Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Evaluation Should Be Winning Probability - Not Pawns

Author: Daniel Clausen

Date: 06:38:01 01/16/02

Go up one level in this thread


Hi

[a bit OT and only loosly related to the topic being discuss ;)]

I remember a friend of mine, who wrote a Connect4 game, where he basically
played random moves from the current position and chose the move which leaded to
the most number of won games. (ie 600 won games out of 1000 random games for
move A meant a winning probability of 60%) It worked surprisingly well. By
playing not completely randon games but discard the obvious nonsense moves (ie
stopping the opponent to get four in a row if you can't win yourself
immediately) it worked even better.

In chess this doesn't work at all because most random games in chess are just
draws and/or endless lines, whereas in Connect4 each game is over after 42 moves
the most. I've tried this idea with some trivial endings in chess like KRK and
it also didn't work there. Too bad. :)

Sargon



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.