🚀 go-pugleaf

David Hillman writes:
> 3)  Today I found myself with a king against its king and rook, playing
>     at level 2.  (No, I'm not very good.)  If I remember, king and rook
>     should be able to mate in this situation.  The program didn't seem
>     to know how to do this, however, moving the rook about aimlessly for
>     quite a while.  

I don't know which version you have but GNU chess is in constant change,
this leading sometimes to bugs which are fixed in the next patchlevel.
The behaviour you describe could well be due to a bug I described and
that appeared in 4.0 pl61 and pl62.  It is also possible that playing
at low levels won't allow the evaluation functions to show their 
intended abilities.

>     Finally it did manage to corner my king.  But what
>     really did surprise me was: at one point it was the computer's move
>     and it had mate in one, but didn't do it!  (The exact situation came
>     up two moves later again, and this time it did do it.)  Can someone
>     explain why this would happen?  (Did it stop searching when it found
>     mate in 3, thereby missing mate in 1?)

Probably.  And it is normal that a chess program with transposition
tables will evaluate the same position occuring twice in a game
sometimes differently.  Think of it as it having a memory.

--
--| Ralf Stephan, Augsburg, Bavaria, Germany |-------| ralf@ark.abg.sub.org |-
--| GO  d* -p+ c++ l++ m* s--/+ g++ w++ t r+ |-------|       IT'S ...       |-