[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [computer-go] Protocol B
My concern is time control. If rounds have a maximum time, the best program
will play random moves very fast and never pass. His opponents will always
lose on time.
David
> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx] On Behalf Of Mark Boon
> Sent: Wednesday, July 27, 2005 11:53 PM
> To: computer-go
> Subject: Re: [computer-go] Protocol B
>
>
> I've not really followed this discussion in full detail, but
> what if the
> score disagreement is in a seki?
>
> David G Doshay wrote:
>
> >
> > (pass - pass)
> > ?score?
> > if agree
> > no problem, game over
> > if disagree
> > resolve-disagreement-genmove
> > if reply is pass
> > resolve-disagreement-genmove
> > if reply is pass
> > game over, everyone alive
> > if reply is not pass
> > continue game with genmove
> > if reply is not pass
> > continue game with genmove
>
> By the way: I'm in favour of playing out the game until the
> very end. If
> both agree on the score after the first two passes, fine, you
> can stop
> early. If there's a dispute (or if one side behaves like an ass) then
> you should have a simple rule-set that leads to an
> unambiguous end. That
> means until all dead stones are captured, where for Japanese
> type rules
> (why use them in automatic tournaments anyway?) there's a one point
> penalty for each pass. Game ends after the 3rd consecutive
> pass (which
> is not penalized).
>
> All other protocols seem overcomplicated and will force one party or
> another to put a lot of effort in resolving disputes. Playing
> until the
> end allows programs to put minimal effort into the
> game-ending phase if
> they wish. Those that complain saying: my program 'knows'
> it's dead but
> may make a mistake in capturing it will have to think again
> about what
> it is their program knows. If a program really 'knows' something is
> dead, the algorithm to safely capture it is fairly trivial
> and something
> your program will need to be able to do anyway. Your program
> will also
> need to be able to capture a 'dead' group when in semeai, and that is
> much more complicated. If your program can do that, it should also be
> able to capture groups at the end.
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx
> http://www.computer-go.org/mailman/listinfo/computer-go/
>
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/