METCALFE HAS A LAW OF computing named after him. Metcalfe's Law states, basically, that the asset value of a computer network increases exponentially as each new node -- or individual user -- is added to it. Because every new user brings along a wealth of linkages and resources, the fund of networked value grows to be far richer than the mere sum of its parts. Metcalfe's Law is at the heart of the Internet's power -- it's the place where (despite occasional courageous, and mostly unavailing, efforts to charge a fee for it) information really wants to be free. And it is the most inspiring model of a decentralized communications network that is likely ever to exist outside of insect pheromone messages in nature.
Metcalfe first became acquainted with the Internet when it was a cozy little loop connecting relatively small numbers of government and academic research outposts. Its users were a closed, purposeful community that invented, out of necessity, the early protocols for online information sharing. Now, he believes, with the community burgeoning toward the billions of users, we are on the brink of having the network connections themselves become despecialized. Already, pagers and cell phones and Palm devices are connected to the Internet.
"PCs are passé," he says, getting at one of the canonical beliefs of the informational new age: intelligent, networked everything -- from refrigerators to blenders to garage-door openers to contact lenses.
At the recent high-tech industry event known as DEMO (a gig where venture capitalists come -- checkbooks open -- to make deals for promising technologies), one of the more revolutionary concepts unveiled was "chips that some geniuses have designed that allow you to run Ethernet on the power line in your house. This is a real step toward ubiquity, being able to put the Internett wherever anything gets plugged in."
But even with all of these great leaps forward so tangibly at hand, Metcalfe sees fundamental problems still waiting to be solved. One thing he would like to tackle next -- whenever next turns out to be -- involves software design. "I have a pipe dream of returning to research," he confesses. "There has been very little software progress in the last 25 years. All the progress we have enjoyed has been hardware progress."
He longs to delve into something he calls "anticiparallelism" -- the ability of software to anticipate what the user will want to do next, and then to do it in the background, invisibly, concurrent with, or parallel to, whatever task is visible to the user.