🚀 go-pugleaf

Barton Christopher Massey (bart@cs.uoregon.edu) wrote:

: In article <2dbq9b$16o@nuscc.nus.sg>,
: Chua Kong Sian <nsrccks@leonis.nus.sg> wrote:
: [...]
: > the PieceList array is now arranged 
: > differently.  (This array contains the pieces that are remaining on the
: > board.  Undoing a capture will yield a piecelist array that has the
: > captured piece in a new location in the array then the original array).
: > With a slightly different PieceList array, the move generation order is 
: > changed.  Subsequently a difference sequence of moves will give rise
: > to different node counts during the search.

: Ouch!  You mean the algorithm isn't sorting the moves according
: to some heuristic, but is instead just trying them in some
: arbitrary order??  You could make the whole thing deterministic
: and also likely improve play if you e.g. try moving the pieces
: which are on high-value squares first...

: 					Bart Massey
: 					bart@cs.uoregon.edu

Well, that is not exactly true.  There is some sorting of the moves.
Namely, killer moves and PV moves are always tried first.   Captures of the
opponent major pieces are also placed first.  So are threats of promotion.
Other than that, most of the other moves are generated depending on which
piece is encountered first in the PieceList array.  Notice that checks are
not considered in the ordering at all (they should be shouldn't they).

There is a flag that can be used to change this slightly.  If -DKILLT is
defined, moves towards the centre of the board will be preferred.

Regards.

Kong Sian