Marshall suggests that 5.Bb5+ or 5.Qxd4 were likely better here. The move chosen transposes to an important line of the Urusov Gambit which usually arises after 1.e4 e5 2.Bc4 Nf6 3.d4 exd4 4.Nf3 d5 5.exd5.
This move is the most active. White can also pursue an advantage by the quiet 6.Bd2 Bxd2+ 7.Qxd2 O-O 8.Nxd4 Nxd5 9.O-O Ne7 10.Rd1 with a slight plus in development, or the tricky 6.Kf1 inviting Black to play a gambit after 6....O-O 7.Qxd4.
The Queen check is important for supporting the Bishop at b4 here, which otherwise would be subject to attack by Qa4+. It is also rather awkward for White at this moment since to block the check he must retreat his Bishop, exchange Queens, or move his King, none of which is particularly appealing.
Marshall writes that "If now 7.Qe2 Qxe2+ and Black stands well for the ending because White's pawn position is somewhat weak."
Not 8.Nxc3 O-O 9.O-O c6! = Estrin and Panov. White needs the c-pawn to support his center.
Some writers blame this move for White's troubles, but it is actually the next that is weak. A different idea is 10.Bg5 when White did well in Skatchkov-Lopatskaya, 1996, after 10....c6 11.c4 Rd8 12.Nc3 Bb4 13.Qb3! Bxc3 14.Qxc3 cxd5 (14....Qxe2? 15.Rfe1 +-) 15.Rfe1 Be6 16.cxd5 Bxd5 17.Bc4 Qc7 18.Bxf6 gxf6 19.Qxf6! Bxc4 20.Qg5+ Kh8 21.Qf6+ Kg8 22.Re5 Nc6 23.Rg5+ Kf8 24.Re1 1-0. But Black's best is probably 10....Re8! to control the important e-file, after which White can play 11.Bd3 Qd6 12.c4 Bg4 13.Nc3 with a good development and chances of play on the queenside with Rb1. Alternately, White could play 13.Nbd2 transposing to the game Kreindl-Franz, Vienna 1998, which went 13....Nbd7 14. Qc2 h6 15. Bh4 Bxf3 16. Bg3 Ne5 17. gxf3 Nh5 18. Ne4 Qf8 19. Be2 Nxg3 20. hxg3 b6 21. f4 Nd7 22. Kg2 Qe7 23. Bf3 Bb4 24. Rh1 Nc5 25. Ng5 Qf6 26. Qh7+ Kf8 27. Ne4 Nxe4 28. Bxe4 Rad8 29. a3 Bc3 30. Rad1 Rd6 31. Bf3 g5 32. fxg5 Qxg5 33. Qd3 Qe5 34. Rh5 Qg7 35. Rdh1 Ke7 36. R1h4 Kd8 37. Rg4 Rg6 38. Rf4 Be5 39. Re4 Rg5 40. Rh1 Rg8 41. Kf1 Qg6 42. Qe3 Qd6 43. Reh4 R8g6 44. Rxh6 Rxh6 45. Qxg5+ Rf6 46. Rh8+ Ke7 47. Kg2 Bd4 48. Qg8 Rxf3 49. Qf8+ 1-0.
An unnecessary loss of time. White
chooses an overly passive set-up for his pieces. Instead, he needs
to centralize his forces beginning with the surprising 11.Nc3! since
Black cannot win material by 11....Bb4?! 12.Qb3! Bxc3 13.Qxc3 Qxe2??
because the Queen is trapped after 14.Re1. The game Gazivoda-Savic,
Correspondence 1979, continued here 11.Nc3 Ne4 (Better is 11....Bg4
12.Rb1 Bb4 13.Nb5!?) 12.Nxe4 Qxe4 13.Bd3 Qe7 14.Qb3 b6 15.Bb2 Nd7
16.Rae1 Qf8 17.Ng5 g6 18.Rxe8 Qxe8 19.Nxh7 Kxh7 20.Bxg6+ Kg8 21.Qh3
Kf8 22.Qh8+ Ke7 23.Re1+ 1-0.
Likely the losing move. Worse, though, was 12.Re1? Bxf3! winning material. Marshall suggests instead 12.Nbd2 Nbd7 with a roughly equal position. According to Marshall's notes, Pillsbury needed too much to win this game to help his standing in the tournament, while Marshall himself was content with a draw, which may explain Pillsbury's unwillingness to simplify the position. Pillsbury would finish in second place, a full two points behind Lasker and a half point ahead of Marshall, who could have taken second himself if he had defeated Maroczy in their last game together. A draw in this game therefore, ironically, would have more solidly guaranteed Pillsbury second place.
This leads to immediate disaster as White's King is compelled to go for a walk. Marshall writes that 13.Bxe4 Qxe4 14.Nbd2 would be playable for White though Black's two Bishops and initiative would give him the advantage.
A pretty move that keeps White's pieces from coming to the defense while preventing his King from retreating.
Marshall writes that 18....h5 is not as good due to 19.Qxe2!? Qxe2 20.Re1 Qd3 21.Re8+ and "despite Black's considerable material superiority the win is not easy with the queenside pieces bottled up."
Marshall gives these nice alternatives: 21.Qxd3 Nxd3 22.Rc2 g5!+ 23.Kg3 g4 24.Nh4 Ne1 25.Rc1 Rxd2 26.Bc3 Rc2! -+; 23.Nxg5 Ne1 24.Rc1 Rxd2 25.Bc3 Rd3+ -+; 23.g4 Nf4+ 24.Kg3 Rxg2#.
Some sources end the game here, which is just as well....