The Azeris seem to have come to this tournament full of surprises. First we saw Radjabov use the black side of a Catalan, an opening he had never used before, to neutralize Kramnik yesterday. Today we see Mamedyarov using this aggressive system for the first time in his life - or at least according to my records on official play.
6...e6 7.Bb3 b5 8.0-0 Be7 9.Qf3 Qc7 10.Qg3 0-0 11.Bh6 Ne8 12.Rad1 Bd7 13.f4 Nc6 So far we've been following several strong games, and the position is well-known to Najdorf players. Here the main move is Nxc6, as in Rublevsky-Bu.
14.f5 This rare move had been used once before in a world class game, about 10 years ago, and never again since. It's interesting, although a little strange, that Mamedyarov is repeating this line. Gelfand doesn't follow the same path as the previous game, which ended in a convincing black victory. The move itself warrants quite an explanation. Playing f5 in a Sicilian is always very committal: White weakens the important e5 square, preventing him from ever achieving an e5 break, as well as allowing Black's pieces (usually a knight, but in some rare cases a queen or a bishop) to occupy the fantastic defensive square e5. Some writers have gone so far as to say that if Black didn't have the e5 square for his pieces after f5, he would lose every Sicilian. White must therefore consistently attack on the kingside from now on until he obtains a material gain, or he will end up strategically worse. [14.Nxc6 Bxc6 15.f5 Kh8 16.f6 A spectacular pawn push, but a well-known one... 16...gxh6 17.fxe7 Qxe7 18.Qf2 And White has compensation for the missing pawn. A few games have seen this position, the most recent being the aforementioned Rublevsky-Bu, 2010 China (rapid).]
14...Nxd4 15.Rxd4 Kh8 This move has not been tried before. The game enters uncharted waters, and the position is full of possibilities. [15...Bf6 was Kasparov's choice, and he further added a "!" in his annotations to the game. Maybe Gelfand was unaware of the Kasparov game, or, more likely, he was afraid of an Azerbaidjani preparation in this line. 16.Rd3 Be5 17.Qg4 b4! 18.f6! g6 19.Ne2 a5! Morozevich-Kasparov (Astana 2001). A fantastic game which Kasparov won after 57 moves.]
16.Be3 [16.f6 is a very logical move. Mamedyarov may have rejected it because of 16...Bxf6 (16...gxf6!? is also an interesting alternative. Black obtains a strong pawn center in return for the lost exchange. However, White can respond by sacrificing material of his own! 17.Nd5!? A crazy move typical of the Sicilian. Black is forced to take this intruder, but opens the fourth rank to the rook on d4. It is almost impossible to give an evaluation of this position without lengthy variations, so let's just say that this position is very unclear.) 17.Rxf6 Qc5! Strangely enough, although both the f6 and h6 pieces are under attack, neither can be effectively taken. However, the rook on d4 is now pinned! (17...gxh6 18.Rxh6 Rg8 19.Qh4 Rg7 gives White some pressure, but might be playable. ) 18.Bxg7+ Nxg7 19.Qf2 Qe5 with a very murky position.]
16...Nf6 17.Qh3 A logical move, upping the pressure on this side of the board. However, it does remove the pressure that the queen on g3 was exerting on the h2-b8 diagonal, specifically pinning the d pawn. Black now achieves another of his strategical goals in the Sicilian: the d5 break!
18.e5! The exclamation mark is not so much for the strength of the move, but for its practical value. White probably already stood worse after a "normal" move such as exd5, so he tries an "all-out" approach to crash on the kingside, or die trying. [18.exd5?! exf5 And White has too many weaknesses, and no coordination. Black is clearly better.]
18...Qxe5 19.Rh4 Now Bd4 is a threat. Black responds with a cool move.
19...Rfc8! Forcing White to spend a tempo.
20.Kh1 [After 20.Bd4 Bc5 21.Bxc5 Rxc5 Black is obviously better.; 20.Bg5 was an interesting alternative, immediately hitting the h7 pawn. Black has a few resources. The more logical ones are: 20...h6 (20...Kg8 21.Bxf6 Qxf6 22.Rxh7 g6 The point is that the f-pawn is pinned due to Bc5+, so White cannot immediately continue his attack. However the position is still far from clear. 23.g4 (23.Kh1 Rxc3 24.bxc3 gxf5 25.Rf3 Bd6 is unclear.) 23...Rxc3! is complicated, but I like Black's chances.) 21.Rxh6+! The only move. White must not retreat! 21...gxh6 22.Qxh6+ Kg8 23.fxe6 Bxe6 24.Bxf6 Bxf6 25.Rxf6 Rc6 White clearly has resources on the kingside, but it's unclear if they are enough. His own king might become vulnerable and the bishop and knight are simply too far away to be of any help right now. Black is probably better in this position, but a lot of play remains after Rf3.]
20...Rxc3! Mamedyarov might have underestimated this strong, strong sacrifice from Gelfand. The sacrifice doesn't seem logical from the beginning: why is Black giving up a perfectly good rook for a knight that in this case had no hope of joining the attack on the kingside? The reason is control: Black will be able to neutralize White's threats, specially Bd4, once his queen gets to c3. Any exchange of queens from hereforth will favor Black, so it's basically a pin. Further, this sacrifice guarantees the burial of the bishop on b3, which will not see a bright future. A strong move indeed!
21.bxc3 Qxc3 22.Rd4?! Mamedyarov panics in a difficult position. Bringing back the rook does little for his position. This is typical of a failed Sicilian attack: even with the extra material it becomes very hard to move the White pieces as they have lost purpose. [22.Bd4 Qxh3 23.Rxh3 a5 24.c3 Ne4 is an ugly, ugly ending to defend. White's rooks have no targets and his structure is weak to say the least, but it may have been the best chance...; 22.fxe6 Bxe6 23.Qf3 Ne4 was also in Black's favor.]
22...a5! Gelfand swiftly punishes White's play. He doesn't spend time with tempting moves such as [22...e5 23.Rd3 Qc6 since 24.Bg5! gives white plenty of counterplay. Probably the point behind Rd4.; 22...exf5?! 23.Rd3 Qc7 24.Bd4 and White's bishops are more alive than Black would like them to be.]
23.Rd3 Qc6 24.c3 a4 25.Bc2 e5-+ Now that the bishop is no longer on the a2-g8 diagonal, the time is ripe to advance the center pawns. Black is strategically winning: he has a crushing center, better piece placement, more targets and better coordination. Mamedyarov must try to create some tricks, but time is running short.
26.Bg5 b4 27.Qh4 bxc3 28.Rh3 This pressure on the kingside would alarm any mortal, but Gelfand plays it cool and simple. He has now obtained a material advantage as he has three pawns for the exchange, and furthermore his center pawns will obviously roll over White in a few moves. Still, the h-file is open and his king seems to be in some danger. Not bothered with this, he simply transfers it to the center.
28...Kg8 29.Re1 [29.Bxf6 Bxf6 30.Qxh7+ Kf8-+ And White simply has no good moves. The king is perfectly safe on e7.]
29...e4 30.g4 White is trying some shenanigans with g5 after Bxf6, but it's just not enough.
30...Kf8 31.Be3 [31.Bxf6 Bxf6 32.g5 Bd8 and again, there is no good move for White. Black just controls too many squares and his pawns are too strong.]
32...Bxf5! Gelfand simply takes all the material that is thrown at him. In this case it doesn't even matter, since it's all about his pawn domination.
33.gxf6 Bxf6 34.Qh5
34...Bg6!? This is an interesting approach. Though Bxh3 was winning, Gelfand is not afraid of being down an entire rook! He has five full pawns as compensation, and furthermore, they are so strong! Very pleasing aesthetically.
35...Qxa2 The fact that Black has the luxury to pick off this pawn speaks volumes of his domination. d5-d4 will come in time, and White has absolutely no counterplay.
36.Bb1 Qc4 37.Qg2 a3 38.Ba2 Qc6 39.Rg3 Rb8 There are simply too many threats in this position, and not enough moves. Despite being up a rook, White's position could not be more hopeless. Gelfand wins a fantastic game that really brings out the very spirit of a Black side Sicilian: He coolly dealt with the kingside threat by transfering his king, he took care to eliminate and restrict White's minor pieces, and he showed how clumsy even an extra rook can be if the army supporting it has no direction. 0-1