Simple as that solution is, it might still be difficult to input the search on a cell phone, so researchers developed a way to let numerals represent the keys. Thus the search for Secretary Rice is accomplished by entering 2 7423, with the space representing a wild card. "The Wild Thing" can also be used in conjunction with MSN's "Virtual Earth" database of maps and aerial photos to quickly search for a business; a few keystrokes entered on a simple form revealed the location of the nearest Starbucks in Mountain View and a map to get there. Microsoft researcher Bo Thiesson says "Wild Thing" will soon be handed over to MSN but won't go live until an extensive round of usability testing -- and likely tweaking -- is completed. Not every project at Microsoft Research is tied so closely to a product. And that's just fine, says Roy Levin, director of the company's Silicon Valley lab. "Advancing the state of the art is the first priority; helping the business is second," he says. Levin is quick to add, however, that his dictum does not imply a lack of concern for commercial applications. "We need to innovate, but innovation is not something you just asset. You must stay on the cutting edge. And to do that, your work has to be subject to peer review," he said during an interview. So staffers at Microsoft Research's five labs (including facilities in Redmond, China, India and the U.K.) turn out a steady stream of academic articles and papers for journals and conferences. Unlike the rest of Microsoft, which reports to CEO Steve Ballmer, the research group reports directly to Chairman Bill Gates. The labs and the business units are linked by a small team of program managers, all of whom have a track record of actually shipping a product. "They're the connectors; they know where our technology might fit the needs of a business unit," says Levin.