How many top-level Microsoft executives have been in the same job for 20 years? By Rick Rashid’s count, he’s the only one. That sort of continuity says a lot about Microsoft Research, the in-house invention factory that Rashid was recruited from Carnegie Mellon to run in 1991.
It’s still growing—up to about 850 PhDs and a total of 1,200 people worldwide, at centers from Beijing to New England. As Redmond lab director Peter Lee said, “the sun never sets on Microsoft Research.”
“If you look at the history of the company, there’s been so much value that has come out of Microsoft research that really it’s now just this pipeline, and no one really questions,” Rashid says. “I mean, no one comes to me and worries about what the pipeline’s going to be in a few years. They look at what we’ve already got and they like that, and they just figure, ‘Well, I guess he knows what he’s doing, he’s been doing it for a long time.’ And I think that’s really the key to making these things work.”
Investors have certainly questioned the business value of having a standalone research organization. But Rashid says that over the years, from Bill Gates to Steve Ballmer and beyond, the company’s executives have seen value in keeping the lights on.
And that’s how Rashid sees the value of a research stable: You can’t stuff its work into a product roadmap or financial timeline. But one day, your company will need to tap into the work that it’s doing.
A prime example, Rashid says, is the Kinect motion-sensing camera for the Xbox—probably the best-known example of a product that came out of the research arm. When executives came calling, Microsoft Research already had been plugging away on the elements that would make it possible. It turned out, he says, “frankly, better than anybody really expected. I had no idea what we really could do in that space.”
“So research, in some sense, 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,” Rashid says. “And that’s really been incredibly successful for us. We built groups in speech recognition and 3D computer graphics and computer vision before there was any part of Microsoft that really cared about any of those things. But when the company needed it, we were there.”
At anniversary celebrations around the globe today, Microsoft showed off some more new ideas that its researchers are producing, from a more efficient search-indexing technology called “Tiger,” to some cool augmented-reality software and sensor systems, which made a stack of virtual blocks rise off a table and even beamed an image of a person onto the surface via the Kinect camera, a la Princess Leia’s “help me, Obi-Wan” moment in Star Wars. (GeekWire’s Todd Bishop was at the Redmond event too, and captured some from-the-bleachers video of the presentation).
One thing they actually refused to let the outsiders photograph was called OmniTouch—Microsoft is apparently demonstrating it again at a conference in the near future. It’s a setup that turns movable surfaces, including your hand or a notepad, into a touchscreen interface like a tablet computer or smartphone screen. The hardware is still a prototype right now, but principal researcher Andy Wilson said miniaturization of the bulky shoulder-mounted camera-and-projector setup could easily follow: “Maybe it’s a kind of thing that’s on a lapel, or maybe on some future Bluetooth headset kind of thing.”
The explosion of real-world data capture, and the ability to crunch those numbers cheaply and quickly, is perhaps the biggest driver of new-frontier research that Rashid sees right now. It’s making formerly arcane and difficult pursuits much easier, and spreading computer science know-how to all kinds of areas of life.
“Those kind of technologies are really going to change the way we think about doing everything: the way we do science, the way we do urban planning, the way we do medicine,” Rashid says.
One very current example of widespread technology that has been given a big boost by big data is language processing, a key feature in today’s upgrade of Microsoft’s new “Mango” version of its Windows Phone system.
“We’ve been doing speech recognition for a long time. Just in the last few years it’s gotten really pretty darn good. You look at something like Windows Phone Mango which has a really good speech engine built into it,” Rashid says. “My 12 year-old—I gave him a Windows phone—that’s how he uses it. Any program he wants to run, he just says the name of the program and it just goes there. He’s just decided the speech interface is the fastest way for him to do things.”
After 20 years, Rashid doesn’t seem to be particularly near the end of his run. Asked how much longer he could be doing this job, he demurred. Although he joked that he wouldn’t be making the 40th anniversary presentation, it seems there are too many surprises to pack it in just yet.
“Some of the stuff you hear about it, and you say ‘wow,'” Rashid says. “I remember when David Heckerman, one of our researchers, first started telling me about some of the work he and his people had been doing with the AIDS community, and some analysis of how the AIDS virus attacks the immune system. My first reaction was, ‘Wow, I didn’t know we were working on that. And my second reaction was, ‘Why are we working on that?’
“Now, some of that work’s actually in a vaccine trial. Not only that, if you go to AIDS conferences, you see a lot of references to the work that’s being done at Microsoft Research—which, again, pretty weird. But it makes sense in that the field of computer science now interpenetrates with almost every other field of science. You almost couldn’t not do that.”
