Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: books

Author: Bruce Moreland

Date: 23:32:41 09/11/98

Go up one level in this thread



On September 11, 1998 at 21:43:46, Mark Young wrote:

>Bruce, I would invite you and anyone else that would care to to please go
>through my thought process with me on a line by line basis with me and agree or
>disagree with it. My intention is not to deal in any way with implications
>against Thorsten but to just deal with the facts and if possible to come up with
>cause behind the facts. I think you might more clearly understand my intentions
>by doing this.

I think that I understand this pretty well.  You insinuate that Thorsten is
fiddling with the book.  It'd be easy to do, it'd be hard to catch, and if you
saw low probability bad moves coming out of Fritz often enough, and you had
reason enough to believe that the operator had an ideological agenda, it might
even seem like the most likely explanation.

I know that you, Enrique, Moritz, and maybe Dirk are frustrated by this
tournament.  But let us try to figure out a way that we can handle this
situation constructively, without calling people cheaters this time around, and
without making posts where people are indirectly called cheaters.  I don't think
this new strategy is much of an improvement over last time.

I have done experiments with wide automatically generated books.  It is hard to
write it properly, so that you don't go down bad lines.  If the book is large,
which I expect Fritz's book is, this problem gets worse, not better.

If you play a move based upon straight frequency, the odds are good that you'll
go down a rare line at least once in a game, and if the rare lines are all bad,
you're going to get screwed up a lot.  As a possibly unrealistic example, if you
have a situation four times per game where you have a bad move played 10% of the
time along the line you play, the odds you will play a bad move at least once
are 1 - 0.9^4, which is about 35%.

If you don't do it by straight frequency, but do it instead by result, you can
still have problems because you'll still go down bad lines sometimes.

You said you looked at those games Fritz played, and analyzed the opening book
choices that Fritz made.  What did you come up with?

bruce



This page took 0.02 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.