Author: Uri Blass
Date: 09:03:17 08/13/01
Go up one level in this thread
On August 13, 2001 at 10:39:36, Robert Hyatt wrote: >On August 13, 2001 at 07:43:09, Uri Blass wrote: > >>On August 13, 2001 at 07:14:11, Eduard Nemeth wrote: >> >>>On August 13, 2001 at 06:44:56, Eduard Nemeth wrote: >>> >>>>Fritz 6b (Engine from 200.12.16) played strong, but he chrashed me not! >>>> >>>>Endposition: >>>> >>>>[D]6k1/1bb2rp1/p3r2p/2p1n1pP/PpPpPpP1/1P1P1P1R/2BN1N2/1K5R w - - >>>> >>>>Fritz played no more a5. He makes a draw with Kg8 and a score +1.80 for >>>>him!? >>>> >>>> >>>>[Event "Blitz 30'"] >>>>[Site "?"] >>>>[Date "2001.??.??"] >>>>[Round "?"] >>>>[White "Nemeth,Eduard"] >>>>[Black "Fritz 6b,64MB,P600"] >>>>[ECO "C20"] >>>>[Result "1/2-1/2"] >>>> >>>>1. e4 e5 2. Qf3 Nc6 3. c3 Nf6 4. Bc4 Na5 5. Bb5 a6 6. Ba4 >>>>b5 7. Bc2 c5 8. b3 d5 9. d3 d4 10. c4 Nc6 11. a3 Bd6 12. g3 >>>>O-O 13. h3 Be6 14. Bg5 Qa5+ 15. Bd2 Qb6 16. Bc1 Bc7 17. Qe2 >>>>b4 18. a4 Qa5 19. f3 Nh5 20. Qf2 Ne7 21. g4 Nf4 22. Bxf4 >>>>exf4 23. Nd2 Rad8 24. O-O-O Rfe8 25. Ne2 Qb6 26. Kb2 Rd7 >>>>27. h4 Qc6 28. h5 f6 29. Rde1 h6 30. Rhg1 Qb6 31. Qh2 Qd6 >>>>32. Ra1 Nc6 33. Qg2 Qe5 34. Qf1 Rde7 35. Rg2 Bc8 36. Ng1 >>>>Bb7 37. Nh3 Kh7 38. Nf2 Kh8 39. Qh1 Re6 40. Qh4 Kh7 41. Rh1 >>>>Qg5 42. Qxg5 fxg5 43. Rgg1 R6e7 44. Rh3 Ne5 45. Rgh1 Rd8 >>>>46. R3h2 Re6 47. Kc1 Kg8 48. Kd1 Rdd6 49. Kc1 Re8 50. Kd1 >>>>Rde6 51. Kc1 Kh7 52. Kd1 R8e7 53. Kc1 Kh8 54. Kb1 Kg8 >>>>55. Kc1 Rf6 56. Kd1 Re8 57. Kc1 Rf7 58. Kb1 Rfe7 59. Kc1 >>>>Bc6 60. Kb2 Kh7 61. Kc1 Rf8 62. Kb2 Ree8 63. Kb1 Re6 >>>>64. Kc1 Rf7 65. Kb2 Rfe7 66. Rh3 Rd7 67. Kc1 Re8 68. Kd1 >>>>Rde7 69. Kc1 Rd8 70. Kb2 Ra8 71. Kb1 Bb7 72. Ka2 Re6 >>>>73. Kb2 Rae8 74. Kc1 Bc6 75. Kb2 Kg8 76. Kc1 Kh8 77. Kd1 >>>>Ba8 78. Kc1 R8e7 79. R1h2 Kg8 80. Kb2 Bb7 81. Kc1 Bc6 >>>>82. Kb2 Re8 83. Kb1 Rf8 84. Kb2 Bb7 85. Kb1 Kh8 86. Kb2 >>>>Rfe8 87. Kb1 Kh7 88. Kb2 Ba8 89. Kb1 R8e7 90. Kb2 Bb7 >>>>91. Kb1 Rf7 92. Rh1 Kg8 1/2-1/2 >>>> >>>>Eduard >>> >>>Fritz 6b makes a draw with 50-rules move 92...Kg8 (score was -1.75!), I can this >>>not understand? >> >>Fritz simply does not consider the 50 move rule in the evaluation. >>I believe that it helps him to search more nodes per second >>and the demage is not big because in most of the cases >>when fritz needs a pawn move or a capture in order to avoid the >>50 move rule draw, >>avoiding the 50 move rule draw is only going to make the game longer >>without changing the result. >> >>It is not clear if knowledge about the 50 move rule is productive >>in terms of rating points. >> >>Uri > > >If you play on ICC very much it will make a difference. Weaker humans will >shuffle pieces and win rating points right and left. 1)It may make a difference at blitz against humans. I do not think that it is very important against computers. I also do not think that it makes a big difference against humans at long time control because most of the positions when Fritz draws by the 50 move rule are practicallly draw positions and programs can win them only at blitz when the opponent lose on time or does a tactical error. 2)I know cases when programs that implemented it in the wrong way suffered against Fritz. old version of Junior in my tests lost a drawn game against Fritz that knew nothing about the 50 move rule. The problem was that Junior simply evaluated positions as a draw without considering the fact that the final position after 100 plies without conversion is a checkmate. The same problem happened to hiarcs in the ssdf games > >Can't imagine a commercial program not knowing about the 50 move rule in the >search. It is _not_ expensive to do this. I guess that the difference in comp-comp games is less than 5 elo rating and it may be counter productive if you implement it in the wrong way so I understand the decision of Fritz not to care about it. I am sure that old version of Fritz did not care about it. I am not sure about Deep Fritz but if I remember correctly it also does not care about it. It is easy to check by looking at the evaluation of Deep Fritz in the final position so people can correct me if I am wrong and the evaluation is 0.00 when the 50 move rule is relevant. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.