But Norman still had the computer bug, and in 1998 he decided to start over, building up a huge library of some 1000 items. Now he was back in New York City, with his collection divided up into 255 lots. The auction had attracted considerable international publicity, especially among the digerati, but Norman knew the potential bidders that were filtering in to take their seats were going to be a tough crowd.

It was a chicken-and-egg situation: rare-book dealers and collectors didn't usually buy computer-related material, because without a market for it already in existence, there was no way to gauge its value. Norman was hoping to break the logjam.

"The book business in general is one of 'follow the leader,' " Norman says. "Collectors and dealers are always chasing the same titles. The only books on computing you normally see are books by John Napier that were published in the 17th century. That's how backward the market is," explains Norman. The technical nature of the collection was Norman's biggest problem. How could he get collectors who weren't computer engineers to understand, say, the value of Claude E. Shannon's 1937 master's thesis, which proved that electronic circuits could be used to do logical and mathematical operations, thereby laying the cornerstone of electronic computing?

Norman's answer was to write and self-publish, with his business associate Diana Hook, Origins of Cyberspace, a book that cataloged his library and explained each item's significance. "Since most other collectors are following the leader, I thought to write a bibliography and create a framework for other people," says Norman. Released in 2002 at $500 per copy, the book wasn't exactly Book-of-the-Month Club material, but Norman hoped it might guide other collectors, and he decided it was time to start trying to sell the collection and get a return on the money spent accumulating it. (Norman wouldn't say how much he had spent purchasing items for the collection.)

Now, on this cold, bright morning, it was the moment of truth. Bidders were given white paddles with three-digit numbers printed on them. Bidders who could not attend in person or who desired maximum privacy were connected to the auction through a squadron of sharply dressed Christie's employees glued to banks of phones along the walls.

Despite Norman's carefully planned strategy, the bidders in the room seemed a little nonplussed by the collection. The first two items failed to sell. Finally, a 17th-century book about a mechanical calculator went for $4800, about twice the valuation in the auction catalog (prices quoted include Christie's 20 percent surcharge). Other items from the 17th and 18th centuries quickly sold, but the 19th century threw a wrench into the works, with the debut publication of the designer of the first real computer, Charles Babbage, left unsold. Then another work by Babbage fetched $38 400—three times the catalog estimate.