Paul Baran was laughing when he told me about AT&T.
This was before I joined the board of Telebit, his company, a seat I would hold because he later asked me to, which sounds like a credential and was really a gift. At the time we were simply meeting. Before he said a word about the idea the company was built on, he wanted me to know what had happened the first time he tried to explain packet switching to the people who owned the telephone network.
It had happened years earlier, in the early days. They heard him out. They decided it would never amount to anything. He told me the story the way you tell a story about an old friend’s worst haircut, with his whole face, delighted. What amused him was not that they were wrong. It was how complete their confidence was, the particular serenity of people who have never once had to imagine the thing that is about to replace them. Amusement, I would come to understand, was his entire theory of the people who said no.
Then he explained the modem.
The idea Telebit was built on seemed so obvious to him that he could not understand why everyone else hadn’t had it first: a modem should not have a speed. It should listen to the line it had been handed and use whatever that line had to give.
A telephone line, he said, is a hostile place. It hisses and drifts and changes its mind between one call and the next. The usual instinct was to design for the worst line you could imagine and live forever with the speed that gave you. Paul’s instinct was to ask the line. What are you today? How much can you carry right now? And then to carry exactly that, and to ask again a moment later, because the answer might have changed.
Somewhere in that explanation, it landed on me that I was sitting across from a modern-day Edison. Not because of any one invention. Because his mind worked differently. We say, all of us, that human beings cannot feel an exponential. We nod at the curve on the slide and then go right on planning as though tomorrow will be a slightly larger today. Paul could feel it. He lived inside it. And a man who can feel an exponential cannot believe in fixed things, because fixed things are only the shape of the present, and the present is the one thing guaranteed not to last. The world he saw was adaptive all the way down.
Later we worked together on Com21, a cable modem startup, the same idea carried onto a different wire. The line was different. The question was the same.
Paul died in 2011. He missed the first SpaceX booster that came back down out of the sky and set itself upright on a concrete pad. He missed the Starlink satellites, thousands of them now, passing packets to one another overhead, every link adjusting itself to the weather and the geometry of the moment, a network with no center and no single point of failure, the very shape he drew at RAND in the early sixties. He missed the large language models, too, and the long argument over whether the curve behind them will keep bending upward, which is really just the old argument about whether people can bring themselves to believe in an exponential.
I do not know what he would have built in this new era of space and AI. I am not sure that is even the right question. Somewhere right now there is a conference room full of intelligent, serious people explaining to one another why all of this will not amount to anything. In all the years I knew Paul, I never heard him say an unkind word about anyone, and I do not imagine he would have started with them. I picture him instead at the back of that room, leaning against the wall, arms folded, listening closely, and smiling.