CCRL 40/40 - A most curious game

[Mike S.]

Yes, I agree too. - If I did create deeper books recently (NP91-09_2500.abk = max. 20 full moves) it was more for reference. IOW, e.g. I look into these books while kibitzing etc. to see, what do masters play here usually. But for engine games, I think the 12 moves limit as CCRL uses it, is a good choice.

[Mike S.]

Shocking!

Indeed, some of these moves are (but most are not) in my book. I guess these come from notation and/or input mistakes, from original sources. As mentioned, they can be prevented to be played from book by game minimum 2+.

There is a special case though, where the same blunder may be contained very often in an otherwise very good quality database, due to a typical notation (just Nd7) and/or input error: Nbd7 where Nfd7 was correct, and the mistake goes unnoticed during computer input because in the correct continuation, the knight is captured and re-captured immediatly:

[d]rn1qk2r/p3bppp/bpp1pn2/3pN3/2PP4/1PB3P1/P3PPBP/RN1QK2R b KQkq - 0 9

9...Nbd7?? If White is aware, he can simply play 10.Nxc6 now, with big advantage.

Problem: 9...Nbd7 10.Nxd7 Nxd7 transposes to the correct 9...Nfd7 10.Nxd7 Nxd7.

I need to admit I forgot about this problem when I created to NP book (10 occurencies... ). But OTOH, if an engine test is being done with same book for all, then it is only a "cosmetic" problem as games will transpose into the correct continuation.

(The "Balanced..." books don't contain this blunder.)

[Event "?"]
[Site "?"]
[Date "2009.10.31"]
[Round "?"]
[White "Book / database problem"]
[Black "E15 9...Nbd7?? (Nfd7)"]
[Result "*"]
[ECO "E15"]
[PlyCount "18"]

1. d4 Nf6 2. Nf3 e6 3. c4 b6 4. g3 Ba6 5. b3 Bb4+ 6. Bd2 Be7 7. Bg2 c6 8. Bc3
d5 9. Ne5 Nbd7 $4 (9... Nfd7 10. Nxd7 Nxd7) *

[Dr.Wael Deeb]

Hi Thomas,
What tool do you use to make these scans
Regards,
Dr.D

[Norm Pollock]

I ran a similar 100 game experiment with slight changes. With the understanding that 100 games is not statistically significant, the result was that using a 16-move book gets +38 better Elo than using an 8-move book. Both used the same engine, Rybka 2.3.2a w32 1cpu, on an AMD 3000+ at time-control: 2' + 1", with the Fritz 9 manager.

The two ("ctg") books, 16-move and 8-move were built by me from the same database of 262,000 quality games taken from my 40H/Grand collection. Neither book had any restrictions or recommendations on which lines to use. The books each used a minimum of 3 occurrences. The two books were of the same quality, with the only difference being their lengths.

Code: Select all

Rybka 2.3.2a 32-bit-16	55.5/100	+38 elo
Rybka 2.3.2a 32-bit-8	44.5/100	-37 elo

My analysis:
Because of the short time-controls, the engine with the 8-move book used up time making many moves (about 4 on average) that turned out to be in the 16-move book. Because of this time differential for the rest of the game, the engine using the 16-move book was on occasion able to search at a higher depth and get better results. If there were bad book moves in the 16-move book, they did not occur often enough to equalize the time differential advantage.

If there were longer time-controls, then the use of the 16-move book would have less of an advantage. But with a short time-control, as used here, the result indicates that a longer book gets a better result.

Games will be made available if anyone wants to see them. The "ctg" books will also be made available if requested.

[Norm Pollock]

I can only imagine that such blunders come into a book by importing a "broken" game. I guess almost all big databases to work with, will contain a couple of such games.

But normally, an Arena book will never play such moves if the book option

Minimal number of games: 2
(or bigger; Fritz' "optimal setting" value is 3 here)

because it is highly unlikely that moves like 6...Ba3 and 9...Qc6 were played more than once. - The min. percentages should be another safety net against such moves, although if a game was broken the result could be wrong, too.

So, even if a book or the database it is based upon, does contain some of such blunders, proper book options should prevent that they are played from an Arena book.

P.S. My most recent free Arena book is based on Norm Pollock's HQ. collection of IM/GM games, quite big and by default with max. 40 plies:

You might be surprised to know that it's usually more than just a couple of games that are broken in these pgn files. I have downloaded a number of pgn files and all contained games with huge errors. This can be of concern if you are using these pgn file to create your opening book.

Norm Pollock produces some of the cleanest and the most error free pgn files that I have come across. Despite Norms efforts his pgn files still contain a surprising number of bad games. Below are the results of scanning Norms pgn file 'SuperGmb.pgn'. I was only looking for very big mistakes (queen suicides), so games that contain rook, bishop or knight sacs went undetected.



File: SuperGmb.pgn

#395
[Event "FIDE WCh KO, Groning"]
[Site "FIDE WCh KO, Groning"]
[Date "1997.??.??"]
[Round "?"]
[White "Korchnoi, Viktor L."]
[Black "Short, Nigel D."]
[Result "1/2-1/2"]
[WhiteElo "2635"]
[BlackElo "2690"]
[ECO "A10"]

1. c4 b6 2. e4 Bb7 3. Nc3 e5 4. d3 Nc6 5. g3 Bc5 6. Bg2 Nge7 7. Nf3 O-O
8. O-O a5 9. Be3 d6 10. d4 exd4 11. Nxd4 Nxd4 12. Bxd4 Nc6 13. Bxc5 bxc5
14. e5 dxe5 15. Na4 Qd6 16. Qd5 Nb8 17. Qxc5 Qf6 18. Qd6 Bf3 19. Qxc7 Bd1
20. Raxd1 Qd6 21. b3 Qxd1 22. Qd8 Nd7 23. Nc3 Nb8 24. f4 g5 25. Qxb8 Qc2
26. Qxa8 1/2-1/2


#1907
[Event "Korchnoi 70"]
[Site "St Petersburg RUS"]
[Date "2001.??.??"]
[Round "1"]
[White "Taimanov, Mark E."]
[Black "Korchnoi, Viktor L."]
[Result "1/2-1/2"]
[WhiteElo "2454"]
[BlackElo "2639"]
[ECO "A36a"]

1. c4 c5 2. Nc3 Nc6 3. g3 e5 4. Bg2 g6 5. a3 Bg7 6. e3 a5 7. Nge2 d6
8. d3 Nge7 9. O-O O-O 10. Nd5 Rb8 11. Nec3 Nxd5 12. Nxd5 Bd7 13. e4 Ne7
14. Rb1 f5 15. Bd2 Kh8 16. Qc1 Ng8 17. f4 exf4 18. Bxf4 Bd4+ 19. Kh1 Qf6
20. Bg5 Qg7 21. Bf4 Be5 22. b4 axb4 23. axb4 b6 24. Bh3 fxe4 25. dxe4 Bxf4
26. gxf4 Nf6 27. Bg2 Rbe8 28. Qc2 Qh6 29. Qc3 Kg8 30. e5 Nh5 31. exd6 Nxf4
32. Nxf4 Rxf4 33. Rxf4 Qxf4 34. bxc5 bxc5 1/2-1/2


#2501
[Event "Madrid ESP"]
[Site "Madrid ESP"]
[Date "1988.??.??"]
[Round "?"]
[White "Gurevich, Mikhail N."]
[Black "Korchnoi, Viktor L."]
[Result "1/2-1/2"]
[WhiteElo "2625"]
[BlackElo "2640"]
[ECO "A52g"]

1. d4 Nf6 2. c4 e5 3. dxe5 Ng4 4. Nf3 Bc5 5. e3 Nc6 6. Be2 Ngxe5
7. O-O Nxf3+ 8. Bxf3 Ne5 9. Be2 O-O 10. Nc3 a5 11. b3 Re8 12. Bb2 Ra6
13. Ne4 Ba7 14. Qd5 Rae6 15. c5 c6 16. Qd2 d5 17. Bxe5 Rxe5 18. Qxa5 Bb8
19. g3 g6 20. a4 h5 21. Rad1 Kg7 22. Rd4 Ba7 23. Rd3 b5 24. axb5 cxb5
25. Rfd1 Rf5 26. Rxd5 Bxc5 27. Kf1 Kh6 28. h4 Kg7 29. Bf3 Kh6 30. Bg2
1/2-1/2


#2843
[Event "VSB"]
[Site "Rotterdam VSB"]
[Date "1990.??.??"]
[Round "4"]
[White "Korchnoi, Viktor L."]
[Black "Gurevich, Mikhail N."]
[Result "1-0"]
[WhiteElo "2625"]
[BlackElo "2645"]
[ECO "A81"]

1. d4 f5 2. g3 Nf6 3. Bg2 g6 4. Nh3 d6 5. Nc3 Bg7 6. Nf4 c6 7. d5 e5
8. h4 Qe7 9. h5 g5 10. h6 Bf8 11. Nh5 Rg8 12. Nxf6+ Qxf6 13. e4 Bxh6
14. Qf3 fxe4 15. Qxe4 Rg6 16. Be3 Nd7 17. O-O-O a6 18. Rxh6 Rxh6
19. Bxg5 Qe6 20. Qd4 Nf6 21. a3 Qe7 22. Rd3 Qe6 23. Na4 Nxd5 24. c4 Nf6
25. Qb6 Qe7 26. Qa7 Qc7 27. Qd4 Nd5 28. Nb6 1-0


#10672
[Event "Fontys"]
[Site "Fontys Tilburg"]
[Date "1997.??.??"]
[Round "9"]
[White "Polgar, Judit"]
[Black "Shirov, Alexei D."]
[Result "1-0"]
[WhiteElo "2645"]
[BlackElo "2690"]
[ECO "C78m"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O b5 6. Bb3 Bc5 7. a4 Rb8
8. axb5 axb5 9. Nxe5 Nxe5 10. d4 Bxd4 11. Qxd4 d6 12. f4 Nc6 13. Qc3 Ne7
14. Ra7 c5 15. e5 Nfd5 16. Bxd5 f6 17. Qh3 Be6 18. b4 Bc8 19. Be4 Nd5
20. exf6 Bg4 21. Rc7 Qe7 22. Kf2 Ra8 23. f7+ Kd8 24. Rxc5 Rb8 25. Bxd5 Rf8
26. Bb2 1-0


#12253
[Event "RS"]
[Site "Ostende RS"]
[Date "1907.??.??"]
[Round "?"]
[White "Janowski, Dawid M."]
[Black "Marshall, Frank J."]
[Result "1-0"]
[ECO "D00a"]

1. d4 d5 2. Nd2 Nf6 3. Ngf3 c5 4. e3 Nc6 5. dxc5 e5 6. Bb5 Qc7 7. b4 e4
8. Nd4 Bg4 9. f3 exf3 10. gxf3 Bd7 11. Bxc6 bxc6 12. Bb2 Rb8 13. Bc3 Be7
14. Ne4 Kf8 15. h4 Ng4 16. Qc1 f5 17. Ke2 Bf6 18. Nd2 g5 19. a4 Qd6
20. Bb2 Be7 21. Ne4 a5 22. Bc3 Bd8 23. Bd2 Kg8 24. Nxc6 Bb6 25. Be1 Ne5
26. Bf2 axb4 27. Rg1 Rb7 28. Ng3 Qc7 29. h5 Qd6 30. Qd2 Bd8 31. Rac1 Bb6
32. Nd4 Ba5 33. Qe1 Rb5 34. c4 1-0


#14504
[Event "DSB-18.Kongress"]
[Site "Breslau"]
[Date "1912.??.??"]
[Round "4"]
[White "Schlechter, Carl"]
[Black "Rubinstein, Akiba K."]
[Result "1/2-1/2"]
[ECO "D37m"]

1. d4 d5 2. Nf3 Nf6 3. c4 e6 4. Nc3 Be7 5. Bf4 O-O 6. e3 c5 7. Rc1 Nc6
8. Bd3 dxc4 9. Bxc4 cxd4 10. exd4 b6 11. O-O Bb7 12. a3 Rc8 13. Ba2 Qc7
14. Be3 Ng4 15. Qe2 Bd6 16. Rcd1 Nxe3 17. fxe3 Ne7 18. e4 Ng6 19. Ng5 Bf4
20. Qh5 Be3+ 21. Kh1 Bxg5 22. Qxg5 Qe7 23. Qxe7 Nxe7 24. d5 exd5
25. exd5 Ng6 26. Rfe1 Rfd8 27. Kg1 Kf8 28. Rd2 Rd7 29. Re3 Rcd8 30. Kf2 Nf4
31. Re5 f6 32. Re4 Ng6 33. Ra4 Bc6 34. Rad4 Ne7 35. Rd1 Bb7 36. R1d2 g6
37. Ra4 Bc6 38. Rad4 Kg7 39. Bb3 1/2-1/2


#14574
[Event "Ch World (match)"]
[Site "Ch World (match)"]
[Date "1978.??.??"]
[Round "21"]
[White "Korchnoi, Viktor L."]
[Black "Karpov, Anatoly"]
[Result "1-0"]
[ECO "D37u"]

1. c4 Nf6 2. Nc3 e6 3. Nf3 d5 4. d4 Be7 5. Bf4 O-O 6. e3 c5 7. dxc5 Bxc5
8. Qc2 Nc6 9. Rd1 Qa5 10. a3 Re8 11. Nd2 e5 12. Bg5 Nd4 13. Qb1 Bf5
14. Bd3 e4 15. Bc2 Nxc2+ 16. Qxc2 Qa6 17. O-O b5 18. a4 Bf8 19. axb5 Bb4
20. Na4 Bd6 21. b6 Bb8 22. Nc5 Bd6 23. b7 Be5 24. f4 exf3 25. h3 h5
26. g4 hxg4 27. hxg4 Bd6 28. e4 1-0


#6952
[Site "Linares ESP"]
[Date "1990.??.??"]
[Round "?"]
[White "Ivanchuk, Vassily"]
[Black "Kasparov, Garry"]
[Result "0-1"]
[WhiteElo "2665"]
[BlackElo "2800"]
[ECO "B97s"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 e6 7. f4 Qb6
8. Qd2 Qxb2 9. Rb1 Qa3 10. f5 Nc6 11. fxe6 fxe6 12. Nxc6 bxc6 13. Be2 Be7
14. O-O O-O 15. Rb3 Qc5+ 16. Be3 Qe5 17. Bf4 Qc5+ 18. Kh1 Ng4 19. h3 e5
20. Na4 Qa5 21. Bc4+ Kh8 22. hxg4 exf4 23. Nb6 d5 24. exd5 cxd5
25. Bxd5 Rb8 26. Nxc8 Rbxc8 27. Rh3 Qb6 28. Re1 Bg5 29. Re6 Qd8 30. c4 Rb8
31. Qd3 Bh4 32. Be4 Qg5 33. Bxh7 Rfd8 34. Qc2 f3 35. Rxf3 Rd2 36. Qe4 Rd1+
37. Kh2 Re1 38. Qf5 Rxe6 39. Qxe6 Kxh7 40. Qe4+ g6 41. Ra3 Kg7 42. Qd4+ Kg8
43. Qe4 Qf6 0-1


#17228
[Event "Ch URS RUS"]
[Site "Ch URS Moscow RUS"]
[Date "1988.??.??"]
[Round "7"]
[White "Vaganian, Rafael A."]
[Black "Beliavsky, Alexander G."]
[Result "0-1"]
[WhiteElo "2625"]
[BlackElo "2645"]
[ECO "E05s"]

1. c4 e6 2. g3 d5 3. Bg2 Nf6 4. Nf3 Be7 5. O-O O-O 6. d4 dxc4 7. Qc2 a6
8. Qxc4 b5 9. Qc2 Bb7 10. Bd2 Nc6 11. e3 Nb4 12. Bxb4 Bxb4 13. Rd1 Rc8
14. a3 Bd6 15. b4 a5 16. bxa5 Ra8 17. Qd2 b4 18. Ne1 Bxg2 19. Nxg2 Ne4
20. Ne1 Ng5 21. h4 Ne4 22. Nc3 Rb8 23. Nd3 Be7 24. Nxb4 Rb6 25. axb6 Bxb4
26. h5 0-1


#17468
[Event "Soviet Union"]
[Site "Soviet Union"]
[Date "1979.??.??"]
[Round "?"]
[White "Polugaevsky, Lev A."]
[Black "Smyslov, Vassily V."]
[Result "1-0"]
[ECO "E11m"]

1. d4 Nf6 2. c4 e6 3. Nf3 Bb4+ 4. Bd2 a5 5. g3 d6 6. Bg2 Nbd7 7. O-O e5
8. Nc3 exd4 9. Nxd4 Nd5 10. Qc2 Ne5 11. Bg5 h6 12. Bf6 Qxf6 13. Nxd5 Qd8
14. Nxb4 axb4 15. c5 c6 16. Rfd1 d5 17. e4 Bg4 18. Re1 dxe4 19. Rxe4 Rc8
20. Rae1 Qd7 21. h3 Bxh3 22. Bxh3 Qxd4 23. Bg2 Qd5 24. f4 Ra8 25. fxe5 b3
26. Qc3 Rd8 27. e6 fxe6 28. Rg4 Qd7 29. Qxb3 g5 30. Qc3 Ra8 31. Kh1 e5
32. Rge4 1-0


#7877
[Site "Rotterdam NED"]
[Date "1990.??.??"]
[Round "?"]
[White "Short, Nigel D."]
[Black "Gurevich, Mikhail N."]
[Result "1/2-1/2"]
[WhiteElo "2635"]
[BlackElo "2645"]
[ECO "C11t"]

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. Bg5 dxe4 5. Nxe4 Be7 6. Bxf6 Bxf6
7. Nf3 O-O 8. c3 Nd7 9. Qc2 e5 10. O-O-O exd4 11. Nxd4 Bxd4 12. Rxd4 Qe7
13. h4 Ne5 14. Ng5 g6 15. Qe2 Qf6 16. Ne4 Qf4+ 17. Nd2 Qf6 18. h5 Bf5
19. hxg6 Bxg6 20. Ne4 Nc6 21. Ra4 a6 22. Qc4 Qe5 23. Qd5 Qxd5 24. Nf6+ Kg7
25. Nxd5 Rfe8 26. Rah4 Re1+ 27. Kd2 Rb1 28. Bd3 Rxh1 29. Rxh1 Rd8
30. c4 Ne5 31. Bxg6 hxg6 32. Kc3 c6 33. Nf4 g5 34. Rh5 Kf6 35. Rh6+ Ke7
36. Rh5 Kf6 37. Rh6+ Ke7 38. Rh5 Nxc4 39. Rxg5 Nb6 40. b3 Kf6 41. Rh5 Rd1
42. a4 1/2-1/2


#9978
[Event "St Petersburg"]
[Site "St Petersburg"]
[Date "1913.??.??"]
[Round "?"]
[White "Alekhine, Alexander A."]
[Black "Bogoljubow, Efim D."]
[Result "0-1"]
[ECO "C64d"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 Bc5 4. c3 Nge7 5. d4 exd4 6. cxd4 Bb4+
7. Bd2 Bxd2+ 8. Qxd2 a6 9. Ba4 d5 10. exd5 Qxd5 11. Nc3 Qe6+ 12. Kf1 Qc4+
13. Kg1 O-O 14. d5 Rd8 15. Qe1 Bg4 16. Bb3 Qf4 17. dxc6 Bxf3 18. Qxe7 Bxc6
19. h4 Rd2 20. Rf1 Re8 21. Qg5 Rxf2 22. Bxf7+ Kh8 23. Rd1 Qxf7 24. Rd2 h6
25. Rxf2 Re1+ 26. Kh2 Qxf2 27. Qg4 Bxg2 0-1



Position summary of above games...

Game# 395, Diff= 1200, Move 18, Score= 785 - rn3rk1/1bp2ppp/5q2/p1Q1p3/N1P5/6P1/PP3PBP/R4RK1 w - - sm Qd6; c0 "Korchnoi, Viktor L. vs Short, Nigel D. 1997.??.??";
Game# 395, Diff= 1088, Move 18, Score=-306 - rn3rk1/1bp2ppp/3Q1q2/p3p3/N1P5/6P1/PP3PBP/R4RK1 b - - sm Bf3; c0 "Korchnoi, Viktor L. vs Short, Nigel D. 1997.??.??";
Game# 395, Diff= 1350, Move 20, Score=-1861 - rn3rk1/2Q2ppp/5q2/p3p3/N1P5/6P1/PP3PBP/3R1RK1 b - - sm Qd6; c0 "Korchnoi, Viktor L. vs Short, Nigel D. 1997.??.??";
Game# 1907, Diff= 873, Move 19, Score=-891 - 1r1q1rnk/1p1b3p/3p2p1/p1pN1p2/2PbPB2/P2P2P1/1P4BP/1RQ2R1K b - - sm Qf6; c0 "Taimanov, Mark E. vs Korchnoi, Viktor L. 2001.??.??";
Game# 1907, Diff= 870, Move 20, Score=-21 - 1r3rnk/1p1b3p/3p1qp1/p1pN1p2/2PbPB2/P2P2P1/1P4BP/1RQ2R1K w - - sm Bg5; c0 "Taimanov, Mark E. vs Korchnoi, Viktor L. 2001.??.??";
Game# 2501, Diff= 1110, Move 18, Score= 1031 - 2bqr1k1/bp3ppp/2p5/p1Ppr3/4N3/1P2P3/P2QBPPP/R4RK1 w - - sm Qxa5; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2501, Diff= 1109, Move 18, Score=-78 - 2bqr1k1/bp3ppp/2p5/Q1Ppr3/4N3/1P2P3/P3BPPP/R4RK1 b - - sm Bb8; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2501, Diff= 1196, Move 19, Score= 1121 - 1bbqr1k1/1p3ppp/2p5/Q1Ppr3/4N3/1P2P3/P3BPPP/R4RK1 w - - sm g3; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2501, Diff= 1204, Move 19, Score=-85 - 1bbqr1k1/1p3ppp/2p5/Q1Ppr3/4N3/1P2P1P1/P3BP1P/R4RK1 b - - sm g6; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2501, Diff= 890, Move 20, Score= 810 - 1bbqr1k1/1p3p1p/2p3p1/Q1Ppr3/4N3/1P2P1P1/P3BP1P/R4RK1 w - - sm a4; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2501, Diff= 887, Move 20, Score=-77 - 1bbqr1k1/1p3p1p/2p3p1/Q1Ppr3/P3N3/1P2P1P1/4BP1P/R4RK1 b - - sm h5; c0 "Gurevich, Mikhail N. vs Korchnoi, Viktor L. 1988.??.??";
Game# 2843, Diff= 1171, Move 14, Score=-1101 - rnb1k1r1/pp5p/2pp1q1b/3Pppp1/4P3/2N2QP1/PPP2PB1/R1B1K2R b KQq - sm fxe4; c0 "Korchnoi, Viktor L. vs Gurevich, Mikhail N. 1990.??.??";
Game# 2843, Diff= 1169, Move 15, Score= 68 - rnb1k1r1/pp5p/2pp1q1b/3Pp1p1/4p3/2N2QP1/PPP2PB1/R1B1K2R w KQq - sm Qxe4; c0 "Korchnoi, Viktor L. vs Gurevich, Mikhail N. 1990.??.??";
Game# 2843, Diff= 1449, Move 19, Score=-928 - r1b1k3/1p1n3p/p1pp1q1r/3Pp1B1/4Q3/2N3P1/PPP2PB1/2KR4 b q - sm Qe6; c0 "Korchnoi, Viktor L. vs Gurevich, Mikhail N. 1990.??.??";
Game# 2843, Diff= 1136, Move 20, Score= 203 - r1b1k3/1p1n3p/p1ppq2r/3Pp1B1/4Q3/2N3P1/PPP2PB1/2KR4 w q - sm Qd4; c0 "Korchnoi, Viktor L. vs Gurevich, Mikhail N. 1990.??.??";
Game# 6952, Diff= 1036, Move 20, Score=-1006 - r1b2rk1/4b1pp/p1pp4/2q1p3/N3PBn1/1R5P/P1PQB1P1/5R1K b - - sm Qa5; c0 "Ivanchuk, Vassily vs Kasparov, Garry 1990.??.??";
Game# 7877, Diff= 1089, Move 20, Score=-1101 - r4rk1/ppp2p1p/5qb1/4n3/3RN3/2P5/PP2QPP1/2K2B1R b - - sm Nc6; c0 "Short, Nigel D. vs Gurevich, Mikhail N. 1990.??.??";
Game# 9978, Diff= 1187, Move 15, Score=-1101 - r1br2k1/1pp1nppp/p1n5/3P4/B1q5/2N2N2/PP2QPPP/R5KR b - - sm Bg4; c0 "Alekhine, Alexander A. vs Bogoljubow, Efim D. 1914.??.??";
Game# 9978, Diff= 1230, Move 16, Score= 129 - r2r2k1/1pp1nppp/p1n5/3P4/B1q3b1/2N2N2/PP2QPPP/R5KR w - - sm Bb3; c0 "Alekhine, Alexander A. vs Bogoljubow, Efim D. 1914.??.??";
Game# 10672, Diff= 1204, Move 17, Score= 718 - 1rbqk2r/R3n1pp/3p1p2/1ppBP3/5P2/2Q5/1PP3PP/1NB2RK1 w k - sm Qh3; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 3237, Move 17, Score=-2531 - 1rbqk2r/R3n1pp/3p1p2/1ppBP3/5P2/7Q/1PP3PP/1NB2RK1 b k - sm Be6; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 3266, Move 18, Score= 735 - 1r1qk2r/R3n1pp/3pbp2/1ppBP3/5P2/7Q/1PP3PP/1NB2RK1 w k - sm b4; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 949, Move 18, Score=-218 - 1r1qk2r/R3n1pp/3pbp2/1ppBP3/1P3P2/7Q/2P3PP/1NB2RK1 b k - sm Bc8; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 799, Move 19, Score= 581 - 1rbqk2r/R3n1pp/3p1p2/1ppBP3/1P3P2/7Q/2P3PP/1NB2RK1 w k - sm Be4; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 32573, Move 19, Score=-31992 - 1rbqk2r/R3n1pp/3p1p2/1pp1P3/1P2BP2/7Q/2P3PP/1NB2RK1 b k - sm Nd5; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 32093, Move 20, Score= 101 - 1rbqk2r/R5pp/3p1p2/1ppnP3/1P2BP2/7Q/2P3PP/1NB2RK1 w k - sm exf6; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 10672, Diff= 32093, Move 20, Score=-31992 - 1rbqk2r/R5pp/3p1P2/1ppn4/1P2BP2/7Q/2P3PP/1NB2RK1 b k - sm Bg4; c0 "Polgar, Judit vs Shirov, Alexei D. 1997.??.??";
Game# 12253, Diff= 1097, Move 19, Score=-1276 - 1r3k1r/p1qb3p/2p2b2/2Pp1pp1/PP1N2nP/2B1PP2/2PNK3/R1Q4R b - - sm Qd6; c0 "Janowski, Dawid M. vs Marshall, Frank J. 1907.??.??";
Game# 12253, Diff= 1030, Move 20, Score=-251 - 1r3k1r/p2b3p/2pq1b2/2Pp1pp1/PP1N2nP/2B1PP2/2PNK3/R1Q4R w - - sm Bb2; c0 "Janowski, Dawid M. vs Marshall, Frank J. 1907.??.??";
Game# 12253, Diff= 1246, Move 20, Score=-1486 - 1r3k1r/p2b3p/2pq1b2/2Pp1pp1/PP1N2nP/4PP2/1BPNK3/R1Q4R b - - sm Be7; c0 "Janowski, Dawid M. vs Marshall, Frank J. 1907.??.??";
Game# 14504, Diff= 810, Move 13, Score=-796 - 2rq1rk1/pb2bppp/1pn1pn2/8/3P1B2/P1N2N2/BP3PPP/2RQ1RK1 b - - sm Qc7; c0 "Schlechter, Carl vs Rubinstein, Akiba K. 1912.??.??";
Game# 14504, Diff= 810, Move 14, Score= 14 - 2r2rk1/pbq1bppp/1pn1pn2/8/3P1B2/P1N2N2/BP3PPP/2RQ1RK1 w - - sm Be3; c0 "Schlechter, Carl vs Rubinstein, Akiba K. 1912.??.??";
Game# 14574, Diff= 1003, Move 19, Score=-1186 - r3rbk1/p4ppp/q4n2/1P1p1bB1/2P1p3/2N1P3/1PQN1PPP/3R1RK1 b - - sm Bb4; c0 "Korchnoi, Viktor L. vs Karpov, Anatoly 1978.??.??";
Game# 14574, Diff= 1101, Move 20, Score=-85 - r3r1k1/p4ppp/q4n2/1P1p1bB1/1bP1p3/2N1P3/1PQN1PPP/3R1RK1 w - - sm Na4; c0 "Korchnoi, Viktor L. vs Karpov, Anatoly 1978.??.??";
Game# 14574, Diff= 1098, Move 20, Score=-1186 - r3r1k1/p4ppp/q4n2/1P1p1bB1/NbP1p3/4P3/1PQN1PPP/3R1RK1 b - - sm Bd6; c0 "Korchnoi, Viktor L. vs Karpov, Anatoly 1978.??.??";
Game# 17228, Diff= 893, Move 20, Score= 801 - r2q1rk1/2p2ppp/3bp3/P7/1p1Pn3/P3P1P1/3Q1PNP/RN1R2K1 w - - sm Ne1; c0 "Vaganian, Rafael A. vs Beliavsky, Alexander G. 1988.??.??";
Game# 17228, Diff= 915, Move 20, Score=-115 - r2q1rk1/2p2ppp/3bp3/P7/1p1Pn3/P3P1P1/3Q1P1P/RN1RN1K1 b - - sm Ng5; c0 "Vaganian, Rafael A. vs Beliavsky, Alexander G. 1988.??.??";
Game# 17468, Diff= 1081, Move 11, Score=-1096 - r1bqk2r/1pp2ppp/3p4/p2nn1B1/1bPN4/2N3P1/PPQ1PPBP/R4RK1 b kq - sm h6; c0 "Polugaevsky, Lev A. vs Smyslov, Vassily V. 1979.??.??";
Game# 17468, Diff= 1397, Move 12, Score= 301 - r1bqk2r/1pp2pp1/3p3p/p2nn1B1/1bPN4/2N3P1/PPQ1PPBP/R4RK1 w kq - sm Bf6; c0 "Polugaevsky, Lev A. vs Smyslov, Vassily V. 1979.??.??";

Looking at the game results, the player who made the "blunder" was able to recover in 8 of the 13 games. The record of the players making the "blunder":

3 wins (10672, 6952, 9978)
5 losses (2843, 12253, 14574, 17228, 17468)
5 draws (395, 1907, 2501, 14504, 9877)

So in those 8 games where the player recovered, either those moves were not blunders, or the other player made a (more) serious blunder.

[pichy]

I ran a similar 100 game experiment with slight changes. With the understanding that 100 games is not statistically significant, the result was that using a 16-move book gets +38 better Elo than using an 8-move book. Both used the same engine, Rybka 2.3.2a w32 1cpu, on an AMD 3000+ at time-control: 2' + 1", with the Fritz 9 manager.

The two ("ctg") books, 16-move and 8-move were built by me from the same database of 262,000 quality games taken from my 40H/Grand collection. Neither book had any restrictions or recommendations on which lines to use. The books each used a minimum of 3 occurrences. The two books were of the same quality, with the only difference being their lengths.

Code: Select all

Rybka 2.3.2a 32-bit-16	55.5/100	+38 elo
Rybka 2.3.2a 32-bit-8	44.5/100	-37 elo

My analysis:
Because of the short time-controls, the engine with the 8-move book used up time making many moves (about 4 on average) that turned out to be in the 16-move book. Because of this time differential for the rest of the game, the engine using the 16-move book was on occasion able to search at a higher depth and get better results. If there were bad book moves in the 16-move book, they did not occur often enough to equalize the time differential advantage.

If there were longer time-controls, then the use of the 16-move book would have less of an advantage. But with a short time-control, as used here, the result indicates that a longer book gets a better result.

Games will be made available if anyone wants to see them. The "ctg" books will also be made available if requested.

Email me the 16 moves Opening and create a 12 moves opening from the same database and I will test them both at time control of 5' + 1" and can you create the books for Arena

[Norm Pollock]

Jorge-

Here are links for the 3 databases (truncated pgn files) that I used to create the 8, 12 and 16-move books. You can create a book in any format you like. Make sure you indicate a ply count 16+ for the 8-move book, 24+ for the 12-move book, and 32+ for the 16-move book. Also use minimum occurrences of 2+ (3 is recommended).

http://www.mediafire.com/file/mjmvmzm0jhe/bobcat-8.7z

http://www.mediafire.com/file/y2xmmhdonmy/bobcat-12.7z

http://www.mediafire.com/file/zjmyuxzyn5n/bobcat-16.7z

-Norm

[Tommy]

Hi Thomas,
What tool do you use to make these scans
Regards,
Dr.D

I developed my own utility called 'Game Analyser', anybody can download it from http://home.pacific.net.au/~tommyinoz/gameanalyser.html

It's primarily designed to for viewing computer vs computer games, but I have recently added the ability to scan pgn files for blunders. After you have started the program look for the menu option called 'Blunder Check', a new window will appear allowing you make adjustments to certain parameters before you start the scan. It took 4 hours to scan the entire 'SuperGmb.pgn' file on my Intel Core 2 3.0 GHz machine. After the scan I then used Crafty to verify the positions found to weed out some false positives.

Cheers,
Tom.

[pichy]

Hi Thomas,
What tool do you use to make these scans
Regards,
Dr.D

I developed my own utility called 'Game Analyser', anybody can download it from http://home.pacific.net.au/~tommyinoz/gameanalyser.html

It's primarily designed to for viewing computer vs computer games, but I have recently added the ability to scan pgn files for blunders. After you have started the program look for the menu option called 'Blunder Check', a new window will appear allowing you make adjustments to certain parameters before you start the scan. It took 4 hours to scan the entire 'SuperGmb.pgn' file on my Intel Core 2 3.0 GHz machine. After the scan I then used Crafty to verify the positions found to weed out some false positives.

Cheers,
Tom.

I found your experiment to be very interesting and can prevent an engine from selecting a blunder move or moves from an unfiltered PGN like in the game Bright vs Cyclone xTreme.

As an experiment I made a change to the way Kanguruh decides which move to play from the opening book. Instead of looking up the current chess position in the opening book database and choosing the move with the highest score, Kanguruh now performs a 8 ply search in the opening book. The search is very similar to a normal search when Kangurhu is thinking except it gets all the scores and moves from the opening book database. The opening book search depth is adjustable by editing the kanguruh.ini file.

[pichy]

Hi Thomas,
What tool do you use to make these scans
Regards,
Dr.D

I developed my own utility called 'Game Analyser', anybody can download it from http://home.pacific.net.au/~tommyinoz/gameanalyser.html

It's primarily designed to for viewing computer vs computer games, but I have recently added the ability to scan pgn files for blunders. After you have started the program look for the menu option called 'Blunder Check', a new window will appear allowing you make adjustments to certain parameters before you start the scan. It took 4 hours to scan the entire 'SuperGmb.pgn' file on my Intel Core 2 3.0 GHz machine. After the scan I then used Crafty to verify the positions found to weed out some false positives.

Cheers,
Tom.

Can you provide a Link where I can download your scanned 'SuperGmb.pgn file