Comments on: Supercomputer slays U.K.'s top chess player

Chess grandmaster comes in a distant second to Hydra, in the latest clash of human versus machine.
Photos: Grandmaster crushed

[Razzl]

Time to answer the fundamental question of chess...

...which is, given the first move, can white always win, or can black always hold for the draw? There must be enough computing power out there now to answer the question, so it's time to get on with it. I agree with Chernev that answering the question may be like losing an old friend forever, but there are rules changes that could enable the game to survive once we know the answer. For a century competitive chess has been about humans trying to see how close they can come to being perfect computing machines, but now our machines can lay bare that lie, so it's time to move forward with altering the rules to reintroduce chance once the machines point out the inevitable pathway to win or draw under the current rules...

Still some other games...

There are still other games where humans
will probably continue to beat computers for
some time into the future. Standard chess is
relatively easy to program. Using the basic
pawn=1, knight=3, bishop=3, rook=5, etc.,
system for evaluating the value of pieces, it
isn't that difficult to teach the computer how to
perform a rough evaluation of the strength of
each side after a combination of moves. Of
course, strong chess programs go far beyond
this basic valuation system in their analysis,
but the point remains that evaluating a board
is relatively easy compared to some other
games. So chess lends itself to "brute force"
programming techniques of simply calculating
every possible combination of moves and
evaluating their outcomes with a relatively
simple board evaluation routine.

But how about a game like the Japanese
game of Go? Here things start to get much
more subtle. Unlike chess, where you can
suddenly lose through the capture of a queen
or king, events proceed much more gradually
and subtly in Go. Furthermore, it requires
some degree of expertise and long-range
vision to evaluate a board position. Unlike
chess, there is no simple pawn=1, knight=3,
etc., evaluation method. Indeed, my father,
who learned Go while stationed in Japan
many years ago, can whip the best Go
computer programs I send to him despite the
fact that he considers himself just an
"average" player. He says that the programs
make a lot of silly strategic mistakes.

[Sam Papelbon]

great...

now when will they get to work on creating a computer that can play monopoly without trying to trade you baltic for boardwalk every 2 turns?

Re: fundamental question of chess...

This chess machine may be extremely good,
but I think it is probably still far from being a
"God-like" machine which is able to play
perfect chess. In a match between this chess
machine and God, I would put my money on
God, even if He plays black and gives the
computer a knight handicap.