Microsoft Research: 15 years of ideas

While corporate interest in basic computer-science research waned then waxed, Microsoft was building a research organization that rivals anything on a university campus.

Microsoft celebrated the 15th anniversary of its in-house computer-science research effort this month, calling it critical to the future of its business.

"The reason we have Microsoft Research is so Microsoft will still be here 10 or 15 years from now," Rick Rashid, the senior vice president who heads Microsoft Research (MSR), said Tuesday.

MSR grew up to be the leader across a spectrum of computer-science specialties in a time of dwindling investment in long-term, basic research by other technology companies.

In previous decades, powerhouse corporate-research outfits such as AT&T's Bell Labs, IBM Research and Xerox's Palo Alto Research Center (PARC) were known as top producers of basic computing research focused several years down the road — rather than on the next product cycle.

PARC, for example, gave birth in 1973 to the first graphical user interface, a concept later used by Apple and Microsoft. The technology didn't make its way into a Xerox computer for commercial sale until 1981, and the machine never sold particularly well.

That kind of higher-risk research withered in the name of short-term profits at many technology companies.

"Microsoft is almost unique in its investment of research that looks out five, 10, 15 years," said Ed Lazowska, Bill and Melinda Gates Chair in Computer Science and Engineering at the University of Washington. "Name your favorite multibillion-dollar IT company, and for the most part they do none of that."

Rashid said MSR, which represents a small fraction of the company's $6.6 billion annual research-and-development budget, first aims to move the state of the art forward before considering whether the work is relevant to a specific company product.

"If we're doing great research, we'll make it relevant to Microsoft," he said.

That approach has been adopted by other companies and the trend away from basic research has started to reverse.

Intel has long had an advanced-research organization tightly tied to its product road map but launched a basic-research effort about six years ago, said James Landay, the outgoing director of Intel's lab at UW and a computer-science and engineering professor.

Yahoo! has been growing its basic research since March 2005, an effort overseen by MSR alumnus Usama Fayyad.

Google's approach has been to integrate research closely with its product groups.

With the resurgence in corporate computer-science research comes greater competition for talent in the field.

"More than a quarter of the people Google has hired from our department have Ph.D.s," UW's Lazowska said.

MSR gobbles up a huge amount of the talent produced in this country and around the world.

Rashid said his organization has added the equivalent of a large university computer-science department every year of its existence and has expanded to China and India to hire computer scientists there.

He's quick to highlight the similarities between his organization and traditional academic departments: Research is open — researchers have published some 3,700 academic papers over the years — and subject to peer review.

Hundreds of interns swell MSR's ranks during summer, increasing the resemblance to a university. It even confers degrees in partnership with Chinese universities.

In the early days, recruiting was a critical challenge at MSR.

Landay, the outgoing Intel lab director, was a computer-science graduate student at Carnegie Mellon University when Rashid left to head Microsoft Research.

"I have to say at the time a lot of people were pretty skeptical about it," Landay said. "Microsoft wasn't really known for leading-edge technology at all."

Rashid said he saw the opportunity to pair research with the business engine revving up at Microsoft in the early 1990s.

"The input to a company like Microsoft was the natural output of a research lab," he said.

Microsoft Research, with 700 employees and growing, today generates more of that output than the company's product groups can use.

One recent response to the surplus is Microsoft's Intellectual Property Ventures Program, launched in 2005.

The program marked a milestone Tuesday with an announcement from one of its early participants. Wallop, a Bay Area startup, launched a trial version of a new invitation only social-networking service based on technology developed by Microsoft Research.

Benjamin J. Romano: 206-464-2149 or bromano@seattletimes.com