Author: Uri Blass
Date: 06:32:03 07/14/04
Go up one level in this thread
On July 14, 2004 at 07:53:07, Ed Schröder wrote: >On July 14, 2004 at 01:50:03, Russell Reagan wrote: > >>On July 12, 2004 at 03:51:44, Ed Schröder wrote: >> >>>The question here is how Crafty being down a pawn, also facing 3 connected >>>passers (h2,g2,f2) evaluates this position. Did 42..a5! show a possitive score >>>for Crafty then hat off for Crafty. Did 42..a5! show a < -1.xx or even lower >>>then I would say Crafty indeed was lucky. >> >>I look at it from a different perspective. Crafty was the better engine this >>game. Dan Heisman says it better than I could. >> >>http://www.chesscafe.com/text/real.txt >> >>"A Chain is Only As Strong As Its Weakest Link >>The best way to introduce the second part of my explanation is to >>make an analogy. Suppose you build a home where the temperature >>is -20 degrees outside. You decide on a one-room home with four >>walls, a roof, a floor, and a heater. You decide to save a little time >>and material by finishing the four walls, the floor, and half the roof, >>but the other half you leave open. Even though you have completed >>over 90% of the structure, the temperature inside your home will >>still be about -20 degrees with half your roof open. If you want >>your inside heater to be effective, you have to enclose all of your >>home. >> >>The cold home analogy is similar to what happens when you play >>Real chess for 90% of your moves, but not for the other 10%. You >>think you are a good player, but weaker players beat you when you >>let down your guard for that 10%. In order to be a good player, you >>have to at least try to play correctly on every move, not just most of >>them. Consistency is important: remember that your chain of >>moves, in many cases, is only as strong as the weakest link." >> >>Crafty may have been outplayed for most of the game, but it was more consistent >>than Falcon, so it ended up the better engine that game. > > >I disagree with the comparison. In computer chess it sometimes happens a >position being so full of dynamics both computers impossibly can't see (predict) >the outcome neither by eval nor by search. In such a case luck moves in. I think >42..a5! is such a case. > >It's like in soccer, 2 equal teams, the ball by one unfortunate clutch coming >right for the foot of the opponent and BANG, you lose. Nobody is to blame. > >In human chess it is different, the human player can always be blamed for not >seeing the strategic consequences. Not always. It is possible that A outsearched B and both thought during the game that A got a big advantage only to discover later that A is losing when A understood that he is going to lose before B but too late. I do not say that it happens often but there are cases when it happens in human chess and the winner is lucky. It's different with computer chess since >strategic understanding hardly is to program and still is an unexplored area. > >To demonstrate my point, put the 42..a5! position in any program, let it think >for 1 ply and I predict no program will report a plus score for black. The question is if humans will report a + score for black only after 1 ply search. Maybe they understand the + score for black after some selective search of more than 1 ply. Uri
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.