Author: Robert Hyatt
Date: 07:12:59 09/14/99
Go up one level in this thread
On September 14, 1999 at 08:27:33, Shep wrote: >On September 13, 1999 at 23:02:36, Robert Hyatt wrote: > > >>That's the best I can do at present... I have _zero_ idea why CM thinks that >>this is +3. One of the two programs is grossly wrong... I hope it isn't mine. > >There are some endgames where CM is completely off. Must be some eval terms >running wild. I've seen it up +5 in a totally drawn ending. >(Although I'd trust its middlegame eval more than anything else. :) > >--- >Shep I know of multiple programs that use a 'anti-human' evaluation. And one big mistake is "queens on = +1.00" which often leads them into getting attacked terribly by another computer because they try at all costs to keep those queens (and tactical threats against humans) alive. IE Vincent (Diep) seems to have a high value for a queen... and I have seen it get into serious trouble in endings where it has a queen for 3 minor pieces, and thinks it is well ahead. Crafty thinks the three minors are murder against the queen and it is generally right...
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.