[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Opponent not passing
From: |
Arend Bayer |
Subject: |
Re: [gnugo-devel] Opponent not passing |
Date: |
Sun, 3 Feb 2002 01:13:25 +0100 (CET) |
> I think that what should really be done is getting the
> owl threat code turned back on near the end of the game.
>
> Arend has been hacking the owl threats and might have
> an opinion on how reliable it currently is in preventing
> endgame tenukis through overconfidence. We've seen
> many cases where it can contribute a bad or pointless
> move, but if we're ahead, safety is the best policy.
I only used the regression tests for tuning, and I don't think there
were examples of this kind in there. However, it does make sense to
me to do an owl-threats analysis just for one dragon if
- the opponent has added a stone to a dead dragon, and
- we are ahead.
Probably this useful not only in the endgame, and should not cost much
time.
Certainly the choice of move that is generated by owl_prevent_threat
has to be improved as well, see 13x13:73 for a very good example, where
GNU Go originally played N12. (Note that this would be a situation where
--owl-threats could really be useful.)
Arend