Microsoft Research is marking its 20th anniversary this week, now encompassing some 850 computer scientists. Its leader, chief research officer Rick Rashid, has been there since the beginning, having been recruited out of Carnegie Mellon University to run the group two decades ago. In fact, he holds the record as the Microsoft executive with the longest tenure in the same position.
It hasn’t gotten old yet. “We have so many people doing really cool stuff that almost every day I can hear about something I’ve never heard before,” he says.
During a conversation on the Microsoft campus, Rashid talked about topics including Microsoft Research’s strategy, the group’s contributions to the company’s products, and his advice for aspiring computer scientists. Continue reading for excerpts.
On the importance of long-term exploratory research: If someone comes to you and says I need an idea to solve this problem, unless you more or less know how to solve it and have just a little extra work to do, you can’t do it fast enough, because products have to ship in reasonable time frames. Research is not done on those time frames. You have to already have been doing the research.
How Microsoft Kinect is an example of that: When (Xbox executive) Todd Holmdahl came to me and said we need some help in order to make this technology really reliable and make it cover all the cases we want to cover … we already had the technologies and the ideas to start from to work with them to really translate those things into their product. If he had said, I need this, and I have to go out and hire people to build a group, that’s not going to happen in a reasonable time frame. Research in some senses is about being ready when someone comes to you with a problem or with an opportunity, it’s not about reacting to a problem or an opportunity, because you’ll be too late.
On the initial skepticism about Microsoft’s research initiatives: I remember when I first came here, I tried to hire a really good friend of mine, a professor at MIT. I just couldn’t convince him to come. This was back in 1992. It basically boiled down to, he didn’t think Microsoft was going to be around long enough to do long-term basic research. He kept saying, look, the software companies don’t last. The only companies that seem to last are the hardware companies. So we made a bet — this was back in 1992 — as to whether Digital Equipment Corp. or Microsoft would still be around in 10 years. Five years later, Digital Equipment Corp. was gone, and I wound up with his check for 25 cents, which I’ve kept. The memo was great. “Rick was right.”
On his tenure in the position: Twenty years is a really long time, especially in a corporate environment. It’s not unusual for professors to be professors for 20 years, but to be doing any function at a high level in a corporation for 20 years is unusual. I’m the Microsoft executive who’s been doing the same job for the longest time. … I partly attribute it to the fact that Microsoft has been both really understanding of and really cared about basic research for a really long time. The usual reason why someone would have changed jobs is that someone else would have decided they were doing it wrong. The approach we’ve taken over the years has been something that has had great support from obviously Bill (Gates) for many years but also Steve (Ballmer).
His advice for aspiring computer scientists: Certainly if you’re in school, and you think you’re going into the field of computer science, these days it’s critical that you learn statistics. That wasn’t true back when I was doing it. I did because I was sort of a mathematician to start with, but the mathematics of statistics are so critical now in so many areas of computer science, that if you haven’t studied that, you’re going to be behind the eight-ball. … Machine learning is an area where there’s a lot of opportunity. If you have that skill set, it’s going to give you the ability to tackle problems and look at them from a different perspective.
Previously on GeekWire: Microsoft Research demo: Augmented reality with Kinect, tablet, distributed computing