Author: Arturo Ochoa
Date: 09:23:16 02/18/05
Go up one level in this thread
On February 18, 2005 at 11:07:14, Sune Fischer wrote: >On February 18, 2005 at 10:20:50, Arturo Ochoa wrote: > >>>>Somehow you must at least factor the engine+hardware strength out of the >>>>equation before you can say anything about the book being killer or not. >>>> >> >>What is your equation? > >Something like > strength = engine + hardware + book > >You say 700 for the book. Correction: I did not say that. Vincent said it. I dont agree with any of the absolute assumptions: Uri's Blass or Vincent's. The Vincent's is just the extreme contrary of the Uri's wrong point. >We already know hardware advantage to be in the order of 60 Elo per doubling, so >for a quad let's give it 100 Elo. > >Shredder system performs at ~2750 at world champs, that means shredder engine >(single game, no repetitions) on normal hardware is less than 2000 Elo??!! > Where is it written that? I did not said that? Why to suppose that I said that? :) >>>>Also these books are not subject for massive statistical testing, they are >>>>private and only used rarely which in makes it all a big load of guesswork. >>>> >> >>How do you know that are not subject of massive statitical testing if such tests >>are private? It is incredible how you suppose and guess without any relevant >>fact. That is the absurd world of Uri Blass. > >We don't have that statistics so to us it is guesswork. > >You claim to have such data but what you say sounds wrong, it would mean among >other things that Frenzee (random book) is about 300 Elo stronger than Shredder >(random book). No, I dont claim that I have the data of 500 engines tested. However, I am not claiming to the 4 winds that my point of view is absolutely correct and all the world is wrong (Uri's extreme). Incorrect assumption: I say, Depending on the engine the elo variation can vary. How much? It depends on the engine and book integration and how the whole components were tested. I am not saying an absolute number for all the engines. About, Frenzee. I dont have any idea. I have never tried that engine. > >Obviously something doesn't rhyme. > >>It seems to be you have got your root with your engine. And you seems not to >>realize what components are really important. > >It couldn't be that you just underestimate the importance of the engine? No. The engine is the core of the chess software component. If you add a tuned book for the engine and well tested (where the engine "feels" Ok with every resultant position after the opening)", the book evidently will help a lot the engine. How much? In Diep about 30% of the games. In Zappa about 25% of the games. In Frenzee, I have not tried that engine. > >-S.
This page took 0.01 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.