Author: Thorsten Czub
Date: 13:42:42 12/13/97
Go up one level in this thread
>I understand what you are trying to say... IE program A can crush >program >B in a match, but you can look at the games and see that B seems to know >more and play better moves. My only counter-point would be that if A >that A's moves are worse is likely flawed. Because I don't believe it >is >possible to play better moves and lose consistently. > >My conclusion is that if A beats B by a score of 200 to 100, then A is >simply better, no matter how you feel about its moves. After all, the >bottom line is winning, not playing "beautiful moves that lose..." Ok - lets say Kasparov wins against Bernd Kohlweyer 200 to 100 ! Now lets take 20 girls and let them kiss Kasparov and Bernd. Let them both drink 20 beer. Let them play again and watch if Kasparov wins still 200 to 100. Now what shall this example tell us ? I tell you what this strange example wants to say: Take Fritz5 on a 486-33 with 1 MB hash and also Cstal on a machine with the same hardware and hash. Let us guess what will happen ! Fritz loses 200 to 100. Now put them both on a k6/200 and give Fritz5 1 MB hash and also CSTal k6/200 and 1 MB hash. Let us see what happens. Maybe CSTal now wins only 170:130. Now give fritz5 128 Megabyte Hash-tables and also CSTal 128 Megabyte hash-tables. And suddenly: Fritz5 wins 200 to 100 or even more. What has happened. Although you have always upgraded BOTH machines in the same way, very fair for both programs, same hardware, same amount of ram, you get different results. The same thing happens often with humans. Although the same guys meet each other year by year, and although Kasparov is stronger than Kohlweyer, maybe he is his fear opponent and maybe yesterday Kasparov had a bad dream about Deep-Blue and is down and maybe, whatever, and Bernd wins "again". I think I can measure the system immanent problems and the CHANCES a program has, the chances to win. With measuring this chances I can etrapolate out of a few samples although it should not be possible. Maybe I don't really know the reasons. But I can feel them and this is enough to extrapolate the conclusion. When chess program X plays program Y, some mechanism work against each other. Similar mechanism, but biological ones, happen when humans have to play each other. And although anything is FAIR and no cheating, strange things can happen. NO measuring system can handle THESE strange effects. But the human (and any animal brain) can handle them- let us call it fuzzy-strategy. Using this fuzzy resources anybody of us can extrapolate or find out if he is used to do it. Don't you have a thing that is non-chess-topic where you are used to do it for 20 years, and you are an unknown "expert" in this area, and you could make bets that you will be succesfully be able to claim in forward what will happen ?? I am sure anybody of us has these FEATURES ! My grandma e.g. was very very succesful in finding out WHEN I WAS NOT SAYING THE TRUTH (in my youth as a child). And she was a succesful in finding out with other persons. Nobody in our family knew HOW she has done it. But she was able to SEE a lie. It was always a very HOT birthday of my grandma, when many guest were invited. Nobody tried to lie. They all had their shameful experiences with my grandma! Only good luck ?? Don't get me wrong. I am not sitting the whole day in my appartment and doing meditation or believing in ghosts nor do I try to walk over water or to fly in my appartment !
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.