I wrote this note originally as an email message to a finishing PhD student in Computer Science and Cognitive Science at IU who was applying for jobs and postdoc positions in spring 1995. Although this letter was written specific for Devin McAuley, most of these comments have analogs that will be relevant for others. Perhaps you will find this letter useful as well.
From: Robert Port
Dept: Linguistics, Comp. Science, Cognitive Science
Inst: Indiana University
Date: February 15, 1995
Subject: Advice about Candidating
Devin, a few weeks ago at our society's annual meeting, with several colleagues from my department, I interviewed 12 candidates for our assistant professor position in just two days. This intense experience suggested many thoughts that I offer to you to think about as you get ready to apply for jobs in computer science and cognitive science. Many of my comments and suggestions are derived from the kinds of questions we found ourselves asking while interviewing and the kind of information we wanted but weren't always able to get from our half hour interviews. Whatever the time constraints on an interview, you basically want to convince them that you fit their needs. The more you have thought about this issue before you start talking with them, the more efficiently your communication with them will go. Of course, you cannot generally know exactly what they want, so you can only try to maximize the likelihood of a fit. These suggestions are just ways of maximizing the mutual discovery of a fit.
A. KNOW THE INSTITUTION YOU ARE APPLYING TO
What do you know about the places you have applied to? Do you know how many faculty they have? Their web page is the first place to look. Who in each dept would be most likely to be interested in your work? and what do they work on? Have you read anything by them? (You can't read everyone, of course, but it is good to know something about the research of one or two of them who will be your closest colleagues. They are the ones whose opinion will probably count the most in departmental discussions.) What degrees does the department offer? How many grad students? Do they have an engineering computer science department too? Or any associated engineering school?
If you don't know the answers to all these questions, then get busy -- at least for the departments you really want to get hired at. If you can't find the information on WWW, then contact the department by email or phone and ask the secretary to send you the information they send to prospective students. Indicating that you know a lot about them is one way you show them you are serious about their job -- that you have thought about how you would fit in and it will greatly help you to demonstrate that you would fit it.
Make sure you have in mind very clearly just what they said they wanted in the job ad. But also listen very carefully to how they describe the job verbally (which will not necessarily be the exactly what the advertisement said). For example, they may or may not be interested in you as their ``cognitive science person'' or AI person or whatever. Think in advance about which of their regular courses you could teach if they really needed you to, as well as the courses you would be most suited for.
At the end of our interviews, we often asked, `Are there any questions you would like to ask us?' The majority of our candidates asked "How many students do you have?" DUMB!! They should have been in a position to ask things like: `Has Fred Householder retired yet?', `Roughly how many of your Phds get postdocs each year?' `What kind of lab facilities are there besides Port's phonetics lab?' `Do junior faculty get some release time to help them get research done before tenure?' `Will the department staff help with faculty grant proposals?' `Would it make sense for me to also have an appointment in department X?' etc. Such questions would show both that they knew us and that they knew where they were going. But few of them asked any questions that revealed they had given any consideration to the kind of work environment we would offer them at Indiana.
B. KNOW YOURSELF AND YOUR NEEDS
I have a couple questions for you to think about because they may be asked by someone. We asked our own variants of these over and over last week - and got very interesting and often revealing answers.
- How would you teach an introductory AI course? Or an introductory AI/cognitive science course? Or introduction to cognitive science? What prerequisites would undergrads or grads need in order for them to take these courses the way you envision them?
- If you could teach any seminar-level course you wanted, what would you do? And how would you teach it? A generalization of your thesis topic is an obvious answer -- and certainly very appropriate for anyone. But if you have time, you might also think of how to give that topic a slant that would especially fit the situation of this department. Is there something clever and novel you might try in assignments, grading, etc. What kind of course work would students need in advance of this? Any engineering or special math? Special lab sections? Suggest how you will get their students actually doing significant research as quickly as possible. That's definitely something they will want to hear about.
- What kind of research do you want to be doing for the next 5 years - between now and tenure? Can you indicate now what your program of research will be over this time frame in general terms? Of course, this latter question may too pointed for most interviewers to actually ASK. But it is something they would probably like to ask. So if you do nothing more that volunteer some remarks suggesting that you are thinking in those terms, it will reassure them of your maturity and your understanding of the importance of producing a body of work that will support tenure at a competitive institution. In other words, THEY are thinking in these terms (specifically: `What is the probability this candidate will be able to get tenure here?),' so they will be deeply reassured to hear that YOU are thinking in these terms too.
If you happen to have any information about it, you might even offer a comment on how and where you would seek federal funding for your research program, or mention any special grants that might be appropriate for you.
- What kind of laboratory facilities will you need for your research program. Will you need a settup for listening and psychoacoustics experiments? Access to special subject populations? It would be best to have a very rough idea of how much your startup equipment will cost and how much space you would need. (Don't worry about asking for too much. Up to $50k should be ok these days in your field if they are serious about faculty who do research and bring in grants.) I know these are very difficult questions. You just may not know answers to these when you do an interview. It is OK to say you don't know yet, but don't be surprised if someone asks.
- Another revealing question for a job candidate (perhaps especially for you) would be `How is your thesis relevant to central issues in computer science?'
- If you had 4 graduate students in artificial intelligence who needed a manageable project to do in a semester's time, what 4 projects would you suggest? Or what about 4 dissertation topics? It would be ideal if the topics were `hot' but also doable.
C. THE JOB TALK
The first rule is, make sure you can give your talk in 50 minutes without being rushed and not a moment more. That's the only way you can be sure to have time for questions. Many people have other things on their schedule and will have to leave. You should give at least two practice runs of the talk in order for it to be smooth and confident with all the bugs worked out. Ideally, I'd recommend that any job candidate do two practice runs for a selected audience of close student and faculty friends and then do one more with a broader local audience before taking it on the road.
My second rule is inspired by a couple job talks I heard recently: Do not give a job talk that `presents your thesis'! That is NOT the point. You should just design a really dynamite research talk that speaks to a very general audience in your discipline. Don't try to tell them too much. Make sure they understand what the issues are early on and what you learned from your research. Present only as much as you need to from your thesis. Talk about `my research' not `my dissertation'. The audience doesn't really care much about what exactly the thesis was. My advice is not even to mention the word in your talk. They just want you to present a clear and compelling research story.
Well, that's about it. Show them you are listening to them when they describe their needs. Convince them you fit in. Then if you can give straightforward answers to questions about your needs, AND if you give them a dynamite, well-rehearsed talk, you will have a really great chance at any job.