Deep Fritz (C) – GM Kramnik [C43] Click here to replay the game
12.01.2006
1.e4 e5 2.Nf3 Nf6 3.d4 Nxe4 4.Bd3 d5 5.Nxe5 Nd7 6.Nxd7 Bxd7 7.0-0 Bd6 8.Qh5 Qf6 9.Nc3 Qxd4 10.Nxd5 Bc6 11.Ne3 g6 12.Qh3 Ng5 13.Qg4 Qf4 14.Qxf4 Bxf4 15.Nc4 Ne6 16.Bxf4 Nxf4 17.Rfe1+ Kf8 18.Bf1 Bb5 19.a4 Ba6 20.b4 Bxc4 21.Bxc4 Rd8 22.Re4 Nh5 23.Rae1 Rd7 24.h3 Ng7 25.Re5 Nf5 26.Bb5 c6 27.Bd3 Nd6 28.g4 Kg7 29.f4 Rhd8 (White is slightly better due to space advantage but Black should not have too much problem holding this position. In fact, I would not be so comfortable as the computer right now. Kramnik has much deeper understanding of endgames than any software out there.)
30.Kg2 Nc8 += (Kramnik may not be the greatest player ever but he is certainly one of the smartest players. He knows how to maximize his game and repertoire to his advantage. In this game, Fritz got basically nothing out of the opening. The Petroff and Berlin are perfect opening choices against the computer. Not many players can do what Kramnik does.)
31.a5 Rd4 (White has to move one of the Rooks to e4 to protect both pawns. My preference would be the Rook on e5 to e4.)
32.R5e4 Kf8 (Black is in no danger. I am not sure who is punishing who here.)
33.Kf3 h6 (Interesting move. I was looking at 33…f5 and Black is more than OK. h6 means that Kramnik is happy with the positioning of his pieces and it is like telling the computer to go ahead and make something happen. If White plays 34.Rxd4 Rxd4 35.Re4 this would lead to an easy draw endgame.)
34.Rxd4 Rxd4 35.Re4 =/+= (Black should not trade the last Rook. Best would be to move back to d6.)
35…Rd6 36.Ke3 g5 (I am not sure if I like this move. This move gives the White Bishop more mobility.)
37.Rd4 Ke7 (This still does look drawish but I think Kramnik is making life more difficult for himself.)
38.c4 (White has space advantage but Black has no serious weakness to cause any problem.)
38…Rxd4 39.Kxd4 gxf4 40.Ke4 Kf6 41.Kxf4 Ne7 42.Be4 (This is one of those lengthy endgames where neither side will make much progress. I still expect Kramnik to draw.)
42…b6 43.c5 bxc5 44.bxc5 Ng6+ 45.Ke3 += Ne7 46.Kd4 Ke6 47.Bf3 f5 48.Bd1 Kf6 (I see no practical chances for either side to win. However, it is a little easier playing this position as White.)
49.Bc2 fxg4 50.hxg4 Ke6 51.Bb1 Kf6 (No reason to keep going on with this position.)
52.Bd4 Ke6 53.Bh1 Kf6 54.Bf3 Ke6 1/2 Finally! Once again, Kramnik held for a draw with ease.
Click here to replay the game.
“how to squeeze a world champion”
“Torturing a world Champion”
Poor Kramnik …
I don’t think Kramnik’s at all tortured here.
30…Nc8 taking time to improve his horsie. At first I thought it was dominated a few moves ago.
Seirawan at first thought the pawn advances by Fritz was bad–now he thinks it’s good.
Finalment!! Good Morning!
After Kramnik’s “bunder of the century” and Fritz lead in the match, if I were the programmers, I would certainly activate the programline which sets “go for draws”. Against the world champion, in a 6 games series, a one point victory is plenty good enough. It will still go down in chess history that another computer defeated another world champion (and for the Fritz people that it was Fritz).
>>
Miguel said…
“how to squeeze a world champion”
“Torturing a world Champion”
Poor Kramnik …
>>
LOL, whatever.
After 37, Kramniks knight looks so weak
Now I can how much removed I am from grandmaster chess. When I thought that g5 was actually good, locking white’s pawn on the same colour as the bishop, she goes on saying that it gives the priest a bit more mobility.
And when I thought Ke7 was a good plan by agreeing to the exchange with the Black King’s position slightly better than before, she says he’s making life difficult for himself.
There’s a big difference between an amateur like me who learns from books and play lesser opponents, and a GM of her calibre, battletested, with more knowledge of how the game should be played in such positions.
There’s still much for me to learn!
Not sure what Susan’s problem with g5 is. Yes, it gives the Bishop more mobility, but it takes away things for it to actually shoot at. If the Rooks come off soon, Black’s going to want as many pawns on black squares as possible.
If the rooks come off, and Black Knight goes to K2, the Bishop has lots of squares but no targets and the White King can’t get through either. What’s not to like about that? White seems to have more empty space, but absolutely no pressure and plenty of maneuvering room for Black.
This game = another dull draw.
Maybe Gabor is right and chess programmers set the engine to “draw mode”.
…but I’m not sure about the reason behind gxf4. Why give up the f4 square? It probably doesn’t do any harm, but still why do it?
Ivan wrote
“This game = another dull draw.
Maybe Gabor is right and chess programmers set the engine to “draw mode”.”
It’s interesting enough. Plenty of maneuvering, and past Move 40. If all that matters to you is to have a decisive result no matter what, irrespective of the quality of the game, you might be playing the wrong game entirely. Chess may not be for you. You might want to try Shogi, which has nearly 100% decisive results.
I wonder what Fritz 10 position estimate is show now (42nd move).
Rybka?
Hey,I never manages to draw my Fritz 7!
So,I guess my games vs machine are more interesting than Kramnik’s ones.
Oh boy,how I would like to see Kramnik one win.Next game for the Human player is to be or to be.I hope Kramnik prepeared something special for that occassion.
One thing for certain now, the Fritz team may be reluctant to agree to a proposed draw from Kramnik… “let’s just play ONE more move, shall we” might be in their mind all the time now!
Yasser Seirawan is doing live commentary? where?
http://mfile.akamai.com/26218/live/reflector:30473.asx
Well, log in to chessbase’s playchess server.
I bet even Kramnik’s wishing that the programmers designed and installed some kind of boredom function into Fritz. This is no endgame which he can’t defend. I’m off to sleep.
Unofficial result 1/2-1/2
Drawnik strikes again!!! We need some like Mikhail Tal to thoroughly thrash this machine. Deep Fritz would never standup to one of Tal’s attacks.
Is there a point to this match, other than to serve as a commercial for Chessbase?
No, this is the most boring match.
“Deep Fritz would never standup to one of Tal’s attacks.”
It would.
the computer would probably find the single narrow and crooked path to prove tal’s sacrifices unsound…
which brings me to an interesting question. people are trying to suit their game according to the comp – cramped positions, more strategy than tactics, etc. all this to amplify their (very questionable) edge in the game.
the comps on the other hand, are trying to play the best lines objectively. which may lead to many draws against careful defenders such as kramnik.
why isnt fritz programmed to make a dubious sacrifice, rip open the position, and by the time human opponent realizes what hit him is just hot air, the comp will be posing threats all over the board.
one risky, dubious move and then go at it with near perfect accuracy. i think very few humans would stand a chance – the chance to find that single saving move after move (even if the line is losing some 16-17 moves down the line).
One anon said this:
We need some like Mikhail Tal to thoroughly thrash this machine. Deep Fritz would never standup to one of Tal’s attacks.
How do you know? Hydra humiliated one of the best in the world, didn’t it?
The decades long argument, whether computers will ever be able to defeat the best chess players has been over since 1997 when Deep Blue defeated Kasparov. Ever since, what we see is the “last gasps of humanity” to slow down the obvious superiority of this technology.And to defeat the best chess programs became a real challenge and now the world champion could be proud to defeat the computer. In the 80’s it was a joke. Now it isn’t.
As I wrote already, there is nothing wrong with that. Humans will return to playing each other, and won’t bother to challenge the computers, just like no 100 meter sprinter would bother to challenge a race car. Cars are racing each other separately and humans separately.
The computer in chess will become an excellent teaching, practicing tool, and scaled down, opponent if a human is not available.
This should no longer be a “pride” issue.
Gabor
Oh yes, and one more thing.
Kramnik would have to win both of the last two games. Very unlikely.
At the likely best he may win one and draw another. I would be very surprised if he could win both.
Gabor
Computerized chess programs today can defeat the most brilliant human grandmasters. In past centuries, several famous automatons—chess-playing robots—wrought similar embarrassment on masters of their own periods. At least, that was what their promoters wanted the ticket-buying public to believe.
First on display was “Turk,” the work of an Austrian, Baron Wolfgang von Kempelen. Debuted in 1769, “Turk” was a life-size figure in Arabian garb, seated behind a chessboard placed atop a sizeable wooden box. Inside the box was a curious system of mechanical workings that duly impressed the audience when it was unveiled to them at the beginning of each game. Whenever “Turk” reached to make his move, observers could hear machinery grinding below. “Turk” earned money for several successive owners, both as a competitor and later, after “retirement,” as a museum attraction. Ultimately exhibited at the Chinese Museum in Philadelphia, it was destroyed in a fire in 1854.
Was it really an “intelligent machine?” Among its skeptics was Edgar Allan Poe, who wrote an exposé, “Maelzel’s Chess Player” (Johann Maelzel, a Bavarian musician, was one of its owners). Poe claimed a man of small stature sat within the exotically clad figure, watching the board through a tiny opening in the clothing at breast level and manipulating the mechanical arm.
Next came a French chess master named Mouret. Operating an automaton in London over a period of months in 1820, he (or the machine) took on 300 challengers and, astoundingly, won or drew all but six games. It’s even more astounding since Mouret gave each opponent the first move and the opening advantage of a pawn. Automaton or no, it was a remarkable chess exhibition.
“Ajeeb” was the 1868 creation of Englishman Charles A. Hopper. Dressed as an Egyptian, it performed on two continents for 60 years. Like “Turk,” it met its end by fire.
“Mephisto” was invented in 1878 by a prosthesis maker, Charles Gumpel. Naysayers of Poe’s persuasion suspected that, as before, a live player was posted inside. At any rate, progress was being made: “Mephisto” may have been the first automaton partly controlled by electricity.
The first legitimate chess machine appears to have been an automaton developed by Torres y Quevedo, a Spanish scientist, introduced in 1890. Quevedo’s player made no attempt to “think” through an entire game, only a simple ending (king with rook versus king). Still, it was ingenious for its day and was a true precursor of the technology that has changed the game of chess forever.
“We need some like Mikhail Tal to thoroughly thrash this machine. Deep Fritz would never standup to one of Tal’s attacks.”
Each of these games seems to point to the same consistant theme. Kramnik is able to carve out an early, slight advantage with his positional play. But as material comes off of the board, Fritz is able to focus on a shrinking number of possibilties with its massive computing power. At that point, the slight advantage isn’t enough be converted into a win against a machine that doesn’t overlook anything. Kramnik’s advantage is when there are 32 pieces on the board and a nearly unlimited amount of possibilities for Fritz to analyze. Omitting the blinder in game 2, I think Kramnik and Fritz have reached a strategic stalemate of play that would probably result in a never ending succession of draws if Kramnik wasn’t governed by fatigue.
All an attacking style of chess like Tal’s would bring would be a more rapid exchange of material, allowing Fritz to get into it’s “comfort zone” that much sooner.
Chess may be far more complex than checkers or tick-tac-toe, but eventually computing power will be able to see every possibility and counter it just like a human can with tick-tac-toe. No matter how big the number is, eventually the computer will be powerful enough to see all of the possibilities and a human GM will have to play a flawless game to achieve a draw and will never be able to win. When that point is reached, the computers won’t be able to beat each other either.
The point is that what makes chess at all interesting as a game, rather than a math problem, is the human side of it. Mistakes are what make the game interesting. Take those away and we might as well join all the kids and start playing Doom III instead.
“Kramnik’s advantage is when there are 32 pieces on the board and a nearly unlimited amount of possibilities for Fritz to analyze.”
Of course. Kramnik needs a programmer to help him, not chess players/trainers.
What is the money at stake in this match??
Kramnik gets $500,000 to show up and another $500,000 if he wins.
“Of course. Kramnik needs a programmer to help him, not chess players/trainers.”
If you are the highest rated person in the world, who, exactly, is supposed to coach you to be better? Someone with a rating 400 points below yours?
Quoting Anonymous:
Chess may be far more complex than checkers or tick-tac-toe, but eventually computing power will be able to see every possibility and counter it just like a human can with tick-tac-toe. No matter how big the number is, eventually the computer will be powerful enough to see all of the possibilities and a human GM will have to play a flawless game to achieve a draw and will never be able to win.
>>>>>>>>>>
No, it’s practically impossible. There are more possible chess positions than electrons in the universe. The exponential growth is not computable to the level of solving chess. You can see this with every level of tablebases, how the storage space required to hold them increases exponentially at each level. According to Wikipedia, six-piece tablebases require about 1.2 terabytes to store, and “a full set of seven-piece tablebases is not realistically expected until 2015 or later.”
“No, it’s practically impossible. There are more possible chess positions than electrons in the universe. The exponential growth is not computable to the level of solving chess.”
Sorry, I don’t buy that. 99% of those “infinite” positions are either theoretically impossible to come up in a game or are so one-sided on the board that it would take little or no thought to win from them. Balanced or nearly-balanced possitions are finite. More importantly, the moment you play 1.e4, the number of possibilities drops exponentially and computing power will reach the point where it can solve the best move for the situation. You can go on believing that it won’t happen, but that’s just wishful thinking. The computers have already reached a point where virtually nobody but the Super GM’s can even give them a game. How much better do they need to get?
The real scary thing is if people playing in tournaments are able to access that computing power while at the board. We already saw the first instance of this at the Open this year. Forget drug testing. If that sort of cheating comes to pass, competative chess will be dead. It’s already impossible to play a coorespondence game without wondering if the other guy did beat you fairly or not. Soon, the same may be said for OTB games, too.
1) As a matter of fact, Kramnik IS seconded by Christopher Lutz, who has a ELO of “barely” 2600. Why him and not a stronger player? Lutz brings two qualities: Firstly, he is an expert in systematic opening preparation. Secondly, he is co-developer of Fritz and has some insights in the heuristics of the machine. Kramnik IS using engineering knowledge against Fritz.
2) I would give Kramnik better odds than Tal, Topalov, or any other player who was primarily a “tactic” player. With the depth of its analysis it can defend itself against any short-term motives. They might have a chance to land a positional sacrifice but only if there was an opportunity for a sacrifice with no short-term compensation. Unfortunately for Tal/Topalow, Fritz gets even positional pawn/exchange sacrifices pretty quickly.
3) Kramnik might have a chance, even in the endgame. With a closed, simplified position, Fritz’ ability to calculate in advance might improve but humans can apply their schemes even better. Fritz overlooked Kramnik’s castle in game three, it might overlook other positional motives. The machines still have problems to calculate the long-term-value of doubled pawns and good/bad bishops. And Kramnik is one of the players I would trust to find a way to abuse it.
Lets be honest folks. We all know that Kramnik Is going to lose this match Don’t we? He already said he lost most of his games against his training version at faster time controls. His love of the draw and general fear of the machine almost gaurantees he’s going to either lose or draw the rest of these games.
When Kramnik blundered into checkmate the most amazing thing is he wasn’t even being lazy. He was paying attention and calculating. Also this isn’t an isolated event. I forget the game but he’s suffer from at least one other very basic hallucination.
Marcellus Wallace
28.g3 is a typo. Should be 28.g4
once again kramnik makes a booooring draw