Author: chandler yergin
Date: 08:34:22 01/30/06
Go up one level in this thread
On January 30, 2006 at 11:24:26, Robert Hyatt wrote: >On January 30, 2006 at 10:29:39, chandler yergin wrote: > >>On January 30, 2006 at 10:18:03, K. Burcham wrote: >> >>> >>> >>> >>> [D] 2Q5/1p2kb2/1q3p1P/2p2P2/3pP3/4b2R/8/3B3K b - - 0 1 >> >>New game, >>[D]2Q5/1p2kb2/1q3p1P/2p2P2/3pP3/4b2R/8/3B3K b - - 0 1 >> >>Analysis by Rybka 1.0 Beta 32-bit: >> >>1...Qb5 2.Qc7+ >>= (0.00) Depth: 12 00:00:16 637kN >>(, 30.01.2006) >> >>How much time did Rybka have to make it's move in the game? > > >It doesn't matter. His analysis shows that the scores for the depths were >different for the same ply. This is commonly caused by transposition table >scores being re-used. And there is absolutely nothing wrong with doing so. >Sometimes it will produce a worse result, but more often it produces a _better_ >result. Just like sometimes searching deeper will cause a program to play a >worse move, although in general going deeper produces stronger moves. > >His point was that comparing scores ply for ply shows a difference, which it >did. My Point... "Rybka played Qb4 in the game, instead of the safer Qb5." In the game! It played the best move found at the time it had to move. I believe that Hyatt shouted that to me more than once. ;) Chan
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.