Re: [computer-go] Protocol B

[Erik van der Werf <erikvanderwerf@xxxxxxxxx>]

On 7/29/05, john tromp <John.Tromp@xxxxxxxxxxxxxxxxx> wrote:
> 
> > My own ;-) AFAIK Migos rules are currently the most accurate
> > programmable approximation of the official Chinese rules (which
> > cannot be programmed).
> 
> Ah, the ones at
> http://www.cs.unimaas.nl/~vanderwerf/5x5/migos_rules_v0.3.txt

Yes, but the wording in the appendix of my thesis is slightly improved.


> You're right that the result of playout then differs from immediate
> scoring.
> This is because in contrast with the Japanese rules, you rule the
> Black group
> as dead, and white wins by 81 points.
> 
> As you can probably guess, my conclusion is that your rules are
> problematic because
> early scoring is not just a shortcut to the playout score.

No surprise  ;-)

> 
> It feels wrong to me that white can claim all of the area occupied by
> black as his own,
> considering that he loses that claim as soon as he undoes the black
> occupancy...

I' aware of that. I showed this position (also in my thesis) because
it is an extreme consequence of my rules.

The question is, if I can prove that I can capture some group, do I
then have to continue play after that capture (which I'm not willing
to do), or am I allowed to go back to the position at the start of the
resolution phase and score the position exactly there where both
programs decided to pass first.

I don't think there is an easy answer and my rules are certainly not
perfect. However until someone comes up with a more accurate
programmable and practical approximation of the Chinese rules I stick
with Migos rules.

Erik
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/