According to Discover magazine, Jonathan Schaeffer has finally "solved" checkers with his Chinook program.
I read Jonathan's book One Jump Ahead: Challenging Human Supremacy in Checkers almost ten years ago when his brother Dan (my neighbor) gave me a copy. It's a great book. The recurring theme was that everyone kept asking him "Isn't Checkers solved, already?"
Well, it wasn't until now. Chinook finally knows every possible position in Checkers and how to win or draw from that position. Which doesn't really make the game "solved"; not until a winning strategy can be explained to a human player who can then implement it perfectly every time.
Oh yeah: who wins? From the starting position, black to play can only draw against a perfect opponent.
Yehuda
1 comment:
A tiny step for AI, a huge leap for checkers!
Excellent post once again Yehuda!
(goes back to lurking)
Post a Comment