In October, man and machine will again compete at chess. This time, says Raymond Keene, the playing field will be level
IN Arthur C. Clarke’s 2001, the astronaut faces HAL, the chess-playing computer. Push each letter of HAL forward by one and you’ll soon see where the name HAL originates. In October of this year — 2001 itself — the new world chess champion, Vladimir Kramnik, will face a challenge from Deep Fritz, the world’s best chess-playing computer program. The contest will take place in Bahrain, with a prize fund of $1 million provided by the Emir’s government. No such clash between the supreme human and silicon brains has taken place for four years. Then, in 1997, Garry Kasparov lost narrowly to IBM’s Deep Blue in New York, in a match which set a record for Internet fascination. Fans of the game itself, followers of the development of artificial intelligence and those who were just plain riveted by the psychological drama of the occasion, accounted for no fewer than 22 million website hits in the final hour of the final game. This made the 12 million hits for the entire three weeks of the previous year’s Atlanta Olympics look utterly puny. Those who want to follow the games this year can do so on the website www.braingames.net.
Unsurprisingly for those who know him, Kasparov, crowned world champion in 1985, and undisputed king of world chess for the subsequent 15 years, did not react charitably towards his defeat. He became convinced that IBM had cheated. They had done so, according to the Kasparov version, by engaging a team of top-class grandmasters, including his old foe Anatoly Karpov, concealing them in a nearby hotel and beaming their collective tactics and strategies direct to the machine.
Indeed, cheating and illusion go back a long way in chess, certainly to mediaeval hucksters who showed punters trick positions which were capable of more than one interpretation. Perhaps this was one reason why the Church periodically took an anti-chess stance. That did not prevent the chess-loving 16th-century Spanish priest Ruy Lopez, after whom one of Kasparov’s favourite openings is named, advising that one should place the board so that the sunlight shines in your opponent’s eyes.
Illusion and delusion in chess surface again in, of all times and places, the rational 18th-century Vienna of the Enlightenment. Here Baron Wolfgang von Kempelen defeated allcomers with his chess-playing machine, the mechanical Turk. There is even a record of a game won against the Emperor Napoleon Bonaparte by this device at Schonbrunn Palace. However, yon Kempelen’s victories owed less to artificial intelligence than to ingenious mechanical technology. The pieces on the board were manipulated by a complicated series of levers and pulleys orchestrated by a hidden chess-expert homunculus.
In the past century the most notorious example of sharp practice was the importation of an alleged Soviet parapsychologist, one Dr Vladimir Zukhar, into the politically charged atmosphere of the 1978 world title contest in Baguio, summer capital of the Philippine archipelago. Anatoly Karpov, the reigning world champion, was the golden boy of the Soviet establishment and a particular favourite of Leonid Brezhnev. Facing him was the Soviet defector Victor Korchnoi, known as the `Leningrad Lip’, famed for his superstition and paranoia as much as for his superlative chess skills. Once the Soviets had realised that Korchnoi thought that their man Zukhar was beaming mind-bending rays at the defector during the games, Karpov was handed an extra weapon in the psywar, winning the marathon match by a single point.
Computers have further complicated the task of the honest chess player. When Kasparov recently espoused the notion of so-called ‘Advanced Chess’ (human-plus-computer faces human-plus-computer in a curious symbiotic match), some cynic announced that this had been going on for ages already and was known as correspondence chess. There is, indeed, no way to police chess from being aided and abetted by computer analysis when it is played by mail, but public tournaments played openly are a different matter. A few years back a young man, going by the name of John von Neumann, whom nobody had ever previously identified as a promising chess player, suddenly leapt into the lead in a strong tournament. He then stunned the onlookers at his crucial game by announcing a complicated checkmate in 27 moves (or some similarly ludicrous number) and officials smelt a mouse. Sure enough, John von Neumann was wired up to a strong computer program which was dictating his moves to him. He was promptly disqualified.
Even chess played on the Web can now be checked for computer patterns in the moves. However, the most pressing question is how to eliminate cheating from the big money match, where a million dollars are at stake. With the flesh-and-blood world champion already facing the inbuilt disadvantage of being pitted against a machine that may see up to 200 million positions per second, it is imperative that he is not handicapped by other factors.
Lessons have been learnt from the Kasparov experience of 1997, where the dice, so to speak, were definitely loaded against the human. To be sure, Kasparov did not actually fear that his arch-enemy and Korchnoi’s nemesis, Anatoly Karpov, was hunched, Turk-like, in a small box under the computer on the stage. Nevertheless, IBM did score cleverly and, I emphasise, legitimately, given the rules in force at that time, on the following points:
IBM’s team changed the program after every game, so that Kasparov was always facing a new opponent; they did not show a single game by the current version of Deep Blue to Kasparov before the match, even though IBM had access to thousands of Kasparov’s games and could prepare accordingly; but the clincher was that adjournments were never permitted. As a consequence, any long game could leave Kasparov exhausted, while the machine happily and without fatigue carried on number-crunching. For the forthcoming clash between man and machine, all these points will be adjusted to secure maximum fairness for both contestants.
IBM executives agreed after their victory against Kasparov that the company had reaped at least $100 million worth of favourable publicity. In Bahrain, though, the playing-field will be entirely level, and it will be fascinating to see what Fritz’s four additional years of computer science can achieve against a human brain that is equally well prepared for the shoot-out. IBM was, of course, invited to enter the qualifying tournament to decide which computer should challenge Kramnik. Doubtless for excellent reasons, the invitation was declined.
Raymond Keene, chess grandmaster and chess correspondent of The Spectator, has been invited to be match director for the Kramnik v. Computer match in October.