Author: Jonas Cohonas
Date: 01:12:20 04/06/02
Go up one level in this thread
On April 05, 2002 at 22:37:53, pavel wrote: >On April 05, 2002 at 22:08:48, Jonas Cohonas wrote: > >>On April 05, 2002 at 21:33:22, Jerry Doby wrote: >> >>>Mr. Berg is a legitamate 2500 elo player and fritz is winning the match, I have >>>yet to see a bad computer result vs humans, yet some people stubbornly hold on >>>to the nonsense that computers are not 2500 elo????? Barbaric ignorance I would >>>say for sure!!! >> >>Even though this is an obvious troll, i will reply. >>I think the argument that computers are below 2500 is valid to some extend, the >>issue here is that comps does not do well in certain positions while in others >>they do extremely well and that makes it hard to say if a program is above or >>below a certain strength, if you don't have the exact position from which it is >>to be judged from. For example in the Van Wely v Reb Cen4, Wely played Bg3!? and >>Rebel went on to take the peice and lose the game (i don't have the position >>here, but i am sure someone could dig it up) then in the last game it won in >>brilliant fashion and played one of the best comp v human games i have seen. Now >>how can you judge a programs overall strenght from that match (as an example) >>being open to the fact that it played below 2500 and above 2500, determining the >>strenghts of computers today are really difficult and personally i believe that >>comps are above 2500, simply because of the fact that if they where to compete >>in human matches they would never get tired or make obvious mistakes+the >>implicit fact that they don't have to take psychology into account. >>When i see discussions on comp strength there is one side arguing that they are >>beyond any doubt stronger than 2500-2700 and on the other side there is the >>complete opposite saying they are below 2500-2300 and IMO i don't think that it >>is that simple. >> >>Regards >>Jonas > >I jonas, > >In regards your post (not the trool), IMO those few (rare!) positions that >computers dont understand these days are slowly become less and less. Also I >dont believe that the best chess program of today (obviously Fritz7), is not >below 2500 on any well known rating system out there against humans. And I am >not going to believe that even if the author of fritz himself comes to me and >tells me that. :) > >I dont think 2500 is a target anymore. > >Regards, >pavs 2500 was just a # and i did say that i belived that they where above that # that could mean from 2500-2800+ depending on the position. You are right positions where comps play less than great are getting fewer or should i say, the programmers are getting better at avioding them :) It is a pretty welknown fact that most comps have a weakness when faced with closed positions, so much that the programmers have implemented code that prevents (or tries to) closed positions, sound or unsound! now this does not mean that they don't do bad in closed positions anymore, they have so far just found a way to avoid them better. Regards Jonas
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.