Author: Wendy M. Grossman (www.pelicancrossing.net)
Publication: The Daily Telegraph
Date: 12/24/1998
Title: Internet pioneer sets sights on new worlds to conquer


Vinton Cerf was ten years old when he became an avid science fiction fan, and you could say that he's been living in his own sf novel ever since. The man most often given the sobriquet of "father of the Internet," Cerf is turning his attention outwards - to the planets. In conjunction with Jet Propulsion Labs (JPL) and several other prominent researchers, Cerf is working on a project to invent the protocols that will be needed for an interplanetary Internet. It is, he says, the fact that TCP/IP has taken 20 years to get from drawing board protocol to world-wide mass medium that tells him it's time to think about what will be needed 20 years from now. As an aside, he comments that attempts to revamp the system by which computers are assigned human-friendly names are thinking too small: what about .earth, .moon, and .mars?

Mars is getting most of the immediate attention. This year's Pathfinder mission is due to be followed up by efforts to map Mars's surface, with a launch in 2001 and more at 26-month intervals thereafter. Other missions will visit Saturn in 2004 and Pluto in 2010, and Cerf is committed to looking ahead at the technology that will be needed, not just for those missions but for manned space stations. Part of his preparation has been returning to some of his favourite Mars authors: Ben Bova, Ray Bradbury, Kim Stanley Robinson, and Robert Heinlein.

"I enjoy the escape," he says, adding that long-term "I think it's also made it easier for me to think about things that weren't quite ready yet but I could imagine might just possibly be feasible."

Cerf's "father of the Internet" tag came from the stint at the Defense Advanced Research Projects Agency from 1976 to 1982 when he and colleague Robert Kahn developed TCP/IP, the protocols which enable all types of computers everywhere to link together and form the Internet. He moved on to MCI, where in the early 1980s he built the first commercial email system to be connected to the Internet, and then served as vice president of the Corporation for National Research Initiatives. In 1994 he rejoined MCI, this year merged into MCI WorldCom, where he is senior vice president of Internet architecture and engineering. He is also chairman of the Internet Society and winner of many awards including the US National Medal of Technology and the Marconi Fellowship; he also serves as an active member of Ireland's Telecommunications Advisory Committee. All this, even though he grew up reading science fiction in an era when it was not only frowned on by parents but looked at askance by other kids.

Cerf remembers it was 1986 before anyone made any money out of the Internet, and then it was the router manufacturers. "Up until then, the only way to build a router was to find a computer and wrap a graduate student around it." 

In 1988, he realised the Net couldn't get any bigger unless it became commercial and self-sustaining, so as a one-year experiment they connected MCIMail in 1989. "As soon as we announced it, everyone else selling commercial email said they had to be connected, and after that no one disconnected," he says. 

Cerf's survey is a reminder that the commercial Internet is still less than four years old; that's why everyone's business models are so confused. "In a gold rush, people make money not by looking for gold but by selling picks and shovels to other people looking for gold," he says. Which may explain how he ended up working for one of those outfits the Net loathes: a phone company. "A lot of us are working for the phone companies," he says, "because a lot of Internet companies are being bought by telcos. In order to continue to grow the Internet, it has to be part of a system that actually has its own capacity." Part of his job is projecting the demands of Net growth (he predicts 300 million users by 2000) and new technologies, from Internet radio and video to high-speed connections and wireless networks. There are, he says, some wild ideas out there, including a scheme to use blimps to provide connectivity. Interplanetary links sound almost sensible.

Creating the technology required for the interplanetary Internet won't be a simple business. TCP/IP itself is unsuitable: the protocol was designed to be robust, but it was not designed to handle the long delays inherent in the vast distances of space. Even at the speed of light, information would take as long as 24 minutes to get from Earth to Mars when they are furthest apart, a length of time that makes our complaints about the "World-Wide Wait" seem kind of, er, terrestrial. Compressing data and transmitting it quickly is, however, something of a NASA specialty, so one concept that's being considered is interplanetary gateways that would hook the terrestrial Internet to the deep-space networks that develop, much as CompuServe and AOL have different (proprietary) systems that require gateways now. Other techniques under discussion include protocols for minimising the amount of data that has to be sent, perhaps by relaying only changes to previously transmitted data instead of the whole pile all over again.


##This is the article as submitted to the Telegraph, since the published version vanished from its site in a content reorganization. Anyone quoting from it should check the official published version, which may have been altered in the production process.##