Author: Thorsten Czub
Date: 05:05:25 10/19/03
Go up one level in this thread
On October 19, 2003 at 04:24:19, Ed Schröder wrote: >>You cannot blame Rebel for Nf6. Others would have played the same move. >>(shredder7). > >ROTFL! > >Give me break :) shredder7 (not really a weak program, isn't it :-) needs to climb into search 17 to change from Nf6 to Qa5. the "score difference" between Nf6 Qe5 Be6 Ng5?! Qc8 Nxe6 Qxe6 and Qa5 Bh6 Qxc5 Qxc5 Nxc5 is from 0.25 to 0.24 ! you cannot really say that shredder has SEEN anything better. so what ?! blame rebel12 for playing Nf6 instead of the "better" Qa5 ? The problem is that (despite SOME programs like THe King) most programs don't see a problem in this position. Because MOST programs have no clue about king attacks. only the king (and few others) have, because they are expert systems for king attacks. >Look, the King was out of book on move 2, better praise the King for playing >12.0-0! all by itself. this is found easily by fritz8, shredder7. so lets praise it :-) the programs don't understand what is going on. they don't understand the difference between a good move and a bad move. they only emulate it. they oscillate between 2 moves, and the quality of the moves is so different, but the programs do not see it. it's really annoying that years and dozens of years after computerchess began the programs still cannot understand the difference between a good and a bad move. in this situation rebel came into a trap because it did not understood WHY and WHAT white was doing with its moves. almost all white moves lead to a plan. it's unimportant WHY the king played it. was it random. was it preparation. was it luck or good score/evaluations. the white moves did something altogether. and the computerprograms do not see what is going on. some see. others not. IMO a sad story. ok. i give you a break. >Today is another today, yesterday is forgotten. > >Take care, > >Ed > >======== the problem was the book. it's not a big problem. jeroen was unable to eliminate the line because rebel came LATER in this line. The king played the line maybe by luck or by chance. whatever. as long as we try to emulate chess in computerchess programs, this will happen. IMO programmers should try to teach chess programs to understand the difference between Nf6 and Qa5 instead of trying to have the strongest chess program in the SSDF list. this brings nothing. it makes you have a BRILLIANT chess program leading ssdf, but the program is not playing chess, it emulates it. it kicks out the other programs, but not because it understood. only because of search tricks or faster algorithms. shredder leads the ssdf list. and evaluates Nf6 with +0.25 for white. then Qa5 is found with +0.24. it sticks to it and later says +0,08. you still cannot really say that it understood the difference. How can we make programs understand what is going on ?!
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.