Re: computer-go: Engineering (was: Most simple Go rules)

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   Date: Thu, 28 Jun 2001 15:35:22 +0100
   From: Nick Wedd <nick@xxxxxxxxxxxxxxxxx>

   In message <200106281344.JAA06481@xxxxxxxxxxxxxxxxx>, Don Dailey
   <drd@xxxxxxxxxxxxxxxxx> writes
   >
   >Yes,  my protocol idea is definitey  broken.
   >
   >I guess it should be this (according to your email) if it's used:
   >
   >--
   >When a player first passes, he either gives the  arbiter his notion of
   >the score, or chooses not to use this protocol.
   >
   >If the second player responds with a pass, he can  do the same.  If it
   >turns out both players reported a score to the arbiter and they agree,
   >the game is over and the agreed upon final score is official.
   >--
   >
   >This is close to Mark Boon's idea which is also  interesting, if I got
   >it right it goes like this:
   >
   >   If either player passes, the arbiter calculates a score and informs
   >   both parties.   If both parties agree, the game is over.

   I am not sure how to interpret this.

   If both parties agree that Black has won by 2 points, but "really" White
   has won, then who has won?

   If both parties agree that Black has won, but they disagree about the
   winning margin, and "really" White has won, then who has won?

   Nick
   -- 
   Nick Wedd


Whatever both parties agree  on is who won.  The  whole idea is to let
the computers come to an agreement, otherwise there is no point.

Even in human go, you play it out until there is agreement.

We could  just  let  a 3rd party    arbitrate which means   we need no
stopping protocol  at all.  This is  supposed to be a  way  to let the
computer take a shortcut in  Tromp/Taylor and not have  to play to the
bitter end.  But if an arbiter is  going to overrule  if they are both
wrong, then why not let  the arbiter just  stop the game when he knows
the correct score?

Don