|DrGridlock: Bronstein includes this in "200 Open Games" and writes after black's move 21 ... g6:|
"During my preparation at home I dared not even dream of getting the position shown in the diagram. The weakening of Black's pawn chain could have best been taken advantage of by 22 Rac1 and 23 Rc7 after which there seems to be no defense against the threat of Qd7 and Qxf7!.
Did I see this move? Of course. What was wrong with it then? I wanted to spare my bishop. Black could have easily defended his f7 pawn by ... Rf8: I was not going to give away my bishop away for this rook! After the game, however, Ivkov and I established that this variation was favourable for White. In short, the move 22 dxe5? was a technical error."
Komodo 27 has a look at this position, and agrees with Bronstein that 22 Rac1 was a winning move for White:
click for larger view
Analysis by Komodo32 3 32bit (depth = 27):
1. (2.20): 22.Qh4 Ng7 23.Ra3 Nh5 24.g4 Qxe4 25.Rh3 f6 26.dxe5 dxe5 27.Be3 Rad8 28.gxh5 Qxh4 29.Rxh4 gxh5 30.Rd1 Rd7 31.Rxh5 Red8 32.Bxa7 Rxd5 33.Rxd5 Rxd5 34.Kg2 Ra5 35.Bb6 Ra6 36.Bd8 b5 37.b3 Kg7
2. (2.00): 22.Rac1 Qxb2 23.Rc7 Qxd4 24.Rfc1 Nf4 25.g3 Qxe4 26.gxf4 exf4 27.Qxf4 Qxf4 28.Bxf4 b5 29.Rb1 a6 30.Bxd6 Red8 31.Rc6 Rd7 32.Kg2 f6 33.Re1 a5 34.Re7 Rxe7 35.Bxe7 a4 36.Bxf6 Kf7 37.Be5 Rd8
3. (1.54): 22.Qf3 exd4 23.g4 Qxe4 24.Qxe4 Rxe4 25.gxh5 Rg4+ 26.Kh1 Rh4 27.Bg5 Rxh5 28.Bf6 Rxd5 29.Rfd1 d3 30.Ra3 Re8 31.Rxa7 g5 32.Rxb7 Re2 33.Kg2 Kh7 34.Bc3 Kg6 35.Bd2 Re4 36.h3 f6 37.b4 Ree5
4. (1.49): 22.dxe5 Rxe5 23.f3 Qxb2 24.Be3 Re7 25.Rfd1 Rc7 26.Bd4 Qe2 27.Re1 Qd2 28.Rad1 Qf4 29.e5 Re7 30.Qxf4 Nxf4 31.exd6 Rd7 32.Ba1 Rc8 33.g3 Nh5 34.g4 Nf4 35.Be5 Nh3+ 36.Kg2 f6 37.Kxh3 fxe5
Komodo, however, fails to attach a "?" to 22 dxe5. White should still have a winning advantage after this move. It's more likely 25 Qh4, 26 g4 and 29 Rbc1 which cost White the win in this position.
Komodo also finds other resources for White at move 22: that Qh4 is the most direct route to a win for White.
It's natural to look for a "technical waste" in analysis directly after throwing away a won position, but such analysis is not likely to be as precise as when one's emotions are taken away from the analysis.