Before long, computers and networks will begin to take on a life of their own as humans essentially become collaborators in a vast worldwide information technology system. So says Intel Corp.'s director of research, David Tennenhouse, who has worked on some of the seminal developments that are transforming the technology landscape.
In this vision, which Tennenhouse sketched at a recent conference at MIT, computers mostly talk to other computers as the line between computer and network begins to blur. Computer users will have to relinquish some direct control in exchange for more-effective, higher-level problem solving.
Tennenhouse sees this development as an organic outgrowth of past trends in the evolution of the computer. Initially, computers were batch processors, spitting out results after a problem and its data were fed in. The next stage in the evolution, in the 1960s, delivered a whole new user paradigm-interactive computing-which set the agenda for technology development for the next 40 years.
During that evolution from mainframe to minicomputer to desktop PC, users increasingly interacted in real-time with the machine, feeding in data, processing it and revising it as a result.
The Internet has begun to mutate that interactive paradigm by letting computers interact as humans and computers do, Tennenhouse said. But for computer-to-computer communications to become efficient, he said, the computer user's role must change. "Everyone is interacting with the network, because that is the paradigm, but how many computers do you really want to interact with? Maybe just one. You certainly don't want to end up trying to direct traffic," he said.
The logical arc would be a shift from an interactive to a proactive paradigm in which the computer and network can anticipate a user's needs, rather than just respond to commands. The first step is to get computers to direct data transfers over the network. "This is a huge business-8.5 billion to 10 billion embedded processors are added per year," Tennenhouse said.
A responsive, anticipatory Internet would be fundamentally different from anything seen so far, said Tennenhouse, a former director of the Information Technology Office at the Defense Advanced Research Projects Agency.
His daring vision would require revolutionary developments in hardware, network architecture and machine-learning software.
