Re: owl improvements (Re: [gnugo-devel] Owl tuning)

[Evan Berggren Daniel]

On Wed, 18 Dec 2002 address@hidden wrote:

> Here is something we could try immediately. Generally E patterns in
> owl_defendpats.db are escape attempts, and s patterns in
> owl_attackpats.db are placement (oki). So the rule would be never to
> play an s attack pattern immediately after the opponent played an E
> defense pattern.

Of course this will mean wasted moves when the defender can try an
ineffective escape pattern, which the attacker then must answer.  This
would waste reading depth in some cases.  However, I think this is a very
minor issue; I just think we should be aware of it.  Essentially we're
treating all escape attemtpts as kikashi.  I suppose Gnu Go could also end
up making mistakes of believing an escape attempt works as a defense,
because it believes that the opponent has to answer, when in fact neither
is the case and the defense must begin playing vital moves instead.  This
could result in Gnu Go attempting to escape and getting killed when the
escape doesn't actually work.  Again, I think this is rare and probably a
workable tradeoff to make for improvements elsewhere.

Sounds interesting, and good luck.

Thanks