Author: Pete Galati
Date: 07:38:59 03/30/00
Go up one level in this thread
On March 30, 2000 at 02:13:04, blass uri wrote: >On March 29, 2000 at 20:46:17, Pete Galati wrote: > >>On March 29, 2000 at 19:53:59, Anatoli wrote: >> >>>Hi everyone ! >>> >>>I read many times that it is better for every chess engine to use it is own book >>>which was prepared for it by experienced players, even IG. OK, I looked through >>>the books which were prepared for Ftitz, Junior or Hiarcs. They are on the CDs >>>and everybody can copy them on your hard drive. But I can't see any point to do >>>it. There is nothing special about these books. There are thousands and >>>thousands games in there from the girl's and boy's tournaments aged under 10-12. >>>Who could chose these games for Fritz or Junior, what is the value of these >>>games in the Fritz's library ? What can it learn from these games ? As for me, I >>>deleted all games of this sort from my DB. So, my personal opinion is that the >>>books of most top engines are just a collection of any games which were squeezed >>>into a tree. Let's better make our own books. >>>Thanks >>>Anatoli >> >>It's very difficult to improve on opening books selected by the programmer, if >>you look here: http://members.xoom.com/avochess/MOREPGN.TXT you'll see a >>summary of what I put in a Crafty opening book that I made, it was a collection >>originally selected for Amy. >> >>This ended up being a 218mb opening book when compiled by Crafty at 60 ply, the >>one you can download at Bob's ftp is only 12mb. >> >>Here's a pgnres of a small tournament where I've been testing it against the old >>opening book, my book is "Crafty" and the original opening book is "Craftyob". >> >>There's only 4 games left in this tournament, between my opening book's Crafty, >>& TCB. The Crafty with my opening book is so far behind in percentage points >>that it won't catch up in 4 games. My book lost I think. >> >>Pete > >I think that your results are not statistically significant and we cannot reject >the conjecture that your book and crafty's book have the same value for crafty. > >Uri You're right, my results are not "statistically significant", but what I'm saying, is that Crafty's book has _more_ value to the program than mine does. Keep in mind, that my book is massive, and designed with other things in mind, while Crafty's book is a reasonable size, and specifically designed for Crafty. My opinion is that Crafty's book is better, not that they're the same value. Pete
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.