Re: computer-go: Fw: Computer Go Tournament Program

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   From: "Wiesiek Grygo" <wiesiek@xxxxxxxxxxxxxxxxx>

    We don't need any agreement protocol!
   Game is over after two pass.

   We need only program wich can score finished game (after filed dame and fini
   shed endgame).
   This program can do all or need help from person (sometimes or always), eg.
   like on IGS.
   Program score game from file in SGF format.

   Is simpler make one program that can score finished game.
   Then make a agreement protocol and score algorithm in every program.

   Rules
   On IGS is used modifed version of Japanese rules, for simplifity scoring alg
   orithm.
   Deatails on: http://home.snafu.de/jasiek/igsrules.html
   One of differences in countig: not filled ko in oryginal Japanece rules this
   give 0 points but on IGS this give 1 point.
   Of course is more differences.


   Wiesiek Grygo


Wiesiek,

What we need is clear, the ability to  play 2 programs together with a
standard protocol that works  without human supervision.  This is what
I believe you are saying right?

Tromp/Tayor  is an extremely clear  cut and simple way  to do this and
seems to define the game in it's purest  form too.  This really should
be the standard in my opinion.

I would strongly  urge people to implement  Tromp/Taylor and not worry
about  the agreement  protocol.   The whole   idea  is that  the  base
Tromp/Taylor ruleset become the "universal" standard for programs that
hope to play unsupervised.

Even  though  I think the agreement  protocol  is a good  thing, there
should  be some consensus  on it's value  (there should  be at least 2
programs that implement it and I'm not sure there is!)


Don