Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Killer Book

Author: Don Dailey

Date: 15:22:30 06/19/98

Go up one level in this thread


On June 19, 1998 at 13:53:11, Bruce Moreland wrote:

>
>On June 19, 1998 at 12:23:56, Ed Schröder wrote:
>
>>#5. Killer books have a bad name because the work (beating the opponent)
>>is not done by the chess program itself (the engine) but by the work of
>>a human at home.
>
>The reason I don't like them is that the program is no longer general purpose.
>
>Some people consider whether to buy programs based upon their performance in
>computer vs computer competitions.
>
>This is not a very good reason to choose one program over another, but when the
>programs have been tuned to play well against each other, it's even worse.
>
>You expect that if you buy something rated 2600 that it will play better
>*against you* than something rated 2500, and a killer book might get the rating
>delta without playing at all better against the customer/opponent.
>
>I think that the average person should buy programs based upon playing style,
>strength against humans, and features, not necessarily in that order.
>
>bruce

I don't like the killer books for pretty much the same reason.
Although I've often argued that the book is an integral part
of the program,  I am much more interested in how a program plays
a variety of positions once it has to calculate on its own.
A mediocre program with a killer book is not very interesting.

I think for practical reasons it's something we probably have
to live with.  It would be interesting to find a technique to
take advantage of the programs that use killer books.  This
would be the ultimate cure.  But I cannot think of any
reasonable way to do this.

- Don



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.