Kasparov-Karpov 8th game : 38. Nc3!

[Vinvin]

I'm comparing computer analyze and human analyze about the World Chess Champioship matches : http://www.talkchess.com/forum/viewtopic.php?t=63296

I can say the level of the players and the level of the analyzes are very low compare as today !

But there are interesting moves as this 38. Nc3!

Showing the weak pawn on g6!

[D]2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38

Evals after 38.Nc3 :

Code: Select all

Stockfish_170305_x64_modern_fast:
 35/52	+1,38	38. ... Qe7 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Qf7 43.Bg6
 35/52	+1,37	38. ... Qf6 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Qf7 43.Re1
 36/52	+1,38	38. ... Qd7 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Bb7 43.Nf5

[pgn]
[Event "World Championship 34th-KK4"]
[Site "Sevilla"]
[Date "1987.??.??"]
[Round "8"]
[White "Kasparov, Gary"]
[Black "Karpov, Anatoly"]
[Result "1-0"]
[BlackElo "2000"]
[ECO "A21"]
[Opening "English"]
[Variation "KID, 2.Nc3 d6 3.g3"]
[WhiteElo "2000"]
[TimeControl "60+0"]
[Termination "normal"]
[PlyCount "99"]
[WhiteType "program"]
[BlackType "program"]

1. c4 e5 2. Nc3 d6 3. g3 c5 4. Bg2 Nc6 5. a3 g6 6. b4 Bg7 7. Rb1 Nge7 8. e3
O-O 9. d3 Rb8 10. Nge2 Be6 11. b5 Na5 12. Bd2 b6 13. O-O Nb7 14. e4 Kh8 15.
Qc1 f5 16. Bg5 Qe8 17. Bxe7 Qxe7 18. exf5 Bxf5 19. Nd5 Qd7 20. Qd2 Na5 21.
Nec3 Rbe8 22. Ne4 Nb7 23. a4 Na5 24. h4 Nb7 25. Kh2 Rb8 26. Ra1 Na5 27. Ra3
Rf7 28. Qc3 Rd8 29. Ra2 Bh6 30. Ng5 Rff8 31. Re2 Bg7 32. Qc2 Rde8 33. Ne3
Bh6 34. Bd5 Bg7 35. Qd1 h6 36. Ne4 Qd8 37. Ra2 Bc8 38. Nc3 h5 39. Be4 Re6
40. Ncd5 Bh6 41. Ng2 Kg7 42. f4 exf4 43. Ngxf4 Re5 44. Nxg6 Rxf1 45. Qxf1
Rxe4 46. dxe4 Kxg6 47. Rf2 Qe8 48. e5 dxe5 49. Rf6+ Kg7 50. Rd6 1-0
[/pgn]

[MikeGL]

I'm comparing computer analyze and human analyze about the World Chess Champioship matches : http://www.talkchess.com/forum/viewtopic.php?t=63296

I can say the level of the players and the level of the analyzes are very low compare as today !

But there are interesting moves as this 38. Nc3!

Showing the weak pawn on g6!

[D]2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38

Evals after 38.Nc3 :

Code: Select all

Stockfish_170305_x64_modern_fast:
 35/52	+1,38	38. ... Qe7 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Qf7 43.Bg6
 35/52	+1,37	38. ... Qf6 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Qf7 43.Re1
 36/52	+1,38	38. ... Qd7 39.Be4 Qe6 40.Ncd5 Rd8 41.h5 g5 42.g4 Bb7 43.Nf5

[pgn]
[Event "World Championship 34th-KK4"]
[Site "Sevilla"]
[Date "1987.??.??"]
[Round "8"]
[White "Kasparov, Gary"]
[Black "Karpov, Anatoly"]
[Result "1-0"]
[BlackElo "2000"]
[ECO "A21"]
[Opening "English"]
[Variation "KID, 2.Nc3 d6 3.g3"]
[WhiteElo "2000"]
[TimeControl "60+0"]
[Termination "normal"]
[PlyCount "99"]
[WhiteType "program"]
[BlackType "program"]

1. c4 e5 2. Nc3 d6 3. g3 c5 4. Bg2 Nc6 5. a3 g6 6. b4 Bg7 7. Rb1 Nge7 8. e3
O-O 9. d3 Rb8 10. Nge2 Be6 11. b5 Na5 12. Bd2 b6 13. O-O Nb7 14. e4 Kh8 15.
Qc1 f5 16. Bg5 Qe8 17. Bxe7 Qxe7 18. exf5 Bxf5 19. Nd5 Qd7 20. Qd2 Na5 21.
Nec3 Rbe8 22. Ne4 Nb7 23. a4 Na5 24. h4 Nb7 25. Kh2 Rb8 26. Ra1 Na5 27. Ra3
Rf7 28. Qc3 Rd8 29. Ra2 Bh6 30. Ng5 Rff8 31. Re2 Bg7 32. Qc2 Rde8 33. Ne3
Bh6 34. Bd5 Bg7 35. Qd1 h6 36. Ne4 Qd8 37. Ra2 Bc8 38. Nc3 h5 39. Be4 Re6
40. Ncd5 Bh6 41. Ng2 Kg7 42. f4 exf4 43. Ngxf4 Re5 44. Nxg6 Rxf1 45. Qxf1
Rxe4 46. dxe4 Kxg6 47. Rf2 Qe8 48. e5 dxe5 49. Rf6+ Kg7 50. Rd6 1-0
[/pgn]

What happens if black plays the obvious human move with 38...e5, which activates the Re8 and Bg7 in addition to winning a tempo because white will be forced to play Nxe5 again. The game could continue as below, which seems to give black some attacking chances.

[pgn]
[Event "Analysis"]
[Site "Talkchess analysis"]
[Date "2017.03.18"]
[Round "?"]
[White "analyze"]
[Black "analyze"]
[Result "*"]
[Time "05:21:05"]
[TimeControl "180+2"]
[SetUp "1"]
[FEN "2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38"]
[Termination "unterminated"]
[PlyCount "29"]
[WhiteType "human"]
[BlackType "human"]

38. Nc3 e4 (38. .. h5 39. Be4 Re6 40. Ncd5 Bh6 41. Ng2 Kg7 42. f4 exf4 43.
Ngxf4 Re5 44. Nxg6 Rxf1 45. Qxf1 Rxe4 46. dxe4 Kxg6 47. Rf2 Qe8 48. e5 dxe5
49. Rf6+ Kg7 50. Rd6) 39. Nxe4 Qd7 40. Rh1 Bd4 41. Re2 Re5 42. Qf1 Rh5 43.
Qg2 g5 44. Kg1 g4 45. Qf1 Bb7 46. Bxb7 Nxb7 47. Nd5 Nd8 48. Qd1 Rhf5 49. a5
bxa5 50. Ra2 Rf3 51. Kg2 Qh7 52. Re1 *
[/pgn]


And then, it seems black is attacking on this line instead of white.

[peter]

Hi!

What happens if black plays the obvious human move with 38...e5, which activates the Re8 and Bg7 in addition to winning a tempo because white will be forced to play Nxe5 again. The game could continue as below, which seems to give black some attacking chances.

[pgn]
[Event "Analysis"]
[Site "Talkchess analysis"]
[Date "2017.03.18"]
[Round "?"]
[White "analyze"]
[Black "analyze"]
[Result "*"]
[Time "05:21:05"]
[TimeControl "180+2"]
[SetUp "1"]
[FEN "2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38"]
[Termination "unterminated"]
[PlyCount "29"]
[WhiteType "human"]
[BlackType "human"]

38. Nc3 e4 (38. .. h5 39. Be4 Re6 40. Ncd5 Bh6 41. Ng2 Kg7 42. f4 exf4 43.
Ngxf4 Re5 44. Nxg6 Rxf1 45. Qxf1 Rxe4 46. dxe4 Kxg6 47. Rf2 Qe8 48. e5 dxe5
49. Rf6+ Kg7 50. Rd6) 39. Nxe4 Qd7 40. Rh1 Bd4 41. Re2 Re5 42. Qf1 Rh5 43.
Qg2 g5 44. Kg1 g4 45. Qf1 Bb7 46. Bxb7 Nxb7 47. Nd5 Nd8 48. Qd1 Rhf5 49. a5
bxa5 50. Ra2 Rf3 51. Kg2 Qh7 52. Re1 *
[/pgn]


And then, it seems black is attacking on this line instead of white.

Did let SF run some Backward of your line, it thinks there are some inaccuracies for both sides, but e.g. what about 42.Kg2 instead of 42.Qf1?

After

1. c4 e5 2. Nc3 d6 3. g3 c5 4. Bg2 Nc6 5. a3 g6 6. b4 Bg7 7. Rb1 Nge7 8. e3 O-O
9. d3 Rb8 10. Nge2 Be6 11. b5 Na5 12. Bd2 b6 13. O-O Nb7 14. e4 Kh8 15. Qc1 f5
16. Bg5 Qe8 17. Bxe7 Qxe7 18. exf5 Bxf5 19. Nd5 Qd7 20. Qd2 Na5 21. Nec3 Rbe8
22. Ne4 Nb7 23. a4 Na5 24. h4 Nb7 25. Kh2 Rb8 26. Ra1 Na5 27. Ra3 Rf7 28. Qc3
Rd8 29. Ra2 Bh6 30. Ng5 Rff8 31. Re2 Bg7 32. Qc2 Rde8 33. Ne3 Bh6 34. Bd5 Bg7
35. Qd1 h6 36. Ne4 Qd8 37. Ra2 Bc8 38. Nc3

38...e4 39. Nxe4 Qd7 40. Rh1 Bd4 41. Re2 Re5 42. Kg2:

[d]2b2r1k/p2q4/1p1p2pp/nPpBr3/P1PbN2P/3PN1P1/4RPK1/3Q3R b - - 0 1


Analysis by pedantFishW_2017-03-15_popcnt:

42...h5 43.f4 Re7 44.Ree1 Kg7 45.Nc2 Bf6 46.Nxf6 Rxf6 47.Rxe7+ Qxe7 48.Re1 Be6 49.Ne3 Qf8 50.Re2 Kh7 51.Bxe6 Rxe6 52.f5 gxf5 53.Nd5 Rxe2+ 54.Qxe2 Qf7 55.Qe3 Nb3 56.Qg5 Qg6 57.Qe7+ Qg7 58.Qxd6 Qf7 59.Qd8 Nd4 60.Qg5 Qg6 61.Nf6+ Kg7 62.Nxh5+ Kf7 63.Qxg6+ Kxg6 64.Nf4+ Kf6 65.Kf2 Kf7 66.Ne2 Ne6 67.Ke3 Kg7 68.Nf4 Nd4 69.Nd5 Kf7
+- (2.81) Depth: 35/60 00:07:37 11683MN

I don't think Black can hold this

[MikeGL]

you are correct, that Kg2 is stronger than my Qf1 above.
thanks. will re-check and repost if something comes up on my analysis.

Hi!

What happens if black plays the obvious human move with 38...e5, which activates the Re8 and Bg7 in addition to winning a tempo because white will be forced to play Nxe5 again. The game could continue as below, which seems to give black some attacking chances.

[pgn]
[Event "Analysis"]
[Site "Talkchess analysis"]
[Date "2017.03.18"]
[Round "?"]
[White "analyze"]
[Black "analyze"]
[Result "*"]
[Time "05:21:05"]
[TimeControl "180+2"]
[SetUp "1"]
[FEN "2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38"]
[Termination "unterminated"]
[PlyCount "29"]
[WhiteType "human"]
[BlackType "human"]

38. Nc3 e4 (38. .. h5 39. Be4 Re6 40. Ncd5 Bh6 41. Ng2 Kg7 42. f4 exf4 43.
Ngxf4 Re5 44. Nxg6 Rxf1 45. Qxf1 Rxe4 46. dxe4 Kxg6 47. Rf2 Qe8 48. e5 dxe5
49. Rf6+ Kg7 50. Rd6) 39. Nxe4 Qd7 40. Rh1 Bd4 41. Re2 Re5 42. Qf1 Rh5 43.
Qg2 g5 44. Kg1 g4 45. Qf1 Bb7 46. Bxb7 Nxb7 47. Nd5 Nd8 48. Qd1 Rhf5 49. a5
bxa5 50. Ra2 Rf3 51. Kg2 Qh7 52. Re1 *
[/pgn]


And then, it seems black is attacking on this line instead of white.

Did let SF run some Backward of your line, it thinks there are some inaccuracies for both sides, but e.g. what about 42.Kg2 instead of 42.Qf1?

After

1. c4 e5 2. Nc3 d6 3. g3 c5 4. Bg2 Nc6 5. a3 g6 6. b4 Bg7 7. Rb1 Nge7 8. e3 O-O
9. d3 Rb8 10. Nge2 Be6 11. b5 Na5 12. Bd2 b6 13. O-O Nb7 14. e4 Kh8 15. Qc1 f5
16. Bg5 Qe8 17. Bxe7 Qxe7 18. exf5 Bxf5 19. Nd5 Qd7 20. Qd2 Na5 21. Nec3 Rbe8
22. Ne4 Nb7 23. a4 Na5 24. h4 Nb7 25. Kh2 Rb8 26. Ra1 Na5 27. Ra3 Rf7 28. Qc3
Rd8 29. Ra2 Bh6 30. Ng5 Rff8 31. Re2 Bg7 32. Qc2 Rde8 33. Ne3 Bh6 34. Bd5 Bg7
35. Qd1 h6 36. Ne4 Qd8 37. Ra2 Bc8 38. Nc3

38...e4 39. Nxe4 Qd7 40. Rh1 Bd4 41. Re2 Re5 42. Kg2:

[d]2b2r1k/p2q4/1p1p2pp/nPpBr3/P1PbN2P/3PN1P1/4RPK1/3Q3R b - - 0 1


Analysis by pedantFishW_2017-03-15_popcnt:

42...h5 43.f4 Re7 44.Ree1 Kg7 45.Nc2 Bf6 46.Nxf6 Rxf6 47.Rxe7+ Qxe7 48.Re1 Be6 49.Ne3 Qf8 50.Re2 Kh7 51.Bxe6 Rxe6 52.f5 gxf5 53.Nd5 Rxe2+ 54.Qxe2 Qf7 55.Qe3 Nb3 56.Qg5 Qg6 57.Qe7+ Qg7 58.Qxd6 Qf7 59.Qd8 Nd4 60.Qg5 Qg6 61.Nf6+ Kg7 62.Nxh5+ Kf7 63.Qxg6+ Kxg6 64.Nf4+ Kf6 65.Kf2 Kf7 66.Ne2 Ne6 67.Ke3 Kg7 68.Nf4 Nd4 69.Nd5 Kf7
+- (2.81) Depth: 35/60 00:07:37 11683MN

I don't think Black can hold this

[yanquis1972]

this position has been bothering me since it was posted...why are todays top engines about a half-pawn off in their initial evaluation? not only does the raw eval end up roughly = to the initially preferred h5 after Nc3 is demonstrated & the engine has gone thru a couple plies, but it also seems more concrete.

what are machines (& me) missing?

[Lyudmil Tsvetkov]

Nc3 gives nothing.
perfect draw.
Karpov defended excellently, picking all the rigth moves( which, btw., engines do not see), until he made a decisive mistake with 41...Kg7 (41...Bb7 should easily draw instead)

It is astounding how weak those 2 guys played that game. It is quite possible that white was lost by move 10 or so.

[Vinvin]

Nc3 gives nothing.
perfect draw.

Are you sure ?

I analyzed some lines and white has +2 everywhere.
Can you defend that ?

[Event "-"]
[Site "-"]
[Round "8"]
[White "w"]
[Black "b"]
[Result "*"]
[FEN "2bqrr1k/p5b1/1p1p2pp/nPpBp3/P1P1N2P/3PN1P1/R4P1K/3Q1R2 w - - 4 38"]

38. Nc3 Qd7 (38. .. h5 39. Be4 Re6) (38. .. Qe7 39. Be4 Qe6 40. Ncd5 Rf7
41. f3) 39. Be4 Qf7 (39. .. Qe6 40. Ncd5 Rf7 (40. .. Rd8 41. f3 Qh3+ 42.
Kg1 Bb7 43. Rh2)) 40. Ncd5 Bf5 41. h5 Bxe4 42. dxe4 g5 43. Rd2 (43. Nf5 43.
.. Rd8 44. f3) (43. Kg2 Kh7 44. Ra3 Rd8 45. Qe2 Qe6 46. Rd1 Rd7 47. Nf5)
43. .. Qd7 44. Kg2 Rf7 45. Rh1 *

[Milos]

Nc3 gives nothing.
perfect draw.

Are you sure ?

I analyzed some lines and white has +2 everywhere.
Can you defend that ?

Are you joking?
Ofc he can't, he's just a troll.
Nc3 is 100% won, period.

[Vinvin]

why are todays top engines about a half-pawn off in their initial evaluation? ... what are machines (& me) missing?

Always the same duality : depth-selectivity.
Sometimes the search has to be wide to find good moves in bad lines (lines that seems to be bad at the beginning).
Sometimes the search has to be deep because the winning point appears when a lot of moves (depth) have been played.

Given the same CPU power :
When the search is wider, the search is less deep.
And the opposite : When the search is deeper, the search is less wide.

[peter]

Hi Vincent!

I analyzed some lines and white has +2 everywhere.

Also don't see good chances for Black to hold that after 38.Nc3!

Maybe yet ...h5 is the best defence, but after that Kasparov might have had an enhancement with 40.f3 instead of Ncd5 too.
If then again 40...Bh6, the other one Knight can go to d5 (41.Ned5), so it cannot be exchanged against the Bishop.
Eval of clear +- again.

E.g.:

1. c4 e5 2. Nc3 d6 3. g3 c5 4. Bg2 Nc6 5. a3 g6 6. b4 Bg7 7. Rb1 Nge7 8. e3 O-O
9. d3 Rb8 10. Nge2 Be6 11. b5 Na5 12. Bd2 b6 13. O-O Nb7 14. e4 Kh8 15. Qc1 f5
16. Bg5 Qe8 17. Bxe7 Qxe7 18. exf5 Bxf5 19. Nd5 Qd7 20. Qd2 Na5 21. Nec3 Rbe8
22. Ne4 Nb7 23. a4 Na5 24. h4 Nb7 25. Kh2 Rb8 26. Ra1 Na5 27. Ra3 Rf7 28. Qc3
Rd8 29. Ra2 Bh6 30. Ng5 Rff8 31. Re2 Bg7 32. Qc2 Rde8 33. Ne3 Bh6 34. Bd5 Bg7
35. Qd1 h6 36. Ne4 Qd8 37. Ra2 Bc8

38. Nc3 h5 39. Be4 Re6

40. f3 Bh6 41. Ned5 Kg7 42. Rg2 Bd7 43. Rh1 Kh8 44. g4

After Backward of this line to the position after 40.f3 (!) :

[d]2bq1r1k/p5b1/1p1pr1p1/nPp1p2p/P1P1B2P/2NPNPP1/R6K/3Q1R2 b - - 0 1

Analysis by Brainfish 190317 64 POPCNT:

40...Bh6 41.Ned5 Kg7 42.Rg2 Ree8 43.Rh1 Be6 44.g4 hxg4 45.Kg1 Rxf3 46.Bxf3 gxf3 47.Qxf3 Rf8 48.Qd1 Bf4 49.h5 g5 50.Nxf4 Rxf4 51.Ne4 g4 52.h6+ Kh7 53.Rf2 Rxf2 54.Kxf2 Bf5 55.Ke3 Qe7 56.Qf1 Bxe4 57.Kxe4 Qb7+ 58.Ke3 Qd7 59.Qg2 Qb7 60.Qe4+ Qxe4+ 61.Kxe4 Nb3 62.Rg1 Kxh6 63.Rxg4 Nd4 64.Kd5
+- (2.31) Depth: 35/69 00:08:50 11369MN

Still difficult to keep best moves and evals in hash over more than 3 moves backward. Too little bandwith of the search for the one of the tree including all lines of interest obviously