Checkers was weakly solved 2007: a question

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Jouni
Posts: 3715
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Checkers was weakly solved 2007: a question

Post by Jouni »

Has anybody verified result so far? With current hardware it should be possible using home PC in hours/days or not? Only 10exp20 positions possible.
Jouni
User avatar
Ajedrecista
Posts: 2145
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: Checkers was weakly solved in 2007: a question.

Post by Ajedrecista »

Hello Jouni:

I am not aware of a formal verification.

5e+20 is more less the search space complexity of the game. As far as I remember, the solution was not to check each possible position, but generate 10-man EGDB (endgame databases), then prove that starting from the standard starting position, Chinook could find a path to a drawn position of an EGDB from a set of openings (the 3-move restriction, widely used nowadays). I do not know if all the 3-move openings were tested; GAYP style (go as you please, play whatever you want from the start) should be proven drawn by Chinook.

I can be wrong in some details, but the key idea was finding drawing paths from the opening book to the EGDB. There is an excellent book from the project leader, Jonathan Schaeffer, titled One Jump Ahead: Computer Perfection at Checkers, which describes the whole project over the years.

Regards from Spain.

Ajedrecista.