I D É E S   F O R T E S    Issue 2.07 - July 1996

Ghosts in the Machine

By Katie Hafner



The Net of the 1970s and '80s - a cozy place for elite computer scientists to schmooze, carry on and carry out their research - has long since been supplanted by something at once more sophisticated and more unwieldy. Yet in dozens of ways, today's Net still reflects the personalities and proclivities of those who built it.

The creators of the original Net, which traces its roots back more than 25 years, weren't just scientists. They were inventors. To make wide- scale computer networking possible, they invented packet switching. Once the concept of this was in place, they came up with an algorithm to perform dynamic routing, and they figured out a way to deal with congestion on the network.

In the field of communication theory in the '60s, this was pretty daring stuff. Back then, the notion of digitising a message, breaking it into discrete packets, sending them every which way through a network and then reassembling the information at the other end, was revolutionary. AT&T executives, too entrenched in the traditional methods of telephony to tolerate new ideas, were convinced packet switching wouldn't work. And IBM believed no computer existed that was small enough (read: cheap enough) to be used as a network switch. But the Net pioneers, young men with strong opinions and nothing to lose, paid no attention and built their network anyway.

So it's not surprising that the Net of today remains imbued with the personality quirks of its inventors. Frank Heart, the BBN engineer in charge of the team that built the interface message processor, the progenitor of today's network routers, was a pragmatist through and through. And that attitude toward technical invention - build it, throw it out on the Net and fix it if it breaks - has permeated Net sensibility ever since. Pragmatism also prevailed in the network's topology. For years, the Net's resident topologist, Howard Frank, was someone with a lot of experience in mapping out natural gas pipelines. A computer network, he figured, shouldn't be so different. And he was right.

Then there was the cabal of graduate students at UCLA, where the first IMP was installed in 1969. In 1968, Steve Crocker, a UCLA grad student of extraordinary courteousness, started a series of notes for the Network Working Group. Crocker named the first note "Request for Comments" because he wasn't sure who was really in charge of Net protocols and didn't want to offend anyone by sounding too official. Now we're on RFC 1936, and RFCs have become the Net's de facto bylaws.

The Net grew by building on existing technology. Bob Metcalfe got his ideas for Ethernet in 1972 while spending the night on Steve Crocker's sofa. Looking for a little bedtime reading, he picked up a paper on Alohanet, an experiment in Hawaii that became the technical underpinnings for Ethernet. Bob Kahn, one of the original IMP guys at BBN, went on in 1973 to co-invent TCP/IP, the suite of protocols that grew from the original host protocol. Kahn relentlessly championed openness in the process of protocol design, and it's that very openness that helped the Internet and TCP/IP pre-vail over other networking technologies simply by making itself available to anyone who wanted to implement it. The shock of commercialisation notwithstanding, openness and ad-hocracy live on in the Net of the '90s.

Vint Cerf, who followed his best friend Steve Crocker to UCLA, is now the éminence grise of the Net. It was Cerf who gave the Net its civility. More than anyone, he cajoled and argued and humoured the rest of the world into adopting TCP/IP, which he and Bob Kahn had sketched out during an all-nighter at the Cabana Hyatt in Palo Alto.

Few of the Net's creators have got really rich off their invention, and they probably never will. They're like the pioneers in tennis, who popularised the game while playing for pocket change - and paved the way for a new generation to make millions. But it's the creators of the Net, now in their late 50s, who still run the Internet Society, and they still go to meetings of the Internet Engineering Task Force. When the Kahns and Cerfs and Crockers finally retire, the Net will miss them. It has done well by them.

Katie Hafner (katieh@zilker.net) is the co-author of Where Wizards Stay up Late: The Origins of the Internet, to be published in the US by Simon & Schuster in August.