Moderators: hgm, Rebel, chrisw
Well, engines don't have access to 14-man endgame tables.Ovyron wrote: ↑Tue Jul 23, 2019 10:38 am
I'm not a believer of such scoring, if the position at move 40 is a mate for white, then surely it should show 100% for white. Then we'd know if at some point White gave up the win, and black gave up the draw with 53 gxf4, which would lead to another 100% score for white.
So they produce approximate evaluations to guide their searches.I agree that Kd1 is a better move.
[d]2q5/p2b2k1/1p1b1ppr/3p1p2/3P1P2/P2BPN2/1P4P1/4QRK1 w - - 0 29MikeB wrote: ↑Tue Jul 23, 2019 6:51 am My impressions and with using some backward analysis, you had totally equalize with move 28. Bd6 - always a good time to offer draw here ;>)
The first weakness was 29 ...b5 which allowed Qa5, the next series of moves are pretty much straight foward, but 36.. Ra1 was not the best move I'm afraid. After that everything was more or less forced until you get to 39. Kf2 by white, 39. Kd1 would have been faster . At this point, it appears you were lost, but if there was any doubt, you were certainly lost after 53 gxf4. That move alone change the scoring % from under 60% to over 70% for white. When one gets over 70% in a ICCF game, the winning side should expect to to win. The game was well played, you were a little bit unlucky in some respects and the position required some very deep analysis that is always much easier to see after the fact using backward analysis. Congrats on your unbeaten streak , that's an accomplishment!
You are right not to offer a draw in equal positions but I do not understand 29...b5Ovyron wrote: ↑Tue Jul 23, 2019 10:38 amThanks Mike, I had a score that was positive for black, higher than what white gets on the opening position (maybe it was wrong, but that's what I thought.) I have many games won from the black side, because if your opponent plays bad enough to allow you to equalize, it may play bad enough to let you win, so I only offer draw in trivially drawn positions.
I'm not a believer of such scoring, if the position at move 40 is a mate for white, then surely it should show 100% for white. Then we'd know if at some point White gave up the win, and black gave up the draw with 53 gxf4, which would lead to another 100% score for white.
The smiley meant I was joking - you were absolutely correct to keep playing for the reasons you stated. Even in ICCF people make mistakes on occasion.Ovyron wrote: ↑Tue Jul 23, 2019 10:38 amThanks Mike, I had a score that was positive for black, higher than what white gets on the opening position (maybe it was wrong, but that's what I thought.) I have many games won from the black side, because if your opponent plays bad enough to allow you to equalize, it may play bad enough to let you win, so I only offer draw in trivially drawn positions.
The white king is going in the wrong direction if Kf2 is played. It loses at least 2 moves which gives black chances to organize a defense on the queen side. But like most losses this one isn't caused by one move. The are several questionable moves. I always like to nip a problem in the bud before it has any chance to do major damage. So I suggest he should have played 36. ... g5. I admittedly haven't looked at this move very deeply but it does seem to solve most of the problems of the position. The only problem I see with this move is that blindly playing it will NOT stop you from playing the same losing moves as was played in the game at some later time. You could transpose back in to the losing lines if you were unaware of what the problem with the position is.todd wrote: ↑Tue Jul 23, 2019 4:56 pmI agree that Kd1 is a better move.
However, what actually changes after Kf2? I have not looked at this in depth, only for a few minutes.
In either case, white is marching the king to the queenside and playing for zugzwang, but after Kf2 black has a little more time to do... something. What does black do, though?
I'm quite happy with a +8.00 score for the position in the OP. I just can't reproduce it, and I wonder if it's because of my lack of 6-men tablebases (can people show big score with just 5-men?) If so, this might be the very first game I see in my life where 6-men tablebases would have made a difference.
Back when I played this game I was using a "full arsenal" of engines analysis method. I analyzed this position with Stockfish and derivatives, Komodo, Houdini, Rybka, Texel, Shredder and Critter.
I have looked at this some more, and I don't see the loss until you played 53 ...gxf4. That is the only point where I am certain that you are lost. But it's clearly possible that my hardware and software is simply not adequate for this game. I think very few people would have access to computer that can find the winning line if you played something other than gxf4. Well, time for you to start that new unbeaten streak!Ovyron wrote: ↑Wed Jul 24, 2019 9:07 am
There's nothing wrong with 29...b5, and it's the only way to tempt white into a line that doesn't exchange the rooks with Rh1, because after Rh1 it's just a dead draw. Here, black fights on. Black messes up later.
Most engines were very useless for these positions (you know that when you have to play a move suggested by Critter...), so I had to rely on Komodo to play for an advantage (at it seemed to pay off for a while, until the opponent played into a mate in N position I saw as drawish, and couldn't see white's plan.)
You may have missed this post by zullil:
White was toast on move 40. I don't think anybody could have drawn the game against zullil at Depth 78 from here. I still don't know if white messed up the win later on. I still don't know what was black's losing move (as it happened before move 40.) I may need to download 6men tablebases to investigate, but I'm not sure it's worth the effort. The lesson for me here was about endgames with same-colored bishops being won for one side (no matter what engines say), this was something new and completely unexpected, specially after seeing endgames where one side has 3 extra pawns and it's a draw, those damn rooks are enough to defend them...zullil wrote: ↑Sun Jul 21, 2019 11:36 am [d]8/3b2k1/1p3pp1/3p1p2/3P1P2/P3P3/5KP1/5B2 b - - 0 40
+7.52 40... g5 41. Ke1 Kf8 42. Kd2 Ke7 43. Kc3 Be8 44. Kb4 Kd8 45. Bb5 Bh5 46. Bc6 Bf7 47. Kb5 Kc7 48. g3 Bg8 49. Be8 Kd8 50. Bg6 Kc7 51. Bh5 Be6 52. Kb4 Bc8 53. Be2 Bd7 54. a4 Be8 55. Bb5 Bg6 56. a5 bxa5+ 57. Kxa5 Bh5 58. Bd3 Kc6 59. Bc2 Be2 60. Bxf5 Bf3 61. Be6 Be2 62. Bg8 Bd3 63. Bf7 Be4 64. Be8+ Kc7 65. Kb5 Kd6 66. Bf7 Bd3+ 67. Kb6 gxf4 68. gxf4 Bf1 69. Bg6 Bh3 70. Kb7 Bg4 71. Bd3 Be6 72. Bb5 Bh3 73. Kb6 Bf5 74. Bc6 Be4 75. Be8 Ke7 76. Bb5 Kd6 77. Bc6 Bh1 78. Be8 Be4 79. Kb7 Bd3 80. Bf7 f5 81. Be8 Bf1 82. Kb6 Be2 83. Bc6 Bd1 84. Bb5 Bc2 85. Kb7 Bd1 86. Kc8 (depth 78, 8:57:10)
Looks like White wins.
Unlikely. I have plans to play some more games on the LSS, there some guys play shocking moves stronger than what I've seen 2400 elo ICCF players play, I expect one of then to trounce me at that point. I'll play in the FICGS World Cup just to train for the LSS guys, but I appreciate the sentimentWell, time for you to start that new unbeaten streak!
Code: Select all
sapeli -fen "8/3b2k1/1p3pp1/3p1p2/3P1P2/P3P3/5KP1/5B2 b - -" -analyze 100000
...
=
[ Time: 145.47s ]
[ Speed: 687.42 games/s ]
[ Score: 1246 - 1316 - 97438 [0.50] 100000 ]Code: Select all
sapeli -fen "8/2k5/5pB1/K2p1p2/3P1P2/3b2P1/8/8 b - -" -analyze 100000
...
=
[ Time: 76.17s ]
[ Speed: 1312.92 games/s ]
[ Score: 571 - 195 - 99234 [0.50] 100000 ]