How IBM's World Community Grid Is Helping Cure AIDs, Cancer, and World Hunger

IBM's virtual supercomputer is tapping the unused processors of half a million people to speed up critical scientific research.

For the past four years, Lauren Moran has devoted herself to groundbreaking cancer research, chronicling the fickle interaction between molecules and proteins. Despite having a full-time job — "stat geek" on the New York Yankees' media relations staff — Moran screens drug candidates daily. And continuously. She conducts experiments while talking on the phone to her parents, attending games at Yankee Stadium, and watching episodes of The Office in her Bronx apartment. Even in the dead of night.

When she's not trying to cure cancer, she's busy cracking other monumental problems: AIDS. World hunger. Clean energy. It's a breathtaking portfolio for a 24-year-old communications major who didn't take a single chemistry or biology class in college.

Moran is a new breed of innovator: a citizen researcher on IBM's World Community Grid (WCG), an unprecedented effort to deploy ordinary people's idle computers to create a free, open-source lab for researchers around the globe. Massive computational research is broken down into discrete problems and distributed across a vast network. Since the tech giant launched the nearly $2-million-a-year project in November 2004, more than half a million people in 218 countries have volunteered some 1.5 million laptops and desktops. In raw computing power, the grid is comparable to a top-10 supercomputer. The average PC would take more than 328,000 years to complete the grid's calculations so far.

The grid, says researcher Alán Aspuru-Guzik, an assistant professor of chemistry at Harvard, "gives you the opportunity to do something nobody else has done. Something disruptive."

Moran's laptop displays a screen saver of her latest WCG assignment, but the science, she admits, is "way over my head. I just know when I'm not using my computer, it's crunching numbers that could lead to a cure."

Most of us use our computers about as efficiently as we use our brains: We scratch the surface, never tapping the full potential. WCG exploits this unused computing power by borrowing — with the owner's permission — a machine's central processing unit to do some serious math. It works unobtrusively, when you aren't working. You download software that takes advantage of any break, from a phone call to a pause while you're thinking of what to type next. The instant your fingers touch the keys, the calculations cease.

At IBM, a full-time staff of seven — dispersed across the country, from Beaverton, Oregon, to Austin — makes sure that the projects' individual applications are running smoothly, that the grid is assigning work and returning results to the appropriate lab, that problems aren't cropping up in the online member forums, and that software for upcoming projects is being vetted.

Using this powerful new tool, AIDS researchers at Scripps Research Institute are generating new drug leads to combat the growing strains of drug-resistant HIV. French scientists are learning more about the proteins behind muscular dystrophy. (Partly because of that project, the lead researcher, Alessandra Carbone, was recently named the "Woman Scientist of the Year" by the French government.) Scientists at the University of Washington are compiling a comprehensive map of rice proteins, which could help developing countries grow more nutritious, higher-yield crops. A team led by the Cancer Institute of New Jersey used the grid to develop algorithms that identify subtle signatures in digitized cancer tissues that could lead to early, accurate, and rapid detection; the results convinced the National Institutes of Health to award the team $2.5 million to expand the database.

WCG, which hosted one project its first year, now runs a half dozen or more simultaneously. The latest: In hopes of discovering new organic electronic materials that could lead to cheaper solar cells, Aspuru-Guzik is screening about 2 million chemical compounds for photovoltaic properties. That's roughly 20,000 times more compounds than he could analyze on a single computer. And the project will take only a couple of years, instead of two decades.

"We're opening up the field of bioinformatics," says Robin Willner, IBM's vice president of global community initiatives, including WCG. "There's nothing else out there like this."

"Looking for aliens is great, but let's cure cancer"
Many companies run in-house grids on their employees' machines, but a network of WCG's size and scope wasn't feasible until recently. "Back in the '70s, we were thinking, What if we could do this?" says Viktors Berstis, a 33-year IBM veteran and WCG's chief scientist. Divvying up data processing to the public around the planet wasn't practical until enough computers were connected to the Web, connections were high speed, and machines' processors were powerful enough to hammer through dense algorithms rapidly.

The first, and still the biggest, public grid is SETI@home, launched in 1999, with more than 3 million members. Its computers analyze radio telescope signals as part of the Search for Extraterrestrial Intelligence. From the beginning, IBM had a different vision. "As someone said early on, looking for aliens is great, but let's cure cancer," says Willner. In 2004, after the company sponsored a modest grid that identified a few dozen possible smallpox treatments in just six months, the systems and technology group proposed creating an enormous grid — using the same open-source grid-management platform as SETI, called BOINC, developed at the University of California, Berkeley — to host multiple humanitarian projects. It would be the first network of its kind and a compelling recruitment tool for employees and clients alike. CEO Sam Palmisano signed off within two weeks.

Certain kinds of research are well suited for grid computing, such as those involving what scientists call "embarrassingly parallel" problems — those that can be solved independently on machines that don't need to communicate. They work in parallel and report results to a central clearinghouse. This is particularly useful in early drug development focused on finding molecules that inhibit a targeted protein — the building blocks for drugs.

Stanley Watowich, a biochemist at the University of Texas Medical Branch in Galveston, uses WCG to model in algorithms what happens between a small molecule and a protein related to dengue fever, West Nile virus, and hepatitis C. His virtual experiments analyze millions of possibilities in months, as opposed to more time-consuming and expensive wet-lab experiments that might examine only hundreds. The first stage winnowed the molecules to several thousand that plug into binding sites on a dozen proteins, like puzzle pieces, says Watowich. The second stage narrows the field again, evaluating the binding energy of those candidates and the proteins, a highly complex computation that's practical only on a large grid. The idea is to reserve lab time for the molecules with the greatest chance of becoming effective drugs. "This has the potential to be a game-changing approach," says Watowich.

Without WCG, he couldn't have pursued this path. "We've worked with supercomputers before," he says, "but there's no way we could say, 'We'd like half your supercomputer for the next six months. Would that be a problem?' " Watowich conducted experiments on a 1,000-computer test grid of his own, but even a network that small took six months to get up and running and proved demanding to maintain. "Now," he says, "IBM does the heavy lifting."

Doing something by doing nothing
"You haven't joined yet?" Willner asks. "Can I scold you?"

The IBM team is acutely aware that its public grid is only as effective as the public allows it to be. The challenge is finding more volunteers and keeping them. As big as WCG is, it represents about 1% of the world's estimated computer population. Often, says Willner, people have to get over any misplaced fear about what the download does. "It's 'dumb' software," she says. "It can't follow you or read your hard drive." All it can do is perform a few specific tasks: determine if the machine is free, run the algorithms it has been assigned, and transfer the results to WCG servers in Toronto. The fact that there have been no security breaches or viruses instills trust.

Rather than use all of the available computing power, the program is set to draw 60%. "We don't ask users to keep their machines on 24/7," says Berstis. "We tell them, 'Use it as you normally would. Everything works the same on your computer. This is running in the background.' " If you want to track the activity, a screen saver provides a running tally of which project the machine is working on and its progress. WCG assigns points for the computations, awards badges for point totals, and ranks individuals and teams. "People go nuts for the points," says Berstis. "There are lots of bragging rights."

Worldwide, there are several dozen large public grids, says Berstis.Their membership tends to top out in the tens of thousands, while WCG continues to grow by several thousand members a week. Although IBM promotes the effort on Facebook, at conferences, and internally, the volunteers themselves are the best recruiters. Some 400 companies and schools that have signed on as partners receive marketing tools to spread the word. Today, people on some 24,000 teams — from Slashdot Users (3,953 members) and Facebook (382) to Team Boulder (413), run by teenage brothers Grant and Max Buster in Colorado — blog and post Web videos about the grid.

For her senior project at Marist College in Poughkeepsie, Moran went door-to-door, installing the WCG program. She and her classmates persuaded the college president to send a campus-wide email encouraging participation. The campaign slogan: "When doing nothing is doing something." It wasn't a tough sell. "I think everybody would like to do more volunteer work, but they don't have the time," Moran says. "This is easy, even for the busiest people."

Berstis and his colleagues realize that for many volunteers, joining WCG is more gratifying than writing a check to fund research. It's an opportunity to participate, to feel a part of the scientific journey toward a life-changing solution. That's why project names have evolved from the esoteric (Human Proteome Folding) to the inspirational (Help Conquer Cancer, the Clean Energy Project, and FightAids@Home). For developer Larry Mezias, one of more than 109,000 IBM employees on the grid, FightAids@Home is deeply personal. His cousin died of the disease in the early 1990s. The screen saver on his home computer means he is doing something on behalf of his cousin — funny, easygoing Tony. "It makes me feel good," Mezias says. "I feel like I'm giving some meaning to the loss in my family." And some hope for a breakthrough in the future.

Readers can join Fast Company's team on the World Community Grid by clicking here.

Follow Chuck Salter and Fast Company on Twitter.