Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The Null Move Killer Killed (and an announcement)

Author: Will Singleton

Date: 08:58:06 09/30/04

Go up one level in this thread


Thanks Stuart.  Chess programming, as you know, is strange.  Sometimes you'll
make a change you just *know* has to help, and it makes things worse.  And
sometimes a logically poor idea can improve things.

I think this is because we aren't dealing with perfect information.  That is, a
code change can affect the execution of other code, which in turn affects other
code, with the result being that the original code change might not always be
the real reason the behavior changes.  So you end up with one guy who has very
good results from a particular innovation, while others report no change or
worse results.

So, I wish I could present a theoretically sound argument as to why
no-nulls-near-leafs works for you and me, but my methods are almost always
empirical.

btw, while I appreciate the prize, I am an amateur, therefore I can't accept.
I'm sure Uri could use it, though. :)

Will



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.