game vs computer3-hard

Bear with me on this analysis. I have found some amazing lines here. Some of them were found by the computer. Anyways, I double-checked all of my personal analysis with the computer.

For starters, the computer playing the 6. Kxf2 line is a bad idea. That line is not good for white. 8. Qf1? is also a bad move. After 9. Nc3? black should be much after 9... Nd6 10. Bb3 Nxf7 11. Nb5 Qd8. It won't be easy for black to unwind, but the extra pawn, coupled with the open position of white's king and the awkward position of white's h1-rook should prove decisive in the long run.

Instead, you played 9... d6?, which is a mistake. I was not quite sure how to punish it at first. I thought 10. g3 was the move, and after 10... Qg4 11. Nb5 I thought black was in trouble. But then I found 11... Ng5!! After 12. Nxc7+ Kd7 13. Nxa8 (13. Nb5? Ke7! {13... Nd8?? 14. Nxe5+! +-} 14.d3 Nxf7 leaves black a pawn up) Qd4+ 14. Kg2 Qe4+ 15. Kg1 Qd4+ and black has perpetual check. The right move for white instead is simply 10. Nb5. At first I rejected  this move because of 10... Kd7. However this runs into 11. Nxe5+! Nxe5 12. Qxf8 Nxc4 13. Qf7+ Kd8 14. d3! (14. Qxc4?? Qf2#) White will win one of the knights and be up a full rook.

Going back, after 10. Nb5, better for black is 10... Qe7. Then: 11. Qe1 %%% (better then 11. Qe2 after which things get very complex: 11... Nf6 12. Ng5 a6 13. Nc3 Nd4 14. Qd1 d5!! This is an absolutely move amazing that I found. Then there goes 15. Nxd5 {15. Bb3 runs into 15... Qc5! I have been unable to find a defense for white in this position. Try to defend white yourself!} Nxd5 16. Bxd5 Qc5. The best way for white to avert mate here is to go into this line: 17. Bf7+! (to block the f-file) Ke7! ### {17... Rxf7? 18. Nxf7 Nxc2+ 19. d4! Qxd4+ &&& [19... exd4 20. b4!! Qc3 21. Ng5!! Bg4 22. Qf1 0-0-0! (22... Qxa1 23. Qf7+ Kd8 24. Qf8+ Kd7 25. Qf7+ and black can't escape perpetual check) 23. h3 (23. Rb1? Ne3 24. Ne4 {24. Bxe3 Qxe3+ 25. Qf2 Qxg5 and black's passed d-pawn is decisive; 24. Qf2 Qd3 25. h3 Qxb1 26. Kh2 Qxb4 27. Bxe3 dxe3 28. Qxe3 Qd6+ 29. Qg3 Qxg3+ 30. Kxg3 Bf5 and black's two extra pawn give him a winning advantage}  24... Qc2 25. Bxe3 dxe3 26. Ng3 e2 27. Nxe2 Bxe2 28. Qe1 Rf8 29. h3 Rf1+ 30. Qxf1 Bxf1 31. Rxf1 Qb2 32. Rf2 {32. Rf4? Qc1+ -+} Qxb4 this position is much better for black, but he will have to display some technique to win this endgame). We are now back to the position after 23. h3. 23... Qxa1 24. hxg4 Qxa2 25. Rxh7 Ne3 26. Bxe3 dxe3 27. Qf5+ Kb8 28. Qe5 Qf2+ 29. Kh2 e2 30. Nf3 Rg8 The computer evaluates this position as equal, which is interesting. White's knight and queen are tied to defending black's e2 pawn from promoting. 31. Qe7 Ka7 32. Rh5 Rf8 33. Rf5 (to stop Rxf3) Rh8+ 34. Rh5 Rf8 and we have an interesting form of perpetual.]

Let us now go all the way back to the position after 17... Rxf7? 18. Nxf7 Nxc2+ 19. d4! Qxd4. This line is marked with &&& signs. 20. Qxd4 exd4 21. Rb1 Be6 22. b3 This position looks better for black, but it's a lot harder to win then just playing 17... Ke7! After Ke7 (marked by ### signs) Then the lines goes 18. d3 Nf3+ 19. Kf1 Nxg5 20. Bxg5+ Kxf7 21. Ke1 Kg8 Although material is equal, black holds the cards because white's king is a lot more exposed then black's. Black has a lot of attacking potential. Fritz 12 actually evaultates this position as -2.03. Try playing this position out as black against a friend. You are likely to score well.

Lets go back to position after the correct 11. Qe1, marked with %%% signs. 11. Qe1 avoids all the crazy lines above. After this, there is 11... a6 12. Nxc7+ Qxc7 13. Qxe4 Rxf7 14. Bxf7+ Qxf7. White is up the exchange here, but his h1-rook is offsided and his king looks a bit shaky. Nonetheless, I would evaulte this position as slightly better for white.

Now back to the game :-) : 10. Nxe4? Qxe4 11. c3 Nd8 12. d3? (12. Bb5+ forces 12... Nc6 as after 12... Ke7 13. d3 Qf5 {any other queen move, such as 13... Qd5, allows 14. Bg5+ Ke6 15. Nxd8+ and black is totally lost) 14. Qxf5 Bxf5 15. Nxd8 and white is up a full piece. Going back to the forced 12... Nc6, White can simply repeat moves with 13. Bc4 and likely have no better choice then just go 13... Nd8 and take the draw by repetition.

Going back to the game. 12. d3? Qg6 13. Bb5+? (13. Nxd6+ cxd6 14. Qe2 and white is likely only slightly worse) c6 14. Bg5? (White must expect being just much worse, if not almost losing, after 14. Bc4 Nxf7 15. Qe2) White is now obviously lost. 14... Rxf7 15. Bxd8? Rxf1+ 16. Rxf1 Kxd8.

A very interesting game (and very, very taxing for me to analyse!) Let me just tell you, the Traxler Variation (or Wilkes-Barre, what ever you call it) is in a bit of a theoretical cloud now. I suggest switching to the main lines in the Two-Knights after 4... d5.