Rick Rashid | Innovation pipeline

Microsoft might not come to mind as a leader in computer science research as readily as, say, the Defense Advanced Research Projects Agency or Mitre. Rick Rashid, Microsoft senior vice president of Microsoft Research, sees this as a misperception. Rashid started the company's research division in 1991 after doing seminal research on operating systems, the fruits of which subsequently were used in Unix, Apple Computer Macintosh and Microsoft Windows NT. Recently, he was in Washington with a number of Microsoft researchers showing off their work (for highlights, see GCN Insider, Page 31). We spoke with Rashid about the difficulties of technology transfer, new operating system developments and why Microsoft just can't get no respect. I describe technology transfer as a full-contact sport. One of the things I did when I first started Microsoft Research was to create a team of people whose sole job was to do technology transfer. This follows along the line of: If you want something done, you need to make it somebody's job, and they need to get reviewed on it. So we have a group of people ' often who have experience in both research and product background ' and they become the glue [between the two].They recognize problems that come up on the product side that the researchers could help solve. They recognize opportunities the researchers create that the product side should know about. They make sure the process completes. A lot of times, when people talk about applied research, they're really talking about advanced product development. That's not really research. Research is something you can publish in a peer-reviewed conference or journal. It needs to stand up to the best people in your field.From my perspective, I don't want to bias the front end of the innovation pipeline. You don't want to cut off possibilities. I want to hire the best possible people to do research, and I want them to do what they find is interesting and exciting about their particular field. As the work matures, I want them to take a hard look at how to use it.The investment we make in basic research is an investment in the future. That's why the country started the National Science Foundation. We recognize that in times of need and stress, having this basic research infrastructure allows agility. In the case of Microsoft, it means that five years from now, when we have a new competitor or the technology changes, chances are Microsoft Research will have been already been working that area. That is what allows us to survive. We've been doing a lot of leading-edge work in the area of being able to prove properties of programs. Some of this is showing up in products. For example, in the device developer list for Windows Vista, there is a software tool that proves whether applications are properly using the [Windows] application programming interfaces. So we can now prove mathematically definable properties of millions of lines of code.[After that work], we began to ask the question, 'Can we think about how to build an entire operating system from the ground up about which we would literally be able to prove certain properties are true?' So that started a research project where we're taking the proof technologies we've developed, along with some of the advanced programming technologies, and trying to build a whole system about which one can prove specific properties. The key advantage is that you'd like to be able to definitely say that a large piece of software does or doesn't do something. Our best ability to do that today comes from testing. We will write the software and then try everything we can to test it. But testing never proves the software does what you want it to do. It simply means we ran a test. If your test wasn't complete, then you will never know whether something is real or not.What the proof technology can do is, given a property that you want to say is true or false, it can literally examine the source code of the program and prove that is true. Part of the work we're doing on the OS side is to see if we can combine proofs of programs and proofs of properties of OS components into joint proofs across the whole system. There are a number of ways you can look at this. Within Microsoft Research, all you really need to do is look at the publication record. Many of the top conferences in computer science have a very large percentage of papers coming from Microsoft Research. By that metric, we're obviously making a significant impact.I think you also need to see what Microsoft as a company is doing. We are innovating in many areas. Look at the work that is going on in the XBox. In the software development kit, we've been incorporating a lot of new technologies in computer graphics. It's allowed developers to do more with our products than with our competitors' products. It allowed Microsoft to get into an incredibly competitive business. Honestly, I don't think very many people gave us a lot of credit for being able to do something like that before we got started.If you look at the work going on in our office products, we're doing many things in the way we handle text and typography, the way we manage documents, the way you share and manipulate information. There is a lot of innovation going on there. I think people don't see the detail of innovation that goes in. In products like the tablet PC, we're building an entire new business.