Author: Uri Blass
Date: 22:45:25 03/20/04
Go up one level in this thread
On March 20, 2004 at 22:59:12, Robert Hyatt wrote: >On March 19, 2004 at 08:25:31, James Swafford wrote: > >>On March 18, 2004 at 12:18:30, Robert Hyatt wrote: >> >>>On March 17, 2004 at 21:19:45, James Swafford wrote: >>> >>>>Just what is a good eval worth? >>>> >>>>I ran Crafty on an AMD XP 2000+ (about 1.5 ghz) on ICC with a stripped >>>>down eval, then with it's regular eval to find out. Check out the >>>>summaries below (the top one is for stripped eval, of course). >>>> >>>>By "stripped down", I mean that the eval counted only material, some >>>>very crude pawn heuristics (doubled pawn, passed pawn, isolated >>>>pawn), and used pawn and knight piece square tables. That's it. >>>> >>>>I'm a bit surprised that the difference isn't a bit more; it's >>>>less than 200 points for blitz and bullet. I strongly suspect >>>>the difference would've been much greater for standard games, >>>>but I have no data to support that. >>>> >>>>Anyone done similar tests? >>>> >>> >>>If you don;t restrict ratings, this isn't as revealing. >>> >>>IE against a 1800 player, tactics will rule the day, regardless of the >>>evaluation. So stripping ot parts of the eval won't make a lot of difference. >>>But against a GM, it will do horribly with no eval... >>> >> >> >>You're clearly right, but I'm surprised by just how much. I >>was also surprised to see that the avg opponenet is so much >>weaker than Crafty when ratings aren't restricted. >> >>Maybe I'll do this again, with a rating window of +/- a few >>hundred points. >> >>What do you guess the difference would've been with that constraint? > >800-1000 at least. GMs will kill it with no eval. With normal crafty they will >rarely win a blitz game... I think that the difference in ICC is not very interesting because it is dependent on the question who are the opponents. I think that there are basically 2 questions: 1)What is the difference against computers. 2)What is the difference against humans. I do not think that ICC rating when you cannot control the opponent is the place to investigate question 1. I also think that rating against humans should be against different opponents because if there are some opponents who know to win you again and again and you play against these opponents a lot of games your rating is too low. The main question is what is the rating of a player in tournament when the player need to play against many oppoents when I believe that part of the humans do not try to play against specific computer weaknesses. You cannot force GM's to play against computer weaknesses and not to use their normal style and if part of them prefer to use their normal style then it is part of the equation to decide about the rating of the computer. If the part who prefer to use their normal style also avoid playing computers Your rating against humans is too low and not realistic. 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.