Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Does any "regular" chess program see White's win?

Author: Uri Blass

Date: 01:18:34 02/13/04

Go up one level in this thread


On February 13, 2004 at 01:47:40, Tord Romstad wrote:

>On February 12, 2004 at 22:17:46, Anthony Cozzie wrote:
>
>>Nullmove may be the best thing to happen to computer chess ever, but its not
>>perfect.
>
>I'm beginning to get a nagging feeling that recursive null move pruning (at
>least the conventional kind) is possibly the *worst* thing to happen to computer
>chess ever.  There has to be a better way to reduce the size of the tree, but
>I have no clue how to find it.  :-(
>
>The zugzwang problem is not too serious, of course, and if you really care
>about it it is not hard to solve.  The real problem with recursive null move
>pruning is that it performs horribly at finding long non-forcing lines.  For
>instance, a human player could take a quick look at a position and see that
>black needs to exchange off white's strong knight on c4, and notice that this
>could be achieved by the maneuvre f7-f6 followed by Bh7-g6-e8-d7-c8-a6xc4.
>A recursive null move searcher needs a huge search depth to find such plans.
>
>Tactically, recursive null move pruning performs really well.  Strategically,
>it's horrendous.
>
>Tord

I believe that chess is mainly about tactics and you can decide about rules when
not to use null move pruning(if it seems that the moves are a plan).

I do not do it but it is one of the things that I should try and the problem is
to have a function to decide if some moves are a plan.

In most practical cases at least part of the moves of the plan threat something
so you do not need a huge depth.

Uri



This page took 0.03 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.