Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Practical lesson for statistics

Author: Sandro Necchi

Date: 23:08:00 12/29/03

Go up one level in this thread


On December 29, 2003 at 22:09:51, Mike Byrne wrote:

>On December 29, 2003 at 19:23:28, Sune Fischer wrote:
>
>>On December 29, 2003 at 19:08:12, Anthony Cozzie wrote:
>>
>>>On December 29, 2003 at 17:28:46, Sune Fischer wrote:
>>>
>>>>On December 29, 2003 at 15:47:05, Vincent Diepeveen wrote:
>>>>
>>>>I'm not exactly sure what your point is, we are not talking about humans.
>>>>
>>>>I guess you are implying that Crafty is 1800 and Shredder is 2830 and so Crafty
>>>>is without a chance.
>>>>
>>>>That is utter nonsense of course.
>>>>
>>>>In REALITY Crafty with a big hardware advantage is maybe 50 points weaker than
>>>>Shredder, so in your human analogy that would be like Kramnik winning in front
>>>>of Kasparov.
>>>>
>>>>If you play only 11 rounds it could happen.
>>>
>>>OK, lets do a little math.
>>>
>>>We know crafty on 2G P4 is about 20-30 elo stronger than Junior on a 266
>>>celeron.
>>>
>>>now, I think that:
>>>
>>>Using Junior's tournament book:        50 elo for Junior
>>>2x hardware advantage rather than 10x: 80 elo for Junior
>>>4xP4 vs 266 celery:                    80 elo for Junior
>>>
>>>Suddenly its game over for Crafty.
>>>
>>>So the question is: which of those statements do you disagree with?
>>
>>I'm assuming Bob can scale Crafty's hardware advantage to keep up, ie. that he
>>could get a 16 or 32 node machine to use against the quads.
>>
>>The only worry is if there is diminishing returns or some pratical problems in
>>using so many CPUs.
>>I do not believe in all that nonsense about tactical barriers, so I think it
>>will mainly be the technical challenge of it.
>>
>>About the book I don't believe in 50 elo.
>>If Bob uses his normal tailor made book full of ICC learning I think it can hold
>>its own, it may even be better than a book full of abstract theory the engine
>>doesn't understand.
>>
>>-S.
>>
>>>anthony
>
>I don't what a good book is worth (elo points) -- but some people get very upset
>when I play engines without their books - the argument they used is the engine
>should be used the way it is sold -- I have no argument with that line of
>reasoning if that is their purpose.  My own personal prefernce is use the same
>(unify)book for my tournaments - becuase I want to know which engine really
>plays better regardless of book. (that's all, personal preference, ymmv)
>Of course in Chess Computer Chanpionship tournaments, each author should be able
>to provide his own book -- as that is part of the preparation and the work of
>each author.
>
>For years, it was my opinion that MChess had the best prepared book for
>tournaments.  I don't know who that was, but he was very good.  Maybe it was
>Marty himself, I am not sure.

The book was made by me when the program was called M-Chess Pro. Previously it
was made by Marty himself.

Sandro

>But you really do need someone to work on a book,
>esp ically in a short tournament  and it has to be updated after every
>tournament.



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.