Andy's Math/CS page

Saturday, October 18, 2014

Making academic contacts (some thoughts for new researchers)

Suppose you're an undergraduate hoping to go into academic research, or a beginning grad student.  It could be very helpful to have academic contacts at other schools---such as professors, but also maybe postdocs and grad students. (I'm concerned here with starting scientific discussions and/or collaborations, not making contacts for graduate admissions per se.  Admissions decisions will be made based on your application.  But a successful collaboration and resulting publication, or letter of support, is one of the best things you can hope to add to your applications.)

Say you've never written a paper, and may not yet have the time or readiness to solve major problems on your own; but you want to have real scientific conversations and perhaps even collaboration with a busy researcher you admire.  What could you possibly talk about?  What can you offer them?   Below is my advice.


1)  In my view, the best approach to making contacts is closely related to how I'd recommend spending your own study time as a beginning researcher:  (a) start reading research papers as soon as possible, and (b) try diligently to ask interesting follow-up questions to the papers you like.

Asking questions is a central research skill.  It is one you can start learning early on because, even in technically difficult fields, simple patterns of questioning often recur in many forms across papers.  E.g., in the theory of computing, if there is a new model of computation being studied, it is likely to have randomized or nondeterministic versions that can be studied next.  Or if there is an algorithmic problem, it might have variant problems to be solved in, say, the communication protocol or decision-tree models.  And so on.

Even though it is not so difficult to produce these variations, few people do so systematically.  If you keep at it, and keep trying to learn and adopt more patterns of questioning, you will hit upon interesting and solvable problems.  You will also start to develop a taste about which problems are likely to be most interesting, which are too hard, and so on.  (Side note---I believe that keeping a research journal with your questions greatly helps this process.)

2)  Now returning to the goal of making contacts, I think that one of the best kind of emails you can send is one that contains a good research question, that's reasonably related to the person's research area.  If you can do this and pique their interest, they may well enter into a dialogue that could become a full-fledged collaboration---regardless of your credentials on paper.  After all, you've already brought something important to the table.

A key advantage of this approach to making contacts is that you're aiming to attract the researcher's own interest and curiosity, rather than just asking them for something.  Another advantage is that there is a lot of freedom in asking and considering research questions.  If you ask someone an interesting, specific question about topic X, there is no presumption that X is your only interest or that future interactions will be limited to that.  And in the course of your correspondence, you might realize there is a better question to ask about X, or you might getting around to asking something about topic Y as well.  That's how scientific interactions go.

So you don't have to worry about choosing the exact right question to represent yourself, as long as it is good and leads to discussion.  In this respect it's less stressful than trying to introduce yourself by defining your whole research outlook.  And it's certainly more promising than suggesting a collaboration based on your GPA or work experience.

Of course, asking good questions isn't easy and takes work. You should think carefully about them before sending, try your best to answer them yourself, and see if there are initial observations or partial solutions you can provide with your question to show you're serious.   Maybe in the end it will be a question they can answer right away; but again, it could still lead to other questions.  There is little risk in trying and it is likely to at least be a useful exercise.


  • Nice to have you back blogging (and thanks for the interesting advice

    By Anonymous Mohammad, at 11:07 AM  

  • Just at the point: find a problem, "one foot from a trivial away", try to solve it by yourself. If you already here face big problems - ask some expert in this stuff. The "one foot from a trivial away" here is important: asking questions experts are trying to answer for years is a very bad deal.

    Another way is to try to solve some special (not-so-trivial) case of a "big" problem, and contact an expert you know is working on it. This will definitely wake an interest. This happened with one student who just came to me with a sketch of a proof that so-called "multilinear" boolean circuits for CLIQUE are no more efficient than DNFs. None of "experts" has probably looked at this question before ("too special", "too simple"). But to make a rigorous proof was not such a simple deal at all. Many hidden aspects of multilinear circuits were detected by us both along the way. He then published a nice paper on it (of course, alone because initial ideas were entirely his).

    Also, one high-school student from Russia has had some ideas on how to prove Sperner's lemma about antichains for multisets. This also immediately waked my attention (and Razborov's - he told me about this). Etc. This way seems to be even better than just asking questions (not to speak about sending CVs or work experience).

    By Anonymous Stasys, at 12:51 PM  

  • I agree, coming up with ideas to solve a standing open question can be more impressive than just asking new questions. But, it's also harder :-) And I am addressing people who might not have the readiness to do this yet.

    I think asking questions is a very good practice for young researchers, and it creates opportunities for them to apply problem-solving on problems that haven't already been scrutinized by experts. And as I say, when sharing a question with an expert it's best to include whatever partial progress you can manage.

    By Blogger Andy D, at 6:39 PM  

  • Andy, I meant not "coming up with ideas to solve a standing open question". Nobody expects this from a beginner. I rather meant "coming up with ideas to solve a SPECIAL CASE of a standing open question". So called "experts" are usually too busy to spend time for "special cases". Hence, they are happy seeing that some student had done this. And want that student have in the team.

    By Anonymous Stasys, at 1:58 PM  

  • Got it, thanks Stasys.

    In some cases identifying a good special case to solve is itself a challenge, and requires the sort of questioning skills I discussed.

    By Blogger Andy D, at 2:49 PM  

  • Great info! Thanks for sharing.

    By Blogger Rakshita Sharma, at 7:48 AM  

Post a Comment

<< Home