computer-go: Huima's hash function

["Antoine de Maricourt" <antoine.de-maricourt@xxxxxxx>]

Hi,

did somebody already used Huima's hash function [1] in pratical application?
or study in more depth the statistical properties of this function?

It seems to me that this function has some very undesirable properties,
that is: some different positions map to the same hash value in a
predictible way that does not depend from the random generator.

Huima points it out at the very end of his paper, and suggests the use
of addition modulo 256 to combine values instead of xor, but
unfortunately this do not solve the problem.

[1] A Group Theoretic Zobrist Hash Function