Author: Tord Romstad
Date: 09:12:07 02/16/05
Go up one level in this thread
On February 15, 2005 at 11:46:11, Anthony Cozzie wrote: >I dont know about 700 points, but the advantages of a good book seem obvious, >and I don't understand why we are even having this debate. As far as I can see, it's precisely the claim of 700 points which is the topic of the debate. Everybody agrees that having a good book can be valuable. >Anyway, I don't claim 700 points, but 100-200 seems definitely possible against >versus a poorly tuned book. > >It would be very interesting to run some test matches to test these theories: >for example Gothmog+Fritz8 vs Crafty+book.bin (according to WBEC there is about >a 100 elo difference between them). I have been staring at the above sentence for a long time, trying to figure out what point such an experiment is supposed to prove. You take a relatively weak amateur engine (Gothmog), take away its own, completely untuned book, and replace it with a book tuned for another, much stronger engine. You match this engine+book combination against one of the strongest amateur engines (Crafty), playing with its own tuned book. How can the result of such a match be used to estimate the Elo gain obtained by using a good book? A better experiment would be to let Crafty+book.bin play matches against several *strong* amateur engines, and repeat all matches with Crafty using some random untuned book. And if you (or some other programmer) decide to use one of my engines in some kind of match or experiment, please don't use Gothmog. Glaurung is stronger, and not nearly as buggy. Tord
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.