Re: [gnugo-devel] min_value revision

[Evan Berggren Daniel]

On Thu, 30 Jan 2003, Gunnar Farneback wrote:

> Um, that's one of the main points of the minimum values, to increase
> the variability in its moves. At least arend:15 and trevorc:1180 can
> be solved by better patterns.

I think that is appropriate in many cases.  However, there are cases like
these where Gnu Go already understands the difference between the moves.
I do not believe we need more patterns that raise the value of the move
higher than it already is -- I think the correct solution is a program
with a much deeper understanding of the situation, but that is obviously a
long way off.  In the mean time, minimum values seem a very reasonable
thing, but adding patterns to deal with their negative side effects seems
counter productive.

> > This patch makes a very simple attempt to treat minimum values more
> > correctly.  It linearly transforms the value of a move from something in
> > the range 0 - min_value + 2 to the range min_value - min_value + 2.  The
> > result is that Gnu Go's decisions about the relative values of the moves
> > are kept, but the minimum value is also observed.  However, it also
> > appears to hurt tunings between moves with a high min_value and moves
> > without.  I am not sure what the correct approach is, but I think the
> > problem needs addressing.
>
> The minimum values are surely problematic, but this is not the way to
> solve it.

I agree.  It just seemed like an easy way to test out the changes and get
some results to work with.

Evan Daniel