Network: Father to the Ethernet

Talent, drive and luck helped Bob Metcalfe to invent the world's most popular networking technology. By Wendy Grossman
Seven-year-old Max has a problem: he wants to go to the computer store as father promised to get him some networking pieces for his computer, but his father won't stop talking to this boring woman who keeps asking questions over breakfast.

Max is patient about this, and draws pictures with a multi-coloured pen, but he wants to know when they can go. "Nine minutes," his father tells him. There's no clock in sight, but Max accepts it. Max's father is Bob Metcalfe, who the night before won an Electronic Frontier Foundation Pioneer award for inventing Ethernet, the most popular networking technology in the world. If your computer at work is connected to other computers in the office, chances are the link is by Ethernet.

Metcalfe also founded 3Com in 1979, turning it into a $400m-a-year company (it's since grown to $1.7bn) before he left in 1990. Unlike some other company founders, he saw the writing on the wall and left amicably before the board could request his resignation. In his award acceptance speech in Boston, he predicted, Cassandra-like, that the Internet would begin to collapse over the next year.

Every successful person in the technology business seems to have drawn on three factors: one, the intelligence and talent to see an opportunity and make the most of it; two, the drive to carry out the plan, and three, the luck to have been in the right place to have the opportunity. "My main luck," Metcalfe says, "was being at Xerox PARC in 1972."

Xerox's Palo Alto Research Center (PARC) was where some of the industry's best and brightest researchers congregated in the early Seventies. The first personal computer graphical interface was developed there for Apple to borrow from; the beginnings of PostScript were invented there and fed into the founding of Adobe. Metcalfe built the world's first personal computer network there.

His career in networking started even earlier while he was leading a confused existence both studying for his PhD at Harvard, and working on Project Mac (now the Laboratory of Computer Science) in the wildly different culture of the Massachusetts Institute of Technology. He got the job at PARC in 1972, then went to defend his PhD thesis back at Harvard - and failed it. In the US academic world, this is not supposed to happen. "Never. Except that one time."

The reason, he says, was, "Failing to have either myself or my thesis adviser careful enough to be sure I didn't fail. And being arrogant for as long I can remember, I didn't see this coming."

The thesis was resubmitted and accepted a year later, and recently republished under the title Packet Communication. It was about the Arpanet, which was the precursor of the Net and which he was already working on, and the Aloha network, a packet radio network in Hawaii. His thesis adviser, by the way, was Marvin Minsky, one of the founders of artificial intelligence.

"Xerox said, come anyway, so I did." At the time, PARC was working on personal computers. "I was the networking guy. And I took the ideas from the Aloha network and transformed them into a thing called the Ethernet, on 22 May, 1973. Approximately."

At the time there were already other networking technologies, but Metcalfe says Ethernet was successful through "a combination of good technology and because it was promoted as a standard." That year, 1973, was the same year the TCP/IP structure that became the backbone of the Net was invented.

It was, as Metcalfe says, never designed to take the load it's now being asked to carry. Hence his prediction of collapse, a word he uses carefully. He doesn't mean complete system meltdown. Instead, he means the equivalent of electrical brownouts and blackouts - intermittent small outages. He figures there are up to 10 ways it could collapse: "I'm just betting one of them's going to happen, because it seems inevitable." He lists things such as overloading, sabotage (whether by hackers or by Net service providers fighting over market share), the advent of video and audio, the bugs in router software, which he is confident are there even if they are unrecognised so far, and the absence of what he calls a "messaging system" between supply and demand that would solve the overloading problem.

This last, he believes, could be tackled with a settlement system, the kind of arrangement that telephone companies have to pay for carrying each other's traffic. It flies in the face of the Net's current "I'll carry your traffic, you carry mine" ethos, but Metcalfe believes this structure is "naive".

After he left 3Com, he spent a year at Cambridge University's Computer Lab. "Loved it," he says. And that's when he decided to become a journalist - he could afford to write 2,000- word articles and get paid $350. He spent two years as publisher of the American magazine Info World, and still writes a column for it. Now he makes his living as a journalist - or rather as a guru who writes - and his thoughts are taken very seriously indeed.

But it's been a lot longer than nine minutes, and Max still wants his computer pieces. "He needs to download," Metcalfe explains, as though it were a phase that all kids go through.