Author: martin fierz
Date: 13:02:07 04/15/02
Go up one level in this thread
On April 15, 2002 at 12:37:38, Chris Carson wrote: >On April 15, 2002 at 11:18:11, Slater Wold wrote: > >>On April 15, 2002 at 11:12:58, Roy Eassa wrote: >> >>> >>>The analysis of the Smirin win over Shredder (round 1) says that Shredder made 4 >>>"?" moves in a row: >>> >>>http://www.chessbase.com/events/games/2002_smirin/smirin.htm >>> >>>See moves 7, 8, 9, and 10. Also "?" on moves 12 and 15. >>> >>>Is this common for today's programs on fast PCs? >>> >>> >>>Shredder 6.02 scores quite well against the other top programs. Yet it seemed >>>to play well below GM level here. How can both of these things be true? >> >>I used Shredder 5 on a Dual AMD 1.4Ghz (much faster than the PC used in the >>Smirin game) against a GM on ICC in a 5 game match. >> >>He said that Shredder played like a 2700 GM in SOME positions, but mostly, like >>an 1800 patzer. >> >>I believe that's why people (like myself and Bob Hyatt) seem to think it'll be a >>while before a computer program on a PC will be deemed "GM strength". Because a >>GM plays like a GM 99.99% of the time, while a computer will play like a >>computer 99.99% of the time. GMs are always GM strength, that's what makes them >>GMs! > >This is not true. GM's put in below 2300 performances (in tournaments) all the >time. See The Week In Chess, any week and you will see GM's with low and high >results for their ratings. A rating is based on a number of games, it is not >based on making one or several good/bad moves in one game. If you look at >individual GM games, you will find plenty of stupid mistakes. If you can not >find them, let one of the top programs show it to you. GMs make DIFFERENT mistakes. no gm would ever play this series of weak moves that shredder played in this game. GMs may blunder a piece in a 2-move combination occasionally. which a computer will never do. >You can find stupid mistakes by 2800+ human players against computers (you >should do this yourself as a project, it is an easy task). that's hard: only kramnik and kaspy have that rating, and since they have it, they never played a comp :-) i know, i'm nit-picking - but you just cannot compare kasparovs ...h6 in game 6 vs deeper blue with this type of computer error shown by shredder (and in van wely-fritz sss, exactly the same). i mean, of course you can compare them, but you should see that they are totally different. > >You should admire how well the GM played/prepared for the computer, that may not >be the case in all the games. > >Rebel has made a 2700 player look stupid, Chess Tiger and Junior have also. no. look stupid is something like what happened to shredder yesterday. LvW did not look stupid in his losses against rebel - rebel played a brilliant attacking game. >few losses that the programs have are always cited as "proof" that programs are >stupid, that is not the case. The top programs win a lot more than they loose >(2 to 1). The programs are GM's, but they are not invincible, they do not play >perfect chess. GM's do not play 2500+ everytime they are at the board or even >over the course of a tournament. That is the nature of chess until it is >solved, GM's and programs will win some you loose some. Ratings increse if you >win more than you loose. > >> >>You can put wool on a wolf, but it's still not a sheep. ;)
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.