MSR Roadshows

MSR Roadshows.

Tomorrow (Wednesday) Microsoft Research is doing open house events in both our Silicon Valley and Cambridge labs.

One of the things that people tend not to appreciate is that good research labs reflect the local and regional tendencies toward particular approaches and domains of research. Computer science research is not the same all over the globe, partly because students and researchers are not infinitely mobile; they often settle in geographic areas and look for research in that area within their chosen domain. Do a research university with a strong domain focus will often draw other research labs and startups with strong ties to those same domains. Microsoft has a research lab in the Valley because there is a wealth of great researchers there working in areas that are important to Microsoft and the industry as a whole, and we want to be able to both hire those people and collaborate with them in other institutions. The same is true for our Cambridge lab, as well as Beijing.

There was a strange newspaper article earlier this week talking about how one of Google's strengths is hiring Ph.D.'s — and how rather than have a “research division” they supposedly let all of their technical employees do “research” on the side for 20% of their time, in contrast to the traditional research lab model with full-time researchers. The reporter (who, by the way, spoke to someone in HR at MS but repeatedly declined opportunities to speak to someone in MSR) thinks this is some kind of magic formula for success. Frankly, I don't get it. A Ph.D. doesn't mean that you're smarter than everyone else, or that you're necessarily more knowledgeable than someone with, say, a B.S. degree, outside a very narrow domain. It does mean that you've been trained to do research, and shown some aptitude for it.

I see lots of problems with their model. For one, why would you hire for a special skill and then only let the person use it 20% of the time? Second, what kind of  “research” are they really doing that only requires 20% of their time? Third, and I've complained about this before, Google employees aren't allowed to publish papers on research results. It's very easy in that environment to convince yourself that whatever you're doing is good research. The only way to know for sure is to have it peer-reviewed and debated by experts in an open forum.

So what do they have? A bunch of very smart people (truly — there is no doubt of that) with advanced degrees, scattered throughout the organization, splitting their time, and telling themselves that they are doing good research work without any real evidence. We have a critical mass of researchers, focused fulltime on research, publishing papers and getting tons of tech transfers (some of which we're showing at the roadshows tomorrow). Maybe they are on to something, but I doubt it and there isn't a lot of evidence to show so far. I'll take our model, thank you. And I hope for their sake that there isn't a mass exodus after the IPO.

[Kevin Schofield's Weblog]

Leave a comment