Draw in Bahrain

Science  01 Nov 2002:
Vol. 298, Issue 5595, pp. 959b
DOI: 10.1126/science.298.5595.959b

Where the board stood just before Kramnik (white) made daring but futile knight sacrifice.


Computers don't get tired, crack under deadlines, or get psyched out, which makes them formidable opponents on the chessboard. Earlier this month, world champion Vladimir Kramnik and chess computer Deep Fritz drew the final game of their eight-game series, and their “Brains in Bahrain” match was declared a tie. But the draw was a substantial victory for the creators of Deep Fritz, who proved that computers can match the abilities of even the best human chess players.

Deep Fritz is very much like other advanced chess computers: It has a vast database of openings and endings, and once the game is afoot, it relies on brute force—with some fine-tunings—to go through all the possible moves and decide on the best one. IBM's Deep Blue used a similar algorithm to defeat then-champion Garry Kasparov in 1997, but many chess fans argued that the match was unfair because Kasparov had never seen Deep Blue play—something that never happens at the Grand Master level. Kramnik was therefore allowed to practice playing with the computer for a few months before the match.

In Bahrain, Deep Fritz lost two of the first four games, but it put Kramnik under constant tactical pressure and wore the champion down. “We could see it in his face. It was rather ghastly,” says Frederic Friedel of Hamburg, Germany-based ChessBase, who is part of the Fritz team. Kramnik lost game 5 through a blunder, and game 6 slipped away despite a brilliant and daring knight sacrifice that would have overwhelmed almost any human opponent. But an exhausted Kramnik quickly accepted draws in the last two games and expressed awe of the computer's abilities. Said Kramnik at a press conference after the match, “It plays like a very strong human. These are ‘human moves.’”

Both sides are looking for a rematch. But although the man's brain will remain the same, the computer promises to get faster, enabling it to play an ever deeper game.

Navigate This Article