Re: [Bug-gnubg] Play mistake

[Jim Segrave]

On Fri 17 Jan 2003 (16:49 -0800), John Doe wrote:
> --- Nardy Pillards <address@hidden> wrote:
> 
> Encountered a similar situation:
> Match ID: MIF0AAAAAAAA
> Position ID: wmfwAWBv2wEAAA

There is something very wrong here. Using today's build, setting
Analysis and Evaluation to 'Expert', I get the following in the hint
window:


In the hint window, trying different plies:
    1. Cubeful 0-ply    5/off 3/2                    MWC: 1.0000
       1.000 1.000 1.000 - 0.000 0.000 0.000
        0-ply cubeful [expert]
A guaranteed backgammon? I find this difficult to believe.
    2. Cubeful 0-ply    5/off 1/off                  MWC: 0.7556 (-0.2443)
       0.998 0.975 0.015 - 0.002 0.000 0.000
        0-ply cubeful [expert]
    3. Cubeful 0-ply    5/4 5/off                    MWC: 0.7286 (-0.2714)
       0.944 0.880 0.015 - 0.056 0.000 0.000
        0-ply cubeful [expert]
    4. Cubeful 0-ply    5/off 4/3                    MWC: 0.7261 (-0.2739)
       0.937 0.880 0.010 - 0.063 0.000 0.000
        0-ply cubeful [expert]
    5. Cubeful 0-ply    5/off 2/1                    MWC: 0.6940 (-0.3060)
       0.883 0.748 0.009 - 0.117 0.000 0.000
        0-ply cubeful [expert]

Let's see what other plies give:
    1. Cubeful 1-ply    5/off 1/off                  MWC: 0.7513 
       0.989 0.958 0.016 - 0.011 0.000 0.000
        1-ply cubeful
    2. Cubeful 1-ply    5/4 5/off                    MWC: 0.7188 (-0.0325)
       0.922 0.845 0.012 - 0.078 0.000 0.000
        1-ply cubeful
    3. Cubeful 1-ply    5/off 4/3                    MWC: 0.7137 (-0.0376)
       0.911 0.827 0.011 - 0.089 0.000 0.000
        1-ply cubeful
    4. Cubeful 1-ply    5/off 2/1                    MWC: 0.7107 (-0.0406)
       0.904 0.811 0.016 - 0.096 0.000 0.000
        1-ply cubeful
    5. Cubeful 1-ply    5/off 3/2                    MWC: 0.6928 (-0.0584)
       0.867 0.765 0.006 - 0.133 0.000 0.000
        1-ply cubeful

Which looks believable

    1. Cubeful 2-ply    5/off 1/off                  MWC: 0.7518
       0.984 0.960 0.022 - 0.016 0.000 0.000
        2-ply cubeful 100% speed [world class]
    2. Cubeful 2-ply    5/4 5/off                    MWC: 0.7230 (-0.0288)
       0.925 0.870 0.012 - 0.075 0.000 0.000
        2-ply cubeful 100% speed [world class]
    3. Cubeful 2-ply    5/off 4/3                    MWC: 0.7122 (-0.0396)
       0.902 0.832 0.012 - 0.098 0.000 0.000
        2-ply cubeful 100% speed [world class]
    4. Cubeful 2-ply    5/off 2/1                    MWC: 0.6908 (-0.0610)
       0.855 0.761 0.012 - 0.145 0.000 0.000
        2-ply cubeful 100% speed [world class]
    5. Cubeful 2-ply    5/off 3/2                    MWC: 0.6903 (-0.0615)
       0.850 0.764 0.012 - 0.150 0.000 0.000
        2-ply cubeful 100% speed [world class]

As does this.

Anyone have any ideas why the 0 ply is so strange for this one move?
If I make this move, then analyse it, I get the same analysis as the 0
ply evaluation (not surprising), again saying that a backgammon was
inevitable (to 3 digits of precision).



-- 
Jim Segrave           address@hidden