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

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




For computers, I suggest the following:
   
  1. Tromp/Taylor scoring.

  2. A SIMPLE but optional early agreement protocol.

  3. A very simple programming interface that can be implemented
     in gmp or anything else.  It should support Tromp/Taylor, the
     early agreement protocol and fully automated matches between
     programs.   

We could argue  about  this forever, but    if we just  got  together,
designed and implemented a very simple  programming interface that did
the right thing, it  would probably get  incorporated into most of the
programs if it was made easy to do so.

I suggest that the programming interface should be so simple it's hard
to get wrong and should  include a test-jig  to  help debug and  prove
correctness.   In tournaments you could  require a program to pass the
run-jig test before being allowed to compete.

The agreement protocol  in my opinion is not  really necessary, but  I
support  the idea if it's truly  made simple.  I am suggesting nothing
fancier that stopping the game when/if both player agree on the score.
The  implementation could be  that you send a score  with the move and
send -1 if you are not ready to come to agreement.

Humans can use this same protocol, it's  not overly artificial.  There
are other possibilities here for  an agreement phase  but they are all
quite a bit more complicated that this.


Don






   From: Nick Wedd <nick@xxxxxxxxxxxxxxxxx>

   In message <3B3DBFB1.387D0EC4@xxxxxxxxxxxxxxxxx>, Robert Jasiek <jasiek@xxxxxxxxxxxxxxxxx>
   writes

   >Christian Nentwich wrote:
   >> I see no problem implementing that in a formal protocol .
   >
   >Instead of perverting the Japanese scoring style it makes
   >more sense to use the Chinese style right away.

   I agree strongly with Robert here.

   Christian suggests creating a new protocol and persuading all
   competitors to implement it correctly.  As most have not yet implemented
   GMP correctly, this seems optimistic.

   Whereas, as Robert points out, we can use Chinese-style counting right
   away.  Even programs which know only of Japanese scoring will not suffer
   by this, unless they omit to fill dame.

   Nick

   -- 
   Nick Wedd