Best license for an applied research project

Question

I am looking to incorporate a license in the source code of a research project of University A, in which I was the last contributor. Historically the computer-science school has not forced researchers to publish the code with any specific license; they mainly care about the ranking affected by publications, and releasing the code publicly is suggested but not mandatory. Obviously researchers cannot sell the code as the commercial use is reserved to the institution (although software patents are very rare in my country).

Given that the project could benefit from a dissertation that I have been developing on University B, what is the best license that fits for this case? I personally prefer to release the new improved version in a compiled format, and keep the new source code private, so GPL is not an option for me. Is the MIT license a better candidate? (It says “and/or sell copies of the Software” though.) The project in University A is an applied research based on state of the art that have existed freely for a long time.

Answer 1

You don't own the source code for the project, so you shouldn't be the one making this decision.

The fact that you were the last one to contribute some code doesn't mean that you have the right to slap a license on the whole thing and dictate its further use. In fact, even if you wrote the whole code, copyright is almost certainly owned by the university. The university legal department is probably who actually owns this decision.

Authority to decide the licensing might be officially devolved to the computer science department. Or it might be that CS is just doing their own thing because nobody really knows what is going on. You also might get away with just putting a license on it yourself. Someone in the department might even tell you that is fine because they don't care.

The problem with such decisions taken without the real authority to take them is that someone might start caring later--for example, if the software becomes a successful product and somebody (not the university) is making money on it. Many open source licenses (including the MIT license) are incompatible with the university retaining rights to commercial use. The university would have to explicitly decide they are ok with giving up this right for such a license to be ok.

Your own self-interest in the matter complicates things.

If I understand you correctly, you want to open-source the project at university A solely so that you can take it to university B and use it (and you don't want to share it further after that). Frankly, this doesn't look great. If there is any ambiguity in the licensing, or any dispute arises, the fact that you did this for your own benefit rather than sharing with the research community is likely to color the whole thing. Any benefit of the doubt will be gone.

I would not do this. Instead, explicitly discuss with the project your desire to contribute (and collaborate).

Rather than open sourcing as a means to an end, get what you really need: permission to work on the code and collaborate with them from another institution.

• They might decide that the best route to doing this is open source--in which case let the department lead that.
• They might decide that simply giving you permission is the right route.
• Or they might even say no--in which case be glad that you are finding out that answer this way rather than in a much more contentious way later.

Answer 2

Yes. MIT should be the first license anyone considers these days for an open source project, and you should choose against it only if you have a good reason to. MIT is very permissive, and so others are allowed to read your code and learn from it.

With many of the copy-left licenses, due to their derivative works clauses, many people in places like national labs are not allowed to even read GPL-licensed source code because their lab doesn't use a GPL-compatible license and thus any "derivative work" would be license breaking (I know a few whose lawyers have been very adamant about this). Thus, you should think of GPL as "closed source" because, except in the GPL-only circles, lots of people cannot ethically read your code and work from it. The GPL license is made to make sure people contribute code back, and thus if this isn't your purpose you should avoid it. LGPL at least lets people link to it safely, so though still closed in some sense, it's not as bad.

Along with MIT, BSD is a very good and popular permissive license. The no-endorsement clause means that if someone uses your code to do something, they can't use your name. A lot of academics closely guard their name as their main product, so in some sense this added protection may be useful.