Attack where you pawns are pointing. Right now White's pawns are pointing rather emphatically towards the Q-side: Black to the King's.

White had to play 21.bxa6 here keeping some sort of advantage by getting the pawn. 21.bxa6 bxa6 22.Qa5, taking the pawn out, after 21.a4 it is too late.

Mythas makes an interesting point, but 21. bxa6 bxa6 22. Qa5 can be met with exf4. Now 23. Nxf4 can be met with Qd4+ and Qb6 or 23. gxf4 Bc4!

If 21....a5? he has 22. Qxa5

@jkarp: but one of his a pawns is taken, so that's no problem for him because it's resolved immediately.

Your rook is also more active now.

FWIMBOI - the pin on the f-file I call a half-pin: the pawn is not completely immobilised...

FWIMBOI?

Find Why I'M Betting On It?  

...and adds its tithe to the attack on a6.

@jkarp: it attacks some squares deep into black's territory.

Not forced, he could play Qb4 and after 28...Nxc5 29.Qxc5 he has pressure against c6.

True, and after 29.... Bb7 we transpose into the game position

@ ldj - Not to mention, pressure on the rook at e7. Seems like a much better line for white.

I am wondering about 30.Rb1...

Yes 30.Rb1 threatens 31. Rxb7 Rxb7 32. Bxc6 forking rooks on b7 and e8. Luckily I have a move first though...

Even if 30.Rb1 seems like a try it is possible that 30.a5 could be even better. There is no immediate drawback from "wasting" a move getting pawn on better square.

@jkarp: I don't agree, the king won't be under immediate attack at this square... However I do agree that it might be a loss of a tempo.

I like this move. If my opponent is down to one bishop and there isn't a move that could lead to advantage, I'll take the move to put the king on a square opposite the color of the remaining opposing bishop.