Annotations by Patrick McCartney.
1.f4 d5 2.Nf3 Nf6 3.e3 Bg4 4.Be2 Nbd7 5.d3 Black has no problems here gaining equality, and White should instead play 5.O-O Bxf3 6.Bxf3 e5 7.d4 with a slight pull.
5...Bxf3 6.Bxf3 e5 7.Nc3 While trading towards the center is usually a good thing, in this line, it tends to open up the dark squares for Black more than anything else
7...c6 8.0-0 Bd6 9.e4 If effect preventing Black from having the luxury of opening the b8-h2 diagonal as 9...exf4? 10.exd5 is far superior for White while after 9...d4 10.Ne2, Black can only open up the diagonal at the cost of trading off the dark-squared bishops or else relinquishing control of it by moving his bishop away. After the move played in the game, the attack on the bishop gives White the needed tempo to close off the diagonal.
9...dxe4 10.dxe4 Qc7 11.f5 Bc5+ 12.Kh1 0-0-0 13.Qe1 h5 14.Na4 Be7 15.Be3 Ng4 16.Bg1 Bg5 17.c4 In hindsight, I don't really like this move for White. It gives Black the d4-square whenever he wants it, and doesn't help to further develop White's pieces. I think better here would have been 17.Rd1 or 17.Qc3. I don't like 17.Qb4 though because Black can then force the Bishops off the board with 17...Be3.
17...Nb6 And in return, I don't like this move for Black. It allows White to trade off the one piece that is hanging out there doing almost nothing. In my opinion, I think Black does best to stall a move and play 17...Kb8, waiting to see what White does next, and use that to determine whether to retreat the g4-knight to f6, or perhaps attempt to break open the center with ...g6, even with White having the Bishop pair
18.Nxb6+ axb6 19.Rd1 Rxd1 20.Bxd1 Nf6 21.c5 White appears to have a fairly significant advantage here.
21...b5 22.a4 bxa4 23.Qb4 Probably the wrong way to go about it. 23.Bxa4 intending 24.b4 is stronger, and should keep some advantage. With White's time-cosuming approach to recapturing the pawn, Black once again equalizes.
23...Rd8 24.Qxa4 Once again, White should take with the Bishop.
24...Rd2 25.Rf2 Rd4 26.Qc2 Qa5 27.Rf1 Rxe4 28.Bxh5 Re1 29.Rxe1 Qxe1 30.Qd1 Qxd1 31.Bxd1 Bc1 32.Bb3 I don't like 32.b3 Ba3 for White at all. Black will evenually net a pawn anyway, and so I'd rather give him the b-pawn, and spend the extra time getting my king out and advancing the kingside pawns.
32...Nd5 33.g4 Bxb2 34.h4 f6 35.Kg2 Bc1 36.Kf3 Nc3 37.Be3 Bxe3 38.Kxe3 Kd7 39.g5 Ke7 40.Kd3 Nb5 After the game, I asked Andrey why he didn't play 40...Nd5, and he pointed out that 41.Bxd5! cxd5 42.g6 forces Black to keep his king on the Kingside and can't make progress due to White's extra pawn on that side. Note that the same thing would apply back on move 38. 38...Nd5 (instead of 38.Kd7) 39.Bxd5! cxd5 40.g5! followed by 41.g6 leading to the same thing. Note that in this case, 40...fxg5 actually loses to 41.f6.
41.Be6 Nd4 Black pointed out that his other major candidate move here was 41...g6, but after my 42.Bc8 gxf5 43.g6! response to that, all we could find was 43...Kf8 44.h5 Kg7 45.Bxf5 Nd4 46.Bc8 Nb3 with a similar perpetual to that which happened in the game.
42.Bc8 Nb3 43.Kc4 Nd2+ 44.Kd3 Nb3 45.Kc4 Nd4 46.Kd3 Nb3 If Black tries 46...Nf3 (or 44...Nf3), then 47.gxf6 Kxf6 48.Bxb7 Nxh4 49.Bxc6 Nxf5 50.Bd7 Nd4 51.Ke4 and it's hard to see Black making progress once the White King gets to d5. 1/2-1/2