Unfortunately for White, 17.Nxc4 seems to be met by 17...Bd5 with a particularly nasty pin. Probably 17.Bxc4 is to be preferred, but the move looks little more prepossessing than the text move. I'm going to have to look back over this game: right now it's hard to figure why it seems to have become sticky.

Also now black's two bishops can play a great role.

Not a bad "in-between" move this one!

A slight inaccuracy in the annotator's analysis: 30.Nf3 keeps White ahead, but not by the whole rook yielded by his main line. In the diagrammed position, the fork set up by White looks nasty, and seems to force Black onto the defensive...

This ending is not so easy for White. The bishop isn't as badly placed as it looks at first sight, and it is certainly an asset on the open board, compared with the White knights. That they are mutually protecting doesn't help White at all. Note, too, that although technically Black's is a "bad" bishop (being on the same coloured squares as Black's central pawns), it is actually quite a good "bad" bishop, as it can find a solid outpost on d5. Finally, White's extra pawn is very hard to protect. Black is threatening, for instance, to play ...Rc6-c2-b2xb4 a plan not easy to frustrate. Maybe, for better or for worse, White had to try 29.Rb1 e.g. 29...Nc3!? (29...Na3? 30.Ra1) 30.Ra1 Bb7 31.Nd4 Rb6? 32.b5 Nxb5? 33.Rb1 or 31...Rc7 32.Rc1. However, Black doesn't have to move his blockading knight, which won't be easy to dislodge. This seems to me already a technically difficult ending for White to win, even if a win is possible.