 |
 |
 |
 |
 |
 | ||||
US law takes aim at E-mail scourge... A product named by its sell-by dat... Take that, you cyber hooliga... Import duties falling through the Ne... Agony aunts and uncles do brisk business... Fore! Mad truckers and alien invaders ar... The truth is out there - on your P... You could be surfing the Web ... E-commerce eyes the consume... Will your E-mail make it past the mille... On-line security system doesn't cut i... Help youngsters become good cyber-citize... Plunging rand to push up PC prices... M-Net plugs its viewers into... PCs made simple for dummies in South Afr... Networking at home means happy families... The tiny powerhouse that changed the wor... |
The man within always wins
Two centuries before Deep Blue, another device was taking the chess world by storm, writes TOM STANDAGEA YEAR ago, IBM's Deep Blue supercomputer defeated Garry Kasparov at chess. It was the first time a computer had beaten the world champion in match conditions, and was heralded as a significant victory of machine over man. But it was not the first time a chess-playing machine had caused a sensation. More than 200 years ago, a mechanical chess player took the courts of Europe by storm. Known as the Turk, it provoked similar reactions to Deep Blue and was every bit as celebrated. It also has a valuable lesson for us today. The story begins in 1769 when Baron Wolfgang von Kempelen, engineer to the Imperial Court in Vienna, was so unimpressed by the "magnetic tricks" performed by a visiting conjuror that he boasted he could do better. The Empress Maria Theresa gave him six months to deliver on his promise. The baron duly unveiled a mechanical chess-playing mannequin dressed like a Turk and capable of beating the court's best players. Spectators were shown into a dark room, and the mannequin, seated behind a large cabinet with a chess board on top, was wheeled forward on castors. Doors on the front and back of the cabinet were then opened and the insides illuminated with a candle to reveal a complex set of gears and pulleys. The Turk would then be wound up and would play chess, moving pieces with its left arm, accompanied by the sound of grinding gears and whirring clockwork. It hardly ever lost a game, and could also solve chess problems. But Von Kempelen soon came to regret the success of his creation. He was more interested in hydraulics and voice synthesis than chess, and built a number of artificial voiceboxes, one of which was said to be able to articulate 30 words. Concerned that the Turk would overshadow his other achievements, he dismantled it, claiming it was broken. In 1783 he was forced to rebuild it on the orders of the Emperor Joseph II, so that it could go on a tour of Europe By this time theories had begun to circulate about how the Turk worked. One person who did his best to find out how the Turk worked was Napoleon Bonaparte, who played it three times in Vienna in 1809. By this time, the Turk had passed into the hands of Johann Maelzel, inventor of the metronome. Napoleon had a grossly inflated view of his abilities as a chess player because his courtiers were reluctant to beat him and, on the first encounter, the Turk easily defeated him in 19 moves. According to Maelzel, Napoleon placed a magnet on the chess board before the second game because he had heard that the Turk relied on magnets for its operation. But Maelzel removed it, and the Turk won. Before the third match, Napoleon wrapped a shawl around the Turk's head and torso, thinking there might be an operator hidden inside. But the Turk won a third time, at which point Napoleon swept the chess pieces to the floor and walked out. During the second decade of the 19th century, the question of mechanical reasoning was being seriously considered by the English mathematician Charles Babbage. Infuriated by the human errors that riddled logarithmic tables, Babbage hit upon the idea of building a mechanical calculating engine that would generate mathematical tables automatically, and began working on a proposal to attract funds to build it. Meanwhile, another mathematician, Robert Willis, heard about the Turk and determined to discover its secret. He collected all the accounts of the Turk he could find and attended several of its shows in London. He was then able to make accurate measurements of the Turk's dimensions using his umbrella, and found that the cabinet was much larger than it seemed - large enough to conceal a fully-grown operator. The clockwork, he suggested, was a decoy whose loud whirring served only to conceal any sound made by the operator. By this time, Babbage, however, had come to the opposite conclusion. He published his plans for a mechanical computer, the Difference Engine, at around the same time that Willis was denouncing the Turk. But while Babbage's proposals won him a medal from the British Astronomical Society, little attention was paid to Willis's pamphlet. The Turk continued to tour Europe until Prince Eugene died, whereupon his heirs fell out with Maelzel. He fled to the US in 1825, taking the Turk with him. To this day, the best-known exposé of the Turk is that written by Edgar Allan Poe in 1836. Actually, Poe lifted most of his article directly from Willis's pamphlet, but in the process he brought its carefully reasoned argument to a mass audience, and further fuelled the growth of the legend surrounding the Turk.
Back in England, the chessplaying automaton and Babbage's proposed mechanical calculating engines were explicitly compared in Natural Magic, a bestselling book published in 1832. By the mid-1830s Babbage had abandoned his Difference Engine in favour of the Analytic Engine, a far more complex design that could, he suggested, be programmed to play chess. He even suggested a crude chess-playing algorithm. The Analytic Engine, like the Difference Engine, was never constructed, but Babbage's work suggested that a mechanical chess player was not entirely out of the question. Maelzel died in 1838 and the Turk eventually ended up in a museum in Philadelphia, where it was destroyed in a fire in 1854. The Turk was, of course, an elaborate hoax, and its movements were controlled by a concealed operator. This was confirmed by Dr Mitchell, its last owner, who wrote an article about it for the Chess Monthly in 1857. Ironically, given Von Kempelen's desire to upstage the magnetic tricks of a conjuror, it relied on magnets in the chess pieces which set tiny, delicate indicators under each square swinging whenever a piece was picked up or put down. The operator, an expert chess-player hidden in a concealed compartment in the cabinet, followed the action on another chessboard inside the machine, and carried out the Turk's moves using an ingenious mechanical arm that was itself a masterpiece of engineering. Although it was a fake, there is no doubt that the Turk got people thinking during its extraordinary 85-year career. Are machines capable of reason? Does being able to play chess count as intelligence? Today, the debate about the definition of machine intelligence is still very much alive. But the yardstick for machine intelligence is now the Turing test, which measures a computer's ability to hold a convincing conversation. By modern standards, neither the Turk nor Deep Blue could be termed an intelligent machine: the Turk because it was a fraud, and Deep Blue because its ability stems not from experience or an innate understanding of chess but from a combination of brute computational force and software algorithms from human chess experts. "I never considered Deep Blue to be intelligent in any way," says Murray Campbell of IBM's Watson ReSearch centre and a member of the team that built Deep Blue. "It's just an excellent problem solver in this very specific domain. It would have been nice if Deep Blue had been able to teach itself to play, and learn from its own games, but it was painstakingly programmed every step of the way." Deep Blue's victory wasn't a triumph of machine over man but the triumph of one kind of human ingenuity over another. The Turk had a human inside it, and in effect, Deep Blue did, too. - © The Telegraph, London
|