Network: First in the academy of wizards

Postel had the vision to see the value of computers talking to each other long before the birth of the PC, Web pages and e-mail

PERHAPS ONE of the greatest testimonials to the life of Jon Postel was the very way I came to know him. I never had the honour of meeting Postel while he was alive. Tragically, he died recently at the age of 55, after undergoing emergency heart surgery at St John's Hospital in Santa Monica, California.

Certainly, I had read about Jon - most notably in the book Where Wizards Stay Up Late: The Origins of the Internet, by Katie Hafner and Matthew Lyon. Postel crops up 14 times in this eight-chapter tale of the founders and the founding of the Internet, alongside pioneers such as Vint Cerf and Larry Roberts.

It can be said that Postel invented the IP addressing scheme and the Domain Name System that underlie the whole of the Internet. As director of the IANA, or Internet Assigned Numbers Authority, Postel oversaw the incredible growth of the Internet, from an original four computers to tens of millions today.

And it was his invention that allowed me to come to know him, in a manner of speaking, shortly after his death. A few minutes after I spotted the assignment in my e-mail to write his obituary for The Independent, I was on a search engine, looking for references to "Jon Postel".

The Net pioneer Vint Cerf had called Postel "our rock, the foundation on which our every Web search and e-mail was built", and sure enough, there was I, searching and e-mailing away for information on the very system that he had built.

The Internet Society's memorial page carried many touching accounts. They described a quiet, unassuming man who was nevertheless a brilliant technologist and passionate advocate. As the Internet grew, many intense, knotty disputes were resolved once Postel had brought his quiet authority to bear on the problem.

Indeed, Postel single-handedly rerouted all of the Internet's domain name servers earlier this year, as an experiment to prove the robustness of the network. It is unlikely that any other human on the planet could have accomplished such a thing, or been allowed to do it.

Postel had, after all, hooked up the switch at the University of California, Los Angeles, in September of 1969, that allowed the first computers on the Arpanet to talk to each other. He was not just a passive observer at the birth of the Internet; he made it happen.

Postel and his peers had the vision to see the value of computers talking to each other long before the birth of the personal computer, Web pages and e-mail. In fact, many of his colleagues didn't see any practical reason to connect computers. After all, they reasoned, if you already have a computer, why do you need someone else's?

But Postel knew that there was always something on another computer - a program, a file, perhaps - that other people would want or find useful. And Postel, who was a harried graduate student in those days, didn't want to hook things together in just any old way.

He and his peers were busy trying to work with the era's exotic mainframes, huge, proprietary beasts that had dissimilar and incompatible operating systems and programming languages. In those days, it took a mighty effort to make a computer do things that we take for granted today, such as word processing. Programs were written by hand by students, and hardware was unique, touchy and expensive.

Bugs were everywhere and failures were common. Postel and others realised that a large network would need a huge bureaucracy constantly to reroute traffic around failures. The phone companies of the world employed hundreds of thousands of people just to keep up the relatively simple voice network.

But early computer pioneers were too busy trying to finish degrees, teaching, and working in part-time jobs to have a lot of time to spend keeping networks going. So they conceived a system, called a distributed network, that would make connecting computers relatively easy to do, and easy to maintain.

They foresaw a system of simple rules that each computer on the network would obey, for forwarding messages one to another, bucket-brigade style. They saw that if they broke up messages into small parcels, or packets, the packets could find their own way from sender to receiver to be reassembled into the original message. "Checksum" data could be included, so that the receiving computer could verify that all had arrived safely and well, or ask the sender to resend any missing or scrambled packets.

Absolutely central to this scheme was an addressing system that would uniquely identify every computer on the system. An equally important component would be a system for allowing humans to type addresses without error.

It was to this challenge that Postel rose, editing some 2,500 RFCs (Requests For Comments), technical white papers that are the foundation of the Internet. That he did his work well can be seen in the growth of the Internet. He worked hard until shortly before his death to see his plan for future Internet growth adopted.

Without this quiet genius, I would not have been able to read this memorial, penned by Vint Cerf, which appeared on my Mac only minutes after I began to search. "Jon has been our North Star for decades, burning brightly and constantly, providing comfort and a sense of security while all else changed," wrote Cerf, current chairman of the board of the Internet Society.

"He was Internet's Boswell and its technical conscience. His loss will be sorely felt, not only for his expertise, but because the community has lost a dear and much loved friend."

cg@gulker.com