Sunday, March 22, 2009

chess solved in a few centuries ??

according to a recent article in the magazine of the Dutch
computer chess association, chess could be 'solved' (like
checkers) in a few centuries, e.g. by approx. the year 2600.

well, if the Fide rules are applied, the result will
a draw, according to the following reasoning:

1) due to the 50-move draw rule- and 3 move repetition,
the maximum length of a chess game is about 6000 moves
(haven't found the proof for this, but by computer
simulation we probably could confirm this, in
fact i suspect 6000 is already quite high)
2) if chess would be a forced win for white, then
in the final stages of the game, i.e. the transition
to the endgame, lets say by move 4000, but probably
much earlier, the evaluation function (or simply
material balance) for white should definitely be
an advantage, otherwise the endgame cant be
won in a sufficiently small nr of moves

but.. 3) due to the fact that all openings can result
in drawish positions -provided black plays
correctly-, in about 50 moves, often earlier,
the above criterium, i.e. a white advantage
cannot be achieved, and certainly not in *all*
cases (ie. when investigating all possible defences
for black after either 1.d4, 1.e4, 1.Nf3 etc.

Ergo, chess is a draw.

A useless exercise to let computers crunch
on this game with these endgame rules for
the coming centuries..

or we should change the rules for computer chess,
which certainly would make endgame theory a
lot more interesting, for the coming future !

Bookbuilder program/download link

Apparently at the site the download link for my chess opening program Bookbuilder disappeared Here it is: Bookbuilder demo ...