SETI@Home is a University of California at Berkeley project founded in 1996 with the intention of searching for extraterrestrial intelligence. Its software, which was distributed through the Internet in June and has been downloaded close to a million times, analyzes clumps of data pulled in from the Arecibo telescope in Puerto Rico for signs of alien communication -- patterns that signify more than just space noise.
Problem is, so many people are using the software -- about 370,000, according to project leaders -- that there's more analysis going on than data being pulled in.
For every new unit of data pulled in by the telescope, about 1.13 units are analyzed. That has led SETI@Home to start sending out its backlog of data to users, but even that might run out -- in about a year.
"It's a complex situation," says Dr. David Anderson, the project director.
It's apparently so complex, and so sticky, that Anderson says he is deliberately holding off on efficiency improvements of the analysis software, at least for now.
That has some software developers upset.
"That gets me mad and it should get you mad as well," writes Armin Lenz, Managing Editor of the website 3Dnow.org in an issue forum on FullOn3D.com. "Let people have the fastest client software possible That is the least they deserve if you want them to support the project."
Lenz, and others, are seeking the source code for SETI@Home software so that they can tinker with the algorithms and optimize the speed for different kinds of computers.
But Anderson says having dozens of different versions of the software floating around the Internet would cause a "software engineering nightmare."
"The programming will have so many customized versions it will be so hard to add on, or debug, or maintain," he said.
Moreover, increasing the speed of the software now would only deplete the data backlog faster.
So what to do?
"The various options are to turn away people, to give the same piece of work to two people simultaneously," or, says Anderson, "increase the amount of computer time for each unit."
It looks like the last option has been chosen. Beginning in a few months, analysis of the signals will be more in-depth and time-consuming.
And if that slows things down too much, then, and only then, will SETI@Home release a faster version of the software.
If anything, the popularity of SETI@Home has given credence to distributed computing. Anderson says its the first time the process has been used for scientific research.
He envisions that in a year from now, people will be able to choose the kind of research they want done on their computers -- whether it be assisting cancer or AIDS research, or crunching data for the Human Genome project.
advertisement
