Author: Jouni Uski
Date: 03:38:24 10/22/01
Go up one level in this thread
On October 22, 2001 at 06:21:07, Uri Blass wrote: >On October 22, 2001 at 05:40:57, Chris Taylor wrote: > >>I got the new Fritz 5.32, and the new Chessbase version of Crafty 18.12. Then I >>had a thought, why not play them over night. Give them 40 minutes per game, >>bang them on the two 1200s. Let us see who is the better. >>Crafty v Fritz 10½-3½ 14 games played. >>crafty +10 -3 =1 >> > > >Something is wrong with Fritz's book > >It seems that it has only 1.c4 in it's book(otherwise I cannot understand the >reason that Fritz starts 1.c4 in every game) >Note that the result was +5 =1 -1 for crafty in the games that Fritz played with >white when Fritz won only the first game. > >I can understand 1.c4 in the first game and 1.c4 in the second game as result of >learning from the fact that it won the first game but I do not understand the >1.c4 in the third game and later. > >Even without learning the probability to play 1.c4 in every game seems to be >small enough in order to suspect that something is wrong. > >Uri Yes You are right - something is wrong. In the end even Crafty start to play only A22 opening. Is is it learning better? But I suspect Crafty learning is not working well as Fritz engine. Jouni
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.