`WHERE WIZARDS STAY UP LATE' NODS OFF TOO EARLY.

Byline: Richard Bernstein The New York Times

Title: ``Where Wizards Stay Up Late: The Origins of the Internet''

Author: Katie Hafner and Matthew Lyon

Data: Illustrated, 304 pages, Simon & Schuster; $24.

Our rating: Four Stars

If you always wanted to know who put the ``at'' sign in your e-mail address, then ``Where Wizards Stay Up Late'' is the book for you.

Katie Hafner and Matthew Lyon have produced a clear, readable history of the Internet, the sprawling electronic mall where you can look for a spouse or read an article from a newspaper across the country or dial up the World Wide Web site of the CIA to see what declassified world maps are available for sale.

The history of the Internet ought to be a great story, and indeed, in Hafner's and Lyon's hands it often is. The authors, a husband-and-wife team, make good use of their material, showing how the Internet's inventors overcame the considerable obstacles that had made it impossible for computers in different places to talk to one another.

The first half of their book is a fascinating narrative, evoking especially the world of early computer scientists - ``an adhocracy of intensely creative, sleep-deprived, idiosyncratic, well-meaning computer geniuses'' - whose energy and vision were utterly amazing.

The problem with ``Where Wizards Stay Up Late'' is that the tale itself turns out, somewhat surprisingly, not to have the legs it needs for so comprehensive and lengthy a treatment.

Certain stories promise more than they deliver, often because the real action takes place inside the heads of people, so the narrative lacks the drama and external conflict that would make it exciting to read.

After describing the background and the early breakthroughs that led to the Internet, Hafner's and Lyon's tale slows down. By the end, it may be only full-blooded devotees of computer history who are still willing to read late into the night.

Hafner, a contributing editor at Newsweek, and Lyon, an assistant to the president of the University of Texas at Austin, do not put it this way, but the origins of the Internet seem to follow the law of unintended consequences.

The Net originally was an effort by the Advanced Research Projects Agency (ARPA) at the Pentagon to find a way for users of computers in one place to communicate with computers in other places.

The initiative, under President Dwight Eisenhower, was one of many by the United States after the Russians put Sputnik into orbit in 1957, and gave rise to the feeling, half truth and half panic, that suddenly the Soviet Union was ahead in science.

In other words, the Internet was not developed to let ordinary citizens exchange messages, buy products, pursue hobbies, look at pictures of nudes or promote scholarly interests.

But once the technology was developed under Defense Department contracts, the ruse of history took over. Indeed, as Hafner and Lyon put it, the original ARPA network ``was a growing web of links and nodes, and that was it - like a highway system without cars.''

This book makes it clear that the creators of the Net were mostly uninterested in Defense Department concerns. The heart of the account lies in the driven personalities of a couple of dozen computer pioneers who are the wizards of the title.

An engineer named Ray Tomlinson hit upon the ``at'' sign for e-mail addresses.

Will Crowther, a true giant of modern technology, was seeking to amuse his daughter after his divorce when he created a game called ``Adventure,'' a simplified version of Dungeons and Dragons that ``demonstrated the appeal of an open networking culture.''

Paul Baran, ``a good-humored immigrant from Eastern Europe,'' devised a revolutionary concept now called packet switching.

In the late 1960s, Baran was trying to make U.S. military communications invulnerable in the event of a Soviet nuclear attack. His revolutionary idea was to break messages into fragments, to send them via a network of redundant lines, and to use a code to reassemble them at their destination.

Later, when Bolt Beranek & Newman, a consulting firm in Cambridge, Mass., got a contract from ARPA to build the first computer links (quickly called the Arpanet), Baran's idea, never used by the military, became the basic idea behind computer-to-computer communication.

It is not easy to make the technological challenges comprehensible to nonwizards, and Hafner and Lyon skillfully use metaphors to make concepts clear. For example, they describe a special control signal called RFNM, devised to keep the links from clogging up, as ``the equivalent of the maitre d' announcing `Your table is ready.' ''

At other times, they slip into computer jargon. While theirs is light compared with the heavier forms, it is nonetheless stilted and unhelpful.

There are moments when the narrative enters a mist of RFCs, TIPs, IPTOs, FTPs, and TCP/IPs, along with ARPAs and Darpas and Elizas, added to unexplained terms like ``logging capabilities'' and ``natural-language programs,'' though, fortunately, those moments are rare.

And certainly, even when the story slows down, Hafner and Lyon have done us a service by rescuing from oblivion the collection of geeks and nerds, bureaucrats and geniuses, who changed everyday life for millions of people all across the planet.