Five Surprises from My Computer Science Academic Job Search
Next in series: The job talk is a performance
I’ve just about settled into a rhythm at Princeton — classes started two weeks ago — and next year’s academic job search cycle is already underway! Indeed, I started my job search in earnest almost exactly a year ago. So I guess surprise number zero from this whole process is how time-consuming it was. There’s been no ‘normal’ or ‘routine’ during this year; each month has been unlike the previous. If you’re starting your academic job search, buckle up, it’s gonna be a wild ride!
There’s lots of advice online about the process; you should read all of it. Instead of duplicating what’s been said, I will focus on the things that surprised me in spite of having prepared as well as I possibly could. So if the rest of this post appears a bit contrarian, it’s just selection bias.
1. You’ll need someone to hold your hand. I can’t overstate how much of a difference it makes to have someone who’s been through the process whom you can talk to on a regular basis during your job search. Whether it’s achieving the right depth-breadth balance in your job talk, or wording emails strategically/diplomatically, or knowing how to best space out your interviews, you can’t figure it out by yourself or by reading online advice.
Typically the person helping you will be your advisor, but if they are busy you should find someone else. The good news is that many people will be willing to help out and pay it forward. It doesn’t have to be one person, you can split it between two or three people. I know that I would have screwed it up many times over if it hadn’t been for my advisor and everyone else who helped me out.
Some candidates networked extensively both to compare notes with other job searchers and to obtain and share privileged information. I avoided this entirely because I didn’t want the stress associated with it, and I’m very happy with my decision. That said, maybe I missed out in some way, I don’t know.
2. It’s not an interview. Perhaps this should have been obvious, but I was taken aback during my first “interview.” People already assume you’re an expert in your subfield, and so they aren’t trying to assess your technical competence. At all. I was tested exactly once in my whole tour — a professor asked me to state and sketch a proof of any theorem (of my choice) from my Netflix paper. Another professor apparently found this egregious, so he later wrote me a rather apologetic email. I found the whole thing rather amusing.
Best as I can tell, what they’re trying to assess is your personality (more bluntly, they want to make sure you’re not an asshole), and whether they can collaborate with you. So everyone was extremely polite to me and never asked adversarial questions or gave me much pushback.
One consequence of this interview style is that no one reads your papers, because they don’t need to. I already knew that no one read my papers in the normal course of things, but before the interview season I thought, “Finally, a few dozen people are going to read my papers!” Didn’t happen. Maybe it has something to do with me, but I think a big part of the reason is that in computer science we don’t seem to have a culture of reading beyond the abstract or introduction of papers (except in reviews, or reading groups, or when directly extending previous work.) Knowing this, authors don’t have an incentive to write in a readable manner, and the cycle is self-perpetuating. But I digress.
A happy side-effect of non-interview interviews was that the process wasn’t mentally exhausting. It was sort of like meeting a bunch of people for coffee and chatting for half an hour with each one. Since all the advice I’d gotten suggested that interviews would leave me dead tired, I was initially worried that I might be doing something wrong — maybe I wasn’t having sufficiently technical conversations? I suspect the real difference is that most people get exhausted due to being pumped full of adrenaline; due to a biological luck of the draw I don’t generate any noticeable adrenaline in these situations (including right before talks, which I’ve found a bit surprising).
3. You don’t have to interview them. Everyone else dispensing online advice seems to think, “you’re not just being interviewed; you’re also interviewing them.” I disagree. In one or two cases it was obvious during my interview that the school or department wouldn’t be a good fit for me without even having to ask them specific questions, but absent any obvious issues, I’m skeptical of how much you can determine by asking. Of course you should discuss areas of possible collaboration with people you meet, but to determine things like how good the students are, how effective the administrative staff are, etc., asking directly is not very useful.
The reason is that people will always spin things in a favorable way (this is not a criticism — most of the time they do it because they’ve been there long enough that they’ve adjusted to the situation and they actually see things the way they spin them.) And you’re not experienced enough to parse what you’re told to figure out what the reality is. You should definitely ask them questions lest they think you’re uninterested, but receive all information with a skeptical ear.
Instead, what was extremely useful for me is to talk to people who’d previously been at the departments I was considering, ask them what they didn’t like, then present that information back to people in my interview loop and ask them for their take, and finally try to reconcile the two views.
4. The job talk is a strategic piece of communication. There is so much subtext it blew my mind. For one, your job talk is all about telling people how awesome your work is, but of course you can’t state that directly. You’ll need to humblebrag without being obvious. For example, in my closing slide, I put up a collage of 24 faces, and said, “Finally, I’m incredibly grateful to my amazing co-authors without whom none of my work would have been possible.” That statement was certainly true, but also important was the subtext: “I collaborate like it’s going out of style.” This one was balanced precariously on the obvious threshold — I got called out in one of my talks!
But there’s more. You have to consider every single thing that you say from the point of view of someone in your field, someone not in your field but familiar with it, and someone not familiar with your field. It has to make sense at different levels to all of them. Also, you have to consider how each statement will sound to someone who spaced out for a bit and just started paying attention. And so on.
Overall, this isn’t going to be like any talk you’ve given. I think I spent 3-4 weeks working primarily (albeit not exclusively) on my talk, with regular tweaking afterwards.
5. You will fall sick. Airports and airplanes spread germs, plus you’re much more susceptible to infection when your sleep and diet are irregular, as they likely will be during your tour. Assuming you have a moderately busy schedule, falling sick is just a matter of time. I mentioned being sick to about 4-5 people, and each of them recalled how they had fallen sick during their own job search.
Naturally, then, you should treat proper sleep and diet as a priority. You should schedule your interviews so that you have time to recover when it happens. Also, try not to schedule your two most important interviews too close together. Finally, there’s a lot you can do in terms of symptom relief (e.g., benzocaine cough drops instead of menthol) to minimize the impact on your thinking and speaking during your interviews, so be medically prepared ahead of time.
That’s it for now. There are several topics that I’d like to address in more detail in separate posts, time permitting: 1. how to prepare for and deliver the talk; 2. what to say in your 1-on-1 meetings; 3. travel tips, and 4. suggestions for interviewers from the point of view of a candidate. If you’ve been through this process recently, I’d love to hear how your experiences matched or differed from mine.
Finally, Princeton CS is hiring this year, and our searches aren’t targeted by subfield, so if you’re on the market you should apply!