Author: Robert Hyatt
Date: 11:13:23 09/15/03
Go up one level in this thread
On September 15, 2003 at 13:38:38, martin fierz wrote: >On September 15, 2003 at 11:28:07, Robert Hyatt wrote: > >>On September 15, 2003 at 09:11:25, emerson tan wrote: >> >>>I played a 24 game match between Shredder 6.02 and Hiarcs 7.32 with the >>>following opening. >>> >>>1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.Nxe4 Nd7 5.Ng5 Ngf6 6.Bd3 e6 7.N1f3 h6 8.Nxe6 >>> >>>Time control was 40/4hrs + 20/2hrs + 2hrs to finish on a 1.8 Ghz processor. >>> >>>Shredder played all the white games and Hiracs all the black games. The score >>>went 13-11 in Hiarcs favor. Only a couple of games were drawn. >>> >>>I choose Shredder 6.02 to have white because it likes whites position and >>>without an opening book, it will play the 8.Nxe6 sacrifice. Also, Shredder has a >>>positional learning. I choose Hiarcs for black solely because of its positional >>>learning. Positional learning is important in this one opening match since it >>>will be able to learn and improve its next play based on the same opening. >>> >>>Maybe 8.Nxe6 is not a forced win for white. Maybe Kasparov can study it and use >>>it in some high profile match against computers since most of the programmers >>>might put it in their books thinking its a forced win for white. >> >> >>The problem is that while Nxe6 probably wins, it is _not_ easy for white to >>find the win. An IM played several commercial programs (he played black in >>every game) several years ago and won every time with black. Black _is_ a piece >>up and white has to justify that on _every_ move or lose the game. > >i disagree. the white attack is definitely not as critical as you make it. Perhaps I didn't word it very well. I didn't mean "critical". I meant that there are _many_ places where white has to find the right move, or black slips away with a piece advantage. If it were easy, the IM would not have won every game with black, against good programs (I think this was 1997-era hardware as this was done within a few months of the match, I believe). > there >are many positions where a single move makes or breaks your attack, but this is >not one of them. basically, white simply develops and throws all pawns forward >in the hope of opening up the position; white wins by exchanging pawns and not >exchanging pieces. >black's main defect, the terrible position of his king, is a long-term defect >which does not vanish just like this. No, but the king sits on the verge of escaping many times. A computer is all too happy to pick up a pawn while the king makes that one critical flight square and the game ends there. >i suspect that the programs are happy to exchange the queen as white if only >they win another pawn, or maybe even two. and then they lose, of course.... That's the problem with computers here. Picking up a pawn is all the minimax search wants to do, but it isn't near enough... > >BTW, nunn's chess openings gives the position after nxe6 fxe6 bg6+ ke7 oo as >"with compensation". definitely not as +-. looking at the results in chessbase, >i guess it's clear that the practical chances among humans are all with white >though. Note I said "probably winning". Way too complicated for me to reach any real decision, and letting two computers play it out certainly won't convince me it is won, lost or drawn either. :) > >cheers > martin
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.