Chess Article #32 Chess Computers and How They Work From: http://www.playe4.com/index.html Playing chess requires acknowledgement of the game rules and strategies, creativity, concentration, abstract thinking, visual memory and other human characteristics. So, how come chess computers often best human chess players, even chess grandmasters as Gary Kasparov? To solve this mystery, we must understand how chess computer work, or in other words, how a computer can imitate the decision-making procedure of a human player (and still step out with the upper hand). Chess Computer Tree Search Today's chess computers are rather complicated, and fast, very fast. To simplify their action, let's use the tree allegory. The tree represents the entire possible moves that can be executed by the computer; the tree's branches illustrate all the legal moves, and the leaves form the final positions. So, if the computer plays White, at the start of the game it has about 20 possible moves: moving either pawn one/two positions or moving any knight to two separate paths. Then, Black has to make one of the 20 possible responses to White's first move, multiplying with 20 we get 400 possible moves for Black, then 8,000 moves for white, etc. In short, you get quite a gigantic tree before you get to the middle game, and even the fastest and most advanced chess computers won't be playing in such conditions, not even the slowest, most novice human player. Computer Chess "Thinking" Instead, the chess computer generates the predictable board positions tree search. A fast computer can generate a ten-level tree of positions (something like 10 trillion positions), a faster one even a twenty-level tree and the fastest computer can manage millions of board position a second. To win at chess, the computer has to do more than just search for possible moves; it also has to evaluate them, i.e. to sieve the ultimate moves out of the seemingly endless forest possibilities. Lacking the human mind, the chess computers have an evaluation function, which can be as basic as chess pieces calculator to as extensive as a formula that measures and weights the value of the position while taking in consideration hundreds of influential factors. Either way, the computer is guided by a certain number that rates the value of the given board position. Known as the "minimax algorithm", on its turn, the computer plays the best possible move, responding to the opponent's worst possible move. Though the computer plays in complete darkness and despite the human brain's alleged superiority, results prove that the program efficiency in beating the intellectual counterparts. ---------- Copyright 2005 LogicEmpire! All rights reserved========== The blind-chess mailing list View list information and change your settings: //www.freelists.org/list/blind-chess List archives: //www.freelists.org/archives/blind-chess =========