Author: stuart taylor
Date: 17:04:16 05/04/03
Go up one level in this thread
On May 04, 2003 at 13:35:14, Joachim Rang wrote: >On May 04, 2003 at 13:06:23, stuart taylor wrote: > >>The real test would be in a position in which there is no king safety, BUT the >>tactics do NOT work. THIS is not a good proof to anything if the tactics DO >>work! >> > >I disagree. There might be exceptional positions like this, where the tactics do >not work and black may escape. I guess, that Shredder may look to optimistic on >such positions but most such positions _have_ tactics. The key to playing >strength is not perfect evaluation but good speculation. Really!? > > >>However, to the naked human eye, this position looks TREMENDOUS for white. >>After only pxp played by white, then all whites pieces are teamed up against >>blacks king except for the rook on e1 which is very easy to get onto the d-file >>quickly. Whereas black is completely blocked up. >> So I believe I would EASILY sacrifice a piece for such a position (if it were >>otherwise equal), and I'd guess a computer would do it even more confidently. >> The tactics are definitely there. BUT a computer needs to see in black and >>white, but it looks SO imminent anyway that I'd expect the computer to >>immeadiately realize multiple new problems in every 100th of a second. >>YES! this position is not the problem! The problem is HOW TO GET into such a >>position. Never mind about a missing piece! >>S.Taylor > > >is some programs will see this position on a leaf node, they may decide not to >choose this line, because they are a piece down. Shredder will choose this line, >bacause alredy his static evaluation of this position is positive for white. Very interesting! Doesn't "static evaluation" get even better after a few seconds, before it really gets deep into analyzing variations? And isn't the "static evaluation" updated on every iteration (I mean on every complete ply, before starting the next)? S.Taylor
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.