computer-go: Discarding the rubbish moves

[Darren Cook <darren@xxxxxxxxx>]

Heikki Levanto wrote:
> Anyway, in any given go position, there are a small number of
> reasonable moves that can be considered, and a great number of
> totally stupid moves that no human would even consider. Any go
> program must be able to discard the 90% of rubbish moves quickly.
> Here a simple move evaluation could make sense. (especially since
> the same move may be totally irrelevant or the best possible,
> depending if there is anything more valuable around...)
This subject interests me a lot at the moment; as it would be off-topic 
I started a new thread.

What rules are there for vetoing moves? I'm interested at two levels:
  1. Proveably bad moves
  2. Pro level at 99.9%

By pro level I mean an algorithm (or database) that would veto moves in 
a board position and at 99.9% of the time it wouldn't veto the move the 
pro played. I'm more interested in type 1 as I'm interested in solving 
endgames, but type 2 is more useful for a general purpose playing program.

As an example of type 1, in a three-in-a-row, the end points are always 
bad moves:
 ?####
 #abc#   ("?" and rest of board are "don't care")
 ####?

'a' and 'c' can always be vetoed. Whether 'b' is a good move for either 
side depends on the surrounding situation (specifically, if the black 
chain can make an eye somewhere else or not).

I think I can generalize to this position:
  ##a
  xy#  x and y are empty points.
  ##b

'y' is always a bad move, by either side, assuming neither 'a' nor 'b' 
is a white stone. (it also works if 'a' is white but 'b' is a black eye 
or otherwise protected, and vice versa).

What other positions are there like this? Has anyone made a list, or 
done research on this?

Darren

P.S. It gets more complicated when allowing for ko threats; however in 
the above example even if 'x' and 'y' could act together as a ko threat, 
'x' would always be at least as good as 'y' as the first move, so 'y' 
can still be vetoed.