Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Analysing while retracting moves

Author: Robert Hyatt

Date: 17:43:22 11/24/01

Go up one level in this thread


On November 23, 2001 at 15:35:57, Tom Kerrigan wrote:

>You make it sound like if a bad score is somewhere in the hash table, that bad
>score will be reported by the search.
>
>I.e., I don't see why a solid position would give you a bad score just because a
>mistake was made later.
>
>-Tom


I suppose you missed the point of the discussion?  Searching from back to
front will move the detection of a bad move back from position A (the point
where the search can find that things are bad with no help) toward position
B (the point were a bad move was actually played in the game.)

If you analyze front-to-back, you find the problem at move A every time,
where the depth is predictable because you know how deeply the engine can
search and how the extensions work.  If you analyze back-to-front, some scores
persist in the hash and move the "bad move here" indication back a few moves
prior to A.  How far?  Varies based on hash table size, what gets overwritten
by chance, etc.

I'm not saying _either_ is better or worse than the other.  I'm simply saying
front-to-back is _consistent_ while back-to-front is not.



This page took 0.01 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.