In the turning point of man versus machine, the 1997 version of Deep Blue -- a chess playing computer designed at IBM's Thomas J. Watson Research Center -- defeated human world chess champion Garry Kasparov by 3.5 games to 2.5. The power behind IBM Deep Blue is an IBM RS/6000 SP parallel supercomputer equiped with chess specific coprocessors. The Deep Blue system is capable of examining 200 million moves per second or 50 billion positions in the three minutes nominally allotted for a single move in a chess game.

The match was a watershed in the history of computing, and, to a certain extent, in the history of technology. Chess has always been considered a grand challenge for computer science because it combines beautiful logic with artistry attained through experience. But until the early 1990s, computers were not really up to the task of challenging a human chess grandmaster such as Kasparov. Why was Deep Blue able to defeat the Russian champion? Because Deep Blue used more than brute computing force. It combined the power of its processors and a highly refined evaluation function that captured human grandmaster chess knowledge -- including Kasparov's. It was able to produce "brilliant" moves, including the famous moment in Game Two when it unexpectedly offered an exchange of pawns instead of simply advancing its queen to an apparently overwhelming position. This move jarred Kasparov, who later described it as brilliantly subtle. For its creators and many of its fans, Deep Blue had, for a moment, used its incredible processing power and the accumulated knowledge of computer scientists and chess champions to engage in something resembling "thought."

The researchers who created Deep Blue are not resting with this victory. While the historic chess duel with Kasparov was the culmination of the first phase of their work, it was not the only reason for developing the computer or studying chess. The IBM researchers are taking what they learned from building Deep Blue -- and from lessons learned during the match -- to tackle other complex problems. Two new areas being investigated by the Deep Blue group are Computational Finance and Data Mining where they hope to utilize the technologies developed in Deep Blue. Many problems in these areas require large amounts of search and evaluation, and the techniques used in Deep Blue (advanced search algorithms, large-scale parallelization, application specific hardware) may be useful in these domains. Currently the group is looking at applying certain data mining techniques to medical databases, and studying ways to improve Monte Carlo simulation in financial applications.

Kasparov vs. Deep Blue: the rematch.