Microsoft's interest in research is healthy, chief says

"When I first started Microsoft Research back in 1991, there were a lot of people that I don't think thought we would still be around 15 years later," said Rashid, who came from Carnegie Mellon University to launch Microsoft's lab effort and remains the unit's chief. "I'm feeling good about that."

He noted that Microsoft has consistently invested in basic research and even expanded its efforts at a time when other companies have scaled back such efforts.

"In today's climate, I think that's a story," he said. "It's something that other people aren't doing."

On the eve of Microsoft Research's 15th anniversary celebration, Rashid reflected on his efforts in an interview with CNET News.com. He talked about the organization's past, current efforts on things like robotics and RFID, and how Microsoft can play a role over the next 10 years in areas such as health care and ecologically sustainable living.

Q: What was the initial budget for Microsoft Research, and how many people were there?
Rick Rashid: Well, there was me. We started out with maybe a handful of people.

At the end of the first year, we had probably 20 researchers working. It was pretty small; we couldn't quite fit in a van together. I guess we could fit in a Volkswagen Beetle, but you can put an arbitrary number of people in a Volkswagen Beetle.

We were also very lucky. A lot of the early researchers we brought on board were really good at what they did, but they were also really entrepreneurial in the way they thought. They really wanted to change the world. It really helped to create an environment here that got other researchers excited.

In the early days of creating a research group, it was hard for me to convince a lot of people in the academic community and the research community that a company like Microsoft...was really serious about doing basic research, or that it would be around long enough to do basic research.

We reached about 100 researchers after roughly five years. It was that first five years that really established us as one of the top research labs. It was the end of that fifth year...that we had a huge number of papers from our graphics group at Siggraph (a computer graphics research conference). I think 20 percent of all the papers had a Microsoft author that year. And from there, we just grew, and it became a lot easier to hire great people.

What are your priorities now? What are the things your teams are most focused on?
Rashid: I don't actually tell my teams what to do. We hire really great people, and we expect them to push forward the state of the art in areas that they do their work. If you are hiring a really great person, in my experience, you are not usually disappointed with the output. We don't really try to direct the research.

That said, if you look at the kind of work we are doing today, there is a huge amount of research that we have been doing in software engineering--really being able to prove properties of large programs, hundreds of thousands or millions of lines of code. We're continuing to invest a lot in that area, because it has such a huge potential for changing the way we really think about creating software.

We have huge investments in areas associated with computer vision, computer graphics. Obviously, these are areas that are now very important to Microsoft businesses. We actually started those groups long before there was any product at Microsoft that could have taken advantage of them.

For me, it's not important that the company be in a particular area for me to have researchers there. We're doing a lot of work these days in mobile. There's a remarkable amount of work in medical areas. We're doing work with astronomers. We're working with oceanographers. Over the last few years, we've built up a very substantial group looking at sensor networks and remote devices.

One of those areas that you guys have moved into that seems to be ahead of where Microsoft has products is robotics. Can you talk a little bit about what you are doing there, and why?
Rashid: We've actually been working with people in the robotics area for some time...Within the last year, we've started to say, 'Can we take more of a role in helping to bring parts of the community together?' We're really trying to bring to robotics some of the same ease of programming and techniques that we've been able to bring into other areas of programming.

You mentioned sensors, too, as another big area. We heard a lot a couple years ago about the promise of radio frequency identification, in particular. Recently we hear that the cost isn't coming down as quickly as needed. What are you guys seeing for RFID and that market?
Rashid: I think you are right, in the sense that the technology is being deployed a little slower than maybe the most optimistic projections were a few years ago. But I think it's inevitable that the physical objects in our world will have sensors on them--they will be able keep track of what's happening to them. I think it's just a matter of time.

Every time there are new areas of technology, there are always, early on, people saying, "This is going to happen overnight." Nothing ever really quite happens overnight. Then there are people who say, "Well, it didn't happen overnight, so maybe it is never going to happen."

I worked back in the early '90s on some of the early efforts at interactive TV. Back then, everybody thought it was going to happen overnight. Well, it didn't. Then there were people who said it was never going to happen. Now, of course, you are really starting to see interactive TV systems being deployed.

(Sensor technology) will have an impact. Increasingly, you are going to be able to put small computers on just about anything. The question is: How do you take advantage of that? What good is it? What are some of the issues that have to be confronted?

People are looking at, for example, being able to create sensor networks just to keep track of the human body. We've been looking at technology for aiding people with memory disorders. There are all sorts of things that will be possible as you deploy these arrays of sensors. It will happen.

So why is speech recognition always five years away from being a mainstream user interface?
Rashid: There are two different answers to that. One answer is, "You weren't paying attention: It's become a mainstream user interface." If you call a lot of customer service departments, the chances are excellent you are speaking to a speech recognition system and not a human being when you first get them.

We're seeing a fairly significant use of speech recognition, for example, in Windows and Office for accessibility reasons. People who can't use traditional interfaces use the speech interfaces because it gives them an ability to use a computer that they might not otherwise have. I think we really are having an impact.

When you set the bar as "another human being," then speech recognition is still going to take awhile. I don't know if it is five years or 10 years. Human beings are really, really good at speech recognition. That's partly because they are not just solving the speech problem. Human beings have knowledge of the world that they have developed over decades. You know this if you spend much time talking to a 3-year-old. It's a very different communications act than it is when you are talking to an adult or a teenager.

I use a smart phone, and I dial it with voice command. There are no buttons on my smart phone for dialing. It's one of those Pocket PC phones that just has the normal four buttons that a Pocket PC has. That's how I call. I don't ever use anything but the voice interface for calling anyone.

Can you talk about some of the ways that Microsoft Research contributed to Vista?
Rashid: One of the least visible ways, but is something I like to talk about, is the work we did on what's called the Static Driver Verifier. People who develop device drivers for Vista can verify the properties of their drivers before they ever even attempt to test that. What's great about that technology is there is no testing involved. For the properties that it is proving, they are either true or false. You don't have to ask yourself "Did I come up with a good test case or not?"

That change is going to have dramatic impact over the next five to 10 years, as we begin to bring these proof tools to bear on larger and larger problems in the software space. We are already doing research on saying, "How would you create an operating system environment from scratch, knowing that you have this kind of proof technology available?" It's a project we have called "Singularity."

Other areas--obviously, we've had a huge impact on the search technologies that are used in Vista. We've been doing information retrieval research within Microsoft Research going back really to the mid-'90s. Those technologies have made (their) way into our Live Search product on the Web but also into our desktop products.

If we sit down and have a conversation on the 25th anniversary of Microsoft Research, what do you hope to be telling me you have accomplished?
Rashid: First off, I think computing technology has a lot to offer in terms of health care and really being able to improve people's lives in a significant way. I certainly would hope that in 10 years, I can look back and say computing technologies and some of the technologies created at Microsoft Research are prolonging people's lives or making the quality of their life better in some significant way.

I think we can dramatically improve the sustainability of our environment through the technologies that we are developing. We have done a lot of work in areas such as traffic analysis. We've licensed some of those technologies to new start-ups that are trying to be able to map in real-time traffic patterns and be able to provide that data to people so they don't have to wait in traffic jams for huge periods of time. Or to provide it to urban planners so they can design the roads better.

I hope that we will be able to say that our software systems are dramatically more reliable because of technologies that we've developed within research.

Do you think you will still be going strong at MSR 10 years from now?
Rashid: I'm sure Microsoft Research will still be going strong. Hopefully, I will too. Everybody gets older, but I hope I will continue to be in good health and doing my job.