26

I have heard from some senior researchers in theoretical computer science that working in a non-research industry job, even just for a few years, will kill your career as a TCS researcher.

However I am suspicious of the claim that the road from being a TCS researcher to a non-research job in the industry is a one-way street. I want to know if this claim is plausible and any implications of making an excursion into a non-research job in the industry in case one later decides to return to a research job in academia.

Do you know examples of people who went to non-research industry jobs after their completing their PhD, worked there for a few years, and successfully made it back into academia as researchers (e.g. got research faculty positions)?

If yes, what fraction of senior researchers (tenured) do they constitute in your department or departments you are familiar with?

What fraction of such candidates who apply for academic research positions fail to obtain one?

Does the number of years in non-research jobs before returning make a difference?

More generally, what role would such excursions play in decisions made by hiring committees?

Since the answer may vary from a region to another one (e.g. North America, Europe, etc.) please mention the region you are talking about in your answer.

For the purposes of this question, let's consider all jobs in which the main task is conducting (publishable) research as "academia" and all jobs in which the main task is not conducting (publishable) research and it is difficult to do research and publish papers as "industry".

Community
  • 1
Holger
  • 975
  • 9
  • 17
  • 1
    maybe it is not so much unique to TCS but that its more a one-way street from academia to industry in general... one hypothesis is that pay becomes more important to individuals later in life eg esp after having families, and the sacrifice can be made younger, without family, but that its too much of pay cut, older. another possibility is an idea circulating/discussed in a lot of academic circles that major contributions have to be made/achieved in youth 1st or they might never be achieved. some prizes/awards ie "restricted under $x$ age only" reflect this. see also misc "age discrimination". – vzn Mar 03 '13 at 18:50
  • 8
    Some people in the past did make this transition. Robin Milner and Tony Hoare both worked in industry before going back to academia to do foundational work. Many of the other Turing award winners, and several of those in theory, worked on practical problems before returning to academia. Many of the senior people I have met have taken time out to start businesses, work in industrial labs, or raise families. Do you have evidence for the chasm you describe, in particular why you make the statement "many senior researchers...seem to believe"? – András Salamon Mar 03 '13 at 20:13
  • 7
    (a) what do you mean by "many" (b) what is "t", and (c) what is "success" ? Hard to answer your question if you don't clarify. I'm assuming you're excluding Google/IBM/Microsoft/Lucent/AT&T/Yahoo ? – Suresh Venkat Mar 03 '13 at 20:18
  • @SureshVenkat: (a/b) I am looking for data and numbers. What proportion of a typical theory faculty at a typical university spent t=.5,1,2,4,8 years in industry after their PhD and before they got tenure? Yes, this is difficult to answer. As a start, what are the numbers at your university? (c) Let's say success just means "got tenure and has sufficient time to do research". (d) Yes, by "industry" I mean non-research positions, businesses, and anything that's not academic research; let's count research labs as "academia". – Holger Mar 03 '13 at 21:59
  • 5
    As Suresh said you may want to clarify your question, do you consider Microsoft research, Google research, IBM research, etc. as industry? I think the main issue is not going to industry or "purity" but rather the kind of work you do. If you go to industry as a researcher and continue to publish good papers in related topics that I don't think is taken negatively at all. – Kaveh Mar 04 '13 at 00:52
  • 2
    +1 for the question. I believe the industry/university distinction in the traditional sense is misleading. I see the distinction as between a job where your main output is papers and your main output is not papers (and could be code, a product, movies, etc). – Vijay D Mar 04 '13 at 02:33
  • 4
    Consider cross-posting to http://academia.stackexchange.com/ – Slaviks Mar 04 '13 at 15:21
  • 2
    I think this is an interesting question, thanks for it. While it is nice to hear of people coming back to academia, most answers so far mention only exceptionally successful academics and/or cases from long ago when the academic market was very different compared to today's situation. It would be more interesting to hear from (1) unsuccessful people who tried and didn't manage, and from (2) some of the successful people: what obstacles did you face and how did you manage? – Standa Zivny Mar 04 '13 at 15:22
  • I spent a year at Yahoo! and am now at a math department -- proof for t=1. Now we just need the inductive case. :) – Lev Reyzin Mar 04 '13 at 16:24
  • But Yahoo doesn't count :) – Suresh Venkat Mar 05 '13 at 00:15
  • 4
    @Kaveh Given how highly up-voted and viewed this question is, it's clearly a question on many tcs researchers' minds. In light of this, it would be a bad idea to close it as off topic, even if it technically doesn't meet the site's guidelines. – Lev Reyzin Mar 05 '13 at 02:56
  • 1
    @Lev, yes, I understand that it has high number of up votes and I am not going to close it using mod powers, but I still think it should get closed. Sometimes posts which are interesting but not good fits for a Q&A get high number of up-votes (e.g. Gorilla vs. Shark type questions on [SO]), that doesn't mean they should stay open. – Kaveh Mar 05 '13 at 03:01
  • re appropriateness, suggest meta discussion—maybe someone needs to attempt to describe how [tag:soft-question] tag (which this question is accurately tagged, and imho is very useful/valuable/relevant on this site, but have seen strong pushback against it on other sites eg cs.se) which is inherently subjective can coexist with the official site/stackexchange policy against discussion and opinions – vzn Mar 05 '13 at 18:28
  • 6
    These questions have answers and are not a matter of opinion. "which researcher came back after industry job" is asking for factual, verifiable data. "how many..." is asking for facts that members of hiring committees ought to have data on. And in the third question, I'm interested in how experience in non-research industry is affecting hiring decisions in TCS (and not how it "should" or "shouldn't" affect them). I'm sorry if my wording caught some people on the wrong foot. -- Also, this should be a CW. – Holger Mar 05 '13 at 21:38
  • @Holger, I think the only questions that by policy we have to make CW are big-list questions. I think a soft-question doesn't necessarily need to be CW and this one seems fine to me. ps: I made a few edits to the post based on your last comment trying to maintain your intention while making the post more constructive, feel free to roll-back or edit further. I also removed some of my comments that didn't seem to apply any more. – Kaveh Mar 06 '13 at 00:26
  • 1
    stackexchange director of community dev robert cartaino on good subjective vs bad subjective – vzn Mar 06 '13 at 06:01

6 Answers6

27

Let me disagree with the other responses.

While there are clearly notable examples of people who can transition to industry and back (see other answers), going to a non-research industrial position, even for a couple years, will make it very hard to return to academia, unless you're already very famous.

The reason is not because academics look down on industry or will think you "impure," but rather that it's very hard to stay active publishing when your job requires all your work time be spent on something else. And if you haven't published anything for a couple years, it will be nearly impossible to convince a hiring committee to choose you over people who have. (Again, this doesn't apply if you already have a Turing award, etc.)

But if you can find the time to stay active in research and publish while working in your industry position, then you can still be competitive. It's just likely to be a losing battle, and your other job responsibilities will realistically probably end up taking priority over research on-the-side.

Lev Reyzin
  • 11,968
  • 13
  • 63
  • 103
21

As someone who helps make hiring and grant decisions involving theoretical computer scientists, I don't care about affiliations. I only care about the quality and impact of the work. If you've been doing high-quality publishable research, you're hirable. If you haven't, you're not.

Lots of theoretical computer scientists work in industrial research labs and then later move to academia. For example, Chandra Chekuri spent eight years at Bell Labs before moving to UIUC, and Suresh Venkatasubramanian spent seven years at AT&T before moving to Utah. And there are many other theoretical computer scientists who have started companies while holding an academic position (Herbert Edelsbrunner at Geomagic; Lars Arge at Scalgo; Tom Leighton, Shang-Hua Teng and many others at Akamai); arguably they're "working in industry" too. And finally, lots of anonymous code monkeys transition back to academia as MS or PhD students (like me).

But if by "industry" you mean "anonymous code monkey", then there's essentially no chance of transitioning to a computer science faculty position. It's not your "theory purity" that you've lost; if anything, having real-world impact makes you more marketable. What you've lost is visibility and impact within the academic research community. A large gap in your publication record, for any reason, raises a red flag.

(I'm on both the faculty recruiting committee and the promotions & tenure committee in a top-10 US computer science department.)

Jeffε
  • 23,129
  • 10
  • 96
  • 163
  • 3
    The "large gap in publication record" certainly seems to be a serious problem in the US academic departments. Other countries may or may not have this problem. In Britain, for instance, we do not mind such gaps, but we would be keen to see what value they bring to the table as a result of such gaps. – Uday Reddy Mar 05 '13 at 10:12
  • 1
    what do you mean with "high-quality publishable research"? Is it good if you have a good paper in arxiv but it hasn't been peer-reviewed? – Marcos Villagra Mar 06 '13 at 01:08
  • 2
    @MarcosVillagra: Well, all else being equal, an ArXiv preprint is certainly better than nothing, but it's not as good as an actual peer-reviewed publication. – Jeffε Mar 06 '13 at 02:03
  • 1
    Are you saying that TCS faculty candidates who had a few years of real-world experience after PhD may even have a relative advantage over fresh graduates, assuming they've published something decent in their recent past to reprove that they're capable to do that? – Holger Mar 06 '13 at 02:32
  • 5
    I'm saying that TCS faculty candidates who have demonstrable real-world impact while continuing to publish have an advantage over candidates with the same publication record, at the same time after PhD, but no real-world impact. I don't think slaving away in the Google mines counts as demonstrable real-world impact. Also: the academic job market is hard, especially in TCS; there are far more amazing candidates than positions. Publishing "something decent" is not enough. – Jeffε Mar 06 '13 at 09:33
  • 3
    By the way, you emphasized "for any reason" in your response, which puzzles me a bit. Would you say that your department implicitly discourages that faculty candidates ever had kids (e.g., with a partner who themselves want to have a career), sometimes chose to prioritize family or two-body situations, or were seriously ill for a long time? – Holger Mar 06 '13 at 17:38
  • I should pull back from that statement a bit. We have both hired and tenured faculty with gaps in their publication record due to childbirth, illness, death in the family, and the like. But people generally don't put those reasons in their CV when they apply. So the gap still raises a red flag, which has to be explained away. – Jeffε Mar 06 '13 at 18:15
  • 3
    "Prioritizing family" is more problematic. Faculty in my department can get a one-semester leave, with a stopped tenure clock, for the birth or adoption of a child. Both parents. A significant fraction of faculty (including me) have young children. But without a further medical/legal reason, leaves longer than a semester are unlikely to be granted. Faculty parents are expected to continue publishing, just as we did before our kids were born. – Jeffε Mar 06 '13 at 18:24
  • "anonymous code monkeys transition back to academia as MS or PhD" very nice, like me :) – Saeed Mar 25 '13 at 13:21
17

Here is one "active" example I know of -- I hope he is not embarrassed...

Andreas Bjorklund has been extraordinarily productive in TCS over the last several years, while maintaining a full-time job in industry. You may wish to contact him, to find out how he does it! At this point, I think his research record is impressive enough to gain a faculty position somewhere, if he wanted that.

Ryan Williams
  • 27,465
  • 7
  • 115
  • 164
  • 13
    @Holger: Faculty positions aside, if your real question is whether you can do research in your spare time while working in industry, the answer is yes. It is simply a question of priorities. – Andreas Björklund Mar 06 '13 at 21:38
  • 1
    @AndreasBjörklund: What spare time do you use? Evenings & weekends? Or some kind of 20% company time? – Holger Mar 07 '13 at 03:25
  • 6
    @Holger: Yes, it used to be evenings, but after I got kids it has mostly been early mornings :-). I went to industry in 2001, long before I got my PhD degree, and it is only in the last 6 months that I've also been fortunate to get the opportunity to experiment with a part time position at the university. Right now I'm testing 20% research position and 80% industry work. Truly understanding people at both instances is of course necessary to make such a deal. To early to tell if it is a better trade-off than no daytime research at all. – Andreas Björklund Mar 07 '13 at 05:09
14

A number of theory faculty (David Karger, Tom Leighton, Shang-hua Teng among others) went to Akamai when it started, and then returned. Rina Panigrahy is not theory faculty, but worked at Cisco for many years before returning to "academia" in MSR. Ken Clarkson was at Lucent the whole time before going to IBM, but spent a number of years "essentially" in a business unit working on a wireless project before "returning" to research full time.

Suresh Venkat
  • 32,071
  • 4
  • 95
  • 271
  • 8
    AFAIK Rina left for industry with a masters and returned first to a PhD program and then to MSR. I think the masters -> industry -> phd -> academia route is less problematic than phd -> industry -> academia. I feel if one were to successfully pull off the second route, the industry job should involve some research element. – Sasho Nikolov Mar 04 '13 at 17:13
  • 2
    yes, that is correct. But tbh I think the question is not very well founded (who are these "many senior theory people" anyway), so any reasonable example is worth considering :) – Suresh Venkat Mar 05 '13 at 00:14
9

A lot of the senior Computer Scientists in Britain have had industrial experience before they came to work in academics. Christopher Strachey, the founder of denotational semantics, was a consultant programmer before entering academics. Tony Hoare, the founder of axiomatic semantics, worked in industry (Eliott Computers) for several years. Samson Abramsky, who holds the Christopher Strachey Chair at Oxford, in fact developed his interest in Computer Science during his work in industry (GEC). Cliff Jones, a Fellow of RAEng, worked in IBM, Vienna for several years before coming to do his PhD, and did another stint at a start-up company called Harlequin even afterwards. I have to say that all of them probably did innovative R&D type of work while in industry, which might be necessary to keep your mind active in thinking about research directions.

Uday Reddy
  • 4,806
  • 30
  • 39
  • 1
    I remember Samson Abramsky telling me that, when he worked at GEC, there was a lot of excitement about theoretical papers by "this guy Plotkin"... maybe Samson will turn up and correct this ill-remembered anecdote. – Ross Duncan Mar 04 '13 at 15:21
-3

Think about it this way: once you start working in industry at a great company that changes the world (think Google), having lots of success, seeing your real-world impact, having much quicker runs of success (as opposed to trying for ages to solve a hard theory problem which in the end nobody will use and will have zero impact on the world), enjoying your life to the fullest, etc., you will not want to go back to "slave away" in academia.

Siamak
  • 11
  • 1
  • 6
    Despite that you didn't address the question, it's nice to see an answer from a different perspective. To be fair, I think neither you nor @JɛffE are right about slaving away. It all depends on a person's interests and preferences -- I know plenty of people who would be happy at Google and hate academia, plenty of people who would be happy in academia and hate Google, plenty of people (like myself) who would be happy at either, etc. etc. – Lev Reyzin Mar 08 '13 at 21:02
  • 10
    or you can open a clinic in a third world country and really change the world for the better. or are we done telling irrelevant stories? – Sasho Nikolov Mar 09 '13 at 16:27
  • 5
    I really don't like the tone on this website. – Holger Mar 10 '13 at 16:55
  • 2
    You are entitled to your opinion. However your post is not an answer to the question. The question is about the situation where one wants to go back to academic research, that is the assumption. You are essentially saying that is a stupid assumption and one would not want to do so. Maybe yes, maybe no, but in any case that is not relevant here. In addition your post is impolite if not rude, please read the [faq]. – Kaveh Mar 11 '13 at 14:42