Ulhumbrus: 9...g5? disturbs the King side pawns without necessity and commits Black to castling on the Queen side.
10 Bd2 obstructs the d file, and with it, the White Queen's defence of the d4 pawn. One justification for 10 Be3 instead is that on 10...Nxe3 11 fxe3 Black's f5 square seems a greater weakness than White's backward e3 pawn. Another alternative is 10 Bc1 eg 10...Be7 11 0-0 Qd7 12 Ng3 Nxg3 13 fxg3 and Black's f5 square remains weak.
11 Qc1 neglects White's development, and although it induces Black to disturb further his King side pawns, Black is going to castle on the Queen side and the weakening effect of the move ....f6 may be insufficient to justify the loss of time for development spent on the move 11 Qc1. An alternative is 11 0-0 Qd7 12 Ng3 0-0-0 13 Nce2 and Black's game seems not have quite attained safety.
12 Nd1 is consistent in one way, occupying the aquare d1 which White's Q has vacated. However it not only neglects White's development but actually undevelops the N. 12 0-0 may be better.
14 Ne3 makes a target of the N for the advance ....f5. On 14 0-0 White may still have the better of it eg 14...Qd7 15 f4 0-0-0 16 fg fg 17 Ne3
15...f5 threatens 16...f4 followed by 17...f3. White seems to have no choice but to block Black's f pawn by 16 f4.
Instead of 17 0-0-0, 17 0-0 may be better because then White has some initiative of his own on the King side. After 17 0-0-0 Black seems to have all the play and White none. A part of the reason for that is that White is unable to find good posts for his Knight and bishop. White is unable to win Black's f pawn, and that pawn advances to f4 and becomes a weapon instead of a target, a weapon which supports Black's attack.
It does not take much to transform an advantage into a lost game. The move 9....g5? contracts a permanent disadvantage. It is such a serious mistake that it takes several errors on White's part, perhaps five errors, to transform finally White's advantage into a lost game.