Re: computer-go: Why Ko

[Robert Jasiek <jasiek@xxxxxxxx>]

Jean-Pierre Vesinet wrote:
> Any ideas about this sort of simplification to the rules?

A few years ago I invented the Basic-Fixed Ko Rules:

Basic Ko Rule: 2-play cycles are prohibited.
Fixed Ko Rule: A play from position A to position B is
prohibited if an earlier play was from A to B.

In practice, your program needs to check 2-play cycles only
because any other repetitive fight is only Ing-disturbing.
The basic ko rule overrides the fixed ko rule only in case
of single basic ko shapes.

A single basic ko is a fighting ko.
A double basic ko is a disturbing ko (double ko coexistence,
perpetual ko, or double disturbing death; also the game end
is as usual).
A triple basic ko is a disturbing ko (with 0, 6, or 12 moves).
Three basic kos are a fight about the most important of them
while the other two are treated as miai, as usual.
A basic ko fight plus a disturbing death (like in a nadare
joseki) are a fight about the basic ko while only the dead
player in the nadare has threats in the disturbing death, as
ususal).
Irregular, long cycles (like eternal life or round-robin ko) 
are a big disturbing ko.
Multi-stage basic kos are like single basic kos fought one
after another, as usual.

The basic-fixed ko rules could be used in tournaments. For
privately developing your program you could use nothing but
the fixed ko rule, if you don't mind a basic ko mouth to be
an eye, e.g. because your study is about neural nets rather
than life and death. Only the fixed ko rule kills ANY ko
fights, i.e. even a basic ko is a disturbing ko.

--
robert jasiek