License for research projects

Question

What is a good way to license your research projects?

Diving into the details of most licences, a lot of their wording and concerns seem centered around 'redistribution', and are not so much driven by the concerns of the spirit of academia.

Personally, I have been active in diverse fields; from fields with great standards in terms of code dissemination, to fields with abysmal or almost complete absence of any standards of code dissemination.

When it comes to making code available for academic use, I am somewhat opposed to 'permissive' licenses, and id prefer to release under the absolute most aggressive 'copyleft' principles. I resent the idea of other so called 'scientists' cranking out non reproducible papers that make use of my carefully packaged and documented work, in any way whatsoever. Yet going over for instance the AGPL, considered one of the most 'copyleft' licenses, it seems to be barking up all the wrong trees.

For instance, technical details concerning 'distribution' like static versus dynamic linking are completely unimportant to me. In fact that whole debate about what constitutes 'redistribution' is rather moot when it comes to research I think. Claiming you did something using some piece of software is arguably always fine under any license that seeks to restrict redistribution; since none is involved.

All I care about is that if you seek to build on my work in any way, that you are bound by the same standards of dissemination. That is, not just some passive duty to make code available upon request (I know how that pans out in practice); but a proactive duty to have your code available on a publicly accessible service, and to have it contain runable examples that I can get going with minimal effort on at least one major platform. That is, without spending days to reverse-engineer your development environment. I suppose that could be formalized by requiring use of some CI service; if you can run some tests automatically, everything needed to run your software must be available as code.

For the specific library I have in mind, I am actually quite happy if people adopt it for commercial use free of charge, and redistribute it in whatever way. Id prefer if they contribute back but whatever. But I am absolutely not fine with perpetuating some of the abysmal code dissemination standards going on in some fields of academia.

Is there any license that covers such an intent? Or is it about time one gets written?

Answer 1

The short and to the point answer is that no currently existing licence does what you want. If you wanted to do as you say, you'd need to write a new licence. That would probably mean employing lawyers.

The long answer is that I doubt whether any such licence would be enforceable in practice. Further, in general scientific software that is released under restrictive or unusal licences is generally not well used by the community. See Lior's Pachter experience. Important here is not that this was about for-profit vs not-for-profit use, but that when people are unsure about a licence, they just won't use your work.

If you don't care that people won't use your work you must ask why you are releasing it in the first place. One hopes that you are releasing it because you wish to further the horizon of human knowledge and believe that your library has a role to play in doing that. If nobody uses your library, you are not meeting your goal any better than if people used the software in non-reproducible science. I you do not want people to use your library then you are not doing science, anymore than those that refuse to release their data or methods are.

Putting all that effort and cost into inventing a new licence, which in the end might lead to people not using your software is a poor use of resources to achieve your ends. Much better to expend that effort convincing journals not to accept non-reproducible things, reviewers to actually review the reproducible of the code and into creating infrastructure that makes it easy to do.

Answer 2

If you want to prevent somebody from incorporating your code in further work without also making their code available, then AIUI the GPL (not LGPL) does that.

But you seem to be saying "any scientist who makes advances based on the work I have done must do what I say". And that isn't your choice to make. If you publish your research, then (short of patents and the like) you are implicitly giving the rest of the world permission to use your conclusions to make further discoveries in the future. That's sort of the point of it all. You don't get to specify who can use the knowledge you provide, or in what manner. If you don't want other people to build upon it, don't publish it.