Live Labs emphasizes the 'search' in Microsoft Research


Redmond (WA) - The principal mission of Microsoft's new Live Labs project - announced last week as a collaboration between the company's Research and MSN divisions - will be to focus on improving the functionality and experience for the Internet search engine user, for as long as the next four years, Kevin Schofield, general manager for Microsoft Research, told TG Daily.

Since 1991, Microsoft Research has employed many of the world's most brilliant engineers, including former Palo Alto Research Center fellow and creator of the Alto computer, Butler Lampson, who also led the development of PKI cryptography; next-generation Internet transport protocol researcher Henrik Frystyk Nielsen; and object-orientation master Luca Cardelli. Some of their work has indeed come to fruition within, or by way of, Microsoft products; but over the years since they were hired - in some senses, hired away from where they were before - a lot of that work has languished. Last week's founding of Live Labs, however, could potentially be the key that unlocks what some observers have characterized - whether fairly or not - as the "basement" of the corporation.

But if these distinguished and admired fellows want to play a role in Live Labs, at least for the foreseeable future, they had better get down to business and think of ways to improve the MSN search engine.

Today, Schofield told us, Live Labs will be "focused on the Internet space, and Windows Live. We certainly haven't precluded that, over time, we may expand it to other things." Those other things might include, as we suggested in our question to Schofield, an open project to improve the functionality and user experience in Microsoft Office. Late this year, the company is expected to release Office 12 in concert with Windows Vista, which continues to appear likely to make sweeping changes in the way its applications look and work.

"Three, four, maybe even one or two years from now, there's certainly a distinct possibility that we'll be engaged in those sorts of projects," said Schofield. One such project may include collaboration with the Office Live team, which is devoted to building Internet-driven applications similar to the Office suite. "Today, at the initial point, it's really focused on the MSN products today and Windows Live."

Last week, we reported on Microsoft's announcement of the founding of Live Labs, culminating from a collaboration between the Research division (MSR) and the company's new chief technology officer, Ray Ozzie. In that report, we stated the work of MSR up to this point has "yielded magnificent research papers, but no practical results - and more importantly, no products."

Dr. Gary William Flake, who is leading Live labs as "Distinguished Engineer" with Microsoft Corp., was hired by the company late last year from Yahoo Labs, where he led research into data mining and search validity algorithms, ostensibly intended to compete with Google Labs. Before that, he co-founded a venture called Overture, whose singular mission appeared to be to beat Google at its own game. So hiring Dr. Flake clearly sends the message that Live Labs is a competitive effort against Google.

But how much of the broader efforts of the whole of MSR will be redirected toward beating Google? Schofield indicated that the work of Live Labs should not be construed as diverting MSR from its original and existing goals. That said, it also doesn't appear likely that Live Labs will do much to expedite those goals that exist outside what he calls "the Internet-scale space," at least for the next few years.

In a research organization, Schofield said, "there's a fundamental tenet that you don't realize until you dive in deep: You can't staff up research like you staff up a product group...I hate to say this, but it's almost a drag-and-drop [process] for software development. Take some people, move them over there, work on that. Research doesn't really work that way, because it's all about solving unsolved problems, and moving the state of the art forward in a particular area where no one has gone before. It's really a bet on the people. First thing you do is, you find people who have domain expertise in that area, and have an interest in diving deeper into research in that area, and you bring them in."

From that statement, one can conclude that MSR won't be arbitrarily shifting key researchers and engineers from existing products in other spaces, over to a new project focused on beating Google. Instead, researchers will continue to engage in the company's existing technology transfer program, which has been described as a conduit for MSR researchers sharing their discoveries with the company's production teams as they happen.

"We really don't think in terms of timelines," Schofield said about his Research division. "Fundamentally, you can't schedule breakthroughs. That's the complementary role of research versus engineering. Engineering, we can schedule it, because people basically understand that with research, you take risks, and some of the things you do are going to fail...[So] you bet on people. You go find the best people you possibly can, give them the best resources you can, the best support you can, and you set them loose. You do this as an act of faith, that if you hire the best people with the best resources, they're going to come through for you."

Schofield drew a kind of verbal map of where he pictures the relationship between Microsoft Corp. and MSR currently, and where he wants it to evolve. On one end of his scale is what he calls basic research, which is the bedrock of his organization, and which he describes as "the steady march of progress in all these different technologies, not necessarily thinking about a particular product." At the other end is engineering, whose processes can be more strictly controlled, he explained, because the people involved are working with the tools and concepts they already know. Engineers can go build something, he remarked, when they have a solid handle on how to build it. This contrasts with basic research, where the how's and even many of the why's have not been determined yet.