Author: Mike S.
Date: 20:54:40 08/19/02
Go up one level in this thread
On August 19, 2002 at 22:52:50, Robert Hyatt wrote: >(...) >So you don't think a top-10 GM makes 5 blunders in a game? >I'll bite. >First, how to define "blunder"? >Major positional error? >Dropping a pawn? >Failing to win a pawn? I think things like that are very difficult to analyse if you don't are 2400+ yourself. I doesn't help much that programs are as good, or even better than GMs in terms of tactics. It may be that he actually didn't drop a pawn, but sacrificed it for positional or dynamical reasons which are very difficult to judge upon, below that playing strenght. Also, he may seem to have failed winning a pawn, but actually decided to reject a pawn offer because he saw the opponents gets compensation (which a program may never see, or only after a very long calculation time...) >Winning a pawn when he could win 2 or 3? This sounds like a more severe case (but will be much less often to find probably). >failing to find a mate although the move played still wins? I think this is not too unusual when the game is really decided already (the GM goes the easy & safe way then, if available). - But can be called a blunder "objectively". But there's a risk of "finding blunders" which actually aren't. >(...) >>>> vincent is talking about john nunn's >>>>excellent book on tactics, "john nunn's chess puzzles" (or something very >>>>similar to that). he compares two tournaments, karlsbad 19-little and the biel >>>>interzonal of about 1990. he used fritz in blundercheck mode to get some kind of >>>>objective measure of the number of errors being commited in the two tournaments, >>>>and the result was that 1920 they were playing abominable chess. I don't know that book, but I'm somewhat surprised that a *GM* would base such analysis on a Fritz blundercheck. It must have pointed him to many moves (amongs some real blunders maybe) which actually were not blunders but moves like positional sacrifices, I assume. I tested the same approach once, analysing Tarrasch's comments of the Nuremberg 1896 tournament. I focussed on tactics though. Either the progs acknowledged Tarrasch's analysis immediatly, or - if it seemed he may have been wrong - the programs just needed more time to see that Tarrasch's lines were correct. I didn't find a single mistake. GMs must not be underestimated :o) Neither today's, nor the masters of the past. Don't trust the computer... :o)) Regards, Mike Scheidl
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.