How to get a job

Question

I'm new to the site. On mathoverflow this would be community wiki, but I don't see how to set that here. Not a research question, but hopefully of interest to professional theoretical computer scientists.

I am a 2nd year grad student in theory, and I was wondering what advice the community had for what I should be doing now to aim for a career in academia. I know I should "do great research" -- yes, I try. :-) I am looking for less obvious advice. How important are social aspects? Going to conferences, knowing great people? Am I at a big disadvantage if my advisor/school are not famous? Does a blog help/hurt my chances?

Thanks!

Answer 1

Ok, let me bite with my own opinions:

How important are social aspects?

I would say that they are very important. Despite popular myth, scientific research is really a social activity -- Your research must interest other people in the area.

Going to conferences,

Very important -- for the previous reason

knowing great people?

Practically it may help a bit if they know you as their recommendation letters may carry more weight - but even this is really second-order.

Am I at a big disadvantage if my advisor/school are not famous?

The truth is that it is often harder to find the "right problems" to work on when you are not at a central department in your area. Human nature being what it is, it may also be somewhat more difficult to get your papers into conferences and journals if you are not from a "famous" school -- but I believe that not by much and that this is quite minor in TCS.

Does a blog help/hurt my chances?

Well, it depends what you write there.... On the average, I would guess that it's a net plus.

Answer 2

Noam and Dana have already given some fantastic advice. Let me echo Dana: Know yourself, work with good people, and be active!

But after $n$ years on faculty recruiting committees, I have to disagree with Noam's response about "knowing great people". Noam is absolutely correct that it is more important that great people know you, but I disagree that this is anything less than a first-order concern. It is not enough to have a great product; you must also convince people to sell it.

It's not uncommon for recruiting committees at strong schools to receive 200+ applications for only one or two faculty positions. They don't have time to read every folder or interview every strong candidate; they certainly don't have time to read your papers. To have any chance at an interview, you need a champion on the recruiting committee, and your champion needs ammunition to push your case. So in addition to a strong research record, you need the following:

• Name recognition. Ideally, someone on (or with connections to) the committee recognizes your name. (One of your advisor's jobs is to bug their friends into reading your application; hopefully, you've made this part of their job easy.) Having a widely-read blog or survey or other community resource can definitely help here. If not your name, the committee should at least recognize the names of your advisor and your other references. If you fail this step, your application may be rejected without even being read; on the other hand, being known for your faults may kill your chances even if your work is fantastic.

• Strong letters from people whose opinions carry weight with the committee. This is by far the most important part of your application package! A letter from your advisor is crucial, but their opinion about the quality of your work will likely be taken with a grain of salt, because they have a personal vested interest in your success. (On average, advisors seem to graduate their "best student in ten years!" every two or three years.) The best recommendation letters come from well-known, well-connected, active researchers at top schools, who know your work in detail and can say great things about it, but who have never worked with you. Of course, to get letters like that, you need work that great people can say great things about.

• Luck. Baruch Awerbuch's observation about conferences applies here: Faculty hiring is a random process whose mean is determined by the candidates and whose standard deviation is determined by the hiring committee. There is absolutely nothing you can do to guarantee success; you can only affect your probability of success. Life is not fair. Let it go.

Answer 3

By far, the most important thing is to do great research. Then, there is making people know you do great research. Here's what I think (the order matters):

Know yourself. What do you like? What are you good at? What's important for you?

Solving hard well-defined problems, or working on an open-ended project? Doing highly theoretic work, or solving problems with concrete real-life applications? Working on small-scale projects, or doing research marathons? How do you learn best? How do you work best?

There are lots of ways to do great work, and the world benefits from them all. Find your own style, and adjust things to make it work.

Also - ask yourself these questions periodically, answer honestly, and plan accordingly: Where do you stand? Where do you want to go? How can you improve, e.g., your knowledge, your skills, your approach?

Work with good people on worthy projects. Work with people that have done successful research in the past, and whom you appreciate, people who are smart, skillful, experienced, motivated, and, very importantly, nice and trustworthy. Work with them on a project in their area of expertise that is exciting both to them and to you. That's the best way to learn.

Be active. After you gained some experience in research and in presentation of your work, talk to more researchers, go to conferences, invite good people to visit and give talks, take internships, visit good places. Write a good survey on your area of expertise. Write a blog or an essay. Teach or co-teach a course or a mini-course. It all depends on what you're good at and what you like doing.

Answer 4

Try to do an internship at a research lab or a software company. An internship can immensely boost the chances of you getting a job. More importantly, spending a summer working with experienced people

There are a few internships that will let you do 100% theory (as in, here's a desk, pencil and stack of paper, go prove theorems), with the goal of writing a research paper at the end.

There are many more where proving theorems isn't in the job description. I believe that the average CS theory student benefits immensely from a couple of months spent in industry. Even if you spend the whole summer polishing doorknobs and writing unit tests, it's worth it.

If you are interested, find relevant people (e.g. authors of papers you like at http://research.google.com, similar exercise for msft, yhoo, ibm, at&t etc). Shoot them a polite email inquiring about an internship. If they want to work with you on research, great. But even if they don't (don't want an intern, don't have headcount, want to work with someone else), look for software engineering internships.

Answer 5

I got another advice. This might sound weird here, but: If you're keen on pursuing an academic career, you might be better off leaving theory and sliding into something more applied, e.g. economics, security, bioinfo, ai. Something where not everybody wants to be in academia, where there are many industry positions, and better paid than an academic ones.

Take a day, research the numbers. Do this with a friend from a different area, and compare results. Go to the CRA website and search for faculty jobs specifically for theorists. Try to find theorists hired on general positions (not specifically listing theory). Estimate the number of theory students the top schools put out every year. Estimate your rank within those. Is this field growing? Shrinking? Look for money, grants, funding. Talk to people about how these effect positions. As you interact with people, become aware of agendas, even on this site.

If you stay in theory looking for an academic position, work with people who are experienced, active and well-connected. Don't go for young, enthusiastic, smart, interesting, positive, with a blog, at least not only.

Answer 6

This has been mentioned previously in one of the other posts but I think it deserves reiterating and that is, looking at combinatorial research in applied areas. There are some very interesting combinatorial problems in bioinformatics (particularly, in DNA fragment assembly, which is essentially a combinatorial problem). Pavel Pevzner, Bin Ma, and several others do very interesting "applied combinatorics". -Boucher