD License. This license goes back to the 1970s.
When I was young, I had to rub sticks together to compute.
To be honest, I think your best option is: "Ask nicely."
Seriously, academia and publishing and citation is a massive reputation system. It almost entirely works on the honor system, with formal inquiries occurring (rarely) when there are major transgressions. Let's say you find or write some complicated open-source license that requires citation. The code will still be available. Unscrupulous people could still use the code and publish without citation. Do you really think you (or your supervisor) would ever bother suing them? I highly doubt it. But you would certainly spread the word that these researchers don't cite properly. You would certainly bring up this issue during peer review. This is where the real damage to them will occur.
So, my recommendation is to just skip the middle-man, and don't bother with the unconventional FOSS license (which would just confuse people who want to use the software but won't ever publish anything). Wherever you post the code, just include a prominent request (on webpage, in README, and code headers) along the lines of "If you publish any work that uses this software, please cite XXX." Most scientists would be happy to add that citation. The only ones who wouldn't are the ones who try to pass off other's work as their own: do you really think they care about respecting copyright?
This is, at least, the procedure used in my field. Publish your paper. Release the code using a standard FOSS license.. Add a citation request. Done.
The BSD license is from UC Berkeley, the MIT license is of course from MIT, llvm is from the University of Illinois / NCSA and uses a license almost identical to the BSD license, etc. For some reason, this sort of "free as in knowledge" type license seems to be rather popular among educational institutions.
My supervisor insists there should be a citation clause, requiring any published article that uses results of the software to cite our paper.
That is a restriction on how it can be used, and I seriously doubt it is at all compatible with Open Source. It certainly wouldn't be compatible with Free-as-in-FSF software.
If they use your software in a manner that, from an academic point of view, requires citation, then they are going to cite you anyway if they are honest.
If they use your software in a manner that, from an academic point of view, does not require citation, then your clause puts them in a difficult position. For example, their editor might insist the citation be removed, but then your license kicks in.
Besides, how are you going to enforce this anyway? Are you going to sue? What kind of damages and remedies are you going to put in there?
Just pick a standard OSI license, tell people who want to cite it how they can, and trust that it will work out. Don't try to force people to use your software in any peculiar way, even if that way does not seem "evil."
I asked a related question here several years ago. I have completed my schooling and released some open source software, some of which has been used and cited.
Copyright licenses generally protect holders from having others distribute their works in a way that they do not want. They do not place many restrictions on how the legally obtained work can be used. You might be able to use an end user license agreement that attempts to mandate citation & worse restrictions (such as not being able to publish software benchmarks) have certainly been imposed. Some authors even mandate registration before others can receive the source code & can then see who may be using but not citing their software. But I think this may actually be counterproductive & it certainly wouldn't be considered free software.
Academic integrity necessitates describing your work accurately in such a way that others can reproduce it. To do this, others will need to say what software they used to obtain the results they publish & they should choose to cite you. This won't always happen, but it will probably happen more frequently than you or your advisor think. It is certainly valid to write or call other academics who you know use your program and ask that they cite your paper in the future. In extreme situations, you can send a note to the editor of the journal that considers such papers that didn't cite your work & most editors will err on the side of strongly encouraging authors to add a citation.
Most other free/open source software that is used a lot in the sciences does not have a EULA of the type you subscribe, yet many are popular & are cited. They may have a FAQ entry or a mention in their README on what should be cited, but they don't try to make it legally binding.
You should ask yourself why you want to release it as free and open source software. Presumably, you hope that others will use it (obscurity is a worse threat than piracy) & maybe even to help you improve it. You also probably want to obtain some kind of academic prestige (which can come not only in the form of citation, but also from name recognition of both the program and the authors of that software). The best way to get this to happen is to write a solid piece of work that can do something that other works that cost (financially, time invested, and responsibilities involved) the same or less can't do as well and that other people want to do. Use a standard FSF/DFSG/OSI license (such as the GPL) & trust that everything else will work out. Getting quirky will discourage use of your software.
You'll give the rest of us a bad name if you actually produce anything. Stop that.
They were right - the revolution did not get televised. It was posted on YouTube instead. All in 120 characters. SLOOSH!
The important thing to have cited is your results in your paper, not your software. Many academic institutions have been writing open source software for a long time, in fact many of the open source licensees that are used every day come from software that was developed in academic institutions. Things like MIT license, much of the motivation behind the GNU license, BSD, the list goes on. None of them require attribution.
Take a look at http://cran.r-project.org/doc/FAQ/R-FAQ.html#Citing-R
Most papers to use R (or other statistical software) will in fact cite it. R makes this easy, by providing built-in citation strings. Do this, and well-behaved researchers will cite your software.
How about you talk to the lawyers your school keeps around for this kind of stuff? Since you're going to be obligated to follow whatever their rules are anyway, you might as well defer to them now instead of after their lawsuit.
If you accept the basic premise of EULA's, which specify a contract, then that contract can do basically anything it likes, within the law.
But once you've gone that route, you've probably already given up the idea of using an OSI approved open source license.
"...what do other academic Slashdotters do?"
I always provide attribution because it's part of the rest of the ethics of science. But for code, don't expect everyone to continue to attribute the paper itself, just the source lab and university. We use the EEG system most common to labs like ours. Everyone mentions the vendor in their work, but nobody mentions the validation papers and review of same. Similarly, everyone mentions what statistical analyses they use, but hardly anyone would even know where to begin to find the original publications validating them. If the code becomes widely accepted, expect it to become commonplace enough that the lab and university get mentioned, but not the paper.
And to be that widely accepted, you'd need to either provide ample validation in this publication, or better plan on doing a validation paper as a follow up. Enlisting similar labs for the latter would give your analysis more weight and them a worry-free pub, and everyone wins. If your analysis doesn't happen to be novel enough to warrant this (say, it's a mash up of previously known analyses) then your validation is reference of the sources, but it's not something you can expect much referencing to.
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
A clause like this is attempting to inflate citation counts beyond what would normally be expected, mainly by forcing even marginal use to result in a citation when often it wouldn't merit one (I don't cite, say, the manual for the Dell computer I use).
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
the GPL does have a detailed description of the attribution issue in their preamble. Asking for attribution on GPLed work as a condition of use is perfectly compatible with the GPL license.
(The above refers to GPL v.2)
--Sam
Two points to keep in mind:
You may be able to convince your supervisor by citing the examples of BSD Unix and X11, which brought fame and money to their creators (the CSG at Berkeley, and project Athena at MIT) while using extremely liberal licenses -- the MIT/X11 license (which is what I use for my research) and the 4-clause BSD license, albeit with the advertising clause not being enforced.
You may also want to cite the following anegdote. Two years ago, I was compiling a Linux LiveCD for our first, second and third year undergrads. One of the pieces of software I wanted to include was a Prolog compiler from a well-known Portuguese university which we use in third-year courses.
Unfortunately, the Prolog implementation was covered by a fairly strict license that would significantly complicate our distribution process. After a few exchanges of e-mail with the copyright holders, they told us that we were welcome to do whatever we wanted, but they'd not change the license for us.
After consulting with our legal department, we decided we could not include the Prolog compiler.
Why GPL? I would expect that when one is publishing academic research, one would want to license any accompanying code under a license that is as compatible as possible with whatever code other researchers are using. That would be something like BSD.
It has to be worked out by you, your employer, your funding body, your university policies, and everybody's egos and lawyers. Any one of them can wind up preventing the sensible others from a sensible answer, especially federal regulatory policy on publicly funded research.
I prefer to use GPL myself, and encourage my employers to do so. It's well documented, easy to follow, and there is now a reasonable body of case law to work with. Whatever you do, don't invent your own, special, unique license. Dan Bernstein did that, and it helped keep djbdns, daemontools, and djbdns from becoming default system components in any OS distribution that I've ever worked with, despite their technical advantages over most other such tools.
BTW is your suoervisor actually the author of any of the software? If not perhaps you could keep him happy by attaching his terms to the copies of the software distributed with the paper and then release it under your preferred terms elsewhere.
As a student/researcher any work the OP did as part of his studies at the university would typically be considered by a court to be the copyright property of the university, at least in as much as they provided the resources (tuition, guidance, computer facilities, access to research grants) that were used to produce it and/or provide the OP with money to live on while it was produced.
IANAL; this is not legal advice. It's true, though.
For my own papers-about-data-analysis-programs, I've had good success with a simple "Please cite this paper" request, but nothing explicit in the license (I use BSD as my university preferred it to GPL). By success, I mean plenty of use of code(s), plenty of citations, and very few cases where I know of code being used and not cited. My observation (as an author and journal referee) is that most authors are happy to cite the papers for the data analysis algorithms with "We used Foo1.2[Barr, et al]" as it is a simple way to explain away details that are important, but not the point of their paper. So my view is that "you MUST cite ..." is best left out of the license.
This is a FAQ about free software licences, see http://people.debian.org/~bap/dfsg-faq.html question 12.g.
Also, enter a comment section in the program code that makes the same request.
Then code into the program a section that prints as a head/footer in the formatted output or as every Nth line in the unformatted output file text such as "Data created by Program X developed by Person Y at Place Z".
You'll have your attribution, and likely only those who weren't going to comply anyway will bother to edit the output or program code to remove it.
QED
This space intentionally left blank.
Since you are asking which license to publish under, it sounds like you haven't done this within the framework of whatever institution where you are working.
It could very well be (probably be?) that the license you have to publish under is already set and that you are legally bound to follow it.
Depending on who funded the research, there could be other restrictions and obligations as well.
Certain funding institutions require there be no copyright at all, while others may have some agreement in place that you might violate if you don't investigate this first.
Stuff like this is how you can lose funding - not just for yourself, but for the institution. And the legal issues, under the wrong circumstances, could end up haunting you.
SocialText, makers of wiki software, created a license that may be just what the OP is looking for and it is OSI approved.
I never really looked into how they might work it into their license, but Gaussian provides a template for their *required* citation when you publish, so they're presumably doing it somehow. http://www.gaussian.com/citation_g03.htm
As several people have said earlier, it would be sloppy not to include a citation to any software you use when you write a scientific text. I've more often experienced the opposite problem; trying to find an article to cite for a piece of software I've needed to use. At times you cannot find such articles and hence you are forced to refer to some web page inststead. Which of course will be down the exact day the referee reads your article. I've never met anyone who are reluctant to cite the software they use. After all, a scientific article on the software frees them from the responsibility of describing every piece of it themselves.
So, to the licensing issue; I'd strongly recommend sticking to one of the standard licenses. It really helps the people who want to use your code (and hence will cite your articles). For every new license your users must consider, they will be forced to decide wether or not they can use your code together with other stuff they need. If they want their program released as free software, and particularly if they want it included in e.g. Debian they will probably steer away from unusual or non-DFSG-compliant licenses. Your supervisor wants them to use your software (even though he might not know that) because usage generates citations.
Trust your users, they will cite you. (And give them the BibTeX entry to your article to make it easier for them.)
Publish your work in a reputable Open Access Journal or make it available as part of the Digitalcommons. Making it visible on the Internet is the best protection against someone taking credit for your work.
http://www.mfa.kfki.hu/~labar/ProcDif.htm
Labar published an article describing the work in a scholarly, refereed journal and then distributed the code(he chose an executable) with a request for citation.
You could also look at the EMAN2 project
http://blake.bcm.tmc.edu/eman/eman2/
EMAN2 is distributed as open source. The authors also published the work in a scholarly, refereed journal.
Compare LaTeX output of math/statistics literate programming examples with bodies of math/statistics papers. The only significant differences are that the literate code is runnable and makes the research reproducible. The question is the academic institution's rights with respect to intellectual property in general and the answer is that science application should not be hindered by rules that encourage academic institutions to capitalize on their research findings. See "reproducible research" "literate programming".