53

I am a PhD Student in computer science. During my my first 2 years I developed an algorithm that seems to work fine in a certain field of applications.

My adviser recently told me that he is starting a spin-off with a colleague of a different department and that he plans, in a couple of years to use my algorithm in it.

Now, considering that:

  • I am not interested in joining my adviser spin off as an employee
  • My advisor can understand the outcomes but not the 100% of the logic of the algorithm
  • In the future I want to make my own company based on that algorithm

I am quite worried.

How fair (and common) is that an advisor plans to use the code of one of his to students to make money (none of which will finish in my pockets)?

The fact that he will use my algorithms, will limit in any way me to use them in the future for the same scope (in a different field)?

May talk to him be a good way to solve this problem?

I don't know how to deal with this fact.

  • 62
    Get a lawyer to read your contract and figure out what rights you have with respect to the intellectual property. (Note that some universities in the US have free Student Legal Aid programs.) – ff524 Oct 7 '16 at 11:18
  • 17
    @UlderiqueDemoitre Is it a public repository? If yes, what is the license that you gave the code? – DCTLib Oct 7 '16 at 12:51
  • 14
    Yesm it is public, and under MIT License, so that piece of code can be reused anywhere. But this is not really my issue, because also without any code, anyone can hire a developer and make him implement the algorithm publicly available on the paper. – Ulderique Demoitre Oct 7 '16 at 13:11
  • 21
    @UlderiqueDemoitre If the code is under a permissive license and the algorithm is public, then anyone who wants to can use the code under the license or the algorithm however they want to. That was the outcome guaranteed by the decision to make the algorithm public and the code available under the MIT license. Anyone can build a company around it. – David Schwartz Oct 7 '16 at 16:45
  • 48
    To be clear: your algorithm is published in a paper, and the implementation is public and published under MIT licence? That literally means anyone can use it for any purpose. – njzk2 Oct 7 '16 at 16:51
104

Actually, I do not really understand your general strategy: If the code is published under MIT license and the details of the algorithm are published, then pretty much anybody can use this code or algorithm for his/her own company (as far as I understand the license, I am not a lawyer). So it is perfectly fine for your supervisor to use this code. It is also perfectly fine for you to use this code in any company you found (again, as far as I understand the license details).

Probably, a scientific algorithm will only be a tiny part of software that you want to sell or for which you want to consult. So I would not worry too much.

  • 13
    What do you mean by "limit" if you don't mean it legally? – J. Fabian Meier Oct 7 '16 at 13:42
  • 47
    As I said: I wouldn't worry too much. To make money with software, you have to have a good algorithm, but there is much more to it. The aspects of implementing, usability, testing, promoting etc. are often vastly underestimated by people in academia. – J. Fabian Meier Oct 7 '16 at 13:46
  • 30
    @UlderiqueDemoitre You already made the decision to give your code away when you posted it on the internet publicly under the MIT License. So any question about “limiting” its usage now is nonsensical. You may choose to keep further revisions to yourself if your legal arrangement with the university so allows. Indeed, your publishing publicly and under MIT license may or may not be a violation of your obligations to the university (I don't know) – All this should be considered before publishing, not after. – Basil Bourque Oct 7 '16 at 20:47
  • 6
    We have a couple of PhDs on rolling loan to our company (part of a knowledge transfer gov't scheme). They're incredibly smart people but utterly clueless when it comes to business realities. The truth is that even if we told the world how we do what we do, it would take years for a 20-man team to turn that process into a robust, tested, accredited piece of business software suitable for sale to Enterprises. We couldn't do what we do without the algorithms, but they're still a tiny, tiny part of the whole operation (and not even the most important). – Basic Oct 7 '16 at 22:47
  • 13
    @emory you can't revoke the rights you granted, unless that's specifically part of the license - and in the case of the MIT license it is not. People who obtained the code through the MIT license can continue to use it and distribute it further under that license. – Martijn Oct 9 '16 at 21:54
28

Taking a slightly different tack from @ff524's comment on your question, I strongly recommend you consult with university legal staff. A previous advisor of mine owned a business outside of the university. He had made particular legal arrangements with the university to ensure that no IP (intellectual property) was being taken between the two businesses, and any IP that was being used was documented and accounted for. Neglecting to do this puts your advisor on the path to lawsuits, and suffice to say, "thar be dragons."

I won't even try to comment on your own liability in this situation, because I Am Not A Lawyer™. This is a legal matter; consult a legal expert.

  • 8
    Yes, he does need to talk to the legal staff, but he also needs to keep in mind that his personal interests may well not align with those of the university, and his own representation might well be appropriate. – Scott Seidman Oct 7 '16 at 13:29
  • 20
    "Consult with a lawyer" and "Consult with university legal staff" are two entirely different things. Let's not pretend the university is a disinterested party, or that the advice would be neutral. – smci Oct 7 '16 at 19:41
  • 2
    @smic - Fair enough. Fact is, though, given that the university (almost certainly) has partial ownership to any relevant IP, and given that (almost certainly) he can't afford his own lawyer, I would definitely start with the university legal staff, as they do this on a regular basis and can—and frequently do—give very sound legal advice. – eykanal Oct 7 '16 at 20:00
  • 3
    @smci +1, if the university has its own agenda, advice given by them can be very bad (I say this from personal experience, unfortunately). – Bitwise Oct 7 '16 at 20:08
  • 2
    @eykanal: not all universities have dedicated legal staff, esp. non-US, and even when they do many of them are crap. In any case, don't expect them to care about, understand, let alone advise in, the researcher's interest. – smci Oct 7 '16 at 21:45
14

Your university will have an office that deals with the transfer of intellectual property. You need to talk to them.

You need to research your agreement as a student with your university. In many cases, the University is the assignee of your intellectual property. This doesn't mean that you are not the developer, and that you're not entitled to any proceeds/licensing fees/etc. that might result. It means that your share might be predetermined by agreement, and you might not have any say as to what the university does with the intellectual property.

My suspicion is that it's likely that your adviser is not entitled to use your code without some licensing agreement with the university, though the university might choose to issue him that license at no charge as part of their spin-off agreement.

I'm not sure if your agreement with the university allows you to just pick up an algorithm that you developed as a student and use it. You might need to license it from the university as well-- in which case, your interest involve making sure that any licensing agreement between the university and your mentor is a NON EXCLUSIVE license.

I am not a lawyer. If this is valuable stuff, I suggest you hire one. In any case, I strongly recommend that all this stuff gets hashed out ASAP, for the benefit of you, your mentor, and the university. For all concerned, this needs to be bulletproof, and the software will need to be able to hold up to a licensing audit. Nobody would like a lawyer to come in and shut down a business with a cease and desist.

If you used significant non-university resources to develop the algorithm (i.e., your own computer, a compiler you paid for, etc.) you should make a list of this and keep it handy.

This is tricky. I suspect that if things go as routine, the university and your mentor's company might not make this bulletproof, leaving them open to legal action on your part. Just by bringing it up, you might force them to cover their bases, and hurt your own interests. Once again, if this is valuable, you need to hire a lawyer, and I suggest you do it BEFORE approaching the university.

  • 4
    +1 for this answer for the last sentence: "Once again, if this is valuable, you need to hire a lawyer, and I suggest you do it BEFORE approaching the university" -- preparing yourself before alerting them that they need to protect their own interests might give you the advantage you need to make sure you still end up with the right to use your own code. – Doktor J Oct 7 '16 at 15:55
  • 6
    @DoktorJ, & ScottSeidman: The time to be thinking about the legal issues was prior to submitting a paper (with algorithm) for publication and making the code publicly available. Once the code is publicly available with a specific license (e.g. MIT), a considerable amount of the legal situation is already defined. There is, of course, the issue as to any preexisting agreement with the university and if the person who made the code available under the MIT license had the right to do so (likely they did). However, if there were such agreement, the principals should already be aware of it. – Makyen Oct 7 '16 at 18:08
  • Ah I missed the comment where he clarified that it was MIT licensed. That said, being so, he shouldn't have to worry about his own ability to use it -- the same MIT license that lets his advisor use the code will also let him use it (this is of course a generalization, but you get the idea) – Doktor J Oct 7 '16 at 18:13
12

If you look at it from purely economic perspective, you have two viable choices here:

  1. To start a company as fast as possible and beat your advisor to it.
  2. Innovate and improve your algorithm over the next two years so the current version becomes obsolete and then use the new version for your business.

Do not waste time and money on legal squabbles. Even if you patented your algorithm or published it under proprietary license, it still wouldn't be worth fighting a legal battle with your advisor for it. Why? Because if there is really any value in it, 10 more companies will gracefully clone your algorithm and flood the market leaving you and your advisor behind. On top of that, suing someone over software patent infringement rarely pays off. This wikipedia page can give you a sense of what it's been like.

  • I think this is the most practical answer. Beat them to the punch. If your personal financial status is like most graduate students, you don't have the money for a legal battle. To avoid future ownership issues, use the algorithm as published in the paper, re-code it from scratch on your own time off-campus, and use a computer that is not owned by the university. – Jesuisme Oct 9 '16 at 13:30
  • And document this entire process in case someone tries to bring it up later. – Jesuisme Oct 9 '16 at 13:44
  • 1
    @Jesuisme This assumes that the publication of the algorithm and release of code were OK in terms of prior obligations to the university. If there is any grey area at all, then the university's first step may well be a lawsuit alleging the opposite, which a startup probably would not have the time or resources to fight effectively. – András Salamon Oct 9 '16 at 21:03
  • @AndrásSalamon I agree it is a possibility. I'm not a lawyer and I can only speak from my experience in the US. The university would be required to send notice of infringement before starting a suit. And in most cases, after receiving the notice of infringement, a licensing agreement is reached. From the start, the right way to do this would have been to license the algorithm according to the school's policies. But the OP is in a situation where the most profitable option (given they want to start a company with the algorithm) is to do it first. – Jesuisme Oct 9 '16 at 22:00
  • Depending on the University's standard licensing agreements, OP may not have the rights to use his or her own code any more. – Magisch Oct 10 '16 at 11:18
9

You wrote an algorithm and published it, and wrote code to implement it with one of the most permissive use licences available - there is nothing you can legally or ethically do to prevent your professor from using and profiting on it.

By using the MIT licence, you effectively said that you are releasing it for free use, without concern for how others use it. Open-source licences aren't a cudgel to prevent others profiting.


Regarding your actual questions:

How fair (and common) is that an advisor plans to use the code of one of his to students to make money (none of which will finish in my pockets)?

Not very fair. But fairness doesn't come into this.

The fact that he will use my algorithms, will limit in any way me to use them in the future for the same scope (in a different field)?

No. You made all the code public so you can use it however you want too. Although if the field is related enough that the algorithm can be used, he may have a better product or better market position.

May talk to him be a good way to solve this problem?

Probably not. There is no problem to discuss.

I don't know how to deal with this fact.

Move on.

  • Best answer I've read so far. Given the information (from comments) that this released under MIT, then it's as simple as this. – GreenAsJade Oct 11 '16 at 5:17
6

Taking another angle based on your comments, having released the code under an MIT licence means that anyone can utilise it for commercial or non-commercial uses (see this other stackexchange question). However, they must include proper attribution and inclusion of your license in any distribution.

Namely, there shouldn't be any ambiguity over the creation of the algorithm and your name should be included in their company's distribution (as included in the license). If this wasn't the case, that would be a legal issue. If you were to implement it separately to your advisor in years to come, you'd have as much right to it as them, as long as you didn't infringe on any of their implementations.

2

My reply is based on the premise that the results of scientific research work should be made available, without restriction, to all of humankind, and thus no one - neither your advisor nor yourself - should be able to benefit financially from monopolizing such an algorithm or restricting its use.

That is not to say that you, or him, or your university, should not start companies which do commercial work based on that research (*) - but certainly don't base them on restricting access to knowledge and insight. You could further improve your algorithm; you could develop useful products based on it; you could offer services involving it; and so on.

As for your questions:

How fair is for an advisor to use his student's code to make money (none of which will finish in the student's pockets)?

If that code is publicly accessible - as fair as it is to "make money" in general. That's doubly true if the algorithm was the result of funded research (which, if you're a Ph.D. candidate, is most probably the case). However, if most of the effort towards making that money was actually producing that code, then it becomes less fair - but then it's more likely that it's not fair for any of you to make money (since you're probably doing it by patenting or close-sourcing etc.)

How common is for an advisor to use his student's code to make money (none of which will finish in the student's pockets)?

It happens occasionally, in my experience anyway. That's not nearly the worst thing advisors do to their advisees.

The fact that he will use my algorithms, will limit in any way me to use them in the future for the same scope (in a different field)?

It shouldn't - but just check the licensing. You should do well to publish the code someplace like on GitHub or BitBucket and with a nonrestrictive license; then if a bunch of people are using it, the cat will be out of the bag and even if someone has some legal argument to make against the public use of this code they'll likely not try to pursue it anyway. Set your code free!

May talk to him be a good way to solve this problem?

The thing is, the question of whether there's a problem at all, and if so, what the problem actually is, is not clearly and consensually resolved, and people who don't agree on nature and the very existence of a problem are unlikely to "solve" it.


(*) - That's another question on which I have strong views which are beyond the scope of this question.

  • +1 for arguing about funded PhDs, most of them do have clauses about the availability of produced work. – grochmal Oct 9 '16 at 0:32
  • @grochmal: I was only referring to the morality rather than the legal obligation, but that's a valid point as well, I guess. – einpoklum Oct 9 '16 at 14:15
1

File a patent first, and then submit your paper for review.

If your algorithm is published before the patent filing, you can't patent the code and use it for profit; this is the case unless your algorithm has been public for less than 12 months. In the USA, there is a 12 month grace period between public disclosures and patent filings. I believe, but do not quote me on this, that the EU has no grace period post public disclosures.

I had a professor publically disclose a material/scheme/application that myself and a few others conceived. The intellectual property was lost because of this motion.

  • 3
    The EU case is mostly irrelevant, as there are no patents on algorithms & software in the EU (see en.wikipedia.org/wiki/…). While such patents have been issued anyway in the past (typically in connection to some claims about hardware), this is a very brittle business, so probably nothing that a private person would want to file. – DCTLib Oct 8 '16 at 11:25
  • 1
    Filing patents take too much time and money for most graduate students. And you'd have to ask an IP attorney if the algorithm is patentable given that it's already published under an MIT License. – Jesuisme Oct 9 '16 at 13:38
  • Furthermore, the OP is probably regarded as a University employee and with that a portion of the intellectual property of his work belongs to the university as well. At least that's the case in Europe... – Dux Oct 10 '16 at 7:15
  • 1
    @Dux This kind of thing really does't generalise across jurisdictions. For example in the UK students are not generally considered employees, principally because the main sources of funding aren't considered to be a salary. Of course the UK won't count as Europe much longer, but that's a different issue. – origimbo Oct 10 '16 at 10:37
  • How can he file a patent on something that's already been published? AIUI patents can only be filed on things that specifically have not been previously disclosed. – GreenAsJade Oct 11 '16 at 5:18
1

This depends completely on the laws your university operates under. In some legislations it is the default that it is the university that owns any legal rights, but in some others it's not true but rather the teacher / researcher who owns those rights exclusively. The fact that the code has already been published could be more of a worry in the case the local laws are so restrictive that the researcher or groups are not in power to make those calls on their own.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.