Computer Chess Club Archives




Subject: Re: Most brilliant novelty from cct7 Witchess-Arasan

Author: Vincent Diepeveen

Date: 08:21:56 02/15/05

Go up one level in this thread

On February 14, 2005 at 19:18:58, Peter Berger wrote:

>On February 14, 2005 at 18:57:07, Arturo Ochoa wrote:
>>Well, if you think that Uri´s posting are not ludicrous, we are alll entering in
>>the absurd world of Uri. That is. It is a waste of time to repeat again and
>>again in this Forum, that a book tuned for a specific engine is a bid advantage
>>against other program or a GM.
>I am afraid that the absurd world of Uri is the one I am living in too.
>There are three kinds of opening books:
>a.) no book
>b.) random book
>c.) optimized book
>b.) should be the worst selection here for a strong engine, worse than a.), but
>it might be hard to prove, and it will not apply to all lines.

A is worst. If you lose 1 game you lose 5 in a row.

Please look at world champs 2004. Even our friend Kure in round 10 against diep
tried to repeat with fritz a line which diep played in round 2 against falcon.

Lucky Arturo had a novelty (Nc5!) somewhere at move 15 or something.

Even with a good optimized book they try to repeat lines.

Now if you play without book, each game will happen the same moves. And you get
in the same bad type of position.

Good example is 1.d4,d5 2.c4,dxc4.

Suppose the engine plays b4 always to cover c4, as soon as they figure that out
you play b5 with black, they will go 3.e3 and the rest of the games you will
lose in a world champs.

>c.) will outscore b.) much worse of course IMHO. That's the best choice against
>the worst.

>Trouble (or art?) is to determine how to do c.) . In the past this was hard,
>hard work, requiring much patience , thousands of moves to be entered in the
>book, but maybe this kind of opening book has some disadvantages by now. For me
>this seems to be Uri's major point, and I think it is a valid one.
>To improve on the engine's own choice it is not enough to just enter moves into
>opening books or to do some blundercheck. Think of the Kramnik-Leko Marshall
>game - this could have been played at a computerchess event too.

Not at all, no computer will consider f3.

>The interesting challenge would be to run a highly optimized opening book
>against one generated from a random source with maybe 20-50 changes altogether -
>is it really clear who would win this challenge ?
>There might be two or three book authors (the famous guys) who go beyond, have
>gotten rid of all automated lines (important!) and are competing with the
>grandmasters themselves with the help of the engines now. But from what I can
>see nearly all still start from some automated base, and this has become a major
>reliability by now.
>It's a bit late - hope the general idea still transports.


I hope that you realize 750 rating points means roughly 0% chance.

What is the chance in YOUR opinion a program without book in 2005 has to win a
world champs event?

If you say 0%, that means 700+ rating points.


This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.