Get your thread on. Consolidate with caution or go for the kill? Kramnik has never beaten Anand with black. Chat and predictions in the message boards. Cheap Shot Dept.: Did Kramnik forget to click the link on the left before game three?
Game four drawn with no hits, no runs, and no errors. Anand again opened with 1.d4 and it looked as though he was trying to rest up from yesterday's excitement. Kramnik didn't shy away from his usual stuff this time and though he avoided the Slav he played an IQP position he's been happy to play before, including in a blitz game against Kasparov in 2001. GM Nick de Firmian tried to make it interesting on ICC Chess.FM, but the game conspired against his best efforts. Anand did get a few tiny threats on the queenside but Kramnik reacted in plenty of time. He spent a while triple-checking 24..g5! (a move Shirov and Moro would play in three seconds) before deciding there was nothing Anand could do to punish the pawn pushes before ..d4 swapped everything off the board. As usual, the pushing of the IQP led to equality and they agreed the draw a few moves later on move 29.
An acceptable result for both players. Anand apparently didn't want to try too hard to kick Kramnik while he was down, instead choosing to play for two results and psychologically consolidate after his big win. The 5.Bf4 system has been played by Ivanchuk, Topalov, and Carlsen recently so it's stodgy reputation is a little unfair. Still, as with the colors reversed in game 1, it's hard to imagine Kramnik losing this sort of position. He prudently dodged any Anand surprises in the main line with 11..Bf6 (which he played against Kasparov) with 11..Bf5. Anand tried to get something going with his queen on the queenside, but there wasn't anything there. Kramnik's fine ..h5, ..g5, ..g4 plan allowed him to get in ..d4 with a fire sale and a draw. The champ's 23.Bh3 looks strange, but Nick showed how the bishop can run into trouble on f3 in some lines. Still, it's worth a second look since Black drew easily in the game.
Overly cautious from Vishy? I don't think so. The game doesn't look like much but this line has been fairly popular lately with aggressive players going for a win. Kramnik simply responded well to neutralize the threats and a draw was the natural result. The match is 1/3 over and Anand leads 2.5-1.5 heading into the second rest day. Kramnik comes back with white on Monday with some problems to solve after he was bitten hard by Vishy's Meran prep in game three. Official site.
I think that if you win with the black in a match like this then you are in a difficult position in the next game because you are supposed to try for a win again but you are not phychologically ready for trying hard.
I won't be surprised if Kramnik presses hard for a win in game 4.
Anand should go for 1.e4 today.
I have a feeling we could be in for a routine draw as players try to get things in control after the excitement of the last two rounds. On the other hand Kramnik may be in a combative mood and essay the Sveshnikov. Anand I think will almost certainly play P-K4.
>Anand I think will almost certainly play P-K4.
Anand himself isn't yet sure what it would be best now. I guess he will wave "to 1.e4 or not to 1.e4 ?" all this morning and made his mind only seconds before the clocks are started.
I think he will go for 1.d4 again, to be able to apply some pressure by avoiding the Petroff and the Wall.
I hope it will be 1.e4 and a fighting Spanish. I mean, Kramnik is like in an awful situation here.
Kramnik will try to equalise all forthcoming blacks and to beat back with white.
The fight is not yet over.
But I think Kramnik's mistake was seemingly to only hone his technique, which is superb and sometimes mind-blowing, and not to try his hand more often at very sharp lines in tournaments...
I still think Kramnik has a chance to equalize...
If I recall correctly, Kramnik did go through a phase of playing tactical openings - people on this very board made fun of the results. No?
"Cheap Shot Dept.: Did Kramnik forget to click the link on the left before game three?" lol
I also believe in 1.e4 today. So Anand will probably play 1.d4.
Derek, I think that was when he was having health problems and was under-performing with whatever he played.
And maybe Najdorf is not his piece of cake, but one does not HAVE TO play Petroff or Berlin. I guess nowadays whichever line you choose in Spanish with black you get descent play. I don't think Kramnik can win a match with avoiding anything complicated and trying to grind Anand in an Exchange Slav endgame. Ain't gonna happen.
P.S. GO Vladdy GO!
Now we are in the small positional advantage department, where Kramnik's skills as a defender are supposed to be best. It will be interesting to see how Anand will handle this. Can he fight for the full point? Or should he save his energy for a better opportunity?
I think this kind of positions don't work against Kramnik. Not even worth watching.
We will have an interesting game tomorrow I hope :)
Looks like a clear draw. Both sides will be relatively happy.....
Playjunior: I agree. This is game 1 with reversed colours.
There is an isolated d pawn, and not much counterplay. Maybe d5-d4 will be possible at some point, maybe not.
I would feel very uncomfortable if I had to defend this.
Kramnik plays for win this.
Yeah, Kramnik at his bloodthirsty best.
Wow, g5 now.
Probably it will all add up to equality in the end, but what a way to find/create counterplay.
Anand's 25.Rad1 is criticized by commentators as allowing g4 (Qc3 preferred).
Draw.
As expected a tame draw. I guess both players were exhausted after yesterdays heroics. Now the next "duo" will be interesting - what has Kramnik prepared and will Anand shift back to e4 ?
This was a typical Kramnik game in which he dried out the position effortlessly, taking no chances. So much for the folks who predicted that Kramnik would throw out his match plan, and play a "balls to the walls" style. That's just not Kramnik's personality when he still has four Whites left.
The open question is whether Anand actually wanted a game like this, or if he expected something different from Kramnik.
>That's just not Kramnik's personality when he still >has four Whites left.
it may cost him dear, he neither worried when was behind Anand during Mexico, he kept drawing and repeating that he still had (theoretical) chances to catch up...that's a bad attitude, the more time passes as this the more one gets indifferent to the forseeable failure
>it may cost him dear, he neither worried when was behind Anand during Mexico, he kept drawing and repeating that he still had (theoretical) chances to catch up...that's a bad attitude, the more time passes as this the more one gets indifferent to the forseeable failure <
But the difference is, in a tournament one can keep drawing and your opponent can keep winning, extending his lead. This is a match, and when Kramnik draws his opponent does also. I think Kramnik won't need to go for broke until Game 10-12. If he can keep Anand's lead to 1, there's no need to panic.
In some of the computer commentary I saw, after the opening, Anand was slightly ahead, with computer evaluations as high as 0.60 when he blockaded the Isolated Queen's Pawn. However, the computer evaluations shifted, and even favored Black later, by -0.20 at the end, when the IQP was dissolved, and the pieces traded off.
Does this say Kramnik outplayed Anand through the midgame, by almost a pawn's worth (delta 0.80)? Or rather, does it say the computer mis-evaluates the isolated queen's pawn?
tja
"I think Kramnik won't need to go for broke until Game 10-12. If he can keep Anand's lead to 1, there's no need to panic."
Anand has a huge plus score against Kramnik at rapid chess. Because of this, a lot of people are regarding a tie in the classical games as favoring Anand. If you subscribe to this theory, then you could say that, in essence, Kramnik is already down two. And that would imply that he cannot wait till game 10 to start pressing hard.
The latter. Only the most optimistic or misguided program would put the eval that high at any point, I'd say. The longer you let the comp think the lower that eval will probably go. Or just walk through its suggestions for 5-10 moves giving it 10 seconds or so on each move. It's eval will almost surely drop steadily.
Not to say that Anand played perfectly. It's likely he had chances to keep the game alive better than he did, otherwise we'd have to admit Kramnik just refuted 1.d4. But it's not a very ambitious line and I'm surprised to hear an eval that high from any popular engine. Black always seems to be able to either back up ..d4 or White has to take the knight on e6, which eliminates the isolani. None of my engines get above 0.25 or so, which is basically margin of error.
ChessOK.com's powerful Rybka 3 at pretty high depth gave +0.4 at the highest, I think -- if Anand had played 21.Rac1 instead of 21.Rd4, or 20.Rac1 immediately.
I've no idea. It seems like White might have a little something. At any rate it seems that would have made Kramnik's plan harder to carry out.
Let's face it: very convincing WC match with two great players.
I left Rybka 3 on my quad for around 10 minutes with two lines and it gives 21.Rac1 0.32 and 21.Qb4 second choice with 0.07. Depth 21.
Mostly I feel the position is the sort that requires great patience and subtlety for White to make any progress. Just keeping it alive by preventing ..d4 is difficult enough. It's the sort of position I could see Kramnik (Karpov, Andersson, et al) enjoying much more than Anand.
You need to bear in mind that engines know nothing about the match situation, the personality of the players, and so forth.
You also need to bear in mind that a 0.4 edge means that the engine sees no concrete way of winning. Those four-tenths of a pawn are an abstraction. Anand knows something the engine doesn't: that there is practically zero chance of Kramnik misplaying such a position.
So when Anand chooses a 0.07 move over a 0.40 move, it does not necessarily mean Anand has made a mistake.
There seems to be a change of policy at the playchess server up until game 3 the broadcast area had been accessible for guests. Suddenly starting with game 3 the server blocked access and says you need a higher rank. I dont know if this is just for the WCC or they are stopping access to broadcasts from now on.......
That Rybka is not necessarily right is obvious. It seems that the players did criticize 21.Rd4 though. Haven't seen the press conference but Dennis Monokroussos writes "The players agreed that Black was always fairly comfortable and never had any real problems. They seemed to agree that each player had one moment when they could have played slightly more accurately, but that it didn't amount to much in either case. Black should perhaps have played 18...d4 rather than 18...Nc5, while White should have preferred 21.Rac1 to 21.Rd4."
Agree about Kramnik feeling more at home in this kind of technical position (from the White side) than Anand. Thus maybe we see the downside of the latter's switching to 1.d4. Even though he got a slight advantage, he was not able to do anything with it. The _type_ of position is in Kramnik's favour, relatively speaking. At least in the Nimzo game it got messy. Maybe should have repeated 3.Nc3.
Andy I noticed that too, and was quite disappointed with Chessbase. I've been watching at ChessCube.com since game 3, which has varied from 30 min delay to no delay (today). Plus I jump all over to various blogs and chatterhouses to get widely varied commentary. (Mig's the best, of course!)
Yah, the fact that Kramnik enjoys playing small disadvantages is a problem for Anand in his switch to 1.d4. If Kramnik were forced to play the Semi-Slav every time out Anand would be in clover. But there are just as many 1.d4 lines that lead to these grind-required positions as 1.e4 lines, and Kramnik's going to know them better as well as feel more at home with them stylistically. Doesn't mean he's invincible there, but he's not going to play sharply with black unless he has to, or unless it's down to game 11 and he's still behind.
Maybe Vishy should do what he did against Karpov back in 1998, play the Tromp! Sure you end up worse half the time, but it's usually an original, dynamic worse!
>Maybe Vishy should do what he did against Karpov >back in 1998, play the Tromp!
that would be psychologically self-defeating since he lost the match with Karpov, and it would make Vishy appear like a coffehouse player improvising openings
Prediction poll openfor the second leg (games 5 to 8).
Please post your guess in the Ninja's Board:
http://www.chessninja.com/boards/ubbthreads.php?ubb=showflat&Number=64202#Post64202
Rumours of Kramnik's death are greatly exaggerated. I'm sure he's got plenty in the bag, despite the Champ's great start..and what's this about Leko being a poor choice cos he mainly plays e4? Players of this calibre can contribute to any opening under the sun, and probably a few others too. Now for which game will the Vish switch to e4, if at all? Prize of a grin for the winner.
Gerhard wrote: "Kramnik will try to equalise all forthcoming blacks and to beat back with white.
The fight is not yet over.
But I think Kramnik's mistake was seemingly to only hone his technique, which is superb and sometimes mind-blowing, and not to try his hand more often at very sharp lines in tournaments...
I still think Kramnik has a chance to equalize..."
Yeah, but Vishy is very comfortable with his lead now. And Kramnik, in order to score, will need to play aggressively, and that is where Vishy is waiting for him to give another blow because apparently Kramnik, the latest edition, does not keep up with the pressure of the real fight and calculations.
In the line of a previous stringtheory's comment.
I wonder if Anand will play d4 in all his white games... Let's imagine it, maybe since six months ago Vishy has been homecooking novelties for d4 openings, how many goodies would he had stored up!
In this light, his last tnmt in Bilbao was only to fool Kramnik pretending loyalty to e4.
This way and as a side effect, he would neutralize most of the Kramnik preparation, just the part he devoted to e4 study; how much do you like? 60%, 70%, 80%? What a wasteland for the Kramnik's team!
I was really sorry to see Kramnik decide to go in for the draw with Rd4 instead of pressing with Nd4. Looked like there were some interesting lines with ... h4 etc. and seeing whether there was any way to expose the White King a little. But if hg then Rg3 and I guess White gets a little attack back on Black's King.
Mig,
I think I saw the .60 evaluation on Chesscube as a Fritz-10 foray around move 15 or 20. Whether that is a bit high, I don't know, I'm not silicon-based!
Two things lead to my question, though.
1. Is the IQP one of those places where we can say organic brains understand chess better than the computer? Is the idea that 'the IQP is always dissolvable' one of those long-range plans that computers don't understand, is it beyond their event horizon?
2. I've long been intrigued by Kramnik's ability to out-play fellow GMs in the middlegame. I thought I saw this in his match with Kasparov - if Kramnik survived Kasparov's preparation, he out-played him for the rest of the game.
There is also a Misha Savinov interview with Peter Svidler where Peter says if he reaches an equal midgame, he fears no GM except Kramnik. This is quite a compliment from a fellow 2700.
Kramnik has a special skill that enables him to win from near-equal positions. Even with your computer, the evals of Game 4 shifted from move 20 to the end by delta 0.5 - not a trivial amount?
I hope this portends Kramnik playing well - but he'll need more than 0.5 to win a game!
tjallen, I guess computers understand pretty much everything these days, except maybe some endgames.
+0.25-0.4 means that black has a structural weakness on d5, and white can prevent d5-d4. As long as this is the case, white has a small plus.
For the computer, making no progress is not a problem, as long as it retains the structural advantage. And that's exactly the way one is supposed to play it as white: prevent d5-d4 and try to induce and error.
The computer's evaluation is more appropriate for my level of play (~2000) than for Kramnik's.
The computer sees the problem in Black's pawn structure. Since there is no concrete line how Black might get rid of the problem, it assigns a pre-defined value. Then this value is added to all the other indicators for the position, which are quite equal, and that's it.
Kramnik's knowledge is deeper. He knows from tons of similar games that in this particular type of position, there is no way White can keep up the advantage in the long run. If White focuses too much on preventing d5-d4, his position will become static, and prone to attacks. On the other hand, if White can't prevent d5-d4, the advantage is only temporary, and minimal. (In computer units: 0.0x).
>The computer's evaluation is more appropriate for my level of play (~2000) than for Kramnik's....
Why would Rybka-3200 be more appropriate for 2000 than for 2800 ?
>Kramnik's knowledge is deeper. He knows from tons of similar games that in this particular type of position, there is no way White can keep up the advantage in the long run. ..
In "tons" of games white does exactly that, look at some "classic" games of Karpov (Karpov-Uhlmann in Tarrasch-French, or Korchnoi-Karpov QGD-Tartakover with IQP-white) to see how difficult can be to escape losing the IQP.
Both Kramnik and Rybka know that there some good chances to dissolve the pawn, say "4 out of 10 games" p=40%=0.4, and if the value of the pawn v=1 , then white is better by p*v=0.4..an objective assesment of the chances based on the previous experience
(every valuation of a chess position can be interpreted as giving the chances to win the game, )
> Why would Rybka-3200 be more appropriate for 2000 than for 2800 ?
Because Rybka and me cannot tell if a particular IQP position will be easy to hold, for a defender who is up to the task. To us, all those positions look alike.
Kramnik can tell this difference, because of his superior pattern matching. He chose to play this line, because he knew: In the particular IQP position, he will be able to achieve d5-d4, or get adequate compensation, 9 times out of 10.
(Of course, Rybka sees much more concrete lines than I do, or Kramnik does, and therefore will be able to correct his evaluation, as soon as all the lines end in simplification or obvious compensation.)
This would be an interesting question:
How much difference must there be between the move Kramnik/Anand makes, and the move Rybka recommends, for one to be 95% certain that Kramnik/Anand made the wrong move? I am not sure of the answer, but I am quite sure it is more than 0.40 of a pawn.
>How much difference must there be between the move Kramnik/Anand makes, and the move Rybka recommends, for one to be 95% certain that Kramnik/Anand made the wrong move? >
Except trivial situations (as Bd3 of Kramnik or forced mate after the not-made Bxd3 of Anand) nobody, not even Rybka, knows what is "the right move", it would be equivalent with having sloved chess.
Different than than Rybka means just worse than
the 3200-Rybka, not the (absolutely) wrong move.
Well...no, I think we can go farther than that. For example, if there's a 3.00 difference between Rybka's recommendation and the move on the board, I'd say there's a better than 95% chance the player has blundered. That's why I said 95%, not 100%, and bearing in mind that not all moves we consider "blunders" lead directly to mate.
Generally, if the engine's evaluation is rooted in a (mostly) forcing tactical line, then its choice will trump the GM or super-GM's choice an overwhelming majority of the time. (There will still be a tiny percent of cases where the engine missed something - see for instance Shredder's loss to Ponomariov in Bilbao, was it?, a few years ago.)
Such forcing tactical lines usually cause "big" swings in the evaluation. That's the logic behind Marc's post. As to how we could rigorously define "big," my hunch is the dividing line is inherently fuzzy. I'd also strongly suspect that the right number (difference in evaluation that might signal the presence of a forced tactical line vs. the engine's often faulty - by 2700 standards at least - positional judgment), would differ by some fraction depending on which engine was used.
As a rule of thumb, I tend to view an evaluation shift somewhere in the 1.50-2.00 range, as an indication the computer sees a tactical refutation. As noted above, such judgments made by an engine are almost always reliable, even vis-a-vis GM choices (assuming of course the engine was allowed enough calculation time that its evaluation won't reverse as soon as looks 1 ply deeper).