Author: Dave Gomboc
Date: 13:12:14 09/06/99
Go up one level in this thread
On September 06, 1999 at 13:18:50, Peter Hegger wrote: >On September 06, 1999 at 06:26:24, Paulo Soares wrote: > >>On September 06, 1999 at 00:07:08, Robert Hyatt wrote: >> >>>On September 05, 1999 at 23:48:31, Paulo Soares wrote: >>> >>>>On September 05, 1999 at 12:49:02, Peter Hegger wrote: >>>> >>>>>On September 05, 1999 at 04:38:57, Paulo Soares wrote: >>>>> >>>>>(Snip) >>>>>> >>>>>Independent the problem happened with Rebel, I think GM >>>>>>Hoffman had the merit of choosing an opening that the programs >>>>>>have difficulty to play. >>>>>> >>>>>>Paulo >>>>> >>>>>I watched the game at ICC yesterday and the general consensus among the >>>>>spectators was that playing a Benko was an *awfully* risky undertaking against a >>>>>computer. I have to agree. A healthy Rebel 10-5 that wasn't playing lemons like >>>>>Ra2 (forget which move) or the terrible Rg4, just to name a couple, wouldn't >>>>>have lost the extra p quite so quickly IMHO. >>>>>Full credit to Hoffman though. He pounced on Rebel's errors swiftly and >>>>>accurately. >>>>>I hope to see them paired up again sometime. >>>>>Regards >>>>>Peter >>>> >>>>The Benko Gambit is usually played for draw by blacks, and great part of the >>>>resulting positions of this defense are difficult for programs evaluation. >>>>Hiarcs7.32 also plays 19.Ra2. >>>>My point in this subject is: coulded any program draw with a GM in the line >>>>that was played? My opinion it is that any program would lose this game for >>>>a GM, unless there was an arduous preparation in the program's book. >>>>Paulo >>> >>>I think Rebel could well have won this game... I saw evals of around +1 for >>>several moves, but then fishy things started to happen. I don't think white >>>was in danger of losing this at all, until hardware problems produced bogus >>>moves. White might not have been able to win, maybe, but it was certainly >>>not going to lose (I don't think) assuming it didn't make some gross mistake >>>later on... >>> >>>The GM made his share of mistakes too... >> >>Ed said that when Rebel was out of the book, after 15..Rb7 the machine crashed. >>So that of there in before it becomes difficult to do an analysis of Rebel's >>moves, but I made some tests with other programs, and I think that they don't >>analyze well the position, because the evaluation of the candidates moves are >>very similar. >>The first move played by white after the opening was 16. Qc1, a doubtful >>movement(I read now in a Ed's post some words that confirms this). 17. Bh6, >>19. Ra2 and 23.f4 don't also seemed good moves. >>I think that soon after 23.f4 Rebel was already in an inferior >>situation, and agaisnt a GM this can means a lose. >>The book used was the same as it was used by Rebel in the World >>Championship in Paderborn, would not be very important for Rebel use a >>specific opening book for the games against GMs? Only a question. >> >>Paulo > >Hi Paulo > >I still think it was risky to gambit a pawn, especially when it gives white a >passer on the a-file. Against another human perhaps it would work, but against a >computer I think a QGD or Queens Indian type set-up might have been safer. >That's just my opinion though. >If you don't mind could you post the other programs evals on 16 Qc1, 17 Bh6 and >19 Ra2? I don't have any programs strong enough to do a trustworthy analysis. >I'm interested to know how the other programs would have fared out of the >opening. Thanks. >Peter What's so risky about gambiting a pawn for long-term positional gains? I don't think that a machine will be better than a GM at deciding whether there is compensation for a pawn, short of a direct tactical refutation (which is unlikely to happen in the Benko!) Black has to go badly wrong before the "passer on the a-file" is going to go anywhere. Dave
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.