Author: blass uri
Date: 03:01:21 08/06/00
Go up one level in this thread
On August 06, 2000 at 05:29:33, Adrien Regimbald wrote: >Hello, > >>Agreed, but certainly it will sell copies of Deep Junior, which is clearly what >>is intended by these statements. For I can't figure out any other possible >>explanation for the claims I see. I mean, clearly 50% against some of worlds >>best is not as good as +3 against the World Champion. There is NO possible > > >I think that being able to beat so called "Anti computer chess experts" >regularly would be a much more tremendous feat. I would rather see a program >which could perform at a "legitimate master level" than one that gets an overall >performance rating of 2700 or so against the cream of the crop. By this I mean >that the computer wouldn't fall into draws with players rated at 1200 (yes, I've >seen some players rated that low manage to hold some of the best computer >programs around to draws, and not by any cheap opening trick either, just by >steering the game into certain types of positions) 1200? The lowest rating in Israel is 1300 so there are no players with 1200 and the Israeli rating is similiar to the fide rating. If I use chessmaster personalities to see the meaning of 1200 then I can see that 1200 players do a lot of tactical mistakes and has no chance against the top programs. You claim that you see 1200 players get draw against the best programs. Are you sure that they are 1200 players? I am interested in games on a real board because it is easy to cheat in games on ICC. , and to be able to >consistently beat any player <2000, not just those that don't know how to play >an anti computer game (I've seen A-class players with a penchant good anti >computer play absolutely rip some of the top programs apart and for the most >part, rarely ever actually lose a game. You won't see this happen with a human >of 2700 strength, EVER). I do not believe that <2000 players will lose most of the games. In the israel league the teams could choose the players to play against the computers and computers won most of the games. There was one player with rating almost 2200 who drew against 3 computers but he is very good at drawing games and he also drew most of his games against humans(I think that he finished with 8 draws and 2 losses when part of the players that he drew against them had rating of more than 2400) > >There is a legitimate argument to be made about the strengths of the players - >the comps are tactical monsters, and if you can subdue them, you win. However, >you simply won't see a human player be made to look like an absolute patzer to >someone rated over 600 points below them, unless something extraordinary happens Programs usually win players with over 600 elo rating behind them. They may sometimes draw in a game with no mistakes and this is not because of the fact that they do not play well but because of the fact that they do not know the opponent. I believe that GM will also draw more often against players with the same style if the GM believe that they play against GM's. When the GM's know that they play against weaker players they may change their style and not allow drawish positions. Uri
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.