1.d4 Nf6 2.c4 e6 3.g3 c5 4.Nf3 b6 5.d5 exd5 6.cxd5 d6 7.Nc3 Be7 8.Bg2 0-0 9.0-0 Re8 10.e4 [This position has occurred a few times on my database, but not this move. White was successful when he played 10.h3 preventing Black's next.]
10...Bg4 11.h3 Bxf3 12.Bxf3 The bishops could come in useful.
12...Nbd7 13.a4 a6 14.Bg2 Ne5 15.f4 Ng6 16.Be3 Qc7 17.Qe2 Bf8 Black seems to have White's standard e4-e5 break well under wraps for now. In view of the fact that there is no finachettoed black bishop staring down the long dark-squared diagonal, White decides to usurp Black's normal position of being in charge of play on the queenside.
18.Rfc1 Qb7 19.Rab1 b5 20.axb5 axb5 21.e5 With the white queen on the same file as the rook, there is some risk in making this advance now. [21.b4 c4 22.Ra1 keeps Black quiet and White may advance his kingside pawns in the near future.]
21...dxe5 22.f5 [22.d6 doesn't appear to gain much for White in view of 22...Qd7 and now: However, the computer comes up with the idea of keeping the f-file closed with 23.f5!? and the game is again unclear after (23.Bxa8 exf4 and the weak white kingside compensates Black fully for the lost exchange; 23.fxe5 Nxe5 24.Nxb5 and after (or 24.Bxa8 Nc4!© and the more useful white bishop is removed) 24...Rab8© (or 24...Ra2© the game is level on material and rather unclear) ) 23...Nf4 (or 23...Qxf5 24.Bxa8 Rxa8 ) 24.gxf4 exf4 25.Bxa8 Rxe3 ]
22...Ne7 [Overlooking the effect of having the queen and rook on the same file, which allows him to play 22...Nf4! 23.gxf4 exf4 when 24.d6 Qd7 25.Bxa8 Rxe3 transposes into the unclear variation C above.]
23.Bxc5? This looks like a miscalculation. [23.d6 Nc6 24.Qxb5 Qxb5 25.Nxb5 Rac8 26.Bxc5 leaves White safe pawn ahead.]
23...Nxf5 24.d6 [After 24.Bxf8 Kxf8 25.Qxb5 White had possibly overlooked (25.Nxb5 Nxg3 ) 25...Qa7+ 26.Kh2 Qf2 when 27.Qd3 Ne3 28.Qe2 Qxe2 29.Nxe2 Nxg2 30.Kxg2 Nxd5 leaves Black a pawn ahead this time.]
24...e4! 25.Qxb5 Nxd6 26.Qxb7 Nxb7 27.Bd4 Bc5 28.Rd1 Rac8 29.Kf1 Nh5 30.Bxc5 Nxc5 31.Rd5 It is understandable that Dan should want to pocket his gains by agreeing a draw against the top seed, but he does stand much better here, with an extra - protected and passed - pawn. [After 31.Rd5 Black plays 31...f5! intending (31...Nxg3+?? loses the knight to 32.Kf2 ) 32.Rxf5?? Nxg3+ ] 1/2-1/2