(1) Topalov,V (2780) - Kramnik,V (2799) [D43]
Corus A Wijk aan Zee NED (9), 22.01.2008
This position is so frequently seen nowadays that it would hardly deserve a diagram under normal circumstances. The real tabyias arise slightly later, but in the present game White deviated from the approved path abruptly.
[During the Elista match, Kramnik played 4...dxc4
in all the 3 games where this position arised, obtaining entirely satisfactory positions out of the opening. After the match, he started employing the sharper Moscow/Anti-Moscow systems, where he seems to feel at home with both colours.]
[Between players that do not shake eachother's hands before the game, the positional 6.Bxf6
is out of question, of course.]
If such surprises, in the true spyrit of the King's Gambit, can arise from once in a while still, we are quite far from the exhaustion of our favourite game still. In the press conference Topalov said Cheparinov found Nxf7 three years ago, and they have been saving and developing it ever since. A huge effort indeed, but the resulting positions cannot be analized properly without considerable investment of time. All engines would consider that Black is just winning in all the lines, which can be quite discouraging for the faint-hearted.
Contrary to the almost unanimous opinion, the move is not a novelty, though. It had been played for the first time by the Romanian correspondemce player Miron Nacu two years ago, as Marius Ceteras (among others, captain of the Romanian Ladies Olympic correspondence team) kindly informed me.
Black is a full piece up and there is no obvious way for White to get at least part of his material back. However, the permanent exposure of the black king to White's pieces' attack justifies the sacrifice from abstract point of view.
[Only this move is new. Both correspondence games continued with 13.f4
(If we spoke about the King's Gambit, this move is quite natural, even if played with a delay of more than 10 moves) 13...b4
(This looks suspicious. Later, Black tried to improve by evacuating the king from the centre with 13...Kg8
when after 14.e5
White's kingside pressure eventually proved sufficient for reaching a draw in Brodda-Zidu, ICCF 2007.) 14.f5
(There is no immediate refutation for 14...bxc3
, but the presence of the king in the centre would be a permanent source of worries.) 15.Bxc4+
with strong initiative for the considerable material disadvantage, Nacu-Brodda, ICCF 2006.]
The next phase of the game consists of natural developing moves, as if nothing extraordinary had happened. Quite logically so, because development should be the highest priority in the first phase of the game no matter what.
[This is the first move after which engines switch their evaluation from better for Black (already not winning, though) to at least equal for White. Which does not mean anything really, it might just be a consequence of the horison effect.
Kramnik's move was probably dictated by the desire to establish some communication between the opposite wings (something that was possible only along the back rank until now). From the computer's suggestions, I would consider 17...Rhg8
as logical, because it develops the last piece, anticipating the infiltration of the white queen at the same time.]
Black's material advance has been reduced to the minimum, but Kramnik probably relied on his stability on light squares as well as on the optical dispersion of White's forces all over the board. The queen and the knight are placed on active positions, but they are not sustained by the rooks, restricted to back rank activity for the time being. At the same time, the g3-bishop is somewhat out of play. Its only function is to keep the essential e5-pawn protected.
The king could not go to c7 because of Nc5, with an unpleasant pin. However, the relatively best king retreat to c8 is not without drawbacks either. White is not at all forced to hurry with the check on d6, when after ...Kc7 Black would reach relative stability on the queenside.
Taking advantage of the fact that the d7-knight is hanging, too, Topalov will leave his own knight on b7 for several moves, keeping Nd6+ in reserve. This is a typical way to increase the force of a determinedpiece. From b7, the knight controls the c5- and d8-squares, but also, indirectly, all the squares that can be reached from d6 in one move. After a premature knight jump to d6, the former area of influence would be lost.
During the game, it is hard to foresee all the cases when a difference would be made by delaying the move Nd6+, but while this possibility will be available anyway, delaying it will (at least theoretically) restrict Black's choices.
Threatening Rxc4! Black has obvious problems maintaining the queenside closed.
[Aiming to maintain the c-file closed. 23...Nxc3
would allow White to coordinate the action of most of his pieces with 24.h3!
would leave both e6- and b4-pawns undefended and White would immediately attack them with 25.Qe7!
when Black would have to find a form of giving up the queen for (probbaly) insufficient compensation, since the natural line 25...Nxd1?
(Black should capture on d6 already. The text move aims to keep the knight and the e6-pawn defended, which is essential in order to avoid decisive attack.) loses the queen for nothing to 27.Nb5+
; In case of 23...bxc3
White has a wide choice, but I like 24.Rfd1
best, because it brings the last piece into play. The concrete threat is Rxd5 followed by Qe7 with a strong attack.]
After the recent structural modifications, Black's central knight has lost stability.
Black cannot afford to open the d-file and has to start chasing the enemy queen.
He could still have waited for one more move.
A first critical moment of the game.
White cannot evacuate his queen starting with 27.Qg6? because of 27...Nf4! 28.Bxf4 Rhg8! followd by 29...gxf4 with a strong counterattack.
Agreeing to the repetion of moves is out of question (they would have had to look into eachother's eyes in order to fix the draw in that case, but this would have been almost as humiliating as shaking hands!) which means, using the method of elimination, that White has to create a threat at least as strong as ...Rxf7.
[Optically speaking, the most natural decision. It is easy to establish that White will get ample compensation for his queen; no complicate calculation is required.
Objectively speaking, 27.h3!
might be better, though. This move was suggested by Garry Kasparov, who was following the game informally (phoning and discussing with people in between) on a notebook without an engine! In fact, the first sequence of moves is not difficult to calculate and I assume that Topalov saw it, too: 27...Rxf7
(The only way to maintain the material disadvantage within acceptable limits) 29.Nxf7
(This move is natural and would be the instant choice of most players. I have awarded it with an exclaim because in a certain line it will be essential not to have the king on the back rank.) 30...Nxc1
Players have reversed their parts and it is White who is a piece up now. However, with the bishop temporarily imprisonned on g3, the b-pawn, sustained by the rook and knight, seems to be very dangerous. Topalov must have evaluated this position as unclear, but further analysis proves that Kasparov's intuition did not let him down. White is able to generate a powerful and somewhat unexpected counterplay on the opposite wing, developing by one tempo faster than Black's simple plan. Here are some possible continuations (part of them provided by Kasparov himself, when confronted with a powerful chess engine by Frederic Friedel) 32.Rb1
(Black should not lose time. In case of an exchange on f4, the bishop will get into play just in time to keep Black's counterplay under control) 34.fxg5
also leads to remarkable play after 36.Nxe6+
. The only possible retreat on an apparently empty area of the board. After any other move, White would play Rxb3! Once again, the direct and indirect action of White's knight keeps under control a bunch of important squares. 37.Rf1
Establishing a nice net around the enemy king. 38...b1Q
(Again the only square. 36...Kd7
would lose the rook; while 36...Kb7
leaves Black without the threat ...Nc3.) 37.g5
Finally, Black has reached his optimal regroupment, but after 38.Rxb2
the pawn is unstoppable.]
[This is the second critical moment and... Black's only chance to save the game!
Kramnik played his last move quickly, apparently without considering any alternative to removing the rook from the attacked square.
By this moment, Kasparov felt somewhat frustrated by the fact that on the server nobody suggested 29...Qe2!
, which he considered to hold the position. The basic idea is similar to that behind his previous suggestion, 27.h3. Instead of parrying the threat Nxh8, Black creates a stronger one! Indeed, in case the knight captures on h8, Black takes on f1 followed by ...b2, with a likely draw by perpetual, because Wite's pieces are not communicating with eachother. Here is a (not entirely forced) line confirming Kasparov's evaluation: 30.Rc3
White is too passive to claim an advantage.) 30...b2
(The rook is instable along the b-file and will have to capture on b2 at some point. However, it is useful to distract from its actual square the knight before doing that. 33.Rxb2?!
allows Black consolidate on the kingside,, while his a-pawn could prove dangerous in the near future.) 33...Nxa4
Apparently, White is in some trouble, but he can maintain some initiative with 37.f4
(Otherwise, Black would have to fight against 2 connected pawns) 38.Bh4
, but the position remains fairly unclear.]
White has a material advantage already, active piece placement, far advanced pawns and the safer position of the king. Black is in big trouble.
[More accurate would have been 34.Rxb3
, keeping both e-pawns on board.]
There seems to be some hope for Black now, since there is no obvious way for White to improve his position.
But after this pseudo-active move, weakening the g5-pawn and allowing White regroup with gain of time, simplifies White's task.
Threatening mate in one.
The h4-pawn has little significance in this moment. Topalov prefers to use his bishop to sustain the advance of his passed pawn.
There is no satisfactory defence against Nc7+ followed by Rb7#. 1-0