Ideas on display at Microsoft's TechFest

REDMOND, Wash.--Microsoft researcher Mike Sinclair wants to teach cell phones to dance.

By holding two phones together and briefly moving them in the same way, Sinclair said, the phones can generate an encryption key that will let them share data in a way that even a determined snoop nearby would not be able to intercept.

There are millions of cell phones out there with Bluetooth wireless technology, ideally allowing for all kinds of ad hoc connections. However, security concerns, as well as the difficulty of creating sharing "profiles," mean that few of these hookups ever take place.

Raman Sarin, a software design engineer in Microsoft's research unit, wants to solve the same problem but has a different approach. His idea, dubbed "Blue Rendezvous," is to have the users of two Bluetooth-enabled phones press the same button to signal that they want to talk to each other. Once that connection is established, the people can then send whatever information they like to each other: business cards, photos and the like.

Separately, both Sarin and Sinclair were pitching their ideas to Microsoft's product units as part of TechFest, a two-day internal science fair that runs through Thursday at Microsoft's headquarters here.

Though such overlap might be seen by some as waste, Microsoft researcher Bill Buxton believes allowing for it is a good idea. Buxton, who joined Microsoft just this past December, says he's encouraged seeing researchers working in similar areas--that means the idea is more likely to become a reality sooner rather than later. He notes that entirely new ideas, like the mouse, take decades to make it to market.

Buxton, the former chief scientist for Alias Wavefront and a TechFest newbie, said he was impressed by the technology on display but was more pleased to see the looks on the faces of the thousands of Microsoft product unit workers who strolled through the booths.

"You need this period to help raise your head," Buxton said. It sends the message to workers that "I'm not going to be (just) fixing bugs in Windows 10.0 or Windows 20.0."

Buxton said that he's convinced such efforts will bear fruit and that he was impressed with what he saw. "I'm convinced in 10 years people are going to say 'When is Apple going to come out with this? When is Google going to come out with this?'"

Unit's growing importance
Microsoft's research unit, now at 700 people, has become increasingly important to the software giant, which is in the throes of competition with both of the companies Buxton mentioned, as well as others.

When Microsoft found itself behind on search after years of underinvesting, it was the research unit that allowed the company to at least try to get back in the game. Though MSN had outsourced most of its search work, the research unit had continued to delve into new areas, including desktop search and approaches to improve Web search relevance.

"We stepped in and we had many components," said Henrique Malvar, director of Microsoft's Redmond lab, the largest of Microsoft's five research facilities. "They were very glad."

Microsoft worked quickly to bring efforts in neural networks from theory into practice in serving up relevant results for search queries.

"We were actually pleasantly surprised how much of the (research project) translated," he said, noting that the company's research projects are not always designed to scale.

But while he's glad to be of assistance, Malvar is also concerned that his researchers not be swayed too much by the desires of product groups and aim too low.

In a recent performance review of one of his researchers, Malvar scanned through a proposed plan for the coming years and decided that it seemed quite doable. But he sent the researcher back to the drawing board.

"We don't want too high of a batting average," he said.

His ideal researcher would have a batting average like a good baseball hitter. "If one out of three crazy ideas works, that's pretty good." But to be good at research, you have to fail more than you succeed. "If you don't do that, you are not trying hard enough to push the limits," Malvar said.

Still, Microsoft is doing more blending of research and product groups, and Malvar said that's probably a good thing. The company recently announced plans for "Live Labs," an effort that will further blur the lines between research and product development.

"I think it will work," Malvar said. He said Microsoft needs to be more nimble, and the research unit can provide a boost.

"Our ability to generate ideas and prototypes is not that bad," he said.

TechFest itself is designed to be a conduit between Microsoft's research and product teams. The shirts worn by the company's researchers make the point: "Techfest: The & in R&D."

When the idea for TechFest first came up several years ago, Microsoft Research boss Rick Rashid thought it was a bad one.

Six years later, he happily admits he was wrong. TechFest, now in its sixth year, has become a wildly popular event. More than 150 booths are packed into Microsoft's main conference center here, with more than 6,000 of Redmond's ranks either having stopped by or expected to stop by before the event concludes Thursday.

The booths run the gamut, but are centered on several areas of particular importance, including search and digital media technologies.

Some are clearly aimed at one product area, such as A.J. Brush's inkable digital calendar that aims to bring the handwritten home calendar and bulletin board into the digital age. Others were more systemwide approaches, such as the Stomp User Interface, designed to let people use their feet as a means of inputting text.

Microsoft researchers also used the event as a chance to sign up human guinea pigs for their latest projects. Richard Hughes was hoping to find Microsofties who were willing to try out his program, Pinpoint, which lets users track their willing friends and get e-mail alerts--say when a friend is nearby, or if their kids have left a 60-mile area.

If all goes well, Hughes will find enough willing participants to tell how well the software works and whether the privacy safeguards that are in place are sufficient.

"At least some people," he said, "think that it is pretty cool."