Much better was 21.b5, attacking the bishop a second time and pretty much forcing it back to d7. Black MIGHT have chosen a bishop sacrifice for two pawns - 21.b5 Bxb5 22.axb5 Qxb5 with perhaps an edge. But it should be borne in mind the material situation would be bishop and two knights (White) against queen and pawn (Black) - a rough material equality. Black's edge would then lie in the activity of her pieces.

This would probably have been better when the White a-pawn still stood on a3, with the threat to exploit the half-pin by ...Rxb4 (axb4?? Qxa1). Having said that, this move prevents 22.b5. On the other hand,if you are contemplating ...Bxa4, you would have to ensure yourself it is a safe option, on account of the bishop being pinned against the queen, and the possible reply, Nc3.

On the other hand, it does threaten a4 which I believe was Erika's goal.

At last mobilizing his stuff.

? Actually, it's his Knight. And, it's a mistake. Why?

OK - especially if ...d5 is the plan. I think I would have played 22...d5 right now, with the idea of closing the centre with 23...d4 if White DIDN'T take on d5. But the point of the move would be to induce the pawn exchange on d5, and thereby open up possibilities for both Black bishops.

The reason for my comment above is because a pawn was left en-prise. Having said that, I would've played 22...Bxa4. Of course, it ties down the Bishop, but that can be solved with 23...Ra7 The idea is that it would've won an additional pawn.

Not a bad idea.

I still would've taken on a4 with the Bishop.

Why not 24.Nxf4, withal? White might also have spared a glance towards 24.a5. The upside of the rook move, though, is that the a-pawn is no longer pinned (because the rook now has protection. So 25.b5 becomes playable.

Yes, I would've responded with 24.Nxf4.

This ought to have been played earlier, but it's as good now as it was then. However, it's something to keep in mind, Erika, because timing is, in general. What might be a good move now, is sometimes a blunder just one move earlier or later. That's something that one of my coaches, kenneth_fiore_3 stressed in one of annotations.

Much more replete with possibilities is 26.Nd5.

Oh, I do like this move!

As do I.

Instead 27.exd5 Qxd5 with all sorts of possibilities - a highly complicated tactical melee in prospect.

It's surely a complicated game.

Fine, since White's move permitted it, but I would also have looked at 27...dxe4.

I would've considered taking on e4. But, after further analysis and thinking about what I really want, this is also what I would've played because it wins an additional pawn which is something that I noticed. I just thought that after 27...dxe4 which can simply answer with 28.dxe4?? However, "it's not so easy" as Jack puts it, and with that, my assessment was wrong here since black has 28...Qxe4+ forking a winning the Rook. Therefore, this would've won a pawn. Having said that, despite what I would've played, I like 27...dxe4 even more because white can't take back. So, black is now threatening a passed pawn. And, even if we can't get it, there's also the immediate threat of 28...Rd4 or 28...Qd5 which puts a lot of pressure against white's King.

Mate would be not far off, if we didn't have the Ra3. But, this move is still good since it liquidates while ahead.

Now Black is making substantial progress.

As a matter of fact, this is also what I would've played.

What would have been a very powerful move here is 30...Qa5!, pinning the unprotected c3-rook against the White king. The only way to defend the rook would be 31.Kd2, whereat Black would win it by 31...d4.

ION's right, and what's bad is that I wouldn't have seen that, either.