Author: James T. Walker
Date: 21:11:59 10/04/02
Go up one level in this thread
On October 04, 2002 at 21:30:53, Joe Pechie wrote: >I have many chess programs but Fritz 6 is the one that prompts me to ask about >book learning. I have had Fritz 6 since it first came out. A few times I have >hit optimize book but changed right back to default. I have never been very >comfortable with book learning. Not that it is not a good thing, I just don't >understand it well. If I used my pc for only long time control match's between >programs I would not hesitate to set learning high. The problem is I tend to >play a mix of slow and fast games. I do not think I want the fast game info in >the learning file deciding moves at the start of a game. Even worse, I play the >program myself and I know I do not want that info in there. Seems as though info >like that could send a program down a bad road. It's like mixing audio for a >band. You can have a $100,000 pa and all the fancy effects in the world but in >the end ---- crap in is crap out. > >If any one would like to give me book learning 101 I would be grateful. > >Thanks >Joe >#20 Hello Joe, I have done some simple test with book learning. I don't think it is bad or has a real down side. It's main accomplishment is to prevent repeating losing games by playing the same bad lines over and over vs an opponent who will find a winning line and repeat it over and over. I have tried using the "learn from database" feature of Fritz to allow it to learn from 1700 games vs various opponents. The performance after learning from 1700 games was no better than the performance of Fritz with all book learning zeroed out in my test. Just my opinion based on my simple test. Jim
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.