The pressure was great. The world was watching. The Olympic veteran exuded confidence for a repeat of past success.
Then the results came in _ or rather didn't _ and the crowd went wild.
As befits the cyber-saturated mid-'90s, one of the biggest problems at the Olympics so far has been a computer system.
And now IBM, which designed and is running the system, is trying to recover from a blow to its image on a grand stage where it has triumphed before.
The company has tried since Saturday to fix a complex system designed to give the news media results from each Olympic event seconds after the event finishes. For some contests, results are fine, but, for many others, the system produces errors or doesn't deliver.
IBM has started faxing results to the big news organizations, which, in turn, have resorted to the time-honored but tedious method of typing them into their own computers. But the Atlanta Committee for the Olympic Games late Tuesday appeared ready to abandon the system.
"The feed for the world news agencies continues to be very discouraging," said ACOG spokesman Bob Brennan. "Efforts to bring it online have not been successful. . . . We're not going to rely on it for the balance of these games."
Six other major IBM systems are performing without a hitch at the Olympics. But the glitches in the seventh have been amplified because of whom it serves.
Created for 13 large news organizations, including the Associated Press, Reuters and United Press International, the trouble with the IBM system has spilled over to thousands of newspapers and broadcasters who rely on those agencies to pass results to them.
The trouble has caught the attention of IBM's senior management in New York, who are updated routinely on progress toward resolution.
"This is the largest sporting event in the history of the world, and we are responsible for the information technology and we take that responsibility very seriously," IBM spokesman Fred McNeese said.
The trouble has left a deep enough scar that the European Broadcast Union filed a formal complaint to Olympics organizers. They and other news organizations may seek refunds for their connections, which cost thousands of dollars.
IBM has not had similar trouble at previous Olympics. The company first began to sponsor the Games in 1960 and has provided computer systems since the 1984 games in Los Angeles.
IBM paid $80-million to be a sponsor and lead "technology integrator" of the Atlanta games, hoping to impress customers with new ideas and skills. Advance promotional materials promised "bulletproof reliability."
Most of its innovations for the Olympics have been successful, including a ticket-selling method that yielded tens of thousands of sales over the World Wide Web. And spectators have had no trouble seeing results on the scoreboards at the events, which are posted straight from the judges who use IBM ThinkPad mobile PCs or the Swatch timing devices that connect through other IBM machines.
Those achievements have diminished the sting of IBM's trouble, said Jed Pearsall, president of Performance Research, a Newport, R.I., company performing market research at the Atlanta games on behalf of several Olympic sponsors, though not IBM.
"My advice for them is there is very little to be concerned about," Pearsall said. "What the consumer sees is being delivered. It would be a much more serious problem if the daily interaction of the people attending the events were affected, such as with ticketing."
IBM had tested the scoring system at Olympic trials and events like the U.S. Open tennis tournament. "We're having no problems at the individual venues because that's the most tested," McNeese said. Unfortunately, he added, "You can't create the entire Atlanta Olympics to test for problems."
In addition to results and commentary systems, IBM created ticketing, worker and volunteer accreditation, incident tracking and a "back office" system for the ACOG.
IBM watcher Bob Djurdjevic, president of Annex Research in Phoenix, said he was surprised the company tried so many new things at the Olympics when so many people would be scrutinizing it.
"Untried technology did what untried technology does, which is fail occasionally," Djurdjevic said. "They have certainly magnified the effect."