Author: Robert Hyatt
Date: 20:22:34 10/14/99
Go up one level in this thread
On October 14, 1999 at 22:40:39, Howard Exner wrote: >On October 14, 1999 at 09:49:54, Robert Hyatt wrote: > >>that was then, this is now. here is current crafty output: >> >> 11-> 3.60 0.59 1. ... Qg6+ 2. Qxg6+ Kxg6 3. Ke3 Kf5 >> 4. Ke2 Kf4 5. Kf2 g5 6. Ke2 Kg3 7. >> Kf1 h3 8. gxh3 Kxf3 >>~ >>so at least now it has an inkling that this is good. > >I tried this out on Rebel 10c and CM6000 and both are totally in the dark. >However when I modify the initial position by removing white's b2 pawn >Rebel finds Qg6 rather quickly. I'm guessing then that it is spooked by the >white pawn majority on the queenside perhaps not having the knowledge that >despite the pawn majority no forced passed pawn can be created for white. > >Why do you think Crafty plays Qg6? Is it such pawn majority code or just >searching and extensions? Table bases? purely the pawn majority analysis. With no material, the bonus is at maximum value (it is basically an exponential function inversely proportional to the material left on the board). It simply thinks that the potential (candidate) outside passer is worth way more with queens off. This isn't perfect. as it is likely that similar positions might be contrived where the outside candidate passer isn't enough to win... but here at least, it is 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.