Author: Robert Hyatt
Date: 12:47:07 11/11/02
Go up one level in this thread
On November 11, 2002 at 11:18:22, Daniel Clausen wrote: >On November 11, 2002 at 10:42:41, Bob Durrett wrote: > >>On November 11, 2002 at 10:24:26, Robert Hyatt wrote: > >[snip] > >>>Depends. In games vs GM players, Crafty averages predicting correctly 70-80% >>>of the time. In games vs computers, this might drop to 60-70 for whatever >>>reasons. But it is generally > 50% for all cases until you start throwing in >>>very weak opponents, as it drops to almost nothing in those cases, for >>>obvious reasons. :) >> >>What's that? : ) >> >>Amateur chess is better than GM chess because there are more surprises. >> >>Bob D. > >Or let your computer play KRBKNN (the famous position with mate in 223) w/o >tablebases against the perfect TBs and see how good the prediction is... My >guess is: a bit below 50%... > >Sargon\ I'd bet it is over 50% still. 1/2 the time the program will probably make a good move. 1/2 it will make a move that increases the mate-in-N depth, but then it will make good moves for a while. I'd be surprised if it was below 50% although the game might take 500 moves to win if it could even win it. I'd more suspect a draw myself, but with > 50% prediction for every move still, just that one critical mistake that takes it from win to draw would almost certainly be made with a mate that deep and the PV being that critical.
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.