Intelligent Machines
Corporate R&D Set Free
Secrecy is verboten at Intel’s network of university “lablets.”
At Carnegie Mellon University in Pittsburgh, Mahadev Satyanarayanan is working on a mobile-computing approach he calls “Internet suspend/resume.” The idea, says Satyanarayanan, is to be free to stop your work, have your files saved automatically over the Internet, and-when you’re ready to resume-find “your world restored” on any computer in any location, as if you were using your personal laptop.
The computer scientist is also experimenting with a new method of transforming his research into real-world technology. Every day, he takes a four-minute walk from his university office to an off-campus research lab funded by Intel. At that lab, the third “lablet” Intel has established adjacent to a leading university, Satyanarayanan has opportunities to translate his vision of Internet suspend/resume into a working prototype-and test Intel’s stated commitment to collaborating openly with academic researchers.
“Most companies have real difficulty” reconciling the academic urge for open communication with the corporate imperative to own and profit from ideas, notes Satyanarayanan, who became director of the facility in August 2001. Indeed, many corporations that fund university research force faculty and graduate students to sign nondisclosure and exclusive-licensing agreements. But “Intel has a collaborative model up front,” Satyanarayanan says. “The right approach is not to tightly control intellectual property but to treat it the way a university does.”
And so far, that’s exactly what the lablets are doing. “The vast bulk of the intellectual property produced by this research will be nonexclusive and licensed to all comers,” says Intel research director David Tennenhouse, architect of the lablet program (see “Intel Revamps R&D,” TR October 2001). At the first Intel lablet, near the University of California, Berkeley, for instance, the operating system behind the lab’s self-organizing networks of miniaturized wireless sensors is accessible to anyone who takes out a free license. Indeed, Intel plans to penalize lablet researchers who don’t share enough. “If a lablet isn’t collaborating with its university, then I’ll close it,” Tennenhouse says.
Support for this kind of openness has nothing to do with charity. “Intel stands to benefit in the long run,” says Tennenhouse. By accelerating progress on new types of mobile computing and other disruptive technologies, Intel hopes to promote ideas that don’t fit within existing business lines but may transform consumers’ computing habits and become critical to the microprocessor market within the next few decades.
Gaetano Borriello, the director of Intel’s lablet at the University of Washington in Seattle, says the openness guaranteed by Tennenhouse was a big factor in his decision to take a leave of absence from the university to start up the facility. “If it had not been a new model, if Intel had just been doing an established corporate-research lab, I wouldn’t have been so interested,” says Borriello, whose lablet focuses on “embedded computing,” the effort to equip working and living environments with small, out-of-sight computing devices. “Instead of doing this at Microsoft, say, and not being able to talk about my work, I’m at a place where I’m encouraged to talk about it.”
For both Intel and the lablets, though, true success might well depend upon accomplishing a different kind of suspend/resume operation-when the time comes to “declare the open’ phase of a project over and shift the research inside” Intel proper, says David Culler, director of the Berkeley lablet. Culler says such transfers will be critically important, not only because they will help Intel collect on its investment, but also because they will create room at the lablets for entirely new projects. Already, though, Culler believes that Intel has achieved another goal with its lablets: “It has stepped into a position of intellectual leadership.”
