[Bug-gnubg] Re: Checkerplay error vs. cube error

[kvandoel]

> last sunday i had asked for a confidence interval of a2(N) and b(N). Maybe 
> you missed my question. I believe this would improve the understanding of 
> your formula.

I haven't computed them but from the configence intervals of the ratings
in my writeup you can sort of  eyeball it; they're pretty small.

Note that this would be the confidence interval of the fit to simulation,
which is not what I think you think a confidence interval would tell you.

Kees