Re: [computer-go] Modern brute force search

[Don Dailey <drd@xxxxxxx>]

Joss,

>   1) White wins and so black has no real "best move", (s)he therefore
>   plays as well as possible, but always knowing that a loss was inevitable
>   assuming that white plays perfect responses.
>   ...
>   ... If not, black may as well resign after the first move 
>   and so the single best line is simply: White moves, black resigns.

Personally, I don't think there is much basis for defining "resign" as
a best move (even in a  losing position) which is essentially what you
are doing.   "Best" means better than  the rest and it  makes sense to
consider it valid for  the "the rest" to be a null  set.  (At least it
makes sense in set theory to consider null sets valid sets.)

As you  point out,  there are  other ways to  define "best."   You can
define "best" based  on opponent modeling, trying to  make things more
complicated for the  opponent.  But then this is  not really chess, it
is a different  game which is no long deterministic.   I admit that it
is  probably more like  the game  as humans  play it!)   It's probably
likely that based on this definition the "best" move may actually turn
a draw into  a loss in some  cases if it can increase  your chances of
winning against a fallible opponent.

This is  just for fun,  but if you  want to extend your  definition of
"best" and make  it a little more "human", I  would try something like
this:

  1.  If position  is a win, the  best move leads to  checkmate in the
      least  number of  moves.  If  you are  on the  losing  side, the
      "best" move delays the inevitable for as long as possible.

  2.  Draws  are little  trickier  due to  50  move rule,  insufficent
      material, etc.  So I would go with this:

      Both  sides  want  to  delay   the  game  as  long  as  possible
      (presumably to  give the opponent  a chance to  blunder.)  There
      are 4 ways  to draw (not including by  agreement which we define
      to always be "bad"):

          1. Three fold repetition.
          2. stalemate
          3. Insufficient material.
          4. 50 move rule.

      OBSERVATION:   All games in chess must eventually terminate, if not by 
                     an outright win or stalemate,  by an eventual draw by repetiton.

      Throw  out  the 50  move  rule  for  this definition,  it's  not
      technically relevant and is  a human concieved device to shorten
      the game.

      Throw out insufficient material.  This is another device to make
      the game  much nicer for human  players and is  not intrinsic to
      the game.  All  of these kind of games  terminate eventually due
      to repetition.

      So the best  move in a drawn position, is  the move which delays
      the  "bitter end"  the  longest,  assuming no  50  move rule  or
      insufficient material!

Please note, even 3 fold repetition  is not a draw unless claimed, but
for calculating  this new hypothetical  definition of "best"  move, we
assume that  it MUST  be claimed when  it first occurs,  otherwise the
game would NEVER terminated  in many positions.  (We could technically
even throw out repetition, but  then our omnicient computer would hang
up in an infinite loop trying to find the end of the game and it would
always play for repetition when a win wasn't possible.)

- Don

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/