Now, for a time, the queenside is fixed. However, Black has switched to the attack of the central pawn P/e4. Black's aim is to eliminate that pawn and make one's central pawns mobile, also opening the diagonals for one's bishop pair.

18 a4 f5! 19 a5 Qc7 is too slow. 18 Bc1 f5 19 Nd2 with the idea of defending the square e4. 19...Rad8 20 f3 Be7! Black has a winning position as White is undeveloped.

! White has several missing tempos to defend the central square e4.

The diagonals for Black's bishops are opening.

? After that move, White's development becomes even worse. Better was 20 Ng5 Bc7 21 Ne6 Rxe6! Black eliminates White's only active piece. 22 fxe6 Qd6 23 g3 Qxe6, but Black still has a winning attack.

! Now, White's King's position is compromised while Black has the so-called "Horwitz bishops" - two bishops on adjacent diagonals are attacking White's king's position.

After 22 Bc1 e2 it is clear why the move 20 Ne1? was not a good idea.

One bishop eliminates the pawn shield while the other bishop helps to create checkmate patterns.

?! Not 23 hxg3?? Qxg3# (White's Queen Q/c2 is cut off from the kingside). Better was 23 Ba5! Qxa5, but after 24 hxg3 Rxe3 Black has still a winning king's attack.

! The defending pawn shield of White's king's position is completely eliminated.