Of Pawns and Programs

  • Share
  • Read Later
ILLUSTRATION FOR TIME BY JOHN CORBITT

At the top echelons of chess, jockeying for advantage before the game begins is a time-honored tradition. In the 1400s the Spanish chess theoretician Lucena recommended placing the board so that the light shines in the opponent's eyes. "Also," Lucena advised, "try to play your adversary when he has just eaten and drunk freely." For the past two months, supporters of Garry Kasparov and Vladimir Kramnik, the two best chess players in the world, have been engaged in a Lucena-style campaign to convince the chess community — and anyone else who will listen — that each is the right person to avenge Kasparov's infamous loss to an IBM supercomputer called Deep Blue five years ago. The culmination of these pregame maneuvers was to have come this week, with each grand master taking on a different computer chess program.

But in an eleventh-hour gambit that set the chess world buzzing, Kasparov's match was abruptly postponed two weeks ago; it won't take place before December. Kramnik's, meanwhile, is still scheduled to start on Oct. 4. Why the fuss? First, because chess is not just another board game; it is the pastime of kings and intellectuals and was long considered the ultimate test of human intelligence. Second, because Kasparov is not just another grand master; he was world chess champion for 15 years and is widely regarded as the best player who ever lived. So when Deep Blue defeated Kasparov in 1997 by a single game — a devastating 19-move blowout — it was viewed as a turning point in the relationship of man and machine. Three years later, Kasparov was dethroned by Kramnik, his onetime protégé. Today both men claim to be the world's best human player; a rematch between them will not take place until at least 2004.

And what about Deep Blue? It is not available for any kind of rematch. It never played again — at least not in public — and was apparently dismantled after the Kasparov face-off. Kramnik's supporters, meanwhile, proclaiming that "mankind has waited five long years for revenge," lined up another computer chess opponent, Deep Fritz from Hamburg, and the King of Bahrain put up a $1 million prize. Rising Middle East tensions after the events of 9/11 delayed the "Brains in Bahrain" competition until this week, giving Kasparov's backers time to arrange for a $1 million match of their own, against an Israeli program called Deep Junior.

In the prematch posturing, each player claimed he had the tougher opponent, yet that's difficult to judge. Deep Fritz won a 26-game match against Deep Junior that went into overtime, but it took place a year and a half ago, and both programs are much stronger today. Although Junior is the current world computer chess champion, Fritz did not play for the title because its creators wanted to keep it under wraps until the Kramnik match. Adding to the difficulty in comparing the programs is the reluctance of their developers to release specifications that they regard as trade secrets. But the programs do have discernibly different playing styles. "Deep Junior is like a computer version of me," says Kasparov. "It plays creatively. It takes chances to wrest the initiative. Deep Fritz is computer Kramnik. It plays methodically, without risk. Frankly, it's a little dull."

Dull or not, it's by no means clear that either program could have beaten Deep Blue. Blue was a chess-playing juggernaut that approached the game with the strategy known to computer scientists as brute force. Employing a design that included 64 custom-made chess-playing computer chips — one for each square of the board — it could examine 200 million moves per sec. Fritz and Junior, by contrast, can ponder only 2 million to 4 million moves per sec. But calculating possibilities is only one aspect of playing chess. Judging which player would wind up with the better board position is also important, and here Fritz and Junior presumably have an edge over Blue.

Whatever the outcome, the two Ks agree on one thing: the days when human players sit anywhere near the top of the chess pecking order are numbered. "I give us just a few years," says Kasparov. "The next challenge will be whether any person can play a long match against a computer and win even one game."