Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: DJ7 and Nolot #1

Author: Daniel Clausen

Date: 01:06:28 04/18/02

Go up one level in this thread


Hi

On April 18, 2002 at 02:07:36, Slater Wold wrote:
>>Note that DT found it for the right reason (winning material).  DJ is finding
>>it by a simple evaluation term that happens to work well here.  The scores
>>obviously show that it doesn't think the move is winning anything...
>
>If computers only played moves they understood, well, you know.  ;)
>
>What you lack in skill, you hope to make up with a little luck.

If you look at hundreds of such positions, I agree that it doesn't really matter
why the engine chose the move. The sheer number of positions where it plays the
correct move (for whatever reason) makes it likely that it will also find the
right move (for whatever reason) in another ("similarly complicated" - whatever
that means) position.

On the other hand, if you take just ONE position, the reason why an engine chose
the right move is crucial.

Having said that, I'm not quite sure whether Juniors evaluation or DT's +1 is
the better reason. Ie it could very well be, that the line which DT gives a +1
is completely lost from a positional POV. Hard to tell since the thing doesn't
exist anymore. :)

Basically, that's the reason why I think all these DT/DB discussions a kinda
useless:

(a) you either have a big number of samples
(b) you have detailed info about the few samples

Since we have none of the above options, any DT/DB discussions are basically 90%
opinion. IMO, of course :)

Sargon



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.