Microsoft shows off research goodies

At an event recognizing the 10th anniversary of Microsoft Research, Gates indulged his fondness for technology, raising hopes for a world where computers will become more useful. "The message you'll get is one of incredible optimism," he said as he described Microsoft's vision for its research labs.

Microsoft follows in the footsteps of more traditional computing companies such as Hewlett-Packard and IBM, which have long had basic research efforts and enjoy making a spectacle of their latest discoveries.

Microsoft Research concentrates on three main areas, Gates said: using computers to extend what people can do; improving digital music, videos and images; and building the company's .Net vision of computing services available over the Internet.

Among the technologies Microsoft showed Wednesday were:

•  A handheld computer that understands which way is up and where it's being touched, technology that lets it reorient the display according to how it's held or understand when a person is holding it like a cell phone to give dictation.

•  The "Mulan" software project for reading Chinese writing out loud or transcribing speech into characters. With about 60,000 characters in Chinese, it's difficult to use keyboards.

•  Automated bug detection that helped make Windows 2000 less crash-prone, which is being used in all other Microsoft product lines.

•  Video compression technology that's less error-prone than the prevailing MPEG-4 standard.

•  Software that's designed not to sap people's emotion when creating narrated slide shows so sharing photos online is more like the storytelling that accompanies the viewing of traditional photo albums.

•  Software that can reconstruct 3D images from a few still photos.

Gates defended research spending during the current economic slump. "The really big advances come from long-term research investments," he said. "We think corporations worldwide should invest more in research," both for themselves and for society overall.

"Our job in research is to make sure there will still be a Microsoft 10 years from now," said Rick Rashid, senior vice president of Microsoft Research and the leader of Microsoft's computer science work.

But science-fiction vision isn't always simple, and research isn't always rosy. Microsoft's aspirations for easy-to-use, crash-proof computers contrasted strongly with the more conservative expectations many experts have for future computers.

"Some challenges are much harder than people had imagined," Gates acknowledged. While hardware such as faster chips and larger hard drives makes new technology possible, limitations such as the slow consumer adoption of high-speed Internet connections continues to be a problem.

Research also has produced some duds, such as "Clippy," the automated help system in Microsoft Office that annoyed many people. "One of the most exciting things we did in Office XP was turn it off by default," Gates quipped.

And making research available publicly doesn't always help just Microsoft. For example, before Rashid joined Microsoft in 1991, he was a Carnegie Mellon University professor who led the development of an operating-system core called the Mach microkernel, software that has become the heart of two Windows operating systems rivals: Hurd from the Free Software Foundation and Darwin, the heart of Apple Computer's new Mac OS X.

Microsoft spends nearly $5 billion a year on basic research and product development, a sum that will increase in coming years. "Research has more than paid off for us. We're continuing to increase that investment even as the economy goes up and down through the next several years," Gates said.

Microsoft Research was founded in Redmond in 1991 and has grown quickly in recent years, with new offices in Cambridge, England; San Francisco; Mountain View, Calif.; and Beijing. The employee count has increased significantly in recent years--from 300 researchers in 1998 to 420 in 1999, 500 in 2000 and 650 today.

In the midst of this growth--and the government's antitrust case against the company--Microsoft Research lost a key employee. Chief Technology Officer Nathan Myhrvold, the man responsible for starting Microsoft Research, left for a one-year sabbatical that later became permanent.

Microsoft tries to balance open-ended research with product-oriented work, Rashid said. The group also relies on close ties with academia, devoting about 15 percent of its basic reasearch funding--$75 million in fiscal year 2001--to projects at universities.

And it takes time for research to make its way into products. The first projects from Microsoft Research emerged in Windows 95 and Office 95, Rashid said.