Chess monster gets human touch
Matt Kwong
Last Updated: October 02. 2008 9:04PM
UAE / October 2. 2008 5:04PM GMT
So infuriated was Napoleon Bonaparte, by some accounts, that the short-fused French emperor knocked the pieces from the chessboard midway through a match, humiliated by his impending defeat.It was outrageous that a machine could outwit him – a distinguished military commander – in a game of strategy.
Napoleon’s formidable opponent, the world’s first chess automaton, was billed as a soulless wooden mannequin encoded with masterly playing prowess. But the turbaned “Turk”, as the machine became known from 1770 on, was not merely all whirring gears. It was an elaborate hoax engineered by Wolfgang von Kempelen. And concealed within the mechanical components was a highly skilled chess player and operator.
Machines are now widely known to have conquered the chess world, overmatching many of the finest grand masters. Although some may still reject the idea that artificially intelligent inventions can best humans at anything requiring imaginative thought, researchers in Abu Dhabi are busy teaching a computer some very human traits.
The PAL group is now trying to code intuition, risk and style into Hydra, a 64-processor behemoth that is considered one of the strongest chess-playing entities ever. It can calculate 300 million positions in one second, according to its Germany-based programmer.
Chess – with its simple rules but infinite complexities – has for years served as an ideal prism through which scientists have measured the advancements of artificial intelligence.
By comparison, the game of checkers is much simpler, having been “solved” last year by Canadian researchers analysing 500 billion billion scenarios over 20 years.
“It is a very complex game because there’s no solution to chess right now,” said Muhammad Nasir Ali, Hydra’s project developer since it began in 2003. “The number of possible moves and positions seems infinite.”
Claude Shannon, the information theorist, calculated that there were more possible chess moves than there are atoms in the universe. That estimate has not daunted the Hydra program, which Mr Ali said was created as a “passion project” upon request by Sheikh Tahnoun bin Zayed al Nahyan.
Hydra dominated the world computer chess scene in its early years, defeating the world’s seventh-ranked grand master, Michael Adams, in 2005, and claiming the title as the top chess machine at the International Paderborn Computer Chess Championships the same year.
Unlike humans, the supercomputer also eliminates the chance of human error, said Abdul Mateeh Khan, the machine’s programmer.
“Usually in a game, there is room for human mistakes. With machines, this cannot happen,” Mr Khan said. They “will just wait for a human mistake”.
Here is the full story.
He was slaughtered.
And it didn’t seem that he cared much. He didn’t change to closed openings. No matter if that’s his repertoire or not: A 2700+ can play any opening good enough to give priority to such variations, which are uncomfortable for the machine.
But Adams played a training match. I was disappointed. Kramnik did far better in his matches against Fritz.
P.S. It seems to me that this article has only collected “old news.” If I look at hydrachess.com, I see no recent Hydra news, and the latest (general) chess news is from July. Doesn’t look very alive to me.
When he says ““The number of possible moves and positions seems infinite.”, it’s a total nonsence: A huge number, even the number of stars, or anything, has nothing to do with the concept of “infinite”, which is not a number. For sure this absurdity does not come from the mathematicians working on Hydra, but from a media, and was then copied “as bad as it is”.
Anonymous says “Kramnik did far better in his matches against Fritz.”
Fritz and Hydra are completely different. Hydra is uncomparabely stronger. So, why this comparaison??
Number of positions is ~10^44.