1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.f3 c5 5.d5 b5 6.e4 0-0 7.e5 Ne8 8.f4 Hard to believe old Nimzowitsch had this in mind when he hatched his opening idea, the Nimzo-Indian. Black's idea is still quite hypermodern though, and is remniscent of the 4-pawn variation of the Alekhine, where white is encouraged to push, push, push, and see his pawn center get ambushed from all sides. Note that at the highest level, Mamedyarov, one of the elite's most adventurous openings players, essayed this against Fressinet, though the French player took on c4 instead of castling.
8...exd5 This move leaves played opening theory, but once again is part of analysis found in the Rybka4 opening book. Shirov is clearly familiar with it, as his sacrificial play against Aronian, officially a TN, was also in it. [The book recommends 8...bxc4 instead of the played exd5, though both are in it.]
9.cxd5 d6 10.Nf3 c4 This is no longer part of the analysis. [10...Nc7 had been the indicated continuation.]
11.a4 Two can play at the undermining game.
11...Bg4 12.axb5 Nd7 13.e6 fxe6 14.dxe6 Nb6 Black feels no hurry to take on e6, as it cannot be protected at the moment, and instead secures a good square for the knight where it supports c4 and d5.
15.Be2 Nc7 Another well placed piece where it exerts a multifunctional role of attacking e6 and b5 while also supporting an eventual d5.
16.Ng5 Not the best, though it looks attractive. White cannot really protect e6, and really needs to look after his development before his weaknesses become his downfall. [16.0-0 was better, after which play might continue 16...Bxc3 (16...d5 17.Nd4 Bxe6 18.Be3 is good for White.) 17.bxc3 Nxb5 18.Bd2 Bxe6 19.Ng5 and the position is balanced.]
16...Bxe2? The engines are oblivious to this error, even approving it, declaring Black's next move as the real blunder. Analysis indicates this is in fact a horizon effect, therefore the error came earlier. [This was Black's chance to rid himself of the e6 pawn and achieve a dynamic equality with 16...Bxe6! 17.0-0 (Or 17.Nxe6 Nxe6 18.0-0 d5 19.f5 Nc7 20.Be3 ) 17...d5 18.Be3 ]
17.Qxe2 d5!? This move is flagged by the engines as a big mistake, mostly because of a horizon effect, and may actually have been the lesser of two evils. [The initially suggested improvement 17...Qe7 still runs into serious problems after 18.0-0 and now if Black tries to boot the knight with 18...h6? (The alternative 18...Bc5+ 19.Kh1 h6 also runs into 20.f5! hxg5 21.Ne4 Qe8 22.Nxc5 (22.Bxg5?! would let Black neutralize the attack by returning the piece with 22...Nxe6! 23.fxe6 Qxe6 24.Rfe1 and Black should survive.) 22...dxc5 23.Bxg5 Ncd5 24.f6! Nxf6 25.e7! Rf7 26.Bxf6 gxf6 27.Rae1 and Black is lost. The threat of Qg4 and Rf3 are fatal.) White has the very strong sacrificial continuation 19.f5! hxg5 20.Ne4 d5 (20...Nxe6? 21.Nxg5 ) 21.Bxg5 Qe8 22.Ng3 Be7 (Not 22...c3? 23.bxc3 Bxc3 and White gets a strong attack after 24.Rac1 Bd4+ 25.Kh1 Rc8 26.Qg4 ) 23.Qg4! and White has a strong attack. For example 23...Qd8 24.Bxe7 Qxe7 25.Nh5 and White is not only threatening f6, but will also bring the other rook into play with Ra3-f3 with deadly effect.]
18.0-0 Qf6 19.f5 Rae8 20.Rxa7? A serious mistake that gives Black chances to survive. [20.Be3! would prevent the tactical trick played in the game.]
20...Nxe6! 21.Nxe6 Rxe6! 22.Qf2 [The point is that White loses if he takes with 22.fxe6 because of 22...Bc5+! 23.Kh1 Qxf1+ 24.Qxf1 Rxf1# ]
22...Qe5? A mistake now, placing the wrong piece on e5, but time trouble was already looming its nasty head. [Black had to play 22...Re5 23.Be3 (23.g4? would be a serious mistake now due to 23...d4 24.Na4 (24.Qxd4 Bc5 ) 24...Nxa4 25.Rxa4 Rxb5 and Black is better.) 23...Rxe3 24.Qxe3 d4 25.Qe6+ Qxe6 26.fxe6 dxc3 27.Rxf8+ Kxf8 28.bxc3 Bxc3 29.Kf2 and Black should make it.]
23.g4 Rg6 24.Qg2 Rgf6 25.Bf4 Qd4+ 26.Qf2 Qxf2+ 27.Kxf2 d4 28.Ne4 R6f7 29.Rxf7 Rxf7 30.Ra1 h6 31.Be5 d3 32.Bd4 Rb7 [32...Nd7 33.b6! ]
33.h4 Bf8 34.Rc1 Nd5 35.Rxc4 Rxb5 36.Rc8 Kf7 37.g5 Ne7?? A final blunder that ends it on the spot, but time trouble had its hand. [37...d2 would have offered better chances, but with seconds on the clock, it was far from obvious. 38.Nxd2 and now Kramnik could have played 38...Ne7 ]