Author: Thorsten Czub
Date: 09:18:20 05/29/98
Go up one level in this thread
On May 29, 1998 at 10:42:34, Enrique Irazoqui wrote: >I have no trouble believing that Fritz 5 does play badly every now and >then. All programs do as we all have seen a zillion times. For example, >take the Kramnik-Shirov position posted in another thread. All programs >play 30.Ng5, a horrible move that loses the game. You can always wonder >about 2500+ players playing so idiotically. Nevertheless, on average >they play much better and score very well, thank you. > >So, what's your point? That F5 plays bad moves now and then? Big news... >:) > >Enrique Dear Mr. Enrique Irazoqui. I was asked to give some examples of WEAK play of fritz. I have given an example. It is - an example. Not more. not less. 1. I understand that you are not interested in games. 2. You are interested in NUMBERS. 3. I am in no way interested in numbers. 4. When I want to play chess and my opponent comes out with telling me numbers, i complain to the referree. My point is exactly what you said above, that f5 plays bad moves NOW and THEN. If this is no big news for you - i am sure it isn't - than be happy. I don't like programs to play those kind of moves/games. If you KNOW that fritz5 plays those moves, I wonder why you believe that THIS IS NO BIG news. But - as I said - talking with you (Enrique and Moritz) about chess or computerchess is senseless. You are only interested in counting Megabytes and numbers and make charts with statistics and give anything a number. This has nothing to do with my way of computerchess. I am NOT trying to find out which program is the fastest one in finding a key move in a certain forced position. Thats maybe YOUR understanding of computerchess. But not mine. These test-suites and the SOLVER programs finding these key-moves do mainly build up how fast their search is. And - since this is their only strength, these programs fail in other situations, where it comes NOT to solve the position in the fastest way, since there is no definite key-move to find, as we have seen in paris where crafty, ferret, dark thought, fritz and others were UNABLE to FIND the right moves in the fastest way, since there were no moves to find... Of course in the BLITZ tournament programs like Ferret HAVE advantages. When I design a program be fast, it will dominate in a blitz-tournament. I mean - it is obvious. Fast moves it the thing i designed it for. So what does all this: interesting king-position stuff here measure, if you don't measure if the program is able to direct into such a position, but instad, you measure how fast it finds the key-move. This has nothing to do with each other. I have not seen that Dark Thought - although computing much faster, although coming much plies deeper, although having a very fast hardware, although using all kinds of databases in the endgame, was able to win against CSTal. In the end the pure FAST/more depths/more NPS race was reduced to the question: does CSTal know that the opponent king has to be cut off at the edge of the board. Obviously CSTal did know about this beginner/novice-rule, meanwhile Dark Thought did NOT know about this rule. So the main-lines showed. And although it was FASTER, deeper and better in tactics, it was unable to come with the king out of the a-file because tals-rook was on the b-file. Of course you are not interested in these silly chess-stuff. You count numbers, NPS, Megabyte of Ram, results, make your statistics. Why should chess intereste you, you have fast computers. Thats enough. You have kyro-technicians. Thats enough. Thats not enough. And that is what my examples wanted to show. But talking with people not interested in games or chess is senseless. So I should better leave this club and talk with people INTERESTED in chess instead of interested in statistics.
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.