Author: Robert Hyatt
Date: 11:40:01 07/15/03
Go up one level in this thread
On July 14, 2003 at 15:05:34, Joachim Rang wrote: >On July 14, 2003 at 09:41:23, Uri Blass wrote: > >>On July 14, 2003 at 08:14:53, Joachim Rang wrote: >> >>>On July 13, 2003 at 19:35:15, stuart taylor wrote: >>> >>>>I find this interview fascinating to say the least! >>>>He also seems to be saying that Junior 8 (or "deep Junior", but the actual >>>>junior/deep junior on the market now is known to be even better, in quality!) is >>>>the first computer to understand quality moves VS. material. >>>>Does this mean that Junior 8 will be immortalized as the first of the real big >>>>super GM programs? >>>>S.Taylor >>>> >>>> >>> >>>this is all nonsense. Kaspy said after his Deep-Blue II - debacle, that now the >>>time has come were computers play better than humans but changed his mind after >>>properly analyzing the games. Now after the Junior-Match he also predicted a new >>>era bla, bla, bla. >>> >>>I think this is mostly to mystify his matches against the machines into >>>something special. >>> >>>Juniors understanding of dynamic advantages and his willingness to give material >>>for initiative is outstanding but there are other programs which play good >>>positional moves too instead of grabbing material. In some position Junior shows >>>a quite poor evaluation and does not find decent moves at all. So the statement >>>that Junior plays "quality moves" and other programs don't is not true. For >>>example the understanding of Hiarcs 8 of the pawn structure is outstanding >>>compared to most other engines. Junior 8 treats his pawn structure rather badly. >> >>I do not have Hiarcs8 but based on 2 games that I watched Hiarcs8 had problems >>of evaluating passed pawns. >> >> >>http://www.talkchess.com/forums/1/message.html?305220 >> >>Hiarcs8 underevaluate a passed pawn of Junior8 and lost the game. >> >>In another game of Hiarcs against GM smirin(in the match that smirin won the >>machines 5-3) I was also not impressed by Hiarcs pawn structure understanding >>and Hiarcs overevaluated it's passed pawn that was trapped by smirin. >> >>I think that passed pawns is part of evaluation of pawn structure and I do not >>know what do you mean when you claim that the understanding of Hiarcs8 of the >>pawn structure is outstanding. >> >>Uri > >AFAIK Smirin played not against the final Hiarcs 8 Version. Mark Uniacke has >completely rewritten the pawn evaluation code for Hiarcs 8, I'm not sure if this >was already done, when Smirin played against Hiarcs. > >Well opinions might differ but I have the feeling, that Junior and sometimes >Shredder make unnecessary pawn advances in certain positions, whereas Hiarcs is >much more cautious with pawn advances and cares more about its pawn structure. >But maybe I have too little examples. > >regards Joachim One thing I noticed _years_ ago was that most commmercial programs are overly aggressive at pushing pawns. But they are very solid players so they can get away with it. I remember some advice from a GM _many_ years ago, that went something like "push a pawn when you need to, but be very careful, because once you push it, it is pushed and can't go back. Every pawn push creates weaknesses in your pawn structure that are not there when all the pawns are on the second rank." Cray Blitz was pretty good about this, but when I wrote Crafty, I found that it could lead to problems, since pushing pawns creates space, and space creates a bind for the opponent, and a bind leads to tactics that are good for you and bad for him. So eventually, crafty became more aggressive, although it does not go too far overboard at the moment. I'm also reminded of the very early deep thought machine, that pushed pawns aggressively and ended up with what looked to me to be an ugly pawn structure in almost every game. But, of course, it won every one of them, so who is to say that "ugly" is the same as "losing". :) I think that so long as a program doesn't sink to 8 isolated pawns or something equally silly-looking, it has enough tactical skill to make exploiting such a weakness very hard. _until_ you run into the right human that is willing to take his time and work on a weak point, rather than going overboard and throwing that advantage to the wind.
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.