Timing a career switch from math to TCS

Question

Short version: Is it better to switch focus as a postdoc or wait until after tenure?

Details: I'm a PhD student in math (combinatorics) graduating next year. Let's suppose that I'm a strong candidate for research-oriented, tenure-track positions in math, but my long-term research program is interdisciplinary with TCS. If I have a really productive 1-2 years of postdoc in TCS, will I have a shot at being hired in a good CS department in the US?

I'd rather be in a CS department, but many people in my area don't share that enthusiasm (or are located in Europe). I'm worried that I'll end up with papers in both general math journals and STOC/FOCS and still be unhireable. Should I wait until after tenure?

Edit: Here's why I'm asking. I have postdoc offers in both math and CS departments - I want to take the CS one, but all of my publications are in math journals, and the CS postdoc won't make me a stronger candidate for math departments.

Answer 1

My 2 cents. I would say that getting hired in the theory area in CS departments, even in good times like now, is difficult and depends on many factors. Counting on getting hired even with a productive postdoc is not a good idea. You should try but keep the expectations reasonable and be prepared to be hired in math and enjoy doing the work you like at the interface. If things work out really nicely you can try moving later in your career as well.

Answer 2

It depends a lot on the kind of research you do and the CS department you are applying to. But my guess is that getting a tenure-track position in TCS will be harder than in math. There are some reasons for this:

• I think that even if you do a postdoc in CS, people will likely see you as a math person (by background and training). In my experience I am not sure that is a positive: while there are lots of strong CS candidates getting hired on interdisciplinary research agendas, usually it comes in the form of applications of their research -- applications to medicine, biology, etc. Even in TCS, it looks really good if you are applying your TCS research to some specific problem domains (say in computational biology, machine learning, systems, whatever).

• To get hired in CS, you have to not just convince the TCS person who is fighting for you, but that person also has to convince the rest of the department. So even if your research is getting published in STOC and FOCS, you also need to have a clear impact in computer science as a whole and that impact needs to be unquestionable.

• Usually part of the hiring process in US universities is a talk that you give to the entire department. These talks are very general-audience and need to have a clear focus on specific applications and problems, with minimal mathematics (aside from boiling technical things down to some very simple ideas that are digestible, and maybe a few slides where you explain some proof in detail but no more than that).

  I have found in these talks that it can be a bit harder for TCS researchers, compared to those who work on systems, security, etc., because you just need a lot of clear applications. This is where it depends a lot on your research; have you worked on finding applications of your research to some other areas of CS?

I'm worried that I'll end up with papers in both general math journals and STOC/FOCS and still be unhireable.

I think here what matters is how to make yourself the strongest candidate, not the specific journals and conferences (assuming they are top journals and conferences). If your research agenda is interdisciplinary, no one will be surprised that you are publishing in both and that should be seen as a positive.

But I think the "interdisciplinary" pitch is likely to work a lot better in math departments than in computer science departments. So if that is the road you are going for, I would stick to math for now (depending on the impact of your work in TCS course -- if you are proving P = NP, your background won't matter).