Categories: A to Z
The question of the longest possible game is more interesting since it is not yet provable whether the Game must terminate (that is, whether a win is inevitable from the rules of the Game), although Game Theory shows that as the game is finite and played with perfect information it must have either a win for one or other player or a pair of winning strategies which must lead to a draw. Informal results suggest that the game will always terminate with a win by one player but a rigorous proof is liable to require several years of computing time. Simpson's results put a probable upper bound of 276,550 moves on any game but this is a less than helpful finding. The Game's fluid structure and huge field mean that even decision trees to decide on longest games by computer are ferociously complex and tend to swamp systems fairly fast. Nonetheless, within the limits of recent research it has been shown that openings at Euston, Camden Town and Shepherd's Bush provide games of 500+ moves each. The first two are obvious, of course, due to Halmeyer's Indeterminacy Theorem but the Shepherd's Bush finding was a surprise. In retrospect, it does seem Shepherd's Bush openings do generally lead to complex games won by exploiting errors.
See also: Shortest Game.