Author: Shane Booth
Date: 17:42:25 05/06/98
Go up one level in this thread
On May 06, 1998 at 15:53:25, Howard Exner wrote: >On May 06, 1998 at 14:25:12, Danniel Corbit wrote: > >>On May 06, 1998 at 03:46:28, Danniel Corbit wrote: >>[snip] >>>>K6n/2P5/1q6/8/8/7k/8/n7 b >>>> >>>>White: Ka8 Pc7 >>>>Black: Kh3 Qb6 Na1 Nh8 > >Snipped analysis. > >This position contains three drawing themes >1.The stalemate if capturing the pawn >2.The two Knight draw if the Queen exchanges itself for the pawn >3. The KQN v KQ draw (at least in this position) > >Crafty and most likely all of todays programs would recognize each >individual drawing them but probably cannot make the human leap of >seeing them altogether in this problem. This could explain why the >eval would remain high. Actually, Crafty doesn't seem to know that KQNvKQ is generally a draw and scores these positions as just below +4. Same for other pure piece endgames like KQBvKQ and KRBvKR (last one if you don't have the (huge) tablebase installed). I notice Rebel Decade 2.0 also gives a huge plus to the side with the extra piece. However, the old Chessmaster 4000 scores these positions below +2. Just add a pawn each and the evaluation jumps over +4.
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.