[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Fwd: Re: computer-go: How many bits are needed to encode an N x NGo position?



Just for  fun, there is another possible  complication.  It's possible
that some  LEGAL positions cannot  be reached in  a real game  from an
empty board.  I don't really know, but if that is the case, someone on
this list  may be able  to construct an  example of a  legal positions
that cannot be reached in a real game.
This is a good question. Are there any legal positions that cannot be reached in a real game? It seems to me that there are not. If a position is legal then it seems to me that the individual stones in that position could be played legally in any order. Kos would not pose a problem, because there need not be any capturing in this game. If one side had many more stones than the other, the other side could pass. There are surely many positions no two sane human players would ever reach, but I think all legal positions should be attainable.

P.S.  If someone  could come  up with  a  practical  way to  "map out"
      illegal positions, then  it might also be possible  to throw out
      huge classes of  positions that are legal, but  can be proved to
      be irrelevant (for instance, easy  to "prove" it's a big win for
      one  side or the  other.)  I'm  skeptical, but  it's an  idea to
      think about.
I would think there would be a way of proving that broad classes of positions are equivalent or worse to others. This wouldn't show that a player playing an ideal game would never reach those inferior positions (since early in the game the stone that looks detrimental may have helped get to that position), but it could be a quick way of eliminating from consideration all the moves going from one position to a demonstrably worse one. For example, it is ever a good move to fill in an eye? Or, similarly, to fill in (or attack) very secure territory?

--

Robert de Neufville
rden@xxxxxxxxxxxxxxxxx