Author: Chris
Date: 07:32:45 01/07/03
Go up one level in this thread
On January 07, 2003 at 07:59:24, Mark Young wrote: >On January 06, 2003 at 16:57:11, Uri Blass wrote: > >>On January 06, 2003 at 14:32:18, Uri Blass wrote: >> >>>On January 06, 2003 at 14:15:00, Chris wrote: >>> >>>>On January 06, 2003 at 14:06:54, Uri Blass wrote: >>>> >>>>>On January 06, 2003 at 12:45:28, Chris wrote: >>>>> >>>>>>Some people here seems to doubt that it is possible to beat the top engines, but >>>>>>I found out that you can still do it with the normal means of setting up the >>>>>>Stonewall as White. >>>>>> >>>>>>These games was of course played without takeback or other kinds of cheating. >>>>>> >>>>>>My OTB ELO is 2230 and I have studied computer weaknesses a lot. >>>>>> >>>>>>After this game, I lost six in a row, and I decided that maybe I need some more >>>>>>time the next time I challenge Fritz :-) I believe I would have a better chance >>>>>>at tournament speed. >>>>> >>>>>I believe that at tournament speed Fritz8 also can play better. >>>>>It is possible that it will not let you to win by the same strategy and the only >>>>>way to know is to try. >>>>> >>>>>It is a mistake to get a conclusion from speed chess about tournament time >>>>>control games. >>>>> >>>>>Uri >>>> >>>> >>>> >>>>I believe Fritz would be stronger tactically in a longer game, but I don't think >>>>it would change so much strategically. I was getting into this position by >>>>following its book. >>> >>> >>>If the book has no errors then the question is simply if Fritz does the same >>>mistakes if you give it some minutes to calculate. >>> >>>I do not know what is the point when Fritz8 starts to show negative evaluation. >>> >>>analyzing with yace and going backward so yace can learn shows that yace see a >>>clear advantage for white after 31...hxg6(1.84 pawns) >>> >>>Yace learned the following line >>>31...hxg6 32.fxg6 Rff8 33.Ng3 f5 34.Nxf5 Qe6 35.Qg5 Kg8 36.Qh5 Rxf5 37.Bxf5 >>>Qxe3+ 38.Kc2 Qxg1 39.Qh7+ Kf8 40.Qh8+ Ke7 41.Qxg7+ Kd6 >>> +- (1.84) >>> >>> >>>Yace does not like 31...hxg6 after learning the line and I have the following >>>analysis >>> >>>[D]4r2k/p3qrpp/2b2pP1/3p1P1N/2pP1Q1P/2P1P3/P1BK4/6R1 b - - 0 1 >>> >>> >>> >>>Analysis >>> >>>31...Rff8 32.gxh7 Rf7 33.Qg3 Kxh7 34.Nf4 Kg8 35.h5 Qc7 36.h6 Rfe7 37.Ng6 Rd7 >>>38.Qxc7 >>> = (-0.23) Depth: 13 00:01:32 33598kN >>>31...Rff8 32.gxh7 Rf7 33.Qg3 Kxh7 34.Nf4 Kg8 35.h5 Kh7 36.Ne6 Kh8 37.Qg6 Bd7 >>>38.h6 Bxe6 >>> = (0.11) Depth: 14 00:04:29 94591kN >>>31...Rff8 32.gxh7 Rf7 33.Qg3 Kxh7 34.Nf4 Qd7 35.Ne6 Kh8 36.h5 Qe7 37.Qg6 Bd7 >>>38.h6 >>> = (0.11) Depth: 15 00:16:59 324004kN >>> >>>(blass, tel-aviv 06.01.2003) >>> >>>Uri >> >>Yace needs a long analysis to see that 31...Rff8 is also bad >> >>The score continues to drop and I stopped it before waiting until it finish >>iteration 17. >> >> >> ² (0.48) Depth: 16 00:45:34 895399kN >>31...Rff8 32.gxh7 Rf7 33.Qg3 Kxh7 34.Nf4 Rb8 35.Qf3 Qd6 36.Qh5+ Kg8 37.Rg2 Be8 >>38.Ng6 Rfb7 39.Qh8+ Kf7 40.Ne5+ Ke7 41.Rxg7+ Kd8 42.Rg8 Rb2 43.Rxe8+ Kc7 >> ± (0.88) Depth: 17 01:41:40 2047157kN >>31...Rff8 32.gxh7 Rf7 33.Qg3 >> ± (0.88) Depth: 17 02:32:48 3034226kN >> >>(blass, tel-aviv 07.01.2003) >> >> >>I can add that yace suggests 30...Rg8 after all this learning so it can reply >>g6 by h6. >> >>After 30...Rg8 it can see 31.gxf6 as the best move. >> >>There are 2 questions here >> >>1)Is it correct to assume that 30...Rg8 is better and 30...Rf7 was the losing >>move? >>2)How much time do programs need to avoid 30...Rf7? >> >>Uri > >Uri the computer (fritz 8) only needs a few seconds to see that Rg8 is better. > >With out this blunder White is not winning and may not ever be better, but worse >in this position. > >I can not agree with what Chris wrote...As the computer was not lost but to a >blunder in a good position.....but still hats off to Chris still a good win. Thank you, but how much is 'a few seconds'? I believe White had a good position, not necessarily winning, but the comp made it possible to build up a strong pressure and then made a miscalculation and so it lost. However, it is not only a question about this game, it also happened in a few of the games I lost, when I was the one missing a tactical point in the end. All things considered, my conclusion that it is possible to build up these attacks fairly easy against this program (I am using the 8.0.0.8 version), and sometimes it wins, sometimes not. However, the winning chances seems to be good in this setup if you can check out the tactics (which is the real difficult part) In the game 31. Qf3 e.g. with the idea of Nf4-g6 is one way to continue the attack. White has compensation, but I believe Black saves a draw by optimal defence. That is mainly a feeling, it will take days to confirm and analyze a position like this to the bottom. Also 31. Rg3 (with the idea of covering e3 and continuing Qg4, Nf4-g6+) has some merit, but 31. Qf3 seems to be slightly better.
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.