Monday, February 14, 2005

From Industry to Academics?

A reader question (anonomized):
How should you decide whether to go into industry? Are there ways back after you have sold your soul? The occasion: I got an offer from a major internet company to work at one of their labs. I have until Monday to decide. At this moment, I don't have any promising leads to do the research I'm interested in inside academia. My advisor says that if I don't like it at this company I can always come back, but I am a bit doubtful of that…what is your take on this?
It depends on the kind of industrial lab and how long before you would go back into academics. If you go to a basic research lab and continue to produce academic papers then you can find an academic job afterwords. You'll lose a little by not having teaching experience but that gets offset by having more time for research.

If on the other hand you go to an industrial job for even a couple of years and don't continue to produce academic research papers or attend conferences it will be very difficult to find an academic job afterwords particularly in a theoretically oriented field.

If you really want to stay in academics take an academic job that might not be to your liking and work hard to produce good research and try again later.

Do any of you readers have stories of people that have successfully gone from industrial jobs back to academics?


  1. Take the job. It's a bad, bad year. I'm in a position with no interviews or leads whatsoever and I never, ever anticipated that this could happen to me. I wish I had an offer similar to yours.

  2. I hope you speak to someone (preferably a colleague or a friend --- your advisor won't necessarily tell you the truth) about what academia is _really_ like before you throw away a potentially good job offer. I am an assistant professor, and between lack of time and the need to do the "right" research to get grants and (hopefully) tenure, I can't do the research I'm interested in, either.

  3. I've done two tours of duty in industry and gone back to academia after each of them. The first one as research scientist in the first comprehensive web search engine and the second as director of Research in the largest non-backbone ISP in the USA. How hard it is to come back depends very much on what your industrial job is about.

    If your job is in an industrial research lab, such as ATT Labs, IBM Almaden or Microsoft Research, going back to academia is easy, and we all know plenty of people who have gone back and forth. If on the other hand your job is research staff, going back is a lot harder.

    If your job does not include publishing, then you are in trouble as most of what you do will be considered proprietary. I was fortunate that I had a second line of research that had nothing to do with my job description, so I could continue to publish while in industry. On the other hand I had to put in double time to get research out.

    After two years in my first stint in industry as a research scientist, I could feel academia slipping away, and right around there I landed a job.

    After three years in academia, I was headhunted and went back to industry. There were plenty of interesting research questions, all of proprietary nature. To this date, a certain percentage of traffic on the Internet is routed using the algorithms we developed there, about which I can publish not a word. (Aside: several years later I got a release from the company to publish one of our most important and strategic studies. It got rejected from a conference because "it had no practical applications." <g>)

    Another piece of advice is that, believe it or not, your job title is important. Make sure that it properly reflects the research aspect of your job. Your job search will be made more difficult if you have to explain in your cover letter that your job actively included research, and you might have a hard time proving the quality of it, if the research is proprietary.

    On the positive side, my time in industry was a rich source of research insights. I know first hand what is in use out there and what are the outstanding problems. To this date a significant portion of my theoretical research is motivated by practical questions.

    Also the thrill of working with "live ammo" is second to none. If our routing algorithms didn't work we could bring down large portions of the Internet. When the algorithms do work there is a great reward in seeing your ideas in action.

    Alex Lopez-Ortiz

  4. I went from an industrial research lab (Digital Systems Research Center) to an academic job. This lab position allowed me to publish a lot, so the switch was not difficult. I agree that if you get to publish in a lab job you can move back to academia fairly easily, if you move after 1-2 (or possibly 3) years; departments can think of your lab time as being like a post-doc, and bring you in as a junior faculty member. I think it's much harder to switch in after 4-7 years, unless you have been very successful in research. At that stage, you're probably not senior enough for tenure, but you are senior enough that a university feels like they can't bring you in as a junior faculty member any more, making it very difficult for them to position you. It's a challenge to "promise" someone they'll come up for tenure shortly at many institutions. After 8+ years, you'll be treated as senior if you look for a position -- challenging, but at least departments know how to handle the situation.

    If you do not get to publish in your lab position, you'd be best advised to try to switch after a year if you decide you don't like lab life; the longer you are out without publishing, the harder it is to move.

    I should mention that I greatly enjoyed working in a research lab. I think many (most?) Ph.D.s think they're supposed to become professors. But working in a research lab is very rewarding. I recommend Ph.D. students try working in a lab at least one summer to get a feel for the environment and decide if it might be the right career path for them.

    Even if you don't decide to make a lab your permanent home, working in a lab does give you a lot of insight into real-world, practical problems, which can have a positive impact on your research agenda. I learned a lot in my 2 1/2 years at a lab; it had a profound and very positive impact on me.

    (Thanks to everyone at DEC SRC, especially Andrei Broder, a great mentor!!!!)

  5. I have a question: In non theory CS fields, is it really that hard to get a teaching job after being in the industry?

    The opportunity cost of grad school is so high, it's tempting to work in the industry for a couple of years to earn back money lost. But since my goal is academia I wouldn't want to risk it.

    Why is industry experience frowned upon? You'd think that might make a professor more attractive to industry-bound students.

  6. I worked for several years towards a Ph.D. I did not receive, and I've been in "industry" (a consultant on diverse subjects like operations research, complex systems, agent-based systems, machine learning and the like) for about a decade. I recently started to go back to school to get my Ph.D in a more appropriate field. So I have a good ten years' observations on both sides of the Great Divide.

    In sum: If you should manage to obtain an academic position, you will be coddled, protected from much stress (beyond committee politics), provided with wave after wave of austerity measures in the workplace, and surrounded by interesting people. In the process of getting there, you will be hazed, prohibited from having any independent social life or existence, oppressed, flattered ad promoted for creativity and independent thought, and treated as a de facto indentured servant.

    If you should manage to obtain an "industry" position you will be: forced to consider applications and interact with customers and provide real results, where you had no such demands before; obliged to have skill inn self-promoting your work successfully to nonexperts; paid ten- to hundred-fold what you did (or could possibly) make as an academic; given little time for rumination and theorizing; exposed to the reality that there is no job security anywhere outside academia. In getting there you will have to learn to deal with diverse problems "outside your specialty" quickly and ably, eschew obfuscatory verbiage, think of the easiest approach as better than the most interesting one, and develop a vigorous and diverse social network and exterior life if you have any hope of advancement.

    Somehow, the barriers to re-entry back from "industry" to academia turn out to be far more intimidating than the reverse. I suspect because so little is at stake.

    If your "industry" job is in business, or some domains of engineering, you may be sought-after upon to academia. But if you're a physicist or a (more theoretical) computer scientist, experience in R&D is not considered much beyond a sad straying from the True Path.

    Keep that in mind. Though it's just my opinion, of course....