additional information regarding the bug:
the sequence: set gnubgid
XGID=---BBBB---A-AA---b--Dbcce-:1:1:1:42:1:1:0:5:10
hint
leads to the absurd evaluation for 13/7

This one leads to a reasonable evaluation:
set gnubgid XGID=---BBBB---A-AA---b--Dbcce-:1:1:1:42:1:1:0:5:10
move 13 7
previous
previous
analyse move
hint


---------- Forwarded message ---------
From: Dirk Schiemann <[email protected]>
Date: Fri, 12 Jun 2026 at 18:14
Subject: Strange evaluations in certain positions
To: <[email protected]>


Hi,

I regularily encounter strange evaluations by GNU in certain position
types, mainly when it is about racing against the gammon.
Here an extreme example where GNU thinks my move 13/7 is 219 blunder where
all resonable moves should yield roughly the same equity:

XGID=---BBBB---A-AA---b--Dbcce-:1:1:1:42:1:1:0:5:10

Two questions: is this a known bug and is somebody still working on the
software?

I am planning to launch a backgammon platform with GNU as its engine, which
is why it would be very helpful for me to receive a reply.

Best regards,

Dirk Schiemann

Reply via email to