WBEC Ridderkerk division 3 is finished!

[Leo Dijksman]

Geschrieben von:/Posted by: Leo Dijksman at 28 March 2004 11:58:48:

3rd Division: (Finished!)
Round 17: (last round)
Hagrid 0.7.56 ½111 3.5/4
Tytan 3.39 ½000 0.5/4
------------------------------------
BlackBishop 0.9.7g 1½½1 3.0/4 ! can be just enough to avoid direct relegation!?
CyberPagno 2.01 0½½0 1.0/4
------------------------------------

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
28/03/2004:
Updated: None found yet
Interresting game from the 2nd Division test, Amateur was leading during the game but the position around move 55 is draw until 77.--Kf4?? (77.--Kf5 is still a draw):
Both engines played without egtbs!
[Event "WBEC7 2nd Division Test"]
[Site "ATHLON-MP2200"]
[Date "2004.03.28"]
[Round "1"]
[Number "44"]
[White "Amateur 2.80"]
[Black "Gothmog 0.4.7"]
[Result "1-0"]
[TimeControl "40/180"]
1. d4 d5 2. c4 e6 3. Nf3 Nf6 4. Nc3 c5 5. e3 a6 6. a3 dxc4 7. Bxc4 b5 8.
Bd3 Bb7 9. O-O Nbd7 10. Qe2 Be7 11. Rd1 Qb6 12. dxc5 Nxc5 13. Bc2 O-O 14.
e4 Rad8 15. Bg5 Ncd7 16. e5 Bxf3 17. gxf3 Qc5 18. f4 Nd5 19. Nxd5 Bxg5 20.
b4 Qc4 21. Qxc4 bxc4 22. fxg5 exd5 23. Rxd5 Nb8 24. Rd6 c3 25. Rc1 Rxd6 26.
exd6 Rd8 27. Ba4 Rxd6 28. Rxc3 Rd8 29. Bb3 Kf8 30. Rc7 Nd7 31. f4 f6 32.
Be6 Ke7 33. Bg4 fxg5 34. fxg5 Kd6 35. Ra7 Ne5 36. Be2 Rd7 37. Rxa6+ Kd5 38.
b5 Ke4 39. h4 Ke3 40. Re6 Kxe2 41. Rxe5+ Kf3 42. g6 hxg6 43. Re1 Rd3 44.
Ra1 Rb3 45. a4 Kg3 46. Kf1 Rb2 47. Ke1 Kxh4 48. Kd1 g5 49. Kc1 Rf2 50. b6
g4 51. b7 Rf1+ 52. Kc2 Rxa1 53. b8=Q Rxa4 54. Qh8+ Kg3 55. Qxg7 Ra3 56.
Qe5+ Kg2 57. Qe4+ Kg3 58. Qe1+ Kh3 59. Qh1+ Kg3 60. Qe1+ Kh3 61. Qh1+ Kg3
62. Kd2 Rf3 63. Qg1+ Kf4 64. Qh2+ Kf5 65. Qh7+ Kf4 66. Qc7+ Ke4 67. Qb7+
Ke5 68. Qb4 Kf5 69. Ke2 Kg5 70. Qe7+ Kf4 71. Qc7+ Ke4 72. Qc1 Kf5 73. Qh6
Rh3 74. Qg7 Rf3 75. Qh6 Rh3 76. Qf8+ Ke4 77. Qe7+ Kf4 78. Qf6+ Ke4 79. Qg5
Rh2+ 80. Kf1 Kf3 81. Qd5+ Kg3 82. Kg1 Rf2 83. Qe5+ Rf4 84. Qe2 Rf5 85. Qd3+
Rf3 86. Qd2 Rf5 87. Qd6+ Rf4 88. Qd2 Rf5 89. Qd6+ Rf4 90. Kh1 Kf3 91. Qd3+
Kf2 92. Kh2 Rf3 93. Qd2+ Kf1 94. Qd1+ Kf2 95. Qg1+ Ke2 96. Qxg4 Kf2 97.
Qd4+ Re3 98. Qf6+ Rf3 99. Qb2+ Ke3 100. Kg2 Rf5 101. Qb3+ Kd4 102. Qd1+ Ke5
103. Qa1+ Kd5 104. Kg3 Ke4 105. Qe1+ Kd5 106. Qe3 Rf6 107. Qe8 Kd6 108.
Qd8+ Ke6 109. Qe8+ Kd6 110. Qd8+ Ke6 111. Qb6+ Ke7 112. Qb4+ Ke6 113. Kg4
Rf7 114. Qc4+ Ke7 115. Kg5 Rf8 116. Qb4+ Ke8 117. Kg6 Rf2 118. Qe1+ Kd7
119. Qxf2 Kc6 120. Qb2 Kc5 121. Qc2+ Kb5 122. Kf5 Kb4 123. Qc6 Kb3 124. Ke4
Kb4 125. Kd3 Ka5 126. Qb7 Ka4 127. Kc4 Ka5 128. Qb5#
{White mates} 1-0

Best wishes,
Leo.



WBEC Ridderkerk homepage.

[AndyC]

Geschrieben von:/Posted by: AndyC at 28 March 2004 17:17:48:
Als Antwort auf:/In reply to: WBEC Ridderkerk division 3 is finished! geschrieben von:/posted by: Leo Dijksman at 28 March 2004 11:58:48:

Interresting game from the 2nd Division test, Amateur was leading during the game but the position around move 55 is draw until 77.--Kf4?? (77.--Kf5 is still a draw):
Both engines played without egtbs!
[Event "WBEC7 2nd Division Test"]
[Site "ATHLON-MP2200"]
[Date "2004.03.28"]
[Round "1"]
[Number "44"]
[White "Amateur 2.80"]
[Black "Gothmog 0.4.7"]
[Result "1-0"]
[TimeControl "40/180"]
1. d4 d5 2. c4 e6 3. Nf3 Nf6 4. Nc3 c5 5. e3 a6 6. a3 dxc4 7. Bxc4 b5 8.
Bd3 Bb7 9. O-O Nbd7 10. Qe2 Be7 11. Rd1 Qb6 12. dxc5 Nxc5 13. Bc2 O-O 14.
e4 Rad8 15. Bg5 Ncd7 16. e5 Bxf3 17. gxf3 Qc5 18. f4 Nd5 19. Nxd5 Bxg5 20.
b4 Qc4 21. Qxc4 bxc4 22. fxg5 exd5 23. Rxd5 Nb8 24. Rd6 c3 25. Rc1 Rxd6 26.
exd6 Rd8 27. Ba4 Rxd6 28. Rxc3 Rd8 29. Bb3 Kf8 30. Rc7 Nd7 31. f4 f6 32.
Be6 Ke7 33. Bg4 fxg5 34. fxg5 Kd6 35. Ra7 Ne5 36. Be2 Rd7 37. Rxa6+ Kd5 38.
b5 Ke4 39. h4 Ke3 40. Re6 Kxe2 41. Rxe5+ Kf3 42. g6 hxg6 43. Re1 Rd3 44.
Ra1 Rb3 45. a4 Kg3 46. Kf1 Rb2 47. Ke1 Kxh4 48. Kd1 g5 49. Kc1 Rf2 50. b6
g4 51. b7 Rf1+ 52. Kc2 Rxa1 53. b8=Q Rxa4 54. Qh8+ Kg3 55. Qxg7 Ra3 56.
Qe5+ Kg2 57. Qe4+ Kg3 58. Qe1+ Kh3 59. Qh1+ Kg3 60. Qe1+ Kh3 61. Qh1+ Kg3
62. Kd2 Rf3 63. Qg1+ Kf4 64. Qh2+ Kf5 65. Qh7+ Kf4 66. Qc7+ Ke4 67. Qb7+
Ke5 68. Qb4 Kf5 69. Ke2 Kg5 70. Qe7+ Kf4 71. Qc7+ Ke4 72. Qc1 Kf5 73. Qh6
Rh3 74. Qg7 Rf3 75. Qh6 Rh3 76. Qf8+ Ke4 77. Qe7+ Kf4 78. Qf6+ Ke4 79. Qg5
Rh2+ 80. Kf1 Kf3 81. Qd5+ Kg3 82. Kg1 Rf2 83. Qe5+ Rf4 84. Qe2 Rf5 85. Qd3+
Rf3 86. Qd2 Rf5 87. Qd6+ Rf4 88. Qd2 Rf5 89. Qd6+ Rf4 90. Kh1 Kf3 91. Qd3+
Kf2 92. Kh2 Rf3 93. Qd2+ Kf1 94. Qd1+ Kf2 95. Qg1+ Ke2 96. Qxg4 Kf2 97.
Qd4+ Re3 98. Qf6+ Rf3 99. Qb2+ Ke3 100. Kg2 Rf5 101. Qb3+ Kd4 102. Qd1+ Ke5
103. Qa1+ Kd5 104. Kg3 Ke4 105. Qe1+ Kd5 106. Qe3 Rf6 107. Qe8 Kd6 108.
Qd8+ Ke6 109. Qe8+ Kd6 110. Qd8+ Ke6 111. Qb6+ Ke7 112. Qb4+ Ke6 113. Kg4
Rf7 114. Qc4+ Ke7 115. Kg5 Rf8 116. Qb4+ Ke8 117. Kg6 Rf2 118. Qe1+ Kd7
119. Qxf2 Kc6 120. Qb2 Kc5 121. Qc2+ Kb5 122. Kf5 Kb4 123. Qc6 Kb3 124. Ke4
Kb4 125. Kd3 Ka5 126. Qb7 Ka4 127. Kc4 Ka5 128. Qb5#
{White mates} 1-0

Best wishes,
Leo.

Gothmog does seem to do that occasionally. Here's another one of Gothmog 0.4.7 vs Abrok 5.00
8/2b5/8/6p1/8/p4P2/Bp2K1k1/8 w - - 0 107generated with GenDiag 1.8 (C) 2002 by Andreas Herrmann
 


White simply has to stop the black king penetrating queen side using his own king and opposition and the rest of the time move his bishop from a2 to b1. This had been going on for some 35 moves since the last capture and I nearly adjudicated a draw only to see Gothmog play Ke3 here and allow the black king over on the queen side leading to a loss.

[Andreas Herrmann]

Geschrieben von:/Posted by: Andreas Herrmann at 28 March 2004 18:04:38:
Als Antwort auf:/In reply to: WBEC Ridderkerk division 3 is finished! geschrieben von:/posted by: Leo Dijksman at 28 March 2004 11:58:48:

3rd Division: (Finished!)
Round 17: (last round)
Hagrid 0.7.56 ½111 3.5/4
Tytan 3.39 ½000 0.5/4
------------------------------------
BlackBishop 0.9.7g 1½½1 3.0/4 ! can be just enough to avoid direct relegation!?
CyberPagno 2.01 0½½0 1.0/4
------------------------------------

Hi Leo,
thanks again for your great work. I have made a big mistake, and have send you before the tournament an untested version. Now after enough test games, i know that the version 0.9.7g is about 21 ELO weaker than the CCT 6 version 0.9.5z.

Program Elo + - Games Score Av.Op. Draws
34 BlackBishop 0.9.7i : 2295 27 24 640 52.1 % 2282 18.6 % IPCCC 2004
36 BlackBishop 0.9.5z : 2288 26 23 650 51.8 % 2278 19.2 % CCT 6 2004
37 BlackBishop 0.9.7i2 : 2285 27 24 650 51.3 % 2278 16.5 %
39 BlackBishop 0.9.7i3 : 2278 27 22 650 50.2 % 2278 20.8 %
40 BlackBishop 0.9.5r : 2274 26 26 580 50.0 % 2275 21.7 %
41 BlackBishop 0.9.1 : 2268 24 27 650 48.8 % 2278 16.2 %
42 BlackBishop 0.9.7g : 2267 23 27 650 48.6 % 2278 17.2 %
43 BlackBishop 0.9.5b : 2262 24 26 650 48.0 % 2278 15.1 %
44 BlackBishop 0.9.2d : 2259 22 26 670 48.9 % 2268 21.6 %
46 Holmes 0.8.29i : 2253 28 22 640 50.1 % 2254 20.8 %
56 BlackBishop 0.9 : 2216 32 28 450 51.7 % 2206 16.7 %
61 Holmes 0.8.20m : 2204 23 27 640 48.4 % 2216 18.8 %
65 BlackBishop 0.9.6w : 2187 24 26 660 47.4 % 2207 15.5 %
66 Holmes 0.8.19b : 2185 25 26 660 47.0 % 2207 14.7 % IPCCC 2003
79 Holmes 0.8.12c : 2106 28 26 580 43.3 % 2154 14.8 %
80 Holmes 0.8.13g : 2099 28 25 580 42.4 % 2154 16.9 %
88 Holmes 0.8.4k : 2039 33 25 520 37.5 % 2129 14.2 % CSVN 2002
92 Holmes 0.8.2 : 2005 36 27 450 37.2 % 2097 14.4 %
93 Holmes 0.8.2v : 1999 38 26 450 36.3 % 2097 11.8 %
96 Holmes 0.8.2n : 1957 41 27 400 34.1 % 2072 13.8 %
99 Holmes 0.8.1 : 1935 45 26 400 31.4 % 2072 11.8 %
103 Holmes 0.8.0 : 1920 46 27 370 31.1 % 2060 13.5 %
104 Holmes 0.4.0 : 1906 47 27 360 29.9 % 2056 13.6 % IPCCC 2001
106 Holmes 0.3.5 : 1873 52 30 290 29.7 % 2025 14.5 %

And with the CCT 6 version i would have had expected a final result around a 50% score after looking to my own test games, before the tournament. But at the end i'm happy with the lucky final result of version 0.9.7g . All opponents between Horizon and Gaviota have about equal strength and luck was not unimportant. In the lower devisions (4th..6Th) the differences in playing strength are much bigger.
Future tournaments in the 3rd devision will become not easier. Other strong engines will enter the 3rd like Djinn and GreKo. I'm sure also Booot 3.1 will enter very fast the 3rd devision. It seems Booot has made very fast an improvement of many hundred ELOs. BB 0.9.7i lost yesterday a test match against Booot 3.1 with 3,5:6,5!!. A new upcomer like Kaissa?
Finally congrats to all, who will start in the next 2nd divison.
best wishes
Andreas

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 28 March 2004 18:25:18:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk division 3 is finished! geschrieben von:/posted by: Andreas Herrmann at 28 March 2004 18:04:38:

3rd Division: (Finished!)
Round 17: (last round)
Hagrid 0.7.56 ½111 3.5/4
Tytan 3.39 ½000 0.5/4
------------------------------------
BlackBishop 0.9.7g 1½½1 3.0/4 ! can be just enough to avoid direct relegation!?
CyberPagno 2.01 0½½0 1.0/4
------------------------------------

Hi Leo,
thanks again for your great work. I have made a big mistake, and have send you before the tournament an untested version. Now after enough test games, i know that the version 0.9.7g is about 21 ELO weaker than the CCT 6 version 0.9.5z.

Program Elo + - Games Score Av.Op. Draws
34 BlackBishop 0.9.7i : 2295 27 24 640 52.1 % 2282 18.6 % IPCCC 2004
36 BlackBishop 0.9.5z : 2288 26 23 650 51.8 % 2278 19.2 % CCT 6 2004
37 BlackBishop 0.9.7i2 : 2285 27 24 650 51.3 % 2278 16.5 %
39 BlackBishop 0.9.7i3 : 2278 27 22 650 50.2 % 2278 20.8 %
40 BlackBishop 0.9.5r : 2274 26 26 580 50.0 % 2275 21.7 %
41 BlackBishop 0.9.1 : 2268 24 27 650 48.8 % 2278 16.2 %
42 BlackBishop 0.9.7g : 2267 23 27 650 48.6 % 2278 17.2 %

You did not play enough games to be sure.
2267+23=2290>2288
I also do not consider 21 elo mistake as a big mistake.
Not that I play more games(I usually play clearly less games than you and use test suites but I do not claim to know if newer version is better).
I do not use test suites to decide about changes in the evaluation but about change in the search and I decide about changes in the evaluation or without testing in games and only testing some positions or with 100 or 200 games against previous version.
I know that I cannot be sure that a new version is better with these methods but time is limited and the best way to make improvement is probably to do something without being sure if you really made an improvement.
Uri

[Andreas Herrmann]

Geschrieben von:/Posted by: Andreas Herrmann at 28 March 2004 19:27:16:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk division 3 is finished! geschrieben von:/posted by: Uri Blass at 28 March 2004 18:25:18:

3rd Division: (Finished!)
Round 17: (last round)
Hagrid 0.7.56 ½111 3.5/4
Tytan 3.39 ½000 0.5/4
------------------------------------
BlackBishop 0.9.7g 1½½1 3.0/4 ! can be just enough to avoid direct relegation!?
CyberPagno 2.01 0½½0 1.0/4
------------------------------------

Hi Leo,
thanks again for your great work. I have made a big mistake, and have send you before the tournament an untested version. Now after enough test games, i know that the version 0.9.7g is about 21 ELO weaker than the CCT 6 version 0.9.5z.

Program Elo + - Games Score Av.Op. Draws
34 BlackBishop 0.9.7i : 2295 27 24 640 52.1 % 2282 18.6 % IPCCC 2004
36 BlackBishop 0.9.5z : 2288 26 23 650 51.8 % 2278 19.2 % CCT 6 2004
37 BlackBishop 0.9.7i2 : 2285 27 24 650 51.3 % 2278 16.5 %
39 BlackBishop 0.9.7i3 : 2278 27 22 650 50.2 % 2278 20.8 %
40 BlackBishop 0.9.5r : 2274 26 26 580 50.0 % 2275 21.7 %
41 BlackBishop 0.9.1 : 2268 24 27 650 48.8 % 2278 16.2 %
42 BlackBishop 0.9.7g : 2267 23 27 650 48.6 % 2278 17.2 %

You did not play enough games to be sure.
2267+23=2290>2288
I also do not consider 21 elo mistake as a big mistake.
Not that I play more games(I usually play clearly less games than you and use test suites but I do not claim to know if newer version is better).
I do not use test suites to decide about changes in the evaluation but about change in the search and I decide about changes in the evaluation or without testing in games and only testing some positions or with 100 or 200 games against previous version.
I know that I cannot be sure that a new version is better with these methods but time is limited and the best way to make improvement is probably to do something without being sure if you really made an improvement.
Uri

Hi Uri,
yes normally you are right. But i have testet the two strength related changes between 0.9.5z and 0.9.7g also in the current version 0.9.7i. Both changes (tested with 0.9.7i2 and 0.97.i3) are showing also that the changes between 0.9.5z and 0.9.7g has been bad. 0.9.7i2 is here 10 ELO weaker and 0.9.7i3 is 17 ELO weaker. Together this is around 27 ELO.
And if i took this together with the 21 ELO, i come up to an average of around 24 ELO ((27+21)/2) which tends the version 0.9.7g to be weaker than 0.9.5z.
And a change of 20 to 25 ELO in a tournament, where most of the opponents are inside +/-30 ELO would not be unimportant.

Good luck with your Movei in the 2nd devision and of course also at the WMCC in Israel.

best wishes
Andreas