This article is printed with the permission of Raymond Keene, OBE
Ever since Baron Wolfgang von Kempelen, Aulic Counsellor to the Empress Maria Theresa, constructed his chess-playing automaton the Turk in 18th-century Vienna, scientists have dreamt of building a machine that could defeat the greatest human chess minds. As Goethe noted, chess is the touchstone of the intellect, and when Napoleon sat down to challenge the Turk at chess, his dismissive behaviour at the board indicated that he believed his intellect to be superior to any machine. In fact it turned out that one of the hidden compartments concealed a very human chess expert.
It was, however, not until 1996 that genuine chess-playing computer programs have been able to compete successfully against the world’s best human players. The first such match was the six-game encounter between IBM’s Deep Blue and Garry Kasparov in Philadelphia in 1996, which Kasparov won. In 1997 an eagerly anticipated rematch was contested in New York. The world champion got off to a good start, but Deep Blue struck back and went on to win sensationally by 3½-2½.
Part of the thrill of any competitive activity is uncertainty of the outcome and the relishing of the infinity of possibilities in play, which have rendered many situations virtually impervious to definitive solution or accurate prediction. In this respect, mental sports fall into three categories. In the first, opponents fight each other with the objective of winning, scoring points, achieving ratings and earning titles. Chess, draughts, bridge and Go all fall into this category. In such games, objective truth is subservient to the overriding imperative of winning.
Alternatively, competitors battle against themselves, seeking to overcome their own limitations while maximising their potential to claw ever improving results out of a fixed set of standards and norms. Memory championships are one example. Finally, there are those activities where the challenge is to reveal what is already known by someone else. This category includes crossword- solving, IQ tests and quizzes.
IBM’s Deep Blue and now the monstrous Houdini, freely available on the internet, demonstrate that chess and draughts can, to a significant extent, be analysed to a correct conclusion. The experience of playing chess in the second decade of the 21st century is beginning to resemble an examination with right and wrong answers.
We now know that Kasparov was wildly inaccurate in his accusations that human players had helped Deep Blue, and also that today even the greatest players stand no chance against computers. Nevertheless, the best human players are learning from the programs, while überchampions, Anand, Carlsen and Kramnik have become remarkably adept at replicating computer decisions in their own games. It is conceivable that human champions may eventually become so good at copying computers that they will begin once again to defeat the silicon brains.
That indeed would be a triumph for the human brain, a brain that, although it cannot as yet match the computer for unerring accuracy, can at its best still challenge and defeat up to 30 flesh and blood opponents at one time without being able to see any of the boards or pieces. This is the feat of memory and calculation that grandmaster Timur Gareev achieved in St Louis last month. Goethe would have felt vindicated.
Raymond Keene, chess writer for The Times