In my department, there are 2 styles for a supervisor to supervise student.

In the first one, a supervisor understands his PhD student's research deep enough to provide technical advices such as which algorithms to try, possible solutions for a problem, or even to be able to read and comment on the source code.

In the second one, a supervisor understands his PhD student's research just enough to comment about the feasibility and novelty of the research and its contribution, but does not understand exactly how the research is carried out.

In my opinion, the second style is easier for a supervisor to not only supervise his students but also looks for new ones, as his candidate pool is larger. It also helps a supervisor to have more diversitized research group.

On the other hand, it is harder a supervisor who has deep knowledge about the research of his students to find a new one, as he can only look for those whose research is close to his. But this makes the research of his group more focused.

So, I wonder which one is a common practice, and if a supervisor should have deep understanding, e.g. technical level, about the research of his student.

My field is computer science but answers from other fields are also welcome.

  • You are assuming that the PhD student's research topic is selected without being strongly influenced by the advisor. Although this is sometimes the case, there's a strong negative correlation between the first style you mention and independently-selected topic.
    – Ben Voigt
    Commented Jun 9, 2015 at 21:55
  • 8
    A good supervisor probably should be capable of understanding the technical details of what their students are doing (apart from the occasional genius, perhaps), so that they can help if problems arise. But to expect the supervisor to know this level of detail at all times seems to invite micro-management (and is completely unfeasible if the supervisor is responsible for more than a couple of students). If my students are making good progress, and getting sensible results, I don't see any need to inspect every line of code that they've written. I expect they would feel insulted if I did.
    – avid
    Commented Jun 9, 2015 at 22:03
  • 3
    Case 1 sounds more like supervising a BSc to me, possibly a MSc. PhD students should not require so close supervision. (It's good if the supervisor understands the PhD student's field so deeply, but I'd say it should not be required.) Commented Jun 9, 2015 at 22:21
  • 8
    @StephanKolassa At least in my field, it's a huge disadvantage to try to do a PhD thesis in an area your advisor is not an expert in. For one, it can easily happen that the PhD research turns out not to be new. Another, my field is very technical, and it is not easy to find good problems/approaches on your own.
    – Kimball
    Commented Jun 10, 2015 at 0:05
  • 4
    A supervisor needs to know and understand enough of the work to accurately assess its worth and quality. He has a personal stake in this. A supervisor risks negative consequences for himself if a student's work is revealed as significantly flawed when he goes to his defense. Commented Jun 10, 2015 at 2:19

4 Answers 4


Both types of mentors can be good. A good mentor of the second type will make sure that there is expertise somewhere that a student can go to for advice and help if called for.


To directly answer the question of what the common practice is: there is no common practice, as the level of expertise depends on the topics at hand and the same advisor may supervise more than one student at a time, each one of them being involved in their own different projects. In my opinion the question is whether it is best for the students to have supervisors who are technically skilled in the area the project is dealing with.

I have got a PhD in mathematical physics and had a supervisor who was by all means deeply expert on the mathematical technicalities of my topic: since a PhD in this area is basically about proving theorems and doing analytical calculations, I found his technical help always very precious, although, on the other hand, sometimes he had the tendency to get bogged down too much on those technicalities. However, do not forget that the purpose of a PhD is also (albeit not only) to eventually produce and publish material: in this respect having a supervisor who is technically skilled does help a lot because it reduces the dead end paths.

I have also had the opportunity to teach exercise classes for graduate students in mathematics and I have always helped them especially in the technicalities, because it is the only way to fully understand what you are doing. Of course nothing is required, but helping on the actual work is always best to have.


In my opinion, the supervisor has no duty to provide technical advice or to control the general direction in which the research is going tightly but it is his duty to verify the results (either himself or by finding an expert who is capable of going into the nitty-gritty of the project). The common understanding is that the PhD student is normally just crossing the line between being a "schoolboy" for whom the ultimate criterion of truth are the grades given by his teachers and being an "independent craftsman" who is solely responsible for the quality of his work. So, if you use style 2, you should be aware that you may have some trouble at the end if the student's maturity and intelligence levels are short of perfect.


There is no "requirement", per se. However, the principle investigator is 100% responsible for work published from his laboratory. In general, (in my opinion) an academic advisor/supervisor should be knowledgeable in all aspects of the research conducted in his laboratory. If the advisor/mentor does not understand everything, how can he be a source of mentorship or supervision? If anyone is unfamiliar with technical aspects of the research, it should be the student. This is a primary reason for going to graduate school in the first place - to learn! That being said, nobody can expect a PI to know every miniscule detail of all experimental protocols in his/her lab. There are other exceptions as well. For example, I know a chemistry professor who hired a molecular biologist to work in his lab. I wouldn't think that the professor is an expert in the technical aspects of the research conducted by this molecular biologist, but he most certainly understand the science.

You must log in to answer this question.

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