Since Noah at some point produced some useful (if not unanimously endorsed) advice on graduate school, and the topic has been on my mind recently, I thought I would write a post on job-hunting. Interestingly, I’m not sure my 3 rounds of job applications have left me a lot wiser on the subject, but being a faculty member in a department doing a job search has been very educational, if only because it got me looking at the problem from the other side. Anyways, I don’t claim that I have a very complete view of how to things work or to give comprehensive advice. But there are some things that popped out at me that maybe candidates don’t know and should. If I think of more, I might add them.
Help your letter writers. Your letters are far and away the most important materials in your portfolio. Really. So do what you can to make them good. One thing is to pick your letter writers carefully. If you can, ask someone in a position to have read their letters about their quality. Definitely ask the writer whether they feel they can write you a strong letter. Provide your letter writers with all your materials, descriptions of all of your research, and anything else they may need to know. Your letter writers are really expected to describe your research in their letter, so giving them some straw to turn into gold is a very good idea. It seems crazy, but your letters will be read more closely than anything else, and praise of any aspect of your work will be more credible in them than coming from you. Helping your letter writers, including help they didn’t necessarily ask for, is actually the best thing you can do to make your dossier better.
This is one of these that makes a lot more sense when you’ve been on the hiring side; when you’re looking at a candidate’s dossier and trying to decide whether you’d like them to be your colleague, what part of that is actually going to give you useful information that you can process quickly? The titles of their papers? Their research description? Their cover letter? No, for better or worse, the most parsible part of an application is the letters, so that’s what gets read, or at least read first.
The other thing that’s a slightly longer term project is cultivating letter writers. By the time you’re applying for tenure-track jobs you will need 4 to 6 people (don’t believe schools that say they only want 3) to write letters for you; if you don’t have such a stable of people, go looking for them. Go to conferences and give talks as much as you can (obviously, that’s a good idea anyways, but let this make it more pressing). Strike up conversations with people after their talks. Email people about their papers. Make sure more senior mathematicians know you exist and know about your work.
EDIT: Since Anonymous asked in comments, let me try to expand on this. I’m probably not the best person in the world to give advice on starting conversations, since I’m pretty happy to just insert myself into them whenever I have something to say (which is reasonably often), but I can still sympathize with feeling of “Oh, jeez, I’m in the lunch line next to person X, I should start a conversation with them. But about what…” I, for example, have been seated at a colloquium dinner next to Sir Michael Atiyah, a situation where I really had no idea what to say.
- First of all, don’t sweat it too much. Sometimes you’ll run into someone it really seems like you should be able to have a good math conversation with, and it just doesn’t come together. Nothing (as far as I know) came of my dinner with Atiyah, but that’s fine; I had plenty of other conversations with people on the same trip. Part of the point of creating many such opportunities is it makes all of them lower pressure.
- Don’t try too hard to be strategic. I know that might sound like the opposite of what I just said, but it isn’t. My point above is that when you’re vacillating between creating opportunities to talk to people about mathematics (“Should I go to this conference? It’s a long way away.” “Should I give a talk? Maybe my results aren’t good enough.” “Should I talk to person X? Maybe I’ll embarass myself.”), you should use the fact that you will need letters of recommendation as a motivator. But when you are in the actual situation, focus on the mathematics (or whatever else the conversation turns to), and the letters and collaborations will take care of themselves.
- Give talks. You might think that the number of talks you give is fixed above by the number of invitations you get, but this is wrong. If you decide to go to a conference, let the organizers know you’d like to give a talk (this is shockingly effective). Make an active effort to get seminar invitations, especially if there are multiple universities you can get to without a plane flight and overnight stay. As general rule, seminars tend to be under- rather than oversubscribed, and seminar organizers will be happy to take a hint. Obviously, if you require travel funding, that’s trickier (I’m not going to suggest you fly across the country for seminars on your own dime), but if you express interest, travel funding has a way of appearing (or if you have a grant, you can make it appear yourself). Note that giving good and interesting talks is great way to get people to talk to you.
- “I really enjoyed your talk” is a good conversation starter (though as per above, sometimes the conversation will happen, sometimes it won’t), as is “I’ve been reading your paper X, and…” Don’t be dishonest about these things; if you aren’t really interested in someone’s work, they will probably figure it out, but if you are, then there’s a ready-made conversation starter. Of course, the best is “I really enjoyed your talk, and I think I know how to prove this conjecture…” but I wouldn’t rely on that one happening too often.
- Don’t be afraid to bounce ideas off people. You can say “Do you know if any one has attacked this problem using…” or “Have you ever thought about applying this to…” Again, the yield will be low, but when you ask the right person the right question, it can work out great.
- If you write a paper that a more senior person will be interested in, you can cold email them with it. Whether they’ll read it is a toss-up, but they might. This has happened to me several times (as the recipient of the email), and in all cases I’m glad I received it. They’ve always been papers I probably would have noticed eventually, but hadn’t yet. Don’t spam people; in most cases, I would guess there a max of 10 people who makes sense to email about a given paper (and often fewer).
Contact people at places where you are applying. This is super awkward, and maybe the professors who are about to get a lot of unsolicited email from candidates will be annoyed with me, but it is absolutely crucial that you have advocates in the department. Any even remotely desirable job these days has hundreds of applicants (for a postdoc at Oregon last year, we had 700). You really can’t depend on the right people happening to notice your application. Somebody has to say in a meeting “Hey, I think we should hire candidate X.” Especially for jobs at large departments in research universities, you probably want someone actively scheming on your behalf. And if you’re really excited about a job, the people there should know you’re excited. While from the candidate’s side, it feels so difficult to find a job, it feels hard from the school’s side as well, so there’s constant worrying about whether a person will really come.
Of course, you shouldn’t have extremely high hopes about this plan. Several people have emailed me about postdoc positions at Oregon, and in basically all cases I have been completely unable to do anything for them for a variety of reasons. But it was still the right decision on their part. After all, you only need it to work once.
Give a good job talk. Seriously. This is probably stupid-sounding, but it’s so important. I have seen quite a few people, some of whom I think are basically good at giving talks, screw these up (hell, I’ve seen me screw up a job talk). The only real interaction most of the department is going to have with you is your talk.
- So practice. A lot. With an audience. Who gives you feedback.
- Make sure it’s accessible. Don’t talk down to people, but aim the first 10 or 15 minutes at a smart mathematician who never thinks about your area, and hasn’t in 20 years. The important thing is to convey the impression that you are a smart person who would be a good colleague and good teacher, not to cover as much of your research as possible. That said, you should have a theorem with your name next to it in your talk. You also want to give the impression that you are a serious researcher, but people will take it from their colleagues who tell them this if you give an appropriate taste. If you really want to tell people research results, try to arrange a separate seminar talk.
- No, really. Practice.
Much of this also applies to talks which aren’t strictly interview talks, but which are at places you want jobs. It essentially all applies to colloquia (if you’re lucky enough to give a colloquium somewhere you want a job). Especially the part about practicing.
Secret Blogging Seminar 
Thank you for the advice!
Do you have any more specific advice (or experiences that you can share) on how to cultivate letter writers / contacts? I’m not questioning if “Go to conferences and give talks as much as you can. Strike up conversations with people after their talks. Email people about their papers. Make sure more senior mathematicians know you exist and know about your work.” is good advice. It certainly is. However, on the other hand, I haven’t heard/seen anyone really successfully carry this out.
So you find yourself standing next to senior person X (who you have never meet) in the coffee line after his talk at a conference. Beyond politely introducing yourself, how would the conversation go from here leading to a collaboration or offer for a recommendation letter? The very few occasions I have seen people try to push for such things, they have always came off as overly opportunistic and (in my view) seemed to have lost more than they gained.
The same question goes for using email. Let’s say I just read person X’s paper and really liked it. Perhaps you have thoughts about improvements, or perhaps I would just like to work on related topics? Is there standard way to try to initiate this via email?
In my experience it’s not the first conference you’re at with someone where you get to really interact with them, it’s the second one. People naturally want to hang out with people they “already know” but the bar for “already know” can be rather low. In some ways this makes things harder, because you have to go to more conferences, but it also means that things get a lot less socially awkward than they might seem at your first couple conferences.
Noah makes a good point, though I think it might be more accurate to say things are much easier if you know at least a few people at the conference. I’ve had plenty of good conversations with people I was meeting the first time, but pretty often they were talking to somebody I knew first, or vice versa.
Anonymous wrote:
Probably the best thing is to say something interesting about math, or ask an interesting question. If someone just politely introduces themselves to me, I’ll probably forget about them in a few days. If someone quickly starts leading the conversation in the direction of a potential collaboration or letter of recommendation, I’ll do my best to avoid them in the future. But if I have an interesting conversation about math with somebody, I’ll remember them and look forward to talking with them again.
Note, the word interesting is important here, and so is the word conversation. Some mathematicians (especially young ones, it seems) engage in monologues where they rattle off definitions and state results they’ve proved, usually too fast to follow, and often not paying enough attention to see if their conversation partner is still interested. That’s really unpleasant. What’s good is when two people find an area of common interest, both take turns explaining things and asking questions, and both leave knowing more than they did before. What’s best is when they come up with a new idea that neither had before.
“common interesting” -> common interest
EDIT (BW): I’ve fixed the typo this is referring to.
Having been on the hiring committee at Toronto for the past two years, I am starting to get a good sense of the “other side of the problem” as Ben calls it.
I would just like to emphasize one piece of Ben’s advice: to supply lots of letters. Though schools may only ask for 3 (research) letters, I think that it is your favour to provide as many as you can, even if they are from people who are only familiar with one part of your work. I think that 6 is a reasonable minimum — I have seen as many as 11.
Wow, that’s one I certainly hadn’t heard. I knew that 3 was too few, but I was a little worried that 5 research letters was too high. (I did get reasonably good advice here: the people I asked about this point told me to have a fifth one, but no one told me to consider more.)
There’s two slightly problematic things here, one is that hidden requirements not in the ad itself are generally bad things. The other is that inflation of the number of letters means it’s more and more of a drain on senior people’s time.
One point to clarify, Joel you’re talking about TT, right? Surely postdoc apps don’t typically have that many…
One point to clarify, Joel you’re talking about TT, right? Surely postdoc apps don’t typically have that many…
Well, I have seen one postdoc application with 8…
I agree with Joel that people seem unreasonably concerned about sending too many letters, though I worry a little about how good your marginal letter after 5 or 6 will be. I’ve had 4 and 5 research letters on my TT searches, and I feel like even finding a 6th would be going to someone who knows my work a lot less well.
I also agree with Noah that this could set a dangerous precedent. Senior people often get asked to write really ridiculous numbers of letters.
Yes, I was talking about tenure-track. I think that for postdoc any more than 4 would be excessive. In some sense, postdocs are usually hired on the basis of one person’s opinion, whereas for tenure track, I think that it is much more important to get a good balanced view of a candidate.
I agree with Noah and Ben that my advice does violate the categorical imperative. If everyone had 11 letters, then it would be a disaster!
Furthermore, I feel like I should note (in part because of the categorical imperative problem) that in my (limited) file reading experience, more than 6 letters is rare, and that 5 or 6 is the mode. I no longer have access to the files for the TT job at Oregon this year, but I don’t think candidates with >6 had many advantages in terms of short-listing, etc.
I have been involved in hiring at Northwestern for many years. My experience there suggests that four letters is (typically) fine, and I wouldn’t suggest trying to get more unless you are sure that your fifth letter is going to be just as strong as the first four. One thing to try to avoid is a letter which is very short and generic. Of course it is hard to know how a letter will look when you are the one soliciting rather than the one reading it, but one way to try to avoid this problem is to be sure that the people you ask to write your letters really do know some details about your work.
This is mostly curiosity, but I imagine an answer also could be helpful in preparing people to write your letters. I am junior and have never been in a situation to read a recommendation letter. What do they usually look like? How long are they typically? How mathematically detailed are they (do they attempt to summarize your results?), compared to high level assessment, are they? Etc.
There’s a lot of variance, but I would say they are usually 2 or 3 pages long. There’s usually an introductory paragraph saying how the writer knows the candidate and giving an overall appraisal. One slightly jarring aspect of these appraisals is they often involve comparison to other mathematicians (particularly jarring has been reading a few letters that compare candidates to me!). They do contain quite detailed discussion of the candidate’s work; usually this is maybe 2/3 or 3/4 of the text. There will also be description of any other relevant qualities of the candidate, say praising their talks, or mentioning if they organized some activity etc., and then closing with a reiteration of the overall assessment.
I’ll just note, at least one colleague has told me they primarily judge a candidate from outside their field by the first and last paragraphs of their letters of recommendation.
I was on the job market the past two seasons. The first time around I had 5-6 letters, and the second time (at the advice of one of my letter writers) I decided to keep it to 3.
There were several other factors at work, but I did much better the second time around than the first. Just based on my own experience, I don’t really believe that 3 letters is too few.
Of course, the exact problem with these things is that there are way too many variables to be sure what effect (if any) one change made. When one gets different job market results a year or two apart, it’s so hard to know what changed.
I think all of us are like the blind men feeling bits of the elephant, but the one thing I’m sure of is that you really want your letters to make an overwhelming impression; if you can get 4 really positive letters, then that’s probably better than 3, but if the 4th one is from someone who knows your work less well, then I could easily imagine it doing more harm than good.
I really appreciate this post. Having been thinking hard about how to get a job for about 6 months, I saw numerous things in this that I hadn’t thought about-all of which I will be attempting to use in the future. Thank you.