Kasparov draws vs. X3D Fritz
I don't know why, but I love these man vs. computer chess matches. They are a perfect, concise metaphor for the conflict between man and machine (Matrix and Terminator movies notwithstanding). The most recent heavyweight bout in this sport was between Gary Kasparov, the chess master known among other things for losing to IBM's Deep Blue, and the reigning computer champ, X3D Fritz.
Why 3D? Because you can play against Fritz wearing these goggles that project a 3-dimensional chessboard that processes moves using voice recognition. Incredibly geeky and silly and unnecessary, and thus immensely appealing to a dork like myself. I myself am a lousy novice chess player, and I know it because when I watch Fritz and Kasparov go at it, I try to guess at possible next moves and am usually wrong.
However, I do appreciate the challenges inherent in this desperate struggle to prove oneself the master of chess. The basic problem is well-stated here. An average chess game involves about 40 moves, or 80 half moves, and each half-move offers on average about 40 options. In a typical game, the number of potential positions (arrangements of the board) is somewhere on the order of 10 to the 128th power (10^128), or more than the number of atoms in the entire universe (10 ^80)! Therefore, no computer can ever win simply by sifting through all possible positions. Even if it could, that would be one slow chess match.
But the goal is simply to defeat a human, and since the human mind is imperfect, the computer need only surpass humans in their ability to look ahead and process various positions. It was not as easy as one would think. Looking ahead just 3 moves into the future offers 4.1 billion possible combinations, and early computers couldn't process that quickly enough. Chess masters are able to look ahead 4 to 4.5 moves. But since then, a variety of advances (notably alpha-beta) and Moore's Law) have made it possible for computers to play at an amazingly high level. They are not quite yet at the level of the top chess playing humans in the world, who can look ahead 7 moves, but humans make mistakes under stress which can reduce their effectiveness.
Humans have not stood still, though. Knowing how computers are programmed, they've devised all sorts of means to neutralize a computer chess program's strengths. One strategy is to quickly exchange queens, depriving the computer of the strongest attacking piece. Another, used by Kasparov to crush X3D Fritz in game three, is to build a row of pawns. In one match, Boris Alterman built an entire solid horizontal wall of pawns against Deep Fritz. This overloads the computer with all sorts of positions because it cannot attack until it has broken through that wall, and breaking through requires a series of moves.
Meanwhile, chess programs having begun working on simulating and storing all possible outcomes of six-piece endings, to the point where if a match gets down to a six piece or fewer endgame, the computer will know immediately whether it can win, draw, or lose and how to achieve any possible victory. In such scenarios, the chess program becomes the perfect player. Humans don't want to go there.
Kasparov ended up with a 2-2 draw in his four game match against X3D Fritz. Kramnik drew with Deep Fritz 4-4 in what was billed as Man vs. Machine 2. So, for now, man and machine stand on opposite sides of the board, bruised and battered, but still standing.
Why 3D? Because you can play against Fritz wearing these goggles that project a 3-dimensional chessboard that processes moves using voice recognition. Incredibly geeky and silly and unnecessary, and thus immensely appealing to a dork like myself. I myself am a lousy novice chess player, and I know it because when I watch Fritz and Kasparov go at it, I try to guess at possible next moves and am usually wrong.
However, I do appreciate the challenges inherent in this desperate struggle to prove oneself the master of chess. The basic problem is well-stated here. An average chess game involves about 40 moves, or 80 half moves, and each half-move offers on average about 40 options. In a typical game, the number of potential positions (arrangements of the board) is somewhere on the order of 10 to the 128th power (10^128), or more than the number of atoms in the entire universe (10 ^80)! Therefore, no computer can ever win simply by sifting through all possible positions. Even if it could, that would be one slow chess match.
But the goal is simply to defeat a human, and since the human mind is imperfect, the computer need only surpass humans in their ability to look ahead and process various positions. It was not as easy as one would think. Looking ahead just 3 moves into the future offers 4.1 billion possible combinations, and early computers couldn't process that quickly enough. Chess masters are able to look ahead 4 to 4.5 moves. But since then, a variety of advances (notably alpha-beta) and Moore's Law) have made it possible for computers to play at an amazingly high level. They are not quite yet at the level of the top chess playing humans in the world, who can look ahead 7 moves, but humans make mistakes under stress which can reduce their effectiveness.
Humans have not stood still, though. Knowing how computers are programmed, they've devised all sorts of means to neutralize a computer chess program's strengths. One strategy is to quickly exchange queens, depriving the computer of the strongest attacking piece. Another, used by Kasparov to crush X3D Fritz in game three, is to build a row of pawns. In one match, Boris Alterman built an entire solid horizontal wall of pawns against Deep Fritz. This overloads the computer with all sorts of positions because it cannot attack until it has broken through that wall, and breaking through requires a series of moves.
Meanwhile, chess programs having begun working on simulating and storing all possible outcomes of six-piece endings, to the point where if a match gets down to a six piece or fewer endgame, the computer will know immediately whether it can win, draw, or lose and how to achieve any possible victory. In such scenarios, the chess program becomes the perfect player. Humans don't want to go there.
Kasparov ended up with a 2-2 draw in his four game match against X3D Fritz. Kramnik drew with Deep Fritz 4-4 in what was billed as Man vs. Machine 2. So, for now, man and machine stand on opposite sides of the board, bruised and battered, but still standing.