Author: Steve Timson
Date: 22:02:46 01/25/02
Go up one level in this thread
> >the +3 pawns looks a tactical score to me. but the +1 is a too optimistic >score for sure. > I checked out the pv for the +3 score. It is actually somewhat interesting. In fact material is even. The +3 is coming from a king safety score. Here is the position at the end of the PV where it is white to move: [D] 2r1k3/p2qr2p/1p4pB/8/4Q3/6P1/PPP2P1P/4R1K1 w - - 0 11 What has happened here is that stand pat in qsearch has missed what is happening. In this position, white chooses to stand pat, finishing the qsearch. My kingsafety penalizes the black king heavily here, because it is undefended by pawns, and attacked by all of white's pieces. Problem is, stand pat isn't valid here. White trades queen and rook into an ending that gets evaluated as roughly equal. My question then becomes, how do other programs avoid this false solution? Do they have more sophisticated stand pat logic? Do they not penalize the king as much because it is defended by friendly pieces? Stuff to think about. - Steve
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.