🚀 go-pugleaf

I was playing gnuchessr on a SUN4 today and had it make a pretty
strange move.  I had just taken a pawn at d6, exchanging rookes and
sacrificing the extra bishop to get a simple, easily won endgame,
shown by this position.

xboard position file -- Fri Jul  5 17:09:19 1991
black to move (gnuchessr)

. . . . . . . .
. . . k . . . .
. . . B . . . .
. . . . p . . .
. . K . P p . .
. . . . . P . .
. . . . . . . .
. . . . . . . .

gnuchessr plays d7e6 instead of d7d6 (KxB) after thinking for about a
minute. What's the deal? Is its endgame analysis strong enough that it
realized that all continuations lose? Or does this reveal some problem
in the algorithm? I've been seeing it do a lot of strange things
lately. Like give up a mate in one when it could have postponed it for
at least two. Most commercial programs won't do that--they usually
rate a loss in 3 as a little better than a loss in 1.

 Paul Vaughan, MCC CAD Program | ARPA: vaughan@mcc.com | Phone: [512] 338-3639
 Box 200195, Austin, TX 78720  | UUCP: ...!cs.utexas.edu!milano!cadillac!vaughan
----------------------------------------------------------------------------------