Re: computer-go: Most simple Go rules

[John Tromp <tromp@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>]

On Thu, Jun 28, 2001 at 08:45:46PM -0700, David Fotland wrote:
> 
> >I agree with Robert that super-ko is NOT too complicated for humans.
> >(At least in real games) Most decent amateurs can replay their games,
> >so it cannot be impossible to detect a super-ko violation.
> 
> I guess we will not agree here :)  I think that a pro playing with superko, 
> could set up a
> position where it would be very difficult to remember which kos had been 
> taken in what order.

Only if his opponent cooperates in setting up many kos:)

But, this seems not to be an argument against superko, since in ANY
other korule, you'll have a similar difficulty of recognizing whether
there is a repetition going on.

You must either allow infinite games (like the basic ko rule does),
or somehow recognize some form of repetition, and if you opt for the latter,
then superko is arguably the simplest way.

Note that the Japanese rules have an

  Article 12. No result

  When the same whole-board position is repeated during a game,
  if the players agree, the game ends without result. 

So both players should be able to remember previous positions,
else they cannot challenge the opponent's claim of no-result.

What do you prefer?

1) superko forbidding all board repetition, or

2) basic ko forbidding 2-cycles + no-result for all other board repetition

regards,
-john