Strength in Numbers

The next time you idly get up to grab another cup of coffee--or lose a few hours in a company meeting--the personal computer on your desk could be working on a cure for AIDS or discovering the secrets of extraterrestrial life.

But for lowly PCs to assist in these enormous scientific tasks, otherwise possible only on supercomputers that simultaneously crunch zillions of mathematical computations, requires teamwork. A single PC must become part of a "virtual" supercomputer, a network of pcs grinding away at small pieces of a problem. Through a process called distributed computing, such proletarian networks enable researchers to bypass the power limitations and huge expense of today's supercomputers. Distributed computing also allows millions of PC owners to donate unused computing power for scientific good: a merger of science and philanthropy that's taking root at research centers like Stanford and in new socially minded businesses.

Joining a distributed computing network is painless public service, says Vijay Pande, a pioneering assistant professor of chemistry. He has drafted thousands of "volunteer" PCs to help simulate the intricacies of the way proteins assemble, or fold, themselves into different structures. Known commonly as protein folding, examining a protein's structure offers clues to its function. Understanding the origami-like configurations of important proteins in the body can help researchers design treatments for debilitating diseases such as mad cow's and Alzheimer's.

During a 24-hour period, roughly 99 percent of the average home or office PC's processing power is wasted. To put that unused power to good use, donors simply download an unobtrusive number-crunching software program to their PCs, often in the form of a special screensaver. When the PC is connected to the Internet, the program automatically receives a small "homework assignment" in the form of a data packet from the research project's central processor. (Like a screensaver, the program immediately jumps out of the way whenever the computer's power is needed.) The program completes its homework while the computer is idle and sends the finished product back the next time the PC is online.

Working together, these clusters of PCs easily leapfrog the processing power of the world's supercomputers with as few as 10,000 volunteers. Case in point: Pande and his colleagues developed equations to simulate how various proteins fold, or perform their specific tasks in the body, but they needed massive computational power to mimic folding enough times to be scientifically useful. "The trouble is, our bodies are infinitely faster than computers," says Pande. With just one PC, a single simulation would take 30 years. Even on a supercomputer with 100 processors, one simulation could last three months. "That's not something we can wait around for," he says.

So Pande's group turned to distributed computing. They developed software to share the workload among thousands of PCs and in September launched the effort as a public service project called Folding@home. Pande's impatience with supercomputers paid off almost immediately. In just six weeks with 7,000 active PC donors, the group succeeded in replicating the invisible--one complete protein-folding simulation. Although Pande concedes, "Once is only a stunt," he intends to achieve scientific validation when he reaches his goal of 100,000 donors.

The enormous potential of distributed computing is exciting entrepreneurs as well as university scientists. San Diego start-up Entropia is one of several companies harnessing unused pc power from both corporations and individual volunteers to speed research for pharmaceutical, life science and technology businesses. Entropia's founders also believe the company has a role to play in philanthropy. "There's such a mind-boggling amount of power [generated by distributed computing] that there's room for both," says CEO Jim Madsen, MBA '89.

In September, Entropia kicked off Fightaids@home in partnership with the Olson Laboratory at the Scripps Research Institute in La Jolla, Calif. This distributed network gives biomedical researchers the computational horsepower to design new drugs needed to fight aids. Donors in this program give 90 percent of their unused computer time to the aids project and 10 percent to Entropia's commercial clients.

Entropia permits donors to increase their AIDS contribution to 100 percent, but many participants opt to share their donation with Entropia's commercial side. "Nobody else can afford to do this research," says Bob Adams, a nonprofit director in Chesapeake Beach, Md., and a participant in Fightaids@home. "We should be grateful that a commercial organization is willing. If they fail [as a company] it won't do anybody any good."

These public-private partnerships make sense, says Larry Smarr, MA '72, director of the National Center for Supercomputing Applications and a member of Entropia's board. "I have spent my life getting grants. This is the end of the line," he says. "This is the new model for funding large-scale ideas."

Researchers, especially in technical circles, have pooled the power of individual computers for decades, but SETI@home, a distributed computing project designed to aid the Search for Extraterrestrial Intelligence, was foremost in putting individually owned personal computers to work. seti@home has drawn 2.4 million donors since its launch in May 1999 but still has not "bagged any aliens," jokes Dan Wertheimer, the project's chief scientist.

Following SETI's lead, both Folding@home and FightAIDS@home recruit donors through their websites, word of mouth and media coverage. seti also encouraged healthy competition among donors--a practice that lives on in the newer projects. Teams of PC owners enjoy competing for the glory of bringing in the most computer time, a somewhat geeky version of online fantasy football leagues.

Folding@home's leading team was founded by 17-year-old Rian Stockbower of Londonderry, N.H. A former SETI competitor, he convinced 50 of his cohorts from a computer programming e-mail list to join him. "I was drawn by the fact that it (the protein-folding research) had real benefit for humanity," he says.

Potential donors need not worry about risk to their machines. Distributed computing programs are secure--they cannot tap into sensitive personal data files--and do not interfere with the operation of the computer, making them acceptable additions to most workplace PCs.

Folding@home has yet to draft the more than 32,000 computers on the Stanford campus. But Pande's group plans a publicity push after a second version of the software is ready and a version for Macintosh computers, about 27 percent of Stanford machines, is finished in early 2001.

For Pande, Folding@home embodies his best scientific instincts--to solve the mysteries of the body and make his discoveries accessible to everyone. Donors learn about a subject otherwise unknown to them, a teacher's dream, says Pande. "It's rare to be able to combine the cutting edge of research with a very useful education tool. It blows away anything else."

For more information or to volunteer, visit www.stanford.edu/group/pandegroup/Cosm/


Deborah Claymon, '92, is a San Francisco freelance writer specializing in workplace and cultural changes spurred by technology.