Karpova: This game was annotated by Tarrasch in the 'Berliner Lokalanceiger'. I won't reproduce all of his comments.Instead of 7.0-0, Tarrasch suggests <7.h3> and awards 7...Bg4 an <!>. He criticizes 8.b3, because it was most urgent to remove the pin, so that all pieces can move freely. Two possible ways were: 1) Develop the ♗c1 to e3, f4 or g5. Then Nbd2, followed by c3 or c4 and Qc2 or Qc3. 2) 1.Re1 2.Nbd2-f1-g3, followed by 3.h3.
12.a3 looks like an inaccuracy, but Tarrasch just notes that White had already a worse position, which will deteriorate further and further. <12.Be2> looks fine, although Tarrasch notes that it is met with the uncomfortable <12...Bf5>, but then White has 13.Ne3.
15.c3 is not annotated, but looks like a mistake. Now was the time to develop the ♗c1 with 15.Bg5. But Black misses the chance 15.c3 Bf5 16.Qd2 and White has problems developing and a weak queenside ♙-structure. The move was not annoated by Tarrasch, though.
16...Bf5 looks fine, Tarrasch also suggests <16...Qf5> and considers <16...Re7> to be best.
Instead of 18...Rae8, Black had the interesting tactical resource 18...Bh3, which is worth a look.
Black plays 20...Bg4 to provoke 21.f3, weakening the e3-square (Tarrasch).
24...Bg5 is a fine move, but Tarrasch suggests Black to cash in on his advantage immediately by <24...Bxb2 25.Qxb2 Bxd3 26.Nxd3 Nxc4>.
It seems that Black gave away his advantage with 27...Qe4 (not annotated), when perhaps 27...Ne5 was in order. This time it was more urgent to remove the pin, so either 28.Qc3 or 28.Qb2. After the text move, Black retains his advantage.
Although he praises White's defense, Tarrasch notes that Black has a winning endgame after 32.Nxb7. In order to realise his advantage, Black should now set his kingside ♙s in motion. He seems to go astray with 39...e3 (not annotated), when 39...Bd2 looks stronger.
43...Kf7 is a grave mistake. Tarrasch suggests <43...f4> to enable <...Bd7>, but White can make use of it by 44.d6. What about 43...h3 instead (e. g. 44.gxh3 Bxh3+ 45.Ke1 Bg2)?
White gives away his advantage with the premature 46.d7 (not annotated). Since the ♗g4 is vital, he should have been removed: 46.Rxg4 fxg5 47.d7 and Black has to give his ♖ for the d-♙.
Tarrasch says that <56.Kg2> could be ignored, because if Kxh2 then ...f3 and Black wins. I'm not sure that this is really the case, e. g. 56.Kg2 Kc7 (56...Kc6 57.Kxh2 f3 58.Nd4+) 57.Kxh2 f3 58.Nd4 and ♔ and ♘ seem to have the situation under control.
Kibitzing
