Author: Uri Blass
Date: 07:14:33 04/03/02
Go up one level in this thread
On April 03, 2002 at 09:59:29, Sune Fischer wrote: >On April 03, 2002 at 09:42:10, Uri Blass wrote: > >>On April 03, 2002 at 09:22:25, Sune Fischer wrote: >> >>>On April 03, 2002 at 08:41:22, Uri Blass wrote: >>> >>>>>I do not think programs has the postional knowledge of a 1400 player. >>> >>>Then you will be so good as to show, e.g. in Crafty, where there are evaluation >>>stuff that could prevent the a4 move :) >>>I see nothing in Crafty, which is know to have a lot of eval, that players above >>>1400-1600 doesn't know, it is basic stuff of passed pawns, connected pawns, >>>material development etc.. beginner level stuff. >> >>The main point is that 1400-1600 may "know" wrong things about evaluating >>positions and I think that not knowing these wrong things is the main positional >>advantage of Crafty relative to 1400-1600 players. > >You can say that, but then again the evaluation scores of crafty will also >sometimes be wrong because the position is an exception from the general rules >(a4 again :). >The human can much more easily distinguish between the minor changes in the >position that can change something from good to bad, the program just mindlessly >gives a score for a rook in an open file, even if the file is completely >irrelevant. And a human may calculate and when he he needs to evaluate if to go to the position he may not pay attention to the fact that there is an open file. I know that there were cases when I calculated some line but did not evaluated the final position correctly because I did not pay attention to some factor in the position. >Also take the piece square tables, how can they ever be as finely calibrated as >the one the human uses. The human will keep changing his table scores, because >in some positions a knight on g4 is better than one on e4, it takes a human to >understand this (or brute force search). > >>A weak human may also forget about a weak pawn in the evaluation of the position >>inspite of the fact that he knows in theory that weak pawns are bad. >> >>Something like this cannot happen to a computer. >> >><snipped> >>>Why would he mostly win against stronger players if the sacrifices are wrong? >> >>It may be a win in most cases if the 1500 opponents blunder when they have to >>defend the position. >> >>a weak human may learn from his experience wrong things because his sacrifices >>may practically work when they are objectively mistakes in most of the cases. >> >>Uri > >I don't think so, it is much more likely the 1400 player will blunder against >the stronger player and lose because of the already bad sacrifice. >IMO weak agressive players are easier to defeat than the ones that just go for >the draw from the beginning. > >-S. I said in previous post that the player is better than 1400. Uri
This page took 0.01 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.