A game from the US Open

[Date "????.??.??"] [White "Wilson, Matthew"] [Black "Watson, IM John"] [Result "0-1"] [FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"] 1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.e5 Nfd7 5.f4 Be7 { This unusual move order threw me off. Almost everyone plays 5...c5 here. } 6.Nf3 O-O 7.Ne2 c5 8.c3 Nc6 9.g3 cxd4 10.cxd4 f6 11.Bg2 Qb6 12.Qb3 Bb4+ { Something has gone wrong for White. At the very least, I lose castling privileges after 13...Bxd2+, since 14.Nxd2 drops the d4-pawn. } 13.Bd2 fxe5 14.fxe5 Rxf3 15.Bxf3 Nxd4 16.Nxd4 Bxd2+ 17.Kxd2 Qxd4+ 18.Kc2 Qf2+ 19.Kd1 { At first glance, it seems like Black is dominating. But after... } 19...Nxe5 20.Be2 Bd7 21.Rf1 Qxh2 22.Rc1 { ...White is suddenly coordinated! } 22...Bc6 23.Qe3 Ng6 24.Qxe6+ Kh8 25.Bd3 { Over the next several moves, a key tactic is 25...Re8? 26.Rxc6! Rxe6 27.Rc8+ with mate to follow. } 25...Qh5+ 26.g4 Qh3 27.Kd2 Qh2+ 28.Qe2 { Is 28.Kd1 better? } 28...Qd6 29.Rce1 { After the game, John suggested 29.Bxg6 as an improvement, since it stops ...Nf4 } 29...Qb4+ 30.Kc2 Nf4 31.Qe7 { A blunder. I was counting on 31...Qxe7 32.Rxe7 Nxd3. Now 33.Kxd3? Bb5+ is bad, but my plan was 33.Rff7! and the doubled rooks on the 7th should provide enough counterplay. But John finds a better reply. After the game, we agreed that 31.Qd2 was superior. } 31...Nxd3 32.Qxb4 Nxb4+ 33.Kc3 a5 34.Re7 Nxa2+ 35.Kb3 Nb4 36.Rff7 Rg8 { It's over } 37.g5 d4 38.Rf2 Nd5 39.Re5 Rd8 40.Rh2 Be8 41.Rf2 Kg8 42.Ref5 Bg6 43.Re5 Bf7 44.Ka3 b5 45.b3 Nc3 46.b4 d3 *