I am a postdoc who was supposed to co-supervise a PhD student along with another academic who invited me for that. However, I ended up fully supervising the student after the supervisor became too busy to attend meetings. Eventually, I also ended up defining a problem for the student, which, after 2 months of work, we discovered it is invalid. How should I deal with this situation? I feel that it's my fault and feel bad for the student. The student is in his first year, and this is his first research problem.

Any ideas on this from experienced supervisors over here? Thank you.

Edit: Thank you, everyone, for contributing your answers and comments. For those of you talking about failure/recovery/publishing what's in hand, I would clarify that the discovered issue is in the problem formulation, where the solution would involve inverting a rank-deficient matrix and differentiating a non-differentiable function. Thus, a mathematical solution is not defined. Therefore, the formulation is invalid. This wasn't foreseeable at the beginning, and only when the student made some deviations, it was possible to reach this conclusion. Now, in terms of publication and thesis content, the efforts exerted thus far don't mean anything.

  • 56
    Tell the student it is 'normal' part of research. As for recovery, I would consult an experienced researcher in the area to see whether he/she can 'rescue' your problem or the work done so far. In my experience, there are usually many branches one can take. So you may backtrack a bit and follow another line of research. Jun 14, 2018 at 1:24
  • 31
    Failures can be discussed in a thesis just as much as successes - keep that in mind. A failure won't necessarily produce a paper, but still provides "thesis material".
    – DetlevCM
    Jun 14, 2018 at 6:17
  • 14
    What exactly do you mean by "invalid"?
    – Vincent
    Jun 14, 2018 at 8:30
  • 3
    Regardless of how you deal with it with the student, you definitely need to clarify your formal position with the university. Postdocs aren't usually supposed to be acting as students' primary supervisors. Even if you're satisfied that the arrangement is "legal", you should make sure you're very familiar with your university's formal policies for PhD supervisors, since you are effectively in that role. Jun 14, 2018 at 12:52
  • 2
    2 months is fine unless it prevented them from making a deadline (like choosing an official adviser)... first-years can easily "waste" more time than this (e.g. a semester) while they try to figure out what they want to work on.
    – user541686
    Jun 15, 2018 at 19:43

5 Answers 5


On the grand scheme of things, what you just described is not the worst I've heard of. There are cases where a student spends years of their Ph. D. on a problem with nothing to show for in the end, for a variety of reasons. Sometimes even experienced advisors will, intentionally or unintentionally, end up letting their students work on ideas that are terrible in hindsight. A couple of months wasted in the first year really is not a big deal (and there are certainly worse ways to waste one's time as a grad student). Besides just learning the lesson and moving on, here are a couple perspectives that I find helpful:

  1. Defining the research question is not the sole responsibility of the advisor. It is true that the advisor sometimes has to act as a quality check for the students' work, but the point of PhD is to take responsibility for your own work. So the fact that the student just accepted the research question without asking whether it's valid or not (I assume) is partly the student's fault as well.
  2. Try to salvage the effort spent as much as possible. Oftentimes it is possible to recast what you've done in light of a different research question, and you should consider whether that's possible in your case. I'd go so far as to say that one should always work on things that will be publishable even if your original hypothesis turns out to be wrong.

It might be worth checking the policies of your university in relation to supervision requirements. At my university a junior academic (e.g., post-doc) cannot be the primary supervisor of a student, and needs to take a secondary position under the primary supervision of a more experienced academic. From your description it sounds like you have been a victim of the bait-and-switch, being appointed as co-supervisor but then effectively doing the job of a primary supervisor. If you are uncomfortable with this, you could raise it with your Head of Department, and you'd get a sympathetic hearing.

As to the present issue with your student, two months working on a badly formulated problem does not seem to me to be that bad. It is possible that the student would have learned something from the experience anyway, and you should try to help the student draw useful lessons from the failed attempt. If this is impossible, and you think it really was a total waste, you should just apologise to the student for the wild-goose-chase and move on. Plenty of students have had a lot more time than that wasted by bad supervision, and two months is very little in the scope of their candidature.

  • 2
    It's also the case in the UK (at least in all the universities I've worked in) that a postdoc can't be a supervisor (or co-supervisor). We can't even be a formal second supervisor. That of course doesn't mean we can't do a lot of looking after , but that the overall direction of the research, and responsibility for the student's progress, are in the hands of the permanent academic.
    – Chris H
    Jun 14, 2018 at 8:35

A few short thoughts:

I eventually also ended up defining a problem for the student,

That's a tad over-controlling. Defining the problem should be at most a joint endeavor of the PhD candidate and the supervisor.

which after 2 months of work, we discovered it is invalid.

Invalid, or too hard to tackle? Or perhaps not as interesting as you had believed?

The student is in his first year, and this is his first research problem.

Yeah, I really dislike this custom of a direct PhD track as the default. It's better IMHO to have a smaller-scope M.Sc. as the default target for graduates interested in trying research, while a direct PhD should be reserved for people who have a good idea of what research they want to do exactly. Your situation is part of the unfortunate but foreseeable consequences. Of course, it's not relevant to suggest to the student to switch...

For more concrete advice I would need more context, such as what has motivated the Ph.D. candidate to go into doing research, what motivated you to choose/define the problem you did, whether this is part of a larger, multi-person single effort or not, etc.

  • I agree with this answer, it brings to attention hard truth that no one wants to admit. The high percent of students are not ready for direct PhD
    – SSimon
    Jun 14, 2018 at 14:13
  • 2
    "For more concrete advice I would need more context" - this sort of thing is precisely what comments are for. You can comment on the question itself to ask clarifying questions. As written, your answer comments on the OP's situation, but doesn't answer the question on what to do now.
    – user8283
    Jun 15, 2018 at 15:52
  • @V2Blast: Well, you're right, but I already had enough for an answer.
    – einpoklum
    Jun 15, 2018 at 15:53

Before you abandon it, dig deeper into the problem to find the nugget of gold.

"Success is failure turned inside out" [Whittier]

"In response to a question about his missteps, Edison once said, “I have not failed 10,000 times—I’ve successfully found 10,000 ways that will not work.”


You should give him a substitute problem. and you should know better. Heck, I knew enough as a grad STUDENT to make sure "you can publish even if it doesn't work". Leave the climbing mount Everest to people with risk tolerance. This is an area where professors over the years have burned students.

Plus, you should understand apprenticeship and progressive learning. Don't give too hard problems to start. Figure out something doable. There is time to move to harder stuff after mastering the easier.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .