Re: [gnugo-devel] Inaccurate estimate_score

[Darren Cook]

> experimental_score's range is now correct, so may be useful to me; I'll
> try it on some more positions.

It didn't take long to find failures: I asserted that final_score should
be >= experimental_score black, and <= experimental_score white. On the
first failure position I'm looking at there are no real moves left but
it misses that black has to capture a dead stone (i.e. fill a point of
territory) once a dame is filled.

I can probably generate some unit tests if there is interest. But
perhaps this level of accuracy is considered outside the scope of
experimental_score?

Darren