While you're asleep, your home computer signs onto the Internet and looks for work. By morning, it's done complex calculations for a scientist in London, designed an ad for an Arizona business--and fattened up your bank account.
This scenario--called metacomputing--could become a reality in the very near future, says Baruch Awerbuch, a professor of computer science at The Johns Hopkins University. He is studying the economics of sharing computer power over the Internet, including pricing and marketing issues. There is money to be earned, Awerbuch says, by tapping the enormous power that lies dormant while computer owners are asleep, at lunch or merely away from their terminals. At such times, these computers could earn extra cash by doing remote work for researchers or business owners in other locations. People who need extra computing horsepower or special equipment for a one-time project will line up to lease time on these idle machines, Awerbuch believes, because renting someone else's computer costs far less than buying a new one.
"It makes huge economic sense," Awerbuch says. "For example, if once in a lifetime you want to use a fancy program that only runs on a particular piece of hardware, why do you need to buy this piece of hardware? Instead, you could lease a cycle on that machine in exchange for some economic favor. The opportunities are endless."
Most of the technology to share hardware over telephone lines or some other network already exists. What's needed, the researcher says, is a system for buying and selling cycles of time on idle machines, along with safeguards to make sure "renters" can't pry into an owner's files. It will also require a change in the close attachment most people feel toward their computers.
"People are accustomed to using only the machines that they've purchased," Awerbuch says. "They're not used to using machines that are not their own. Yet it seems silly to invest lots of money to buy more computing power when all you have to do is utilize what's already out there. Think of it this way: 90 percent of the computing power in the world is inaccessible to people, simply because we haven't set up the right mechanism to help one another."
Here's how that mechanism might work, Awerbuch says: An accountant owns a PC. It crunches numbers perfectly but can't handle the elaborate graphic design work needed to create a splashy business brochure. The accountant posts a note on the Internet, offering 50 cents a minute any time that week for remote use of a Macintosh computer with graphics software. If the brochure is needed within the next hour or two, the accountant might offer $5 a minute for immediate access. A Macintosh owner, or an unattended machine that searches the Internet for work whenever it is idle, responds. The accountant takes temporary control over the Mac terminal, which might be located anywhere in the world. When the project is completed, the accountant issues an electronic payment for the time.
This same system could allow a hospital to lease time on powerful computers elsewhere for demanding jobs such a medical imaging. Other companies could break complex tasks into small pieces, then farm them to small outside computers. Hardware owners could also post notices on the Internet, describing their machines, hours of availability, memory capacity and software.
People who are reluctant to loan their lawn mower to a neighbor may be even less inclined to let perfect strangers use their computers by remote control. But Awerbuch believes these transactions can be sound and profitable.
"This sharing should not be viewed as an altruistic thing, nor as some kind of a Communist idea, where you don't own anything personally and you must share with other people," he says. "To the contrary, this is purely an entrepreneurial arrangement, where you trade what you own in exchange for cash or some other economic benefit. The policies and approaches will be totally up to the individuals or companies or whoever is doing it. It could be quite interesting to see this electronic marketplace in action."
This research, by Awerbuch and co-principal investigator Yair Amir, assistant professor of computer science at Hopkins, is funded entirely by a $1 million three-year federal grant from the Advanced Research Project Agency, Technology Management Office.