[blml] Thai braking

[Marvin French]

From: "Steve Willner"

> > An problem with the standard WBF Victory Point
> > scale is ... the
> > majority of imps scored are worth nothing, but
> > a minority of particular imps are worth a full
> > victory point.
> ...
> > Another problem with the standard WBF Victory
> > Point is that the scale is too compressed.
>
> I think you mean "too nearly linear."  The WBF scale is
much closer to
> linear than logarithmic, and log is probably not the right
functional
> form anyway.
>
> For alternatives, you might want to look at the "Bethe" VP
scales, which
> have the merit of a clear mathematical justification.  I
believe the EBU
> scales are similar but not identical to Bethe.
>
> Bethe scales for 3-10 boards are in Appendix V of
> http://usbf.org/docs/COC/usbc2003final.pdf
> I'm fairly sure the underlying derivation was posted here
on BLML, no
> doubt before 2003.

Bethe's approach gives every imp a vp value (up to a certain
maximum). This "experiment" is now mature enough that it
should be universally employed. Some on the C&C committee
were worried that lower-level players would object to seeing
decimal points in the vp scale. That's easily solved by
rounding the scale shown on the ACBL convention card (but
not in the computer), telling players that the computer will
be more exact. The vp total that players get will be a
fairly close approximation to what the computer comes up
with. I doubt that anyone would object to this.

The ACBL continues to use the wrong divisor to make total
scores more meaningful. My suggestion to use the number of
results instead of the number of comparisons received no
reply.

Marv
Marvin L. French
San Diego, California
www.marvinfrench.com