Re: computer-go: Computer Go Tournament Program

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   From: Robert Jasiek <jasiek@xxxxxxxxxxxxxxxxx>

   Don Dailey wrote:
   [...]
   > If there is something bothering you about a possible ambiguous result,
   > give us an example scenario in which there  would be confusion that is
   > not handled by the protocol.

   If it were so simple to predict what agreements will actually do,
   then why do you not simply describe it? You presume the 
   possibility of disagreements so you have to handle them. All
   involved entities can fail, so you have to handle each possible
   case of failure. Here is a straightforward solution:

   1) Unless four successive passes have been made, if both programs 
   agree by submitting the same score with their successive passes
   to the program arbiter and if both programmers agree that the
   program arbiter displays both scores submitted by both programs
   correctly and the prior handling by the arbiter program is 
   correct, then the final result is given by those agreeing
   scores.
   2) Otherwise if four successive passes have been made, then the
   arbiter program submits a score to both programmers and if
   both agree that this score and the prior handling by the arbiter
   program is correct, then the final score is given as this score.
   3) Otherwise if in either of the two preceding cases the programmers
   disagree, then the referee decides, subject to any higher instance
   court.

   --
   robert jasiek



This is unduly  complicated and  solves no  problems.  You are  saying
that not only do we agree on the  scores, but we implement yet another
protocol to determine if the arbiter verified our agreement correctly.
This is going too far, it's paranoia beyond  what is reasonable.  What
if I don't trust the "agreement of  the agreement" protocol?  It's far
more likely the agreement protocol will break  than that the game will
end  up being scored incorrectly.   You are  more philosopher than you
are engineer.

The whole point of this is utter simplicity.  If  it's not simple then
no one implements it and no one gets any  benefit out of it.  It's not
only  simple, it's very effective and   far superior than the japanese
tournament system of "winging the endgame."

Don