|Photo by Anne Fishbein|
Im holding a doctoral dissertation titled Message Delay In Communications Nets with Storage. The original, not a photocopy, and even though Ive just been handed it by its author, UCLA Professor of Computer Science Dr. Leonard Kleinrock, I still feel as if I should be wearing white cotton archivists gloves. Because what Im looking at is no less than one of the blueprints for the Internet.
Kleinrocks modest office at his Internet-gateway company, Nomadix, in Santa Monica, is filled with various plaques, paperweights, lucite blocks and something on a brass stand that looks like an upended suspension bridge, all of which pay tribute to his status as one of the fathers of the Internet. That thing on the brass stand is no less than the Marconi International Fellowship Award, one of the highest honors in telecommunications, presented by the prince of Belgium. In degree of prestige, the Marconi is runner-up to the L.M. Ericsson Prize, the Nobel Prize of telecommunications, presented by the king of Sweden. Kleinrock has one of those, too.
Yet what makes him light up is the Dodgers baseball bat inscribed with his name. After more than three decades in L.A., New York is still in Kleinrocks voice. And one of the more fun aspects of having the City of Angels realize that it had one of the founders of the Internet in its back yard was Kleinrocks tribute at a backslapping ritual in Dodger Stadium.
Kleinrocks work is so fundamental, many of his explanations are followed by the phrase but they werent calling it that yet. Take that 1962 dissertation for example. It introduces the idea of packet-switching, the underlying technology of the Internet. The term itself would not be coined until 1966 by Donald Davies, a physicist at the British National Physical Laboratory.
Kleinrock came up with his breakthrough because of a phenomenon that is painfully familiar to graduate students everywhere: The good problems had already been solved, and the ones that were left were hard and mostly unimportant. While searching for his own topic at MIT, he recalls, I was surrounded by computers, and I said, One day, these machines have to talk to each other. Looking more closely, he observed how his fellow students were working. I saw that when people sat down at their keyboards, they spent most of their time scratching their heads.
Scratching your head is fine when computers are plentiful and cheap, but not back in the day when computing resources were precious. Within living memory, although most who went through the experience would rather forget it, computers were enormous, expensive machines shared by many users. Since one job would not take up all the resources of the computer, a system was created that ran many jobs simultaneously, with the computer deciding which one had priority. The idea was called time-sharing, and it inspired Kleinrock to reconsider how resources were allocated across networks. In time-sharing, you chop things into little bits, and the short messages get through more quickly, the short jobs get through more quickly. Thats what you want with packet-switching, explains Kleinrock.
With packet-switching, Kleinrock applied the resource-sharing concept behind time-sharing to communications. Each message is broken into packets. Each packet has a header that indicates where it is in the message, rather like the disassembled London Bridge each piece of the bridge was numbered, so that it could be put back in the proper configuration when it arrived in Lake Havasu. Most importantly, because of the headers it doesnt matter when or in what order the packets are sent they will all be reassembled correctly on the receiving end, so you do not need to have a continuously open connection, as with circuit-switching, the technology behind traditional telephone networks.
Granted, this is not the sort of stuff that makes laypeople hyperventilate. But at least the engineering community got it several years later. I published it, and nobody cared. Nobody cared. The people who in particular did not care was AT&T. Id say to the telephone guys, You dont understand, you want to charge me for a three-minute call, it takes me 25 seconds to dial up the call, and I want to send a millisecond of data. And their answer was, Little boy, go away. The reason was I didnt represent a source of revenue, and they were absolutely right data wasnt a source of revenue, because there was no data around, there were no data networks to speak of. So in that sense they were right, but in the big sense they were dead wrong.
Kleinrocks work would eventually be recognized by Larry Roberts, then the head of computer and communications research at ARPA, the Defense Departments Advanced Research Projects Agency, formed by President Eisenhower using Sputnik as the excuse for the interdepartmental science agency he had long advocated. Simply out of a desire to solve the problem of how to more efficiently use the resources of highly expensive computers, Kleinrock and the other ARPA researchers created the ARPANET precursor to todays Internet.