Re: computer-go: perfect play

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   From: "Vlad Dumitrescu" <vladdu@xxxxxxxxxxxxxxxxx>
   Date: Wed, 27 Sep 2000 15:12:50 +0200
   MIME-Version: 1.0
   Sender: owner-computer-go@xxxxxxxxxxxxxxxxx
   Precedence: bulk
   Reply-To: computer-go@xxxxxxxxxxxxxxxxx
   Content-Type: text/plain;	charset="iso-8859-1"
   Content-Length: 520

   > Looking at a simple binary tree, where 1 is winning positions and 0 are
   losing, we could have
   >
   > 0
   > 1 1 1
   > 110 100 000

   A wining position is a position where one wins whatever the opponent does,
   and similar for a losing one. There is no way to "turn around" the tables...
   So there are actually 3 kinds of positions: winning, losing and undecided.

   Otherwise the perfect game would be one of the following two

   Black: <winning move>
   White: resigns

   or

   Black: resigns

   Not a very interesting vision, right? :-)
   Vlad



Your examples  would be a  perfect  game.  Maybe  we should also  view
perfect moves  as requiring the  utmost resistance, even  when losing.
In other words, you should delay your progress to the  end node of the
tree.  Make your  opponent beat you.  Doesn't  that introduce a bit of
GoDevil behavior?  I would think that GoDevil  would tend to delay the
game in hopes of an error.

Don