Kasparov Wins

Monday, Feb. 19, 1996
  • Share
  • Read Later
PHILADELPHIA: World chess grand master Garry Kasparov won his match against IBM's Deep Blue computer. Kasparov had started the match slowly, losing the first game in just 34 moves. With the six-game match tied at 2-2 heading in Friday's game, Kasparov roared back, winning the next two games. The match has if nothing else served as a demonstration that brute computer power can at least equal the best that humans have to offer, says TIME's William Dowell. "Kasparov personally evaluated Deep Blue's performance as ranging somewhere between 2300 and 3000 depending on circumstances. Kasparov's own chess rating is 2750. The best computer before Kasparov was somewhere around 2300." Although Kasparov -- who spent much of the week grimacing and holding his head in frustration as he sat across the board from some stone-faced IBM scientist taking instructions from the computer -- worked hard to win, Dowell notes that this was really a collaborate effort: "After each game, Kasparov goes through the game in reverse, explaining to the IBM team where he or the computer had gone wrong. This has surprised some people -- why would he give away secrets -- but Kasparov himself uses a computer chess program as a tool to help improve his game, and he should be able to learn from working with the IBM scientists." Grand master Yasser Seirawan, ranked as one of the top 30 players in the world, concurred. "We international grand masters will take these games to our home laboratories and analyze them to a T. The best grandmasters will learn a great deal from this contest."
The Real Game"Chung-Jen Tan, who heads IBM's parallel processing research department as well as the Deep Blue team, sees chess as merely a convenient, off-the-shelf benchmark to test cool technology, an effective proof for a tool that will soon be assigned to managing other facets of our lives," Dowell notes. The most immediate application? A process for extracting meaning from a flood of apparently chaotic information known as 'data mining.' "For example," says Dowell, "if you have a sudden run on beds and axes in Minnesota stores, and can't figure out why, data mining might tell you that there has been a sudden forestry boom and that lumber jacks are now setting up new camps. Applications could range from air traffic control to playing the stock market, or taking chess as a model, even to the next war."