Some scientists, including Egill Hauksson, a senior research associate in geophysics at Caltech, who oversees the Southern California Seismic Network, have doubts about the quality of that data. Nonetheless, Hauksson says, "If you have hundreds of thousands of these computers reporting, maybe you will see something interesting."

Sensors in quake-prone areas such as California are miles apart, and Davis says if there were more QCN-linked computers in an area, they could provide information on how the shaking varied across the affected area. "It's obviously a very limited seismometer," Davis says, "but it would indicate where the biggest shaking concentrated in a way we've never done before."

The devastating Northridge earthquake, which hit Los Angeles in 1994, had some unexpected effects in parts of Southern California, Davis says, so scientists deployed seismographs in backyards to try to figure out what was going on. "Had there been all those laptops measuring at the time, that could have been worked out much quicker," he says.

Lawrence's hope is that the network might even be able to give an early warning of quakes, based on the relatively gentle waves that occur before the more brutal ones. Even just a few seconds of warning may be enough time for people to take cover and automated systems could slow trains and divert traffic from vulnerable bridges. There's no such system in the United States, but in Japan, high-speed trains are stopped when a major earthquake is detected.

However, Caltech's Hauksson says he's "very skeptical about using this kind of network for warning."

Although David Oppenheimer, a seismologist with the U.S. Geological Survey, who isn't involved with the project, sees "significant problems" with the notion of using laptops as quake sensors, he's intrigued by the idea of equipping desktop computers with inexpensive seismic sensors.

"To me, that's very exciting because there are large portions of the world where we don't have adequate seismic monitoring," Oppenheimer says. If USB accelerometers were attached to internet-connected computers in those regions, they could detect a quake more quickly than more-conventional sensors located hundreds or thousands of miles away.

If something like the QCN had been in place in Indonesia in 2004, when a huge quake triggered a devastating tsunami, it could have helped in warning emergency workers. "Thousands of kilometers of laptops could have lit up in Sumatra," Davis says. "They could know within a couple of minutes that it happened," rather than waiting for the half-hour it took for the quake to be picked up on sensors farther away. That earlier notice could allow response teams to mobilize more quickly during the next quake, and tsunami warnings could be issued in time to make a difference.