How did it get there?

Saturday, February 26, 2005

Chess vs. Go

Adam Cadre, host of the Lyttle Lytton contest, has this to say on chess and go:

"I prefer go to chess for two reasons. One is that go feels less arbitrary. Instead of having six different types of pieces, each moving in a different manner and arranged in a fairly arbitrary fashion, go presents the player with a bowl full of identical stones and a very simple set of rules. Players alternate placing stones on the intersections of a grid. Stones or groups of contiguous stones entirely surrounded by stones of the opposite color are removed from the board. Board positions cannot be repeated. When both players pass, points for which all paths along the grid reach only stones of one color or the edge of the board count as territory for the player using that color. That's pretty much it. And from these simple principles comes a game much deeper than chess: computers can now beat the world's best chess players, but the best go programs play on the level of a low-intermediate amateur. So that's one reason. The other is that while chess is a game about destroying things -- pieces are steadily removed, and at the end of the game not much remains -- go is about building: even a player who's lost quite badly will carve out at least a small territory."

I concede that go has a minimalist appeal, but it should be noted that from a mathematical perspective, chess is actually a simpler game. It has a wider variety of pieces, but a far smaller total number, and the chessboard has only 64 squares, in comparison with go's 361 grid intersections.

Second, while it's true that go starts with a blank board and you take turns putting pieces on it, it still takes the form of a conflict, a struggle for control, and the "destructive" nature of chess means it has a decisive ending. I can't help but feel that "checkmate" is more aesthetically satisfying than "time to count up how many squares we control".

Third, the computer thing. Computers ARE worse at go than chess. This is due to two factors. One, go has a board over five times as large, so there's hundreds of possibilities for each move. Two, it's played with a large number of low-value pieces, making it significantly harder to measure who's winning, and consequently making the decision tree harder to prune. If you take 36 chessboards and stitch them together into one big giant game of superchess, humans will be better at it than computers again. Being "deeper" in this sense has essentially nothing to do with whether a game is good.