Re: [Bug-gnubg] Removal of absolute fibs ratings

[Albert Silver]

I didn't mean it was close with my example. Perhaps I've
misunderstood, but I was under the impression that these 'close calls'
really meant what range of equity loss it would use to consider a
double into the number of decisions it judged a human made for the
rating. In other words it would count how many cube decisions the
human had to make, and the ones it counted were the 'close decisions'.
My point was to show that even if that blunder in my example hadn't
been made, I'd prefer it if GNU considered it as one of the decisions
made, and not skip over it simply because the size of the equity loss
made it not worth considering. It *is* worth considering because a
human would likely debate it, and doesn't matter whether GNU thinks it
a big mistake.

Albert

On 9/5/06, Christian Anthon <address@hidden> wrote:
> I agree assigning skill levels to backgammon players is a very complex
> matter and what is difficult varies from player to player. As to your
> position you must remember that gnubg counts close doubles and not
> non-obvious cube actions. In the position the cube actions may be
> non-obvious, but they certainly aren't close.
>
> Christian.
>
> On 9/4/06, Albert Silver <address@hidden> wrote:
> > I think it is rather complicated. Admittedly, some doubles with such
> > an equity difference can seem absurd, but many will leave players very
> > much in doubt, especially if they are dependent on the score. Here is
> > an example:
> >
> >  GNU Backgammon  Position ID: 7LYBAGPN3QYgIA
> >                  Match ID   : MAFgAAAAAAAA
> >  +12-11-10--9--8--7-------6--5--4--3--2--1-+     O: GNU
> >  |                O |   | O  O  O  X  O  O |     0 points
> >  |                O |   | O  O  O  X  O    |     On roll
> >  |                  |   |    O  O          |
> >  |                  |   |                  |
> >  |                  |   |                  |
> > ^|                  |BAR|                  |     3 point match (Cube: 1)
> >  |                  |   |                  |
> >  |                  |   |                  |
> >  |                  |   |       X          |
> >  |                X | X | X  X  X  X       |
> >  |             O  X | X | X  X  X  X     O |     0 points
> >  +13-14-15-16-17-18------19-20-21-22-23-24-+     X: Albert
> >
> >
> >
> > Cube analysis
> > 2-ply cubeless equity  +0.3741 (Money:  +0.2537)
> >    51.93%  29.02%   1.08% -  48.07%   8.47%   0.14%
> > Cubeful equities:
> > 1. No double            +0.3045
> > 2. Double, pass         +1.0000  ( +0.6955)
> > 3. Double, take         +0.0365  ( -0.2681)
> > Proper cube action: No double, take (27.8%)
> >
> > This situation occured in a game of mine, and my opponent sent the
> > cube. It is a blunder to send, over 0.250, and a larger one to pass. I
> > had no idea what to do, and analyzed quite some time before taking.
> > I've analyzed it and understand better, but even if it is in blunder
> > territory, I don't think that it is so obvious to all, despite the
> > potential size of the mistake.
> >
> > Albert
> >
> >
> > On 9/4/06, Christian Anthon <address@hidden> wrote:
> > > The threshold of 0.25 seemed too large for common sense, far too many
> > > doubles very counted as being close, but also the function was plain
> > > stupid and hard to understand. For example all too-good positions were
> > > counted in for some reason. The threshold could be set back to 0.25
> > > for counting of the doubles, but it would still be difficult to guess
> > > the average relation between the old and current versions of the
> > > function.
> > >
> > > Christian.
> >