Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To NON-believers in EGTB benefits... (some engines benefit greatly...)

Author: Tord Romstad

Date: 00:05:16 11/21/05

Go up one level in this thread


On November 21, 2005 at 01:48:25, enrico carrisco wrote:

>I understand that -- but my mainpoint was that EGTBs help in the same position
>and *do* announce mate.  Unless Fruit can "auto-disable" its permanent NullMove
>Pruning in such a situation than it benefits from EGTBs.

This is true, of course.  Nobody would disagree that for any given program,
it is possible to find a position where it would find a win with EGTBs, and
where it would not be able to find a win within reasonable time without EGTBs.

It does not automatically follow that adding EGTBs is the best thing to do,
however.  Instead of adding EGTBs, you can try to understand why the
program is not able to find the win on its own, and to solve the problem
by improving the search and/or evaluation (a trivial example is disabling
null move search in the position you gave us in the beginning of this
thread).

I think trying to play basic endgames perfectly by using search and
evaluation only is a better approach, for several reasons.  The most
obvious one is speed:  Probing EGTBs is a very time-consuming
operation.  It also strikes me as an extremely inelegant solution.  I
really don't want to use multi-gigabyte lookup tables just in order
to play the most basic endgames correctly.

The most important reason of all is that the techniques we discover
when trying to master basic endgames with search and evaluation
are likely to be useful even in more complicated endgames.  In my
opinion, a better way to use EGTBs would be to use them as an oracle
when implementing endgame evaluation rules.  We could analyse all
the positions in an EGTB and try to find general rules and patterns
which enable us to classify the majority of positions as won, drawn
or lost without consulting the EGTB.  With some luck, the principles
we discover can be applied to a wider range of endgames.  It could
even turn out that they can be formulated in a way understandable
to human players, and help us all to improve our understanding of
the endgame.

Tord



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.