[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: Most simple Go rules



The biggest issue   for  me is  that computers  should  be requred  to
determine their own  fate.   There needs  to be  a logical irrefutable
conclusion  (like   two   succesive   passes)  with   no    room   for
dis-agreements, and arbritation.   Scoring the final result should not
be a   complicated procedure involving  a room  full  of entities.  It
should be  handle  by the  computers alone, it's   their game.  If the
computers disagree on the results,  then one (or both)  has a bug  and
the actual score can be determined via Tromp/Taylor.

The Tromp Taylor rules are   ideal for this  and  are logical.   Tromp
Taylor doesn't provide for komi, but this can be  added.  Even komi is
slightly illogical because it  illogically   assumes we know what   it
should be set  at.  A simple  suggestion that makes  it  logical is to
assign it something  small (like 1/2  or nothing at   all) and play  2
games  per match, with   colors  reversed.  I  don't  know if  this is
PRACTICAL, but it is logical.  There are ways to break ties reasonably
in case of 1-1 results if that is what you need.

But komi is not  worth an argument.  Any komi  can be used if everyone
plays by the same rules and has the same chances for color.

It doesn't  have to be Tromp Taylor,  but it should be  something that
does not require human intervention to be successfully applied.

Don


   >    From: Robert Jasiek <jasiek@xxxxxxxxxxxxxxxxx>
   > 
   >    What really worries me is usage of illogical rules in CG
   >    tournaments while everybody knows about the advantages of
   >    logical rules for programming. Sponsorship is a weak argument
   >    because the AGA gets much attention from Asian professionals
   >    despite their logical rules. Cultural tradition is an even
   >    weaker argument these days when professionals from all big
   >    go playing countries spread go. So why must every tournament
   >    organizer and programmer suffer from implementing more than
   >    one ruleset, one worse than the other?!
   > 
   > I totally  agree.  They should attempt to  use  a ruleset for computer
   > competition that leads to the least problems and is most logical.
   > 
   > Don

   I would appreciate hearing both of your thoughts on 

   1) In what respect do you consider the AGA rules "logical"?

   2) What ruleset would you consider ideal for a computer go tournament?

   --
   Bob Myers