In 1994, Marion Tinsley, the world’s best living checkers player, took on Chinook, the world’s best unliving checkers player, in the Second Man versus Machine Championship. Tinsley withdrew due to health concerns, and the man who replaced him, Don Lafferty, split the twenty game series at one win a side (with eighteen draws). In 1995, Chinook beat Lafferty in a thirty-two game series (one win, thirty-one draws) to become the undisputed world checkers champion. But two hundred years earlier, a Hungarian inventor named Wolfgang von Kempelen, had created a mechanical chess player that could compete against the Europe’s strongest players. Von Kempelen’s chess-playing mannequin, carved to look like a seated Turk, with hands that moved the pieces, was one of the many mechanical marvels (link via Eclogues) produced in the eighteenth and nineteenth centuries: Vaucanson’s celebrated duck, which walked, ate, and shat pellets; Droz’s writer; Kempelen’s own bellows-powered speaking machine. Making automata to dazzle nobles was a relatively common endeavor for the mechanically inclined and socially ambitious; their marvels served as a way to introduce themselves to court. Later, smaller models began to be made for the haute bourgeoise, of which many examples survive today. (Cosma Shalizi has compiled a automata bibliography for those seeking more information.) No other automaton, with the possible exception of Vaucanson’s duck, was as celebrated as the Turk, however.
Baron Kempelen displayed the Turk to Emperor Joseph II and Grand Duke Paul of Russia, then took it for a successful tour across Europe. It would play — and win — exhibition endgames in front of crowds. Kempelen would select members of the audience, theatrically wind the Turk, and then stand back to let the Turk play until it gave its satisfied double nod that indicated victory; among the many players it defeated was Benjamin Franklin, then ambassador to France. When Kempelen returned to Austria, however, he packed it away and never displayed it again. It was left to Johann Nepomuk Maelzel, inventor of the panharmonicum and one-time friend of Beethoven’s, to bring the Turk to lasting fame. Kempelen, a legitimately skilled inventor (who had designed landscapes, built fountains, worked with hydraulics, and applied for a patent on the steam turbine in addiition to creating his speaking device) who had been ennobled not for amusing Maria Theresa but for military service to the Hapsburgs, seems to have been faintly embarassed by the chess-playing automaton, as it was, of course, a fake.
Maelzel had no such qualms; he bought the Turk from Kempelen’s family, shipped it to London, and began displaying it once again. In Maelzel’s hands, it made quite a stir: it defeated Napoleon in 1809; E.T.A. Hoffman wrote a story about the "talking Turk"; Maelzel brought it to the United States, where it toured up and down the East Coast, playing to large paying audiences. Newspapers gladly fueled the Turk’s reputation, speculating — just as Edgar Allen Poe and David Brewster did — about the true nature of the automaton. The truth of the matter was revealed (to very little notice) in June of 1827, when two boys reported to the Baltimore Gazette that they had seen European chess grandmaster master William Schlumberger hasten from his hiding place and out of the Turk’s back, gasping for air and suffering from heat exhaustion.
James Cook’s The Arts of Deception devotes a chapter to the Turk and argues, in effect, that the public took no notice not because they wanted to believe but because they wanted ambiguity. They wanted to wrestle with the problem of how the automaton worked — Was it under Maelzel’s control, perhaps through a system of magnets? Was there a midget inside? Did it genuinely play? — themselves. Cook examines this phenomena in the wake of the arrival of mass communication and mass entertainment, but people are still fascinated by game-playing machines. Not long after the development of the electronic computer, researchers started turning their attention on computer chess; forty years work has led to Deep Blue, a computer capable of beating the world champion.
Chess is a deep, difficult game, one associated with intelligent players (think of the villians of any number of bad James Bond ripoffs, or consider HAL’s chess style in 2001). One might expect chess-playing computers to throw the very idea of rationality into doubt. But it’s clear that abstract games — familiar ones like tic-tac-toe, checkers, chess, or go, and unfamiliar ones like Twixt, Octi, or Salta — are almost custom-made for computer play. There are a fixed, finite number of moves at every position; a computer can simply play the moves for both sides for several rounds, then weigh the ending positions to find which current play yields the best future result. Games like tic-tac-toe are trivial to solve — a chicken can play them. Games which provide many more possible moves per position than chess may not have skilled computer players for decades. Octi’s designer took some care to make the game difficult for computers, but the exponential progression of computing power means that some day almost all abstract games will have fallen. What then? Computers can beat Kasparov at chess, but have immense difficulties parsing language at a level a five-year-old can understand. Years of AI research have made it more clear that language is intensely difficult — there’s probably not a computer on earth that could handle Jeopardy, not because of the difficulty of the trivia but because of the difficulty in parsing the questions. Perhaps a game like bridge or poker, where so much is dependent on reading the expressions of the other players, will remain the domain of humans. And if that’s the case, some canny showman is probably building a box that will play exhibition matches, and give two self-satisfied nods as it lays down the ace of trumps.