FRAMINGAM: Microsoft Research, the software company's university-style pure research arm, celebrated its 15th birthday by showcasing some of its recent greatest hits.
While Microsoft publicly trumpeted the many incremental innovations that have streamed out of Microsoft Research's labs since 1991, its 750-researcher group, which Microsoft said is the largest computer science department in the world, has yet to produce a revolutionary technology to match research giants such as Palo Alto Research Center (mouse, graphical user interface), Bell Labs (Unix, six Nobel Prize winners) or the current leader, IBM (relational database, RISC chips, Deep Blue).
Microsoft also continues to wrestle with how to handle the fruits of its laboratory's labours. In May 2005, it began licensing patents from Microsoft Research to outside companies. That has resulted in a handful of spin-offs, such as the social networking site Wallop, which last month announced that it had received $US10 million in venture funding.
But it raises the question of whether aggressively pursuing lucrative licensing deals would divert Microsoft Research from its core mission of serving Microsoft products and boost competitors.
"Microsoft hasn't shown any inclination to becoming a major IP licensing company," an analyst at Directions on Microsoft, Paul De Groot, said. "Could they make more money? Maybe, but maybe not."
Microsoft has never been burdened by the "not invented here" syndrome. Its first product, MS-DOS, was created by licensing another version of DOS and tweaking it. And some note the many similarities between the original Windows and Mac user interfaces. At the celebration at its headquarters, Microsoft Research demonstrated many of its more recent inventions to combat the company's lingering "me-too" image. For instance, the Touchlight 3D holographic display lets users select and manipulate data with their hands, similar to the movie, Minority Report.
Microsoft also showed its research into Bayesian-style artificial intelligence, which relies on probability and statistics. A principal researcher at Microsoft Research, Eric Horvitz, said such technology could some day be incorporated into Office Communicator messaging software to predict how long users would be tied up on phone calls and pass that information to would-be callers or to screen calls and instant messages.
Horvitz also showed off a feature called SuperFetch built into Vista that allows programs to better manage their use of system resources.
In the long term, one of Microsoft Research's biggest contributions could be in improving how operating systems are developed after the much-delayed Vista. Vice-president of research, Dan Ling, said that his group was creating tools to "componentize" the development of large software.
While Ling acknowledged criticism Microsoft Research had so far failed to produce paradigm-shifting technology, he argued that few high-tech advances in the past decade had been research driven.
"Search, via AltaVista, did come out of a research lab. But you can't say that a Web browser is revolutionary technology. Or Google's Web advertising model. These are not technological innovations; they are business ones," he said.
Microsoft's 746 US patents in 2005 ranked it 18th, far behind IBM, which led for the 13th year in a row, with nearly 3000 patents, according to IFI Patent Intelligence. However, De Groot said there was a greater degree of difficulty to winning the software patents that Microsoft sought, compared with hardware or chip ones.
Microsoft's patent total was up 18 per cent over 2004, while most companies, including IBM, had double-digit drops.
Moreover, Ling, who was a senior manager at IBM's flagship Thomas J. Watson Research Center before joining Microsoft, said the company trumped IBM. "Working with product groups was a difficult thing at IBM," he said. To improve relations at Microsoft, Ling has embedded researchers into product groups to foster teamwork and hired developers for MSR so that his team can create slicker tech demos to impress product managers.
Slick demos, however, did not help Microsoft see the potential in Wallop, a social networking site that languished for three years before being spun out to a start-up. In those three years, sites such as MySpace.com and Facebook emerged out of nowhere and gathered tens of millions of users.
Senior vice-president for research at Microsoft, Rick Rashid, said that there would always be some tension resulting from product teams at Microsoft wanting to keep promising technology in-house and Microsoft Research's desire to see it released, even if it is with a start-up. "This is just the first year of the program," Rashid said, referring to the company's licensing efforts. But for the technology licensing team, he said, its goal was to get our technology in other peoples' hands and it was very motivated to do so.