The Investor vs. the Machine
By Jack Uldrich
December 12, 2006
Although it received little mention in the popular press, Vladimir Kramnik, the world’s reigning chess champion, was defeated last week 4-2 by the computer Deep Fritz, in a best-of-six-game match.
Advantage computerThe defeat was not really that surprising to me. When Kramnik last met Deep Fritz four years ago, he battled the computer to a 4-4 tie. Since then, however, the computer has only gotten better.
In fact, since 2002, Kramnik’s “Elo” rating — which measures the strength of a chess player — has dropped, while Deep Fritz (an improvement from IBM’s (NYSE: IBM) famous Deep Blue program) increased the number of positions it could calculate per second from 2.7 million to 8 million.
What this meant in more practical terms is that Fritz could now look ahead a full nine moves. In spite of this advantage, Kramnik figured to be competitive because humans still excel at long-range planning. More specifically, he figured to minimize Fritz’s processing advantage by striving for “quiet” positions and exchanging queens early.
Kramnik employed this strategy beautifully in the first game. Alas, he lost his patience in the second game and suffered a humiliating defeat when be blundered into a checkmate, which many analysts immediately ranked as one of the most amateurish moves in the history of competitive chess.
The full article can be read here.
Do you agree with the author?
I was disheartened to read some of the blog comments regarding computers and chess. How easily people are willing to conceed defeat. Even the people arguing against computer dominance sort of imply the opposite by their wording. I suppose it makes sense because Krammniks defeat is an echo/reinforcement of the experience of most of the people posting and their experience with computers. Instead of being just a reflection of his own incompetence. In Kasparov’s games against Fritz and Junior he took a completely different approach. He actually entered complex, sometimes open positions and played tactics and combinations.
So I get more concerned if I see grandmasters talking this way. Does Susan believe she is inferior to these machines? The very clear fact is that:
1. computers do not play perfect or even near perfect chess. They do make mistakes.
2. Anything that bleeds can die.
Real grandmasters shouldn’t be intimidated. But there more people holding the title Grandmaster than there are holding the description Grandmaster.
I don’t have a lot of respect for Krammnik. And with the blunder in game two I realized just how far my opinion of him and Topalov who lost to him has fallen.
Marcellus Wallace
It’s known that Susan challenged the computers many times including Deep Blue. They don’t want to face her as it’s embarassing to lose to a woman.
it’s embarrassing to lose to a woman?
why is that? please tell.
Susan still doesn’t like VK: the article is much more than just the comment in bold about the blunder.
Kramnik vs Fritz 10 should have ended with a tied match. Surely noone takes seriously Kramnik’s loss in the second game or attributes that loss to the intrinsic superiority of computer programs to human players.
Kramnik’s loss in Game 2 to a one mover was a freak accident and that’s all there is to it. It has never happened in any serious game Kramnik has played and it never will happen again in a serious game Kramnik plays, end of story.
If Kramnik had drawn Game 2, then he would have played a defense in Game 6 that he was comfortable with instead of the Najdorf. Kramnik played the Najdorf because he had to win GAme 6 to draw the match. So bottom line is that Kramnik lost the match due to a freak accident in Game 2.
Having said the above I think there is no question that chess playing software for personal computers and personal computer hardware had improved over time. But it is not clear to me that Fritz 10 plays better chess than Kramnik. Maybe Fritz 11 will play better than Kramnik and of course it is inevitable that computers, both mainframe and PCs, will be better than humans someday and probably someday soon.
What are we agreeing or not agreeing to?
a). the full article and what’s posted here is simply informative. i see nothing to agree or differ with.
b). i think one of ms. polgar’s bots created this blog entry soley due to its chess content. the full article has little to do with chess.
c). what makes you think ms. polgar does not like kramnik? please site a reference or two. i am only curious (and neutral. i will not defend kramnik or polgar).
d). how many games could susan win against judit in a 6 game match?
Fritz has nothing to do with Deep Blue at all. They are completly different programs written for a total different architectures.
While Deep Blue was written for a special computer Fritz just runs on an average Windows system.
Have there been any games where the computer beat the human but the human did not make an obvious error? What about VK-DF10 Game 6?
If I was pitching $1m into man-machine-chess sponsorship, I’d want a blunderguard to prevent the likes of Game 2.
For example, a ‘play by wire’ system might yellow-flag any move deemed to lower the score by more than x ‘pawn points’ (e.g. x=1). How small does x have to be to filter out ‘obvious errors’.
VK’s moment of blindness spoiled an interesting end to Game 2 and the contest: both would otherwise have ended in a draw.
Given that I just felt that was a sad and fruitless end to a game, I clearly don’t want to see that sort of outcome again because it proves nothing.
At the same time, I’d give back the machine the Endgame Tables, and certainly not give the humans the ‘right to the Endgame-Table Draw’.
g
“Deep Fritz (an improvement from IBM’s famous Deep Blue program) “
what a bunch of bollocks!
>>Kramnik vs Fritz 10 should have ended with a tied match. Surely noone takes seriously Kramnik’s loss in the second game or attributes that loss to the intrinsic superiority of computer programs to human players.
>>
Depends what you mean by “superiority”. Does it reflect a greater understanding of the game? No. Does it reflect the fact that human performance is more variable in quality, and subject to sudden drops due to fatigue, distraction, or whatever? Yeah. Is it a superior aspect in a player to be able to maintain focus? Yeah. Does it mean we could expect the same result if the match were played again? No.
So, take your pick.
>>it’s embarrassing to lose to a woman?
why is that? please tell.
>>
Women are more likely to use floral mousepads, to use fluffy mouse covers, and to stick smily faces on their monitors. What self-respecting computer would want that? A Macintosh might like it, but not a rugged manly-man PC.
Yasseer wants comps to be handicapped in future matches with humans, My Goodness ,Kramnik got the most favourable conditions ever, how much more handicap should the comps give? Every other GM did not get the same privileges Kramnik got. I really wonder how Topalov or Kasparov would perform with such privileges. With these advantages Kramnik had some superior positions but still could not win, whose fault is that? All this flimsy excuses for Kramnik is unnecessary. As a complete package Kramnik’s best years are behind him. As a complete package Kramnik simply is NOT the best HUMAN player in the world, PERIOD.
Kramnik, did not just lose , but he got COMPREHENSIVELY outplayed POSITIONALLY for that matter. The time has come to accept comps UNDERSTAND the game better that humans. Humans progress slowly because they stick to rules and pattern recognitions, anything outside that box baffles them e.g. Silly static rules like exchange bad Bishops or Do not move pawns in front of your King tries to reduce chess to simple irrational silly rules which is TOTALLY far from the truth in actual practise…. comps DO NOT have this limitation. In a way I am happy the comp won, it will open all our eyes. chess players can learn alot from this match and computer thinking in general
>>
Kramnik, did not just lose , but he got COMPREHENSIVELY outplayed POSITIONALLY for that matter.
>>
Did you see the same match? He got tactically outplayed in Game 6, played a bit better positionally in Games 1 and 2, though not enough to win, and held it even in 3, 4 and 5. Computers haven’t passed humans by positionally yet, though they do have other advantages.
And they fail to understand even some things that low level players would know, such as that Kramnik’s exchange sac in the one game would lead to an inevitable draw.
>>As a complete package Kramnik simply is NOT the best HUMAN player in the world, PERIOD.
>>
Looks like you wanted to believe that so badly that you forgot to come up with any kind of argument for it. Very dangerous when the conclusion comes before the facts.
Obviously he is the best, or should be considered as such (since nobody has exactly the same performance each and every day). He just beat the World’s Top Rated Player in, not one, but two separate matches, a Classical one and a Rapids one. Now, if Anand were to come out and roll Fritz up, that might cause some serious questions to be asked, but I think that’s wishful thinking.
Who cares if Kramnik is the best player in the world or not? 1.) Suppose Fritz 10 might lose against five grandmaster in the world. But it can still beat the rest of the 6 billion people in the world. Based on this, who is better Fritz or the homo sapiens…? 2.) Fritz is not the best existing chess program. Rybka is much better, as well as the supercomputer Hydra.