A strong positional decision! Black gives up his lovely bishop for the knight, ensuring himself the ‘eternal’ blockader on c6, as the white dark-squared bishop can never attack the Nc6.

Depriving White of the two bishops, after which he faces a long and unpleasant defence. The engines assess the endgame as equal, but from the player’s viewpoint, Black’s position is clearly preferable.

Black’s plan is simple – to advance his kingside pawns, cramping the white pieces, and to bring his king into the centre. White has no sensible counterplay.

The position looks roughly equal. The e5-square is the wrong colour to serve as a convenient outpost for the white knight – Black simply takes on e5 with the bishop, and begin play on the light squares. Therefore, White instead tries to seize the e-file, and is able to assist this with play on the kingside, by exploiting the barely-noticeable weakness of the g6-square.

19...Nc6 20.Ng4 Ne7 21.Qh5 (21.Nxh6+ Kh8 22.Nf7+ Kg8=) 21...Nf5 22.Rhe1, and White also keeps the initiative.

The black pawns are fixed on the same colour squares as his bishop, which is useful in the long term. White has also realised that after the exchange on g6, the white pawn appearing on that square is a strength, not a weakness. It seriously cramps Black, and cannot easily be attacked.

20...Be7 21.Qxf7+ (21.Ng4!? Qf3+ 22.Kh3 Qxc3 23.Nxh6+ Kh8∞) 21...Nxf7 22.Ng4±.

The g6-pawn disturbs Black and in the end, decides the game.

Surely Alexei Shirov isn’t counting on queening his pawn? No, of course not. By advancing to d7, the pawn cuts the black position into two halves, and is also prepared to lay down its life in due course. The other important thing is that he opens a group of squares for the attack – d5 for the knight, d6 for the white queen, and also the a2-g8 diagonal.

The knight heads for d4, but this does not bring great benefits, since the other black pieces are uncoordinated, especially his rooks.
It was stronger to open up the bishop with 14...e4, or stop the white pawn by 14...Nd7.