Game Over at Last?
This month marks the 10th Anniversary of Garry Kasparov's infamous match with Deep Blue. That match marked the first time that a chess computer beat a reigning world chess champion in a match. Of course, much recrimination followed that result, not least from Kasparov and his understandably bruised ego. Years later, the computer logs were published by IBM and the last vestige of doubt was removed. It was a fair match and Kasparov lost, which in an unofficial sense at least, made Deep Blue the world chess champion.
Of course, Deep Blue then did a "Fischer" as IBM packed up their bags and decommissioned their chess playing monster, having attained their objective and received a mountain of publicity that even they could not afford to buy.
It's tempting to view the match as a tipping point at which computers finally fulfilled the dreams of the original programming pioneers to surpass all human chess players. It's true to say that the last barrier had been broken for a chess computer, but the success was mainly achieved through brute force processing power rather than programming advances.
The real advances came later as more and more subtle features were built into the evaluation functions of chess engines. As computers 'understood' more, they gained in strength and no longer needed specialist hardware to defeat the best grandmasters.
Which leads us right up to the present day and the remarkable Rybka. Even at a conservative estimate, Rybka is stronger than any other chess engine by around 60 Elo points and now appears to have broken the 3000 Elo barrier.
A challenge has been issued by a poster at the Rybka forums, offering $1000 to anyone who can beat Rybka (under conditions which handicap the computer as much as is reasonably possible without actually giving a pawn odds).
I sincerely hope that no-one attempts to take up this challenge. The time for computer v human chess matches has surely passed. Let's face it - computers can now play better chess than humans. The real point to understand is that it doesn't matter. Chess computers are best used now to help us improve our understanding of chess through analysis of human games.
Can we declare the game over at last?
Of course, Deep Blue then did a "Fischer" as IBM packed up their bags and decommissioned their chess playing monster, having attained their objective and received a mountain of publicity that even they could not afford to buy.
It's tempting to view the match as a tipping point at which computers finally fulfilled the dreams of the original programming pioneers to surpass all human chess players. It's true to say that the last barrier had been broken for a chess computer, but the success was mainly achieved through brute force processing power rather than programming advances.
The real advances came later as more and more subtle features were built into the evaluation functions of chess engines. As computers 'understood' more, they gained in strength and no longer needed specialist hardware to defeat the best grandmasters.
Which leads us right up to the present day and the remarkable Rybka. Even at a conservative estimate, Rybka is stronger than any other chess engine by around 60 Elo points and now appears to have broken the 3000 Elo barrier.
A challenge has been issued by a poster at the Rybka forums, offering $1000 to anyone who can beat Rybka (under conditions which handicap the computer as much as is reasonably possible without actually giving a pawn odds).
I sincerely hope that no-one attempts to take up this challenge. The time for computer v human chess matches has surely passed. Let's face it - computers can now play better chess than humans. The real point to understand is that it doesn't matter. Chess computers are best used now to help us improve our understanding of chess through analysis of human games.
Can we declare the game over at last?
