washington

SETI goes home: a screensaver searches for patterns in intergalactic radio noise.

Three months after it began, an innovative scheme to enlist public help in the search for extraterrestrial intelligence (SETI) already has more than a million volunteers taking part in 224 countries — making it the largest distributed computing project in history.

SETI home is a downloadable screensaver programme that uses idle time on home and office computers to look for patterns in radio-telescope data. Its spectacular success now has its creators thinking about ways that scientists in other disciplines could harness lots of small computers to process large amounts of data.

The programme was initiated by the University of California at Berkeley's SERENDIP SETI project, which every day records 35 gigabytes of data collected by the Arecibo Radio Telescope in Puerto Rico. The idea behind SETI is to search all this radio noise for non-random patterns that might possibly be deliberate signals from another civilization.

But analysing large volumes of data with the desired sensitivity and frequency resolution is a daunting task. A distributed system “was really the only way we could solve this computation problem”, says SERENDIP director Dan Wertheimer. “Together, the million participants have formed the largest supercomputer on the planet.” The combined force is capable of six teraflops, or six trillion floating operations, per second.

Participants download the screensaver programme from a Berkeley web server (http://setiathome.ssl.berkeley.edu) along with 350-kilobyte ‘work units’ representing 107 seconds of Arecibo data. When one unit finishes processing, it is uploaded to Berkeley and another one is downloaded.

Wertheimer, who expected only 100,000 participants, admits that there were growing pains, including early difficulties handling the data traffic and providing customer support. He also says he was worried about quality control in the data returning from users. But very few people have been filing fraudulent work units, and these have been easy to spot. The project is expected to run for two years.

David Anderson, the Berkeley computer scientist who leads the project, says he is now looking around for other problems to work on that could benefit from such large-scale distributed computing. He has already had queries from biologists working in fields ranging from drug design to genetics, and hopes to set up a centre at Berkeley to offer the public a choice of projects in which to participate.

The challenge for scientists — and an obvious advantage for SETI — is coming up with computational problems that capture the public imagination.