Stockfish Elo setting is way OFF ....................

[Fritz 0]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

[Chessqueen]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

I believe Mr. Kaufman was referring to game at 15/10, since I was testing at game at T/C 15/10 using snitch vs Komodo at Knight Odds

[lkaufman]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

I believe Mr. Kaufman was referring to game at 15/10, since I was testing at game at T/C 15/10 using snitch vs Komodo at Knight Odds

Yes, not only because of the game in question, but because the CCRL Rapid (40/15) time control is very close to 15 minutes plus 10 seconds increment. Even at classical time limits I suppose a 2450 or so CCRL Rapid engine would perform well over 2500 FIDE, maybe 2550 or more. So perhaps my 2650 estimate for human FIDE rating playing at rapid TC was too low, maybe 2700 would be a better guess.

[Fritz 0]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

I believe Mr. Kaufman was referring to game at 15/10, since I was testing at game at T/C 15/10 using snitch vs Komodo at Knight Odds

Yes, not only because of the game in question, but because the CCRL Rapid (40/15) time control is very close to 15 minutes plus 10 seconds increment. Even at classical time limits I suppose a 2450 or so CCRL Rapid engine would perform well over 2500 FIDE, maybe 2550 or more. So perhaps my 2650 estimate for human FIDE rating playing at rapid TC was too low, maybe 2700 would be a better guess.

So basically CCRL Rapid rating + 100 Elo is about FIDE Classical rating. And what about CCRL Blitz ratings? How can we convert them to FIDE Classical ratings?

[lkaufman]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

I believe Mr. Kaufman was referring to game at 15/10, since I was testing at game at T/C 15/10 using snitch vs Komodo at Knight Odds

Yes, not only because of the game in question, but because the CCRL Rapid (40/15) time control is very close to 15 minutes plus 10 seconds increment. Even at classical time limits I suppose a 2450 or so CCRL Rapid engine would perform well over 2500 FIDE, maybe 2550 or more. So perhaps my 2650 estimate for human FIDE rating playing at rapid TC was too low, maybe 2700 would be a better guess.

So basically CCRL Rapid rating + 100 Elo is about FIDE Classical rating. And what about CCRL Blitz ratings? How can we convert them to FIDE Classical ratings?

It's not quite like that, CCRL ratings are more spread apart (at least in the 2000+ range) than human FIDE ratings, so they may be about equivalent at classical time limit at 3000 or so but a 2000 CCRL engine is MUCH (200 or 250 elo maybe) stronger than a 2000 FIDE human at classical. For CCRL Blitz to convert to FIDE blitz you need to add some huge number. I posted some formualae for all this a year or so ago, but don't remember exact details. I think I concluded that Hikaru Nakamura would get a CCRL blitz rating of somewhere in the 2200s, about 600 below his actual FIDE blitz rating. It also depends on time limit; 3 + 0 blitz is a lot different than 5 + 3 blitz.

[Fritz 0]

I think that weak engines should perform worse than humans against strong engines because they do not know that they need to simplify and are going to allow simplifications.

I believe that at TC of 15+10 many humans can win with knight odds even when they have fide rating 2400-2500
Part of the humans who won against Dragon with knight odds in the match that I organized did not use most of their time and they are below 2400.

You are correct, but some of the games that the human in that range lost are due to tactical errors under pressure that engines between 2440 thru 2460 will never make. So it is hard to balance which evil is worse small tactical blunders or NOT knowing when to simplify when ahead, but if you trade every pieces and pawns you can NOT mate with an extra Knight or Bishop at the end unless you end up with King and Rook vs King. Anyway as I thought without the b1 Knight Komodo Dragon was going to have a tough time versus Snitch 1.6.2 rated 2442.

[Event "Computer chess game"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.05.22"]
[Round "?"]
[White "Dragon-2.6.1-64bit-avx2"]
[Black "Snitch"]
[Result "1/2-1/2"]
[BlackElo "2442"]
[ECO "B27"]
[Opening "Sicilian"]
[Time "07:55:41"]
[Variation "1.e4 c5 2.Nf3"]
[WhiteElo "3600"]
[TimeControl "60+3"]
[Termination "unterminated"]
[PlyCount "3"]
[WhiteType "program"]
[BlackType "program"]

1. e4 c5 2. Nc3 Nc6 3. g3 Nf6 4. Bg2 Bc5 5. d3 d6 6. h3 Be6 7. Na4 O-O 8.
O-O a5 9. Kh2 Bd4 10. c3 Ba7 11. c4 Bc5 12. Bg5 h6 13. Bd2 Nb4 14. Be3 Bxe3
15. fxe3 Qd7 16. Nc3 c6 17. a3 Na6 18. Rc1 Rfe8 19. Ne2 Nc5 20. b4 axb4 21.
axb4 Na6 22. Qd2 b5 23. cxb5 cxb5 24. Rb1 d5 25. Rfd1 dxe4 26. dxe4 Qxd2
27. Rxd2 Rac8 28. Rdb2 Rc4 29. Ng1 Rb8 30. g4 Bxg4 31. hxg4 Nxg4+ 32. Kg3
Nxe3 33. Nf3 Rxe4 34. Bh3 f5 35. Nd2 Rd4 36. Nf3 Rd6 37. Nxe5 Re8 38. Nf3
g5 39. Bf1 Rb8 40. Be2 f4+ 41. Kf2 g4 42. Ne5 g3+ 43. Kf3 Nd5 44. Nd3 Rf8
45. Rg1 Re6 46. Kg4 Re3 47. Ne1 Naxb4 48. Bf3 Re5 49. Rd2 Ne7 50. Kh3 Nbd5
51. Ng2 Rg5 52. Rd4 Kg7 53. Ra1 b4 54. Ra7 Kh8 55. Rd7 Rff5 56. Rd8+ Kg7
57. Ra8 h5 58. Ra7 Kg8 59. Rd7 Kh8 60. Rd2 Re5 61. Rd3 Kg8 62. Rd4 b3 63.
Rb7 Ref5 64. Rxb3 Kf7 65. Rb7 Kg8 66. Rd7 Re5 67. Nxf4 Nf6 68. Rb7 Re3 69.
Rd8+ Kf7 70. Rd3 Rxd3 71. Nxd3 Nd5 72. Bxd5+ Rxd5 73. Nf4 Rg5 74. Kg2 h4
75. Kh3 Rg8 76. Ng2 Rh8 77. Nxh4 g2 78. Kxg2 Rxh4 79. Kf3 Ke6 80. Rb5 Nd5
81. Rb2 Ra4 82. Re2+ Kf5 83. Rd2 Ra3+ 84. Kg2 Nf4+ 85. Kf2 Ke4 86. Rd6 Ra2+
87. Ke1 Ng2+ 88. Kd1 Ne3+ 89. Kc1 Nc4 90. Rd7 Rh2 91. Kb1 Ne5 92. Rc7 Kd3
93. Rc8 Nc4 94. Rd8+ Kc3 95. Rc8 Rh7 96. Rc5 Rg7 97. Rc8 Rf7 98. Rc5 Ra7
99. Rc8 Re7 100. Rc5 Re1+ 101. Ka2 Re2+ 102. Kb1 Re6 103. Rc7 Ra6 104. Rc5
Rf6 105. Rc7 Rf1+ 106. Ka2 Rf2+ 107. Kb1 Rg2 108. Rc5 Rg6 109. Rc7 Rg1+
110. Ka2 Rg8 111. Kb1 Ra8 112. Rc6 Ra5 113. Rc7 Re5 114. Rc8 Re1+ 115. Ka2
Re2+ 116. Kb1 Re3 117. Ka2 Re4 118. Kb1 Re5 119. Rc7 Re8 120. Rc5 Rf8 121.
Rc7 Rf5 122. Rc8 Rf3 123. Ka2 Rf6 124. Kb1 Rf5 125. Rc7 Rf2 126. Rc8 Kb3
127. Rb8+ Nb6 128. Rxb6+ Kc3 129. Rc6+ Kd4 130. Kc1 Kd3 131. Rd6+ Kc3 132.
Kd1 Ra2 133. Rc6+ Kd3 134. Ke1 Ra4 135. Rc8 Rc4 136. Rd8+ Ke3 137. Re8+ Re4
138. Rc8 Re7 139. Rd8 Rc7 140. Re8+ Kf4 141. Kd2 Rc6 142. Kd1 Rd6+ 143. Ke2
Rd7 144. Ra8 Rb7 145. Ra4+ Kf5 146. Kf3 Rb2 147. Rf4+ Ke6 148. Ke3 Ke5 149.
Re4+ Kf6 150. Kd3 Rb5 151. Rg4 Rd5+ 152. Ke2 Rd7 153. Rg8 Re7+ 154. Kd2 Rg7
155. Rd8 Rg2+ 156. Ke3 Rc2 157. Ke4 Ke6 158. Rb8 Re2+ 159. Kf3 Ra2 160. Rd8
Ke5 161. Rf8 Rb2 162. Re8+ Kf6 163. Rf8+ Ke5 164. Rd8 Ra2 165. Rh8 Rc2 166.
Re8+ Kf6 167. Rb8 Ra2 168. Kf4 Ra4+ 169. Kf3 Ra2 170. Rb4 Kf5 171. Rb7 Ke5
172. Rc7 Rb2 173. Rc3 Kd5 174. Ra3 Rb6 175. Ke3 Re6+ 176. Kd2 Re7 177. Ra4
Re4 178. Ra5+ {50 moves rule} 1/2-1/2

You keep conflating FIDE and CCRL ratings, as if they were comparable. They are not; a CCRL engine with a 2442 40/15 rating would crush a 2442 FIDE rated human in a match. Engines of the same actual strength as 2500 FIDE GMs do much worse with knight odds than the GMs did.

You are correct it is better for Komodo Dragon to open with 1.e4 no matter if it is b1 or g1 knight missing but even if CCRL rating are not comparable to FIDE that is the reason why I chose snith only rated 2441 which I thought it would be in the ballpark of FIDE 2550 at least.

The disparity is bigger than that, maybe FIDE 2650 might be a good estimate.

Do you mean 2650 in rapid or classical?

I believe Mr. Kaufman was referring to game at 15/10, since I was testing at game at T/C 15/10 using snitch vs Komodo at Knight Odds

Yes, not only because of the game in question, but because the CCRL Rapid (40/15) time control is very close to 15 minutes plus 10 seconds increment. Even at classical time limits I suppose a 2450 or so CCRL Rapid engine would perform well over 2500 FIDE, maybe 2550 or more. So perhaps my 2650 estimate for human FIDE rating playing at rapid TC was too low, maybe 2700 would be a better guess.

So basically CCRL Rapid rating + 100 Elo is about FIDE Classical rating. And what about CCRL Blitz ratings? How can we convert them to FIDE Classical ratings?

It's not quite like that, CCRL ratings are more spread apart (at least in the 2000+ range) than human FIDE ratings, so they may be about equivalent at classical time limit at 3000 or so but a 2000 CCRL engine is MUCH (200 or 250 elo maybe) stronger than a 2000 FIDE human at classical. For CCRL Blitz to convert to FIDE blitz you need to add some huge number. I posted some formualae for all this a year or so ago, but don't remember exact details. I think I concluded that Hikaru Nakamura would get a CCRL blitz rating of somewhere in the 2200s, about 600 below his actual FIDE blitz rating. It also depends on time limit; 3 + 0 blitz is a lot different than 5 + 3 blitz.

Yes, this is the topic:

forum3/viewtopic.php?f=2&t=75988

Human Elo scaling is not the same as engine Elo scaling (not just offset but actual value of Elo point is different). So the fact that Naka is 600Elo stronger than you in "human" Elo, might mean only 300 "engine" Elo.
At 2+1 Naka is probably stronger than 2050 CCRL but not by much, maybe around 2150-2200 Elo. Chess.com ratings in blitz are certainly overinflated. Probably difference of 1000Elo on Chess.com is less than 500Elo on CCRL.

I'm trying to find a corelation between CCRL blitz ratings and FIDE classical ratings. In that respect, it seems reasonable to me that Nakamura playing at 2+1 is about as strong as a 2200 FIDE player playing classical (550-600 points below his classical strength). So maybe CCRL blitz ratings are what the engines would get playing 2+1 against humans playing classical?