HOME
E-MAIL

 
Computers don't care where they sit or what kind of donuts you stick in front of them. They just flip their little bits and bytes this way and that and come up with answers.
  technofile
Al Fasoldt's reviews and commentaries, continuously available online since 1983

SETI@Home project uses an old technique of neighbors helping neighbors


Sept. 12, 1999

By Al Fasoldt
Copyright © 1999, Al Fasoldt
Copyright © 1999, The Syracuse Newspapers
    When three dozen trees blew down on the east side of our yard a few years ago, all the neighbors came over and started cutting and hauling. They helped clean up the mess in a single day.
    Neighbors helping neighbors. That's what I call it. But the nerds out there might call it "distributed tree cutting." The work goes faster when it's distributed among a lot of people. That's the idea.
    It's the same approach scientists use when they need to get a whole lot of work done in a short time. Suppose it takes one scientist one year to figure out one problem. All other things being equal, two scientists should be able to finish the job in half the time. (All other things aren't ever equal, of course, but let's pretend that everyone gets along perfectly for this example.)
    If you have 365 scientists, you should be able to get that work done in a day. Why stop at 365? If you have 8,760 scientists, you should be able to get things done in an hour.
    Hey, get with the plan. If you have 525,600 scientists, take the rest of the day off. You get the job done in a minute. Or, with 31,536,000 scientists, the work goes by in a blink. In a second.
    Of course, you'd expect 31 million scientists to put in a normal workday. Or at least a normal work hour. So 31 million of them could get a year's work done in a second or 3,600 years of things done in an hour.
    Wow! Makes sense, doesn't it? The only problem, as any parent or psychiatrist can tell you, comes from the fact that all those scientific types would be feuding and fighting from the start of that hour to the finish. They'd disagree on the kind of donuts they wanted. Some would want coffee, some would want tea. My mother, if she were one of these distributed science types, would want Postum. They wouldn't be able to agree on where to sit. And on and on. It would be a mess.
    So you let computers do all the calculations instead. Computers don't care where they sit or what kind of donuts you stick in front of them. They just flip their little bits and bytes this way and that and come up with answers. Put a lot of computers into the same room and wire them together and make them work on the same problem and you get distributed computing. Computer scientists have been doing this for years. It works great.
    Except, of course, when you don't have a room big enough. Or a wire long enough. Or enough computers. Scientists who do distributed computing talk about 16 computers as if that's a lot. Tell them they can use 64 and they faint away from excitement.
    Pshaw. How about a million? It's nothing. We've got a room big enough and a wire long enough. The room is the plain old ordinary planet, and the wire is the Internet. And the million computers? They're yours and mine. You just have all those million computers run the same program to work on some big problem and tell them to send the results of their calculations back to a central location.
    And that, in a nutshell, is the way the SETI at Home project works. Computers of all kinds run their own versions of the SETI program and chew away at computations that look for signs of intelligent life far off in the universe. They get the stuff they work on over the Internet and they send back their finished calculations the same way.
    Right now more than a million computers are doing this. It's the world's biggest number-crunching project. Calculations that would have taken years on the world's fastest supercomputer are being done in a few seconds. They're being done on a million computers instead of one, and that makes all the difference in the world.