Baran’s design began, like so many innovations, with an idea that hadn’t worked. The Pentagon, he’d thought, might broadcast thousands of coded messages over AM radio frequencies all at once as an attack approached. Missile-silo commanders and bomber commanders would cluster by their transistor radios, collecting their “launch” codes with the ease of listening to a late-night baseball game. That target-shaped, “just aim here” web of phone lines would be replaced by something far harder to wipe out with a single R-7 Semyorka missile shot. But this approach had its own problems. It relied on broadcast towers and on insecure AM radio waves. Yet the idea of such a widespread, untargetable network got Baran thinking. Sending out the messages and letting them find their own way had a lot of appeal, if it could be done. There would be no central hubs. Information would sail over linked lines in the same way radio signals moved in the air. Military communications, in Baran’s system, would bounce from point to point on this tapestry, at each stop being redirected toward their intended destination. The resulting network, if you drew it out, would look like a fishnet: lots of links connected to a few knotted nodes. And because the bundles of data—Baran called them packets—could be moved by the network itself, you could cut or nuke or sabotage the net in a few places and still use it. The packets would find another path. Even a badly ripped-up and irradiated network could, in theory, carry a launch—or a recall—message safely from the White House to a bomber pilot.
“The early simulations,” Baran recalled, “showed that after the hypothetical network was 50% instantly destroyed, the surviving pieces of the network reconstituted themselves within a half a second.” In other words, his messages were finding new routes on the network, even after huge parts of the system had been taken off-line. And they were doing it nearly instantly. Better still, as he began to model these “distributed” networks, Baran discovered that they were not only capable of surviving attack, they were also incredibly efficient. “If built and maintained at a cost of $60 million (1964 dollars),” he calculated, his design would “handle the long-distance telecommunications within the Department of Defense that was costing the taxpayer about $2 billion a year.”
Baran traveled the country for most of 1961 and 1962, classified presentation and slide rules in hand, trying to persuade skeptical generals and engineers. It was a nearly impossible task. He recalled a visit to the towering AT&T switching headquarters on Broadway in lower Manhattan, an implacable temple of the high priests of hub-and-spoke network design. That one building handled more telephone and telex traffic than nearly any other single point on earth. No doubt the place ranked very high on Moscow’s first-strike list. So Baran expected a friendly reception. After all, he’d be telling a bunch of men with a uranium death sentence that he’d found a way to get them off the Soviet target plan. His new “mesh” network would mean that bombing AT&T would be largely pointless. It wouldn’t blind U.S. commanders. If only they’d redesign their network, the AT&T engineers might save their own lives.
They thought he was insane.
“I tried to explain packet switching to a senior telephone company executive. In midsentence he interrupted me,” Baran recalled. “The old analog engineer looked stunned. He looked at his colleagues in the room while his eyeballs rolled up, sending a signal of his utter disbelief. He paused for a while, and then said, ‘Son, here’s how a telephone works.’” Of course Paul Baran knew how a telephone worked. You jacked one point to a switch to another point. That was the problem. This was why AT&T’s design would prove useless in the face of the catastrophe he’d been told to prevent. Baran was, nerve and blood and bone, as an analyst, and even as a refugee, perhaps, alive with the imperative of survivability, of how connection might mean the difference between war and peace. Those calm, slide-ruling engineers in the AT&T building? What could they be worrying about?
But it wasn’t just that $2 billion annual check from the U.S. Defense Department that those wizened phone wizards were seeing vanish in Baran’s fishnet; it was a whole way of thinking. The AT&T scientists wanted to control the addresses, the routes, the timing, of messages from the center. This authoritarian design appeared more efficient to them; perhaps it was even more psychologically comfortable, since it matched their own experience of being commanded and controlled. Karl Wittfogel, the historian who identified the water totalitarians of ancient China and Egypt, would have recognized them: Switch Despots. “It was a conceptual impasse,” Baran reflected. He moved to the next stop. Same result. And the next. Same result. Eventually Baran’s engineering colleagues back at RAND were so affronted by the routine dismissal of his logic that they spoke up. They had seen the classified briefings. They knew just how easily the nation could be hobbled—and their Santa Monica building was surely on some target list too. RAND’s scientists demanded a detailed, critical study of the distributed network model. By the time they were finished, the air force was preparing to begin construction.
Survivability. Plucked from the impossible-looking puzzle of how to communicate during a nuclear war emerged the first honestly distributed network. Other scientists had been chasing the idea as well, but the design suited Baran’s problem particularly well: a network with no central control, survivable, uncuttable. The earliest large network built on Baran’s principles became known as ARPANET, the Advanced Research Projects Agency Network—a mesh of connections that even today serves as the backbone for parts of the Internet.