Wednesday, September 09, 2009

Research Labs "vs." Academia

Muthu points to a blog entry (by Danah Boyd of Microsoft Research) on research labs vs. academia. It's a good read -- especially for graduate students thinking about career paths. It got me thinking about the topic as well.

I spent two+ years at Digital Systems Research Center out of graduate school, and I loved it. It was great training (Andrei Broder was, as I always rave, a great mentor to me there, but there were many other more senior scientists there who were very helpful to me as well), it really cemented my interest in connecting systems and theory, and my experience matched, to a great extent, the positives mentioned Danah Boyd's blog entry: freedom to pursue what I wanted, a great density of great people, many chances to interact with students, and so on. All without the hassle of grant-writing.

If I hadn't also liked being a professor, I would have gone back. But I liked being a professor, and ended up staying at Harvard. I don't think I'd say one is better or worse than the other. I think it depends on a lot on the person, and for many people, both are possible answers.

The one looming aspect of the research lab (perhaps underestimated in the linked blog post) is the issue of job security. I moved to Harvard just after Digital got bought my Compaq, later bought by HP, later... well, the Systems Research Center just doesn't exist. Similarly, IBM and AT&T labs have gone through massive changes over the years. Microsoft seems like a safe bet in terms of a stable research lab environment for the next decade or so, but it's hard to prophesize beyond that. Now, the "insecurity" of the corporate environment is maybe not such a big deal -- people from SRC for the most part just moved on to different jobs, at Microsoft or Google or Yahoo or HP or a startup or academia or wherever, and continued to be very successful. But I do know from firsthand and secondhand experience it can be an unpleasant, even if only temporary, external disruption if your company situation changes significantly.

I hadn't thought much of the research lab vs. academia issue much in recent years since as a family we're quite happy in our present location, and Boston hasn't really been a hub for research labs until recently. Now, I suppose, there are more options. While I don't imagine leaving academia any time soon, I have been spending some time this summer at the Microsoft New England lab, and have been really enjoying it. But more on that will be a topic for another post.

9 comments:

Anonymous said...

Michael,
It is easy for Danah to say that working at Research lab is as good as (or even better in some cases) than working in academia. She is fortunate to be working at MSR. But there are not many industrial labs that give so much freedom to the researchers as MSR. The future for CS PhD students does not look very good.
-Sushant

Anonymous said...

I, too, can vouch for what Sushant says. I know several people who left jobs in research labs recently because of stifling research environments. Only the elite few can make it into labs like MSR that compete with academia.

I think the best bet for CS PhD's is to hope for a place that is naturally in alignment with his/her own interests. My guess is that it is easier said than done.

Anonymous said...

MM,
You are forgiven for talking only about elite outcomes, after all you went to elite schools and teach at an elite school. I imagine most of your readers share this background - don't be something you're not.

The real issue here is how long Microsoft can support such a fabulously unproductive enterprises. The fact that MSR is so posh is a direct reflection of its economic unsustainability.

Sounds like the researchers there have flipped things around and concluded their trappings are so lavish because of the value they create. This is somewhere between wishful thinking and willful delusion.

Unknown said...

I can't speak for MSR, but other research facilities have, as a whole, been greatly beneficial to their enterprises. The first time I encountered a company research lab, I was a little taken aback at the idea that one out of every ten projects (or, often, careers) would make the lab money. While I came to understand the virtues of such a system, most labs have gone the opposite direction and demanded that all researchers produce modestly lucrative work rather than that all researchers produce work with the potential of being extraordinarily lucrative. (Hopefully the last anonymous post is wrong and MSR isn't just a vanity lab with no commercial upside.)

This story suggests that only our tax dollars can come to the rescue here and, perhaps in the form of write-offs, that doesn't sound half bad. Unfortunately, in my community as in yours, we're moving towards a world in which fundamental research and business are diverging, the only folks doing both being those working for the few remaining permissive labs (like MSR) and those doing research on their own time and dime while also working a "real job" (like me).

Michael Mitzenmacher said...

Anon #3: I will recognize I've had a blessed career path. I'm not sure I'd go with the phrase "elite outcome"; let's reserve that for people who signed up as one of the first 100 Google employees...

To be clear, I quite recognize that MSR offers substantial freedom to its researchers. I pointed out that that was also my experience at SRC -- but I left before the big Compaq/HP changes, and my post mentions IBM and AT&T, which have, at various periods, both laid off researchers or had cycles where research was supposed to show its (short-term) "relevance" in very clear terms. I'm not blind to the issues, nor do I think I gave them short shrift, but I do appreciate commenters who clarify and point out these potential issues with labs.

Anonymous said...

Isn't there another lifestyle between research labs and academia, namely vacation? Many universities and NSF insist that faculty take a month vacation (at least a month during which there are no explicit demands, if this is not strictly vacation) not counting mid-winter breaks. In many research labs you are on the same vacation schedule as all other employees; 2 weeks your first year, rising to 3 weeks in year 5.

I had distant family I needed to visit and this aspect was a big turn-off for industrial research labs.

Anonymous said...

In 12 years in academia, I have never had any real vacation. The university doesn't pay you over the summer. So if you have grants to pay your summer salary, you have to forfeit your vacation. Winter vacation is when you get time to finish the papers for the January-February conference submission deadlines.

(Some universities do require you to take one *unpaid* summer month).

Anonymous said...

In 12 years in academia, I have never had any real vacation.

Then you are doing something wrong. You are certainly granted vacation days by your university -- why aren't you using them? Moreover, one of the advantages of being a professor is that you can work when you like -- there is nothing stopping you from working 50% harder two weeks in a row and then taking the third week off.

Anonymous said...

there is nothing stopping you from working 50% harder two weeks in a row and then taking the third week off.

Well, one thing that might be stopping you is your competitor who is working 50% harder every week. Once you have tenure, you can afford to let someone else work 50% harder than you if they want to, but until then you really can't (at least if you want a job at a strong research university).