Re: computer-go: Superko rules and computer go programs

[Robert Jasiek <jasiek@xxxxxxxxxxxxxxx>]

Peter.Smith@xxxxxxxxxxxxxxxxx wrote:
> For terminology refer to Robert Jasiek's excellent pages around
> http://www.snafu.de/~jasiek/superko.html.

This was the 1996 page about superko.
I have now uploaded my 1998-12 update, which I held in reserve 
for checking the details and making it part of several updated
pages. Unfortunately, I have lacked time for too much updates.
Due to interest I have uploaded that superko update with some
new information, however, I do not guarantee 100% perfection of
that page. For this reason I leave the date at 1998-12 for now:)
Still you might find parts of the new information interesting.
Corrections and comments are welcome.

> So far I have tried to implement situational super ko but have been worrying
> about passes.

You need not worry about passes except when playing under superko
against other programs that might handle passes differently;(

> There are pathological Go positions in which passes would be
> illegal under a situational superko rule if passes counted as moves.

Positional as well as situational superko restrict board plays only.
Thus there should be no illegal pass problems.

> I had
> been assuming that players can always pass.

AFAIK, there are no CG tournaments with primitive ko rules, so
passing is always allowed until the game end.

***

You can forget about positional/situational superko differences.
In practice they simply do not occur:) My estimate is 1:100,000.
Of course, a programmer might devise to test the opponent's ability
to understand rules by playing a useless cycle in a 
sending-2-returning-1. So maybe it is better not to forget about
intricacies after all...

-- 
robert jasiek
http://www.snafu.de/~jasiek/rules.html