I'll grant you that the N on b2 looks like a problem but hmmmm maybe we can save him despite this. Are chess games really lost when the N is "on the rim." What do you think? ....

What if 18....Bf8 or 19...Bf8; is this the last chance for black to exchange? if you are ahead in material you should always give exchanging a serious consideration (sometimes you cannot, admittedly).

I may not be able to beat Sneaky, but I know if you violate this rule the chess gods will get mad.

I am looking at Shadout's line now but I am pessimissistic because of the above. And look at move 23. not Bxh7 but 23 Bxe8; and if you try to cover g7 its BxQ.

So I have not worked it out deep, but it seems like once he violates the rule on move 18 or 19; the rest of the game is pretty much a mess.

Do others have the same experience? I lose games all the time and try to figure out what happened. You stick the last 10 moves in the computer and each move it says "okay" but suddenly it's a mess.

You are convinced your opponent can see 15 moves in advance, you will never beat him.

The new try (1941-42!) 16.Qe3, succeeded in the game Armati-Koshnitzky, correspondence, 1941-42: 16...Re8 17.Qf4 c5 (perhaps f5) 18.Ng5 Qb6 19.Rae1 d4 20.Ne4 Bf5 21.Nd6 Bxd6 exd6 d3 23.Bb1 b4 24.Re5 with the advantage.

Averbakh played the new move 16.Qe3, but then at move 17, he played 17.Nd4. Averbakh noted: <White has created two threats. One is obvious - to exchange the opponent's dark-squared bishop by 18.Nc6, while the other is a latent one - to begin a pawn attack on the kingside with 18.f4..>

Averbakh stated that if Zak now played 17...c5, there would follow 18.Nc6 Qd7 19.Nxe7+ Qxe7 20.Bg5 Qf8 21.Bf6, followed by 22.f4.

During the game, Averbakh considered 17...Bf8 to be the strongest reply. He was then intending to play 18.Bxf8 Rxf8 19.Qh6, and if 19...Qd7 20.f4 f5 21.g4 Rae8 22.gxf5 Bxf5 23.Nxf5 gxf5 24.Kh1 with an attack on the g-file.

Only later Averbakh discovered that Black could reply in this line, 19...c5!, Averbakh then states that after 20.Bxg6 hxg6 21.Nxe6 fxe6 22.Qxg6+ Kh8 it is doubtful if White has more than perpetual check.

Averbakh stated that during the game he was prepared to reply to 16...Nc4 with 17.Qf4 Re8 and then 18.Ng5. According to Fritz 9, Black has the advantage in this line, even if White does not play 18.Ng5, so 16...Nc4 would be an acceptable line for Black.

Fritz 9 considers the game to be about equal if Black had played the 17th move Averbakh thought, during the game, to be Black's best; 17...Bf8 18.Bxf8 Rxf8.

At move 17, Black could also play 17...Nc4 18.Qf4 Qd7 19.Qg3 c5 20.Nxe6 Qxe6. This variation was given by Fritz 9. After 17.Nd4, Fritz evaluated the position as being in favor of Black, and recommended 17...Nc4 as best (-.84) (16 ply).

Fritz also evaluated Averbakh's suggestion of 17...Bd7 as being in favor of Black (-.55) (16 ply). Fritz suggested 18.Bb3 Nc4 19.Qf4 c5 as the best continuation.

Averbakh considered 18...Nc4?? to be the losing move. Fritz 9 agreed. Fritz gave the following as the best continuation (16 ply): 19.Qg3 Nd2 20.f5 Nxf1 21.Rxf1 Bc5 22.Kh1 and wins or 22.fxg6 fxg6 23.Bxg6 and wins.

Averbakh believed that Black could have held the position at move 18 with 18...c5, and if 19.f5 cxd4 20.cxd4. At this point Averbakh indicated Bxf5 21.Bxf5 Qa7 as best for Black. However, Fritz 9, indicates 20...Ra8c8 21.Kf2 Bh4 is stronger. Fritz evaluated the position after 18.f4 as: (.00) (18 ply).

Fritz even considered 18...f5 to be a fully acceptable line for Black. After 18.f4, Fritz gave an evaluation of (-.06) (18 ply), 18...f5 19.exf6 Bxf6 20.f5 Bxf5 21.Bxf5 gxf5 22.Qg3+ Kh8 23.Rxf5 Qd6 24.Bf4 Qe7.

click for larger view

Analysis by Rybka 2.1c mp:

17.Qf4 Re8 18.Rfe1 Qd7 19.h4 b4 20.cxb4 Na3 21.Bb3 Rab8 22.Qc1 Bxb4 23.Bd2 Be7 ³ (-0.31) Depth: 25 08:26:20 5786344kN