PhD advisor is pushing me to do a project which is doomed to failure

Question

I am in a STEM field, my advisor is a theoretician, and my work is mostly experimental. My job is to implement an idea that exists purely in theory and see how it holds up in practice.

Recently, my advisor asked me to look into doing an experiment based on theory which he had published several years prior (but which was itself not substantiated by any experiments).

I initially had reservations, as I felt the paper he had published used a model which was simply not realistic enough to be useful in practice, and that it would be difficult, if not impossible, to run experiments to prove his theory. I did not voice these, as at the time, I was a very junior student who knew very little, so this was just my gut telling me that this probably wouldn't go well, but I might as well try and see what happens.

After spending a lot of time (roughly 8 months) implementing his ideas, and testing them first-hand, I came to the conclusion that his model does not work in reality for a multitude of reasons. He seems convinced that I am doing something wrong and has even harshly compared me to others, saying "if I gave this task to X, they would have easily done this experiment in a week". While I could have believed such a statement at the start of this project, I am now convinced, based on my own expertise, that he has no idea what he is talking about, as all the work he does is in theory-land, and he has never done experimental work before, and doesn't understand fundamental differences between theory and practice. But he is adamant and told me if I don't finish this project with good results to meet a certain deadline, then he would probably have to stop supporting me with funding.

Three big problems:

1. He has a big ego, and is convinced his theory holds in real life, even though he has no experimental experience.
2. It's just me and him on the project, so there is no unbiased third party that is also an expert on the topic whose advice we could use.
3. He is operating from a sunk cost fallacy: "I spent so much time and effort on this student so he better produce something rather than spend all this time to get a negative result. Might as well get rid of him, so give him an ultimatum".

Is it time that I finished working with this professor and his doomed projects ?

I see little recourse, given the circumstances, and am curious what course of action other intelligent people would take in such a situation.

Answer 1

I would change advisors. You two are clearly not compatible.

Leaving aside the particulars of the case, I think you need to find an advisor who is also an experimentalist.

Of course, you want to do this as diplomatically as possible, especially if your advisor has a lot of power in the department. You might want to talk to the department chair. But programs want graduate students to succeed. Even if not for emotional reasons, then for practical ones. Having students not graduate looks bad.

Answer 2

I hate to suggest it, since I doubt it will help you, but showing definitely that some approach fails can be as valuable to the research community as showing that something succeeds. It is the truth of it, as best we can determine, that matters.

I worry, though, if your advisor is implying that you have to verify their underlying assumptions, rather than truly researching the question even if it results in a refutation.

You don't seem to have a trusting relationship, otherwise I'd suggest you talk it over with them. You can judge the wisdom of that approach.

But, I have to agree with Peter Flom that a new advisor might be your best route to success.

Answer 3

In the long run, academia will not honor your work if it somehow looks bad, even if this is not your fault. Your advisor should not be the only person in the room with an ego.

Don't just comply with what you think is an unreasonable demand. But be tactful. Both of the above comments contain good advice, I think.

Answer 4

A negative answer is also a useful answer itself. Why could you not just provide some 90% C.L. rejection of the theory or its certain rules as the output of your experiments?

Describing the theory itself, detailing experimental methods, providing statistical analysis and summary, and giving conclusions and future outlooks is already good research. Moreover, you can complete this task with ideas of experimental setups, how to verify results using other approaches, etc. It's hard to recommend what to do without domain information, but overall, work is done; you just need to validate results, discuss with your supervisor in a good manner, and publish a paper. Who knows, perhaps the experiment was incorrect and will be proven in the next 20 years, or perhaps you will be proven correct soon.

Negative results are also results; science is not only about positive results.

Some physics' domains publish exclusion regions with some C.L. because they didn't find anything, but what they found was something where the region is not possible to happen (exclusion region).

Good luck with your research. Let us know how things evolved at the end.