Re: [Bug-gnubg] snowie equivalent bug?

[Tore Fredriksen]

No this sounds like a snowie problem. But atleast it's nice to know
that the snowie equivalent number in gnubg is slightly softer (perhaps
about 0.4 in a medium length match) than snowie4 would grade you.


On Wed, 10 Nov 2004 13:02:38 +0000, Joern Thyssen <address@hidden> wrote:
> On Wed, Nov 10, 2004 at 01:39:44PM +0100, Tore Fredriksen wrote
> 
> 
> >
> > Problem description : I recently had 7 matches recorded from a live
> > tournament and analyzed by snowie 4 and then my own gnubg. With
> > snowie4 my errorate was X.6 while with gnubg the snowie equivalent
> > errorrate was X.1
> >
> > I initially thought this to be because I have a bias towards gnubg
> > playing style since I don't have a copy of snowie4, but then I noticed
> > a worrisome thing. Snowie4 lists a smaller amount of total moves in
> > each match than gnubg, and that might just be the reason that my
> > errorate was estimated lower by gnubg than by snowie4.In that case
> > this should be considered a bug. Examples from 6 of the matches :
> > Snowie 4 number of moves -351,329,174, 240,213,375. Gnubg number of
> > moves- 385,374,193,269,232,413.
> 
> Snowie does not include moves made after breaking contact, due to, I
> think, bad play in bear-in and/or bearoff.
> 
> I was aware of this bug, but I don't see an easy way to fix it.
> 
> Jørn
> 
> 
>