Author: John Timm
Date: 21:33:39 01/03/99
Go up one level in this thread
On January 03, 1999 at 18:58:35, Phil Dixon wrote:
>Has anyone had the problem with the analysis moving from a positive to a
>negative score at 11 ply depth? The numerical evaluation is the same but it
>switches from + to -. This happens with all engines and it is annoying.
I have not seen a problem with
changes from positive to negative evaluations (or vice-versa) because depth of
search increases. However, there is a
possible change from positive to negative (and vice-versa) which can happen at
any depth. If an analysis engine is "locked" on a specific position (so that
the engine continues to analyze that position even if the position on the board
diagram changes), then, when the analysis engine reaches a new conclusion, the
sign (positive or negative) of the analysis which is displayed is displayed from
the standpoint of the side to move in the board diagram, NOT from the standpoint
of the side to move in the "locked" position which is being analyzed.
This
is guaranteed to drive the user slightly crazy, because there is no way to know
whether the engine thinks the evaluation is positive or negative (except if (1)
the position is so one-sided that the correct sign is obvious, (2) the user
happens to notice which side is to move in the board diagram at the moment the
new analysis is displayed, or (3) multiple variations are displayed, so the
correct sign can be inferred in some cases - for example, if, with White to
move, the three best moves are rated -0.40, -0.20 and -.0.10 in descending
order, it is clear that the sign for the first two variations (but not
necessarily the third) should be positive, not negative).
In my opinion
CB7 would be much improved if all evaluations which favor White were shown as
positive and all positions which favor Black were shown as negative.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.