AT LARGE: The best of mates

AT LARGE: The best of mates

Garry Kasparov is at it again. In a couple of months, he's going to play a chess match against Deep Junior, the world's champion computer chess program. In the meantime, there's a match on between someone called Vladimir Kramnik and something called Deep Fritz. No-one has ever heard of either, so consider it a lead-up to the main event.

Last time Kasparov did this, in 1997, the reigning computer was called Deep Blue and was built by IBM. The match was closely scrutinised by the world's chess-loving media.

(This is actually a slight exaggeration. The world's media tend to ignore chess, but "man versus machine" contests are generally good for a ratings point or two. Coverage of the Kasparov-Deep Blue match involved much commentary along the lines of "the little horsey one moves in an L".)The stakes were actually a few measly dollars, but the contest was reported as if whichever species won the match would then lay claim to dominance of the world.

Kasparov lost, IBM did not take over the planet (any more than it had already) and the media went back to not caring. Kasparov went on to win lots of chess games against people, while Deep Blue got reprogrammed. Not sure what it's up to these days - perhaps, like most five-year-old computers, it's been handed down to the kids for homework.

Now, five years later, Junior (a program that anyone can buy for their PC) is apparently much better than Blue, and some of the key disadvantages Kasparov faced (such as his opponent's lack of a playing history - the computer had been programmed with every move Kasparov had ever made in every game he'd ever played) have been overcome by the growth of computer-vs-computer tournaments. Deep Junior is officially the world's champion computer chess program.

Why anyone thought that Deep Blue required improvement is a bit beyond me. To my mind, the ultimate achievement of any technological development is to perform the function for which a device was designed, and to do so reliably. Deep Blue was designed to beat Kasparov, and it did it - just the once, but reliably.

So why make it better? So that it could beat Kasparov again, in fewer moves? So that it would be so good, and so knowledgeable about the ways of chess, that it could handily beat any human chess player? This seems self-defeating.

Allow me to clarify. Machines, by their definition, do things better than people do. If they didn't, we would never have invented them. Shovels move dirt better than bare hands do, just as computers do Fourier transforms faster than the smartest human engineers. When a team of chess-loving engineers set out to design a chess computer, the goal of that project is necessarily a computer that will be able to play chess better than any human.

The thing is, humans have a bit of a problem with being beaten all the time. I find it very hard to get anyone to play Trivial Pursuit with me for this very reason. If the engineers are successful in the attempt to build a machine that can beat any human chess player, no human chess player is going to want to play against it - the machine is then useless and may as well be melted down for scrap.

Deep Blue was designed and programmed specifically for Kasparov. Because the challenge was so specific, Kasparov was almost obliged to respond to it.

The goal in that case was clear: to prove that a group of computer programmers, none of whom was particularly brilliant at chess, could make a machine that could beat a player none of them could. It's a weird, but in its way noble, goal.

Making a machine that can beat anyone, on the other hand, is madness. The reason people devote themselves to the pursuit of chess (a pursuit which, I assure you, requires much devotion) is because it offers the opportunity to prove intellectual superiority. There is no element of chance, no degree of physical prowess, no advantage can be had beyond the ability to think. (This is, I suspect, a large part of the reason the game is largely ignored by the media. If it were made an "extreme" sport, with mud and injuries and maybe nudity, it would rate better.)The ultimate goal of that devotion is to be the best. To build a reputation as a powerful abstract thinker and overwhelming competitor. No-one wants to dedicate themselves to being the best "for a human". If the best chess player is a machine no human can beat, chess will become a pointless endeavour.

Maybe that's the point of this match: so the human can win and save chess for everyone.

Go Garry.

Matthew JC. Powell e2-e4. Opposing gambits to

Follow Us

Join the newsletter!

Error: Please check your email address.
Show Comments