Ulhumbrus: 21...Bxg4?! concedes both the bishop pair and the h file to White. It looks at first sight like a losing strategic mistake.
After 22 hxg4 White has available a well known plan of g3, Kg2, Rh1, Nf5 and eventually checkmate on h8, a plan that may have brought White many victories.
How is Black to answer this threat?
Svidler's answer is to open the a file.
23 Nf5 may be premature. Instead of 23 Nf5, 23 Qe2 prepares Be3 in the event that Black replies to f4 with ...Bd4+
Is 24 Ne3 necessary? 24 g3 continues the plan of Kg2 and Rh1. Perhaps Giri wants to avoid exchanging pieces too soon. However as Miroshnichenko indicated Black's N on e7 turns out to be very useful for defence, as in the event of ...Kf1 it may go to g8 if a White rook or queen lands on h8.
However does White have to play for a king side attack? An alternative plan is to play for a central pawn advance of f4 and e5 and from e3 White's knight may be able to blockade Black's c pawn
After 24...b4, 25...a5 and 26...a4 Giri exposes his a pawn by 27 ba. Svidler said that if White were to allow the invasion ...Ra2 that would make f4 more dangerous. However White does not manage to keep his a pawn after 27 ba.
Does White need to play for a king side attack at this point? Instead of 27 ba, 27 Bd3 prepares to blockade Black's c pawn by Nc4 getting ready for a central pawn advance. This also seems consistent with Ne3.
30 Bxa4 concedes the bishop pair although Black is on the verge of playing ...Nc3.
Svidler's explanation was that Giri miscalculated something.
It may be that there was something that Giri saw too late.
Another possible explanation is that after 22 hxg4 Giri thought that he had an automatic attack and that Black had no defence, and Svidler's plan of opening the a file created greater difficulties for White that Giri had foreseen.
This begins to suggest that Giri underestimated the difficulties in the way of White's attack in the event that Svidler were to find a brilliant defence as Svidler did, the plan of opening the a file.
If we consider that the move 17 b3 creates a target for Black's a pawn this suggests that the move 17 b3 makes the king side attack more difficult for White and that after 17 b3 White acquires increased justification for thinking of an alternative plan eg that of a central pawn advance
So the result of the game suggests that 17 b3 gives Black a target for his counterattack and so makes the plan of a king side attack more difficult to follow so that any alternative plans for White become correspondingly more attractive
We can imagine that when the true answer which represents the best play is found eventually it will be something like this: after 22 hxg4 White does not have available an automatic attack but rather an extremely difficult plan of attack which will appear to an ordinary player to be an automatic attack, so that one can quote the saying <An expert makes it look easy>