Can We Solve Chess One Day?

That's the title of a guest blog by computer scientist Ken Regan, who also happens to be an IM and a regular commenter on this site. Have a look!

Comments

[Kenneth Regan]

The comments below that item give the correct link to Eiko Bleicher's 6-piece EGT server at www.k4it.de, and fix the definition of "DTC". They also have other discussion. Has anyone picked up Anand definitively saying he saw the K+P ending was winning when he played 40...Kg7! in the last game?

The title question wasn't mine, and elsewhere I've given a simplistic answer: If by "solving Chess" one means the same kind of explicit tabulation of an optimal strategy as was done for 8x8 (Western) Checkers, represented in the manner of the current tablebases, one can definitively say No. The number of bits in such a tabulation would exceed the Bekenstein Bound for human scales---meaning any toilet in which it were hidden would collapse into a black hole.

However, an implicit tabulation by giving rules---rules which might have nothing overtly to do with chess---is still distantly possible, and that's where deep unsolved questions in my field enter the picture. Can one leverage the chess engines' own search functions to regenerate full information about a position while providing only a fraction of the table?