Sunday, September 26, 2010

For grad students who want a research career

Following my last post regarding how to get invited to give talks as a graduate student, a commenter asked:
How many days-spent-at-seminar talks+conferences/year is a good number for a grad student interested in a career in research? Obviously it depends on a whole bunch of factors, but just wondering what your best generalized estimate would be.
This is a great question. I may even rephrase it to be, "If I want a research career, what are some good ways to spend my time as a graduate student?"

I have some colleagues who get ideas just sitting in their office, alone, reading papers and thinking. This strategy sometimes works for me, but the truth is, I get more ideas from attending seminars/workshops/conferences and from talking with other people. A recent paper I wrote came entirely from seeing something wild at a conference a few years ago coupled with a recent talk I attended given by someone in a different department at my university. Somehow, these two very different things clicked in my brain, and off I went.

Coming up with research ideas is only half of the problem, though. You also need support to turn them into reality. For this recent project, I consulted with the person from the other department for tips on how to use some specialized equipment which was new to me. I then consulted with another person in his department for ideas on how to formulate my research questions. I ended up with a really solid paper having had these consultations.

So from an idea generation and cultivation perspective, I think spending time attending talks and conferences and chatting with people is a great use of your time(*).

But attending these events are also a great use of your time if you want to get employed some day. :) As I alluded to in the last post, I think being a superstar / having a superstar advisor is rare, and even if you are a superstar, a little extra insurance doesn't hurt. So the other advantage to spending time attending talks and conferences is being able to meet other people.

In many fields, the research world is a small one. That person you chatted with at the coffee break may one day be reviewing your paper/grant proposal, or may one day be offering you a chance to come give a talk at their lab, or even could one day be on the other side of the hiring table. This doesn't mean every interaction with someone matters, it just means more doors may open for you if you put yourself out there.

The goal of attending these things is not to go up to every person you meet and say, "HI GIVE ME A JOB PLEEEEZE". It's a way to casually chat with people and show them that you are a friendly person with good ideas. You form relationships that last throughout your research career. Your advisor may be able to help start you on this path of building these connections, but at some point you need to take initiative yourself. And I think it's good to start building these connections as early on in your career as possible, not two months before you go on the job market.

This all being said, you of course don't want to spend your life on the road. I suppose I did spend a lot of time attending seminars/workshops early on in graduate school, but I was geographically selective. A few times I traveled far, but usually I just went to seminars at my university or at other local universities, or workshops/conferences that were < 5 hours away. Pretty much everything turned out to be useful in one way or another. Even one conference which was a total dud introduced me to a great collaborator who I still work with today, as well as an invitation to give a talk.

(*) I should note, some departments will only fund attendance at conferences/workshops if you have a paper accepted. And if you're a first or second year graduate student, you may not have had any time to do anything publishable yet. However, it's super easy to get a short ideas-only paper accepted at a workshop so long as you can write coherently.  (e.g., a 2-4 pager like - "Here's an idea on X. Here's some background which makes us think X is true. We're going to implement and test X in the coming months"). Also, many big conferences have doctoral consortia, poster submissions, works in progress, etc. These are also definitely worth trying for, because it gets you a plane ticket. Most university accountants aren't going to realize/care that the poster acceptance rate was 99% while the conference acceptance rate was 12%. :-)


  1. First, I think grad students should go to all on-campus, open seminar talks in their field and even allied fields. It's easy to get in a mindset where you feel like you can't spare even 1.5-2 hours away from your research, but it really is valuable to know what's going on outside your own corner of the world, even if just at a relatively abstract level. I've realized this as I've settled into a faculty position (now two years out from my PhD) and started interacting with more people outside my field.

    Conferences are a lot more time-consuming. The problem is that there are so many of them and they can be expensive. I don't know how it is in your part of CS, but in mine the number of conferences has exploded to the point where I could probably spend more than half the year on the road if I wanted to. Even limiting to 2-4 a year, I'm traveling to Europe and East Asia more frequently than students who are originally from those places. I basically am resigned to spending about a month each year on the road at conferences, so I try to pick the ones that the most important people in my field will be at. I don't attend them for idea generation or even to keep up with the field, really; for me it's almost all about networking and visibility.

    In every place I've been at the professors pay for their students to go to conferences, not the department. I have been able to afford to send my students to nearby conferences even if they have nothing to present, but more expensive trips require first-author full papers. That was actually a motivator for me as a grad student---do good work, go see a faraway exotic place. And that's a big part of how I got my current faculty position. By the end of my grad studies I was traveling about as much as I am now.

  2. Great post, and I also agree with Anon's comment above.

    For getting sexy new ideas it's probably more important to be exposed to work outside your primary area than in it, because cross-disciplinarity is where the best new thrusts come from. For that, the student should foremost make an effort to attend as many of the in-house seminars as possible/meaningful. For my students, this means department seminars in mine and 3 other departments, and several larger topical group meetings where talks are the norm.

    My best student now has too much of an isolationist streak (must sit at desk and work, other humans are not worthy of his time); I am trying to get it into his head that he must open himself up to other people's work and their connections... I have had mixed success so far.

    I also find, like Anon, that getting to see faraway places on the PI's nickel is a good motivator for the students to really get some good work out and fast (especially if I say they only get to go if they get a talk, not poster).

  3. Great post and I also agree with everything GMP and anon said.

    Regarding invitations to seminars to other places however (which you wrote about in your previous post), in my field those are much more prestigious than presenting at a conference - external invited seminars seem to be "reserved" for senior academics. In my experience, it just does not happen that a PhD or a postdoc would be invited to give a seminar somewhere else and have all expenses paid. We have an internal seminar series at our department where our PhD students/postdocs are all expected to present, but in our external seminar series we only invite senior people (and then we pay their travel expenses to get here + accomodation and food, etc.). This is more or less standard for most places in my field. Even I, as a junior academic, had only ever had one invitation to a seminar (and that was within the same country). You are lucky to be invited to external seminars at your stage of career!

    And funnily enough, same as GMP, my best student is an introvert who does not want to "do" conferences and networking. He's also turning out to be a perfectionist, who won't show anything "in progress" to anybody (which is a problem, because most of the conferences in my field have deadlines 7-8 months ahead and so normally whatever you submit is always "in progress" at the time of submission) and so I am getting grey hair over trying to get him to submit stuff.

  4. Hi, this is the anon from the previous post. Many thanks for addressing my question. I have an add-on:
    as others said, there are so many interesting conferences/seminar/etc. in my field that I simply can't go to them all. (By seminars, I mean the sub-field variety, nobody is interested in me talking at a departmental colloquium, as far as I can tell.)

    So how do you pick and choose which are worthy and where do you draw the line and say this looks like fun but I can't go because I need to actually do some research?

  5. Thanks everyone for the comments.

    I agree that there are an insane number of conferences/symposia/etc. I can't speak for all fields, but at my lab many of these are seen to be bogus. For example, everyone scoffs at these mega multi conferences. They're kind of like those takeout places that have Chinese/Greek/German/Ethiopian food. I'm all for fusion, but come on. There's no way a restaurant can cook all those foods well. It's the same thing for conferences.

    My general metrics for choosing conferences are publication venue, people, and location. For venue - will my paper be published in the IEEE/ACM/other conference proceedings? A journal? Is it just going to go on a USB flash drive they give out to participants? This is less appealing for me at this point in my career. I know with papers on webpages /google scholar it doesn't matter as much as it used to, but something that shows up in the ACM Digital Library / IEEE Explore / etc holds more weight to some people I think. (Though there are plenty of bogus IEEE conferences! :))

    My other metric is the people. Who are the organizers? Are they well known in the field, doing interesting work, or affiliated with top labs? And if none of the above, have they invited interesting speakers to come to the workshop?

    For conferences, I always look at the acceptance rate. If a conference has a small acceptance rate, that (usually) means the papers are good and the talks are good and it'll be worth my time. A 50% acceptance rate is what it sounds like, some amazing stuff and some horrible stuff. Anything above that and you're looking for trouble. (Unless you know the organizers are good, in which case it doesn't matter. I regularly attend one event that has a high acceptance rate, but I know the organizer and they do an amazing job every year).

    I also factor location in. I have skipped a few events that I know would be super useful to have attended, but I knew I just couldn't handle the travel. Travel takes time - not just getting there and back, but settling back into a research frame of mind after the trip/getting over jetlag/etc. Early in my career I was lured by the exotic places thing, but now I honestly just hit "delete" when I see a conference in Australia. It's just too much. (Maybe I'm getting old :))

    As for which department / university seminars to attend, I read the abstracts and look up the speakers. If either sound interesting, I'll usually go. Otherwise I skip it. It's usually pretty easy to tell from the abstract if the topic is interesting/useful, though I have been misled a few times!

  6. Oh, also -- be really careful about industry-run conferences. In general they're fine, but I went to one which I thought was a technology conference but was actually was run by a single company who had marketing people *everywhere*. The conference attendees were really people they were trying to convince to buy their products. As a researcher they did some really sick and scary bullying things to me which I'll maybe write about some day, but the short version is that I am supremely cautious of such "conferences" now.

  7. FCS, on an unrelated note, did you see you got blogrolled at FSP place? :) Enjoy the traffic!

  8. I did see that, and am incredibly flattered. :)