🚀 go-pugleaf

In article <1993Feb19.001614.17278@leland.Stanford.EDU> 928s4@leland.Stanford.EDU (Benjamin Sloss) writes:
>Further examination has shown that the spurious move problem is not related
>to the hash tables.  I'm not sure what is going on - Gnu will play a totally
>random move (hanging its queen in one game!).  If you save the game and reload
>it, gnu plays normally from the same position.
>
I have had this same problem with Gnuchess.  In fact, many of the wins I have
scored in blitz against Gnuchess have been because it hung mate-in-one when it
could have defended against it (albeit with loss of material in some cases).

This behavior didn't seem to occur with earlier versions of Gnuchess (I tried
to figure out what patch level I am currently running, but Gnuchess doesn't
seem to announce it or provide it... the README file mentions pl44, but that
file may be out of date.  My xboard version is 2.0, pl23, and I retrieved
Gnuchess at the same time.)

>gnu for these mistakes (which show up about every other game), so these 
>results should be taken as an indication of the strength of a fully-debugged
>Gnuchess.  It is also possible that this problem is being caused by the
>compiler I'm using (IBM's XlC).
>
I am running on a Sparc 2 using the gcc compiler.  The problem shows up once
every 10-20 games.

john//
jrb@ingres.com