Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Has anyone done a Crafty version tournament?

Author: Robert Hyatt

Date: 13:52:19 09/08/04

Go up one level in this thread


On September 08, 2004 at 15:58:49, David Dahlem wrote:

>>Even if your strange book lines put the program into positions/openings that the
>> author _never_ intended to try to handle???
>>
>>I know lots of very strong chess players that don't play the French, for
>>example.  Do they _have_ to play it to prove they are strong?  Or can they prove
>>they are strong by playing _other_ openings and never playing something that
>>will transpose or go directly into a French as black???
>>
>>If I want to write a "Karpov emulator" must I _still_ make it an expert at
>>attacking using the KID as black?  Even though it is never going to play the KID
>>as black?
>>
>>That's the point that is getting missed here.  Engines do not need (nor do they
>>try) to excel in _every_ kind of opening position.  IE Crafty can play
>>Fischer-random chess.  But not very well because it is specifically developed to
>>handle the normal starting position.  Eval terms need to change in ways unknown
>>to me to handle oddball starting positions.
>
>Once again, i don't use strange book lines, they are all standard eco lines, all
>chess rules apply.

Once again, I define "strange book line" as "any book line crafty will not play
on its own."

I do not see why that is so hard to understand, but apparently it is.

Strange or oddball doesn't necessarily mean openings like 1. f3 and 2. Kf2.
Strange can easily mean 1. e4 c5 if the program will not play 1. ... c5 on its
own, but is forced to by a poorly tuned book (poorly tuned for that particular
opening since the program doesn't prefer such a line.)

What such matches proves, I have no idea.  But it absolutely doesn't prove much
about the overall strength of the engine.  Perhaps is says something about the
engine's versatility in handling every type of opening, but who cares when the
engine is not going to play such openings.

For example, would anyone really care when two track stars run the 100M dash in
a contest, when one of them only competes in the 5000M race?  The 100M dash
won't mean a thing except for the one that actually runs the 100M race at meets.

Yet that is _exactly_ what you are doing.  Forcing one of the players to compete
in something it won't normally be doing.


> There are lots and lots of testers using the same method i
>use, based on messages in the forum.

Fine.  There are lots of people that do drugs.  Rob banks.  Drive insanely.  I
don't necessarily copy their behavior...



> You can't change my mind, and i'm sure i
>can't change yours, so i see no need to continue this back and forth dialog. :-)
>
>Regards
>Dave



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.