Re: [gnugo-devel] 9 stone games

[Arend Bayer]

Dan wrote:

> This follows Gunnar's suggestion of adding a scale_randomness
> field to struct move_data.  The randomness is only increased
> for j and t patterns.

> - new scale_randomness field in struct move_data, function scale_randomness()
> - j and t patterns get increased randomness

As far as I can tell this does not clash with the logic in value_moves.c
around lines 2490 (because this logic does not apply to patterns with
fixed move values). It might be worth pointing this out with a comment
somewhere.
(I briefly thought that your patch might change the tendency in choosing
between j and Fj pattern, but I think this is hardly the case.)

Arend