Re: ScoreQ issue

[Stuart Cracraft]

It's the other way around. Since QUEEN_NOT_PRESENT is -25,
the program is being penalized for swapping off queens. The point of
this is to retain possible tactical possibilities. A lot of players just try
to trade-down against computers and the computer should be wary
of this as the strength of most programs is in the tactical semi-open/open
middle-game, with queens present. This may not be perfect but does
add a significant quarter-pawn bias against not having the queen if
it is the computer's side being evaluated., at any phase in the game.

Here's an ed diff listing to make the fix.

679a

   if (board.b[side][queen] == NULLBITBOARD) {
      if (side == computer) {
        s += QUEEN_NOT_PRESENT;
      }
      return(s);
   }

.
677,678d
~

--Stuart

At 12:35 PM 12/20/2002 +0000, you wrote:

> Tim Bedding wrote:
> >
> > I have noticed a possible problem with the
> > function ScoreQ in the eval.c file.
> >
> > The QUEEN_NOT_PRESENT bonus will never be awarded
> > because you never get that far. You always hit
> > the
> >    return(0);
> > line.
> >
> > Perhaps a comment would be a start at fixing this.
>
> Thanks Tim,
>
> I think we can probably fix that one in one go. Question is are
> we sure swapping queens off is good - the bonus is -25.
>
>  Simon
>
>
> _______________________________________________
> Bug-gnu-chess mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-gnu-chess