Kramnik will take on Deep Fritz starting tonight in a six match game. I expect he’s got very little chance – especially the way he played against Topalov. He played better than Topalov and Topalov is a great player, but . . . Topalov didn’t play that well against him (except in a couple of games) and Kramnik made quite a few bad mistakes. But we’ll see. Troppodillians will be amongst the first group of people not particularly interested in chess to find out!
Kramnik has this to say.
Fritz examines millions of moves per second. It is extraordinarily difficult to play against such a calculating monster. Right from the start you are walking on a very narrow ridge, and you know that any inattentiveness will be your downfall. It is a scientific experiment and I will have to fight very hard for my chance
Well, I guess we can be grateful that we live in an age in which humans do beat the best computers . . . in some games anyway.
A matter of some interest is that chess computers never got anywhere until they began aping humans ability to recognise patterns. In other words their strength in analysing millions of moves per second didn’t rate against humans’ ability to recognise patterns until they worked out ways to train computers to recognise patterns and added to that their monster calculating abilities.
But what does ‘recognise patterns’ mean? If that’s the key concept in your post, isn’t it worth spelling out a bit? Does it mean seeing something familiar in a particular configuration and thinking, ‘I’ve come across this situation before, and it lead to the knight taking the bishop four moves later, so I should avoid it while I still can’? If so, does it mean the computers are fed a database past matches which they can scan? If not…
Sorry James. I agree that ‘pattern recognition’ versus ‘raw computing’ does beg a few questions. But I’ve taken the distinction at face value. No doubt more could be said following up the relevant google links, but doing so quickly turns up more of the same kind of question begging implicit assumptions that the distinction isn’t particularly problematic. No doubt I’d have more to say with a bit more time on it – but it was a bit of a throwaway line, not a post about patten recognition.
Well how about game two?
Play through the game if you like here.
There have been no mate in ones missed by a player at Kramnik’s level. But one move which looks worse, where Reschevsky took a pawn with his queen in what would have been mate, but it was guarded by a bishop!
It’s all very sad as in both games Kramnik got the better of the computer (no mean feat with black in this game) and had winning chances.
I thought the monster had Kramnik when I went to bed last night. Kramnik played into a position that I think he thought would favour his style of play, but it became clear very quickly that it favoured Fritz. I expected Fritz to come down the board and squeeze the life out of Kramnik. But he defended well, and came up with a plan – to swap a rook for a bishop and a pawn and then it was easy to draw (for Kramnik that is!)
The official blurb on Chessbase says this.
Play through here if you want.
Meanwhile, anyone want to participate in the inaugural Troppo chess challenge – to be played on http://www.freechess.org ?