Content area
Full Text
The lead in the New York Times front page story was dramatic: "In brisk and brutal fashion, the IBM computer Deep Blue unseated humanity...." That was Monday, May 12, of this year. Two weeks later, the defeated Garry Kasparov published a letter in Time magazine. The banner headline declared, "IBM Owes Mankind a Rematch."
Mankind had been defeated at one of its most complicated games. Kasparov's frustration at losing brought to mind his warning back in 1989 when he beat a younger Deep Blue: "I don't know how we can exist knowing that there exists something mentally stronger than us." Now it had happened-and where did that leave us?
Not as losers, certainly. Not if you total all the results. The beneficiaries of Deep Blue's research and development are widespread including businesses such as Charles Schwab, Eastman Kodak, ShopKo, Lloyds of London; companies doing environmental modeling (Pacific Northwest Laboratories); even the Department of Energy, which is seeking a computer-generated model of the radioactive decay in our nuclear weapon systems rather than risk human data gatherers. And these are just a few of the winners.
Deep Blue is not a short-term project created by IBM for the sole purpose of capturing pawns and public attention. It is an evolving experiment in computerized problem solving.
The game begins. How did it start, and why was a game chosen to be an IQ test for our computers? To answer these questions, we need a closer look at the game of chess and IBM's RS/6000 SP.
Fewer than 10 years ago, credible experts claimed that a computer could never be programmed to defeat a current world chess champion. One reason for this assumption is that chess can be very complicated. For example, there are only 32 pieces on a board that has 64 squares. The field looks small until you calculate the total number of possible moves in a game. That number is 25 x 10^sup 115^-a substantial figure when you consider there are only an estimated 10^sup 79^ electrons in the universe.
Game theory, a serious branch of math and computer research, defines chess as a "finite, chance-free, two-person, zero-sum game with perfect information." (Oxford Companion to Chess) "Perfect information" implies that it is theoretically possible to look...