Computers of the world, unite! Analysis/Commentary Part of the strength of any network comes from the number of nodes it can support. Like a chain, the better distributed the weight, the stronger the chain. Several private organizations and companies have begun exploring the concept known as "distributed computing," or "community computing." The theory is simple. A powered-up computer sitting idly on someone's desk still has a capacity for work. The CPU, or central processor, remains turned on and idling. Distributed computing harnesses idle CPU cycles to solve intricate data problems for medical research, mathematics, astronomy, cypher decryption, and e-business. Participants in distributed computing communities download a software program that runs in the background, and solves problems any time a computer is not actively running major applications. The software parcels out a small portion of a collective problem -- such as sifting through SETI's SERENDIP findings for "intelligent" radio noise from outer space. (seti.planetary.org) It runs on its own desktop location and doesn't interfere with a user's personal data or the ability to run programs. If he or she decides to use all the CPU's resources, the software simply shuts down. It also works when a user has disconnected from the Net. Enthusiasts say this kind of computing beats renting high-performance servers or Cray supercomputers. Already, Envive, an e-business management company, has partnered with Entropia, a distributed computing network, to poll Web usage and performance for its customers. Non-profit organizations, like Distributed.net, perform tasks that harness up to 2 million participants. Skeptics say distributed computing lacks commercial potential because of the current physical limitations of the Net. They also ask, as time passes and bandwidth gets cheaper and cheaper, why bother jerryrigging second-rate boxes? Other critics, Internet programming purists, suggest that distributed computing misses the point of the Net, which aims at device independence -- independence from clients as well as servers. In effect, they prefer to distribute coding, not bandwidth, but this seems unrealistic today. Distributed computing obviously works. Now it needs to show how it can make money. November 22, 2000