[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ga] STV voting method



I accidentally sent the following note on STV to Elisabeth rather
than the list.  




Elisabeth Porteneuve wrote:

> Michael,
>
> I was explained that when external considerations like geographical
> diversity (which is essential in ICANN process) are added to the
> STV algorithm itself, the method became distorted.

True.  Geographic diversity rules can, if mechanical, distort
everything.  That is why I have always suggested we allow that form
of
diversity to be accomplished through the same natural processes as
we
use to achieve other forms of diversity and not put it on a higher
pedestal.  I.e. I would achieve it through STV, as well.

> ...
> I believe the electorate size shall have some minimum to be
> considered for STV, otherwise it is meaningless.

What is the basis of this conclusion?  What is that number (i.e. do
we
have more members than that?)  How significant is the skew
(especially
given the distributive nature of STV vs. the absolute prejudice
resulting from "first past the post" elections)?  What is the nature
of
the skew in the alternative systems?  How does this number differ
under
STV from what it might be in the alternative system you would
suggest?
What mechanisms should be employed to counteract any skew (such as
super majority requirements)?

> There should be at least two parameters to be taken into
> account simultaneously:
>         some ratio between Number-of-people-to-be-elected
>         to Electorate-size *and* Electorate-size
> 	  to have some mathematical significancy.

What is that ratio?

> Last not the least, the understanding of voting procedure
> by everybody is essential -- and this is not a case for STV.

Please explain this.  As I understand the system, the voters are
only
told to pick their top X choices in order of preference.  The
computer
does the rest.

Here is an explanation from someone familiar with the system:

     > Using STV, if there are P places, a voter assigns each of
     the P more
     > preferred candidates a rank from 1 to P.  The ballots are
     run through
     > a computer which automatically selects the most preferred
     candidates
     > one after another, assigning any unused votes after each
     round to the
     > next more preferred candidate.
     <http://nersp.nerdc.ufl.edu/~grapski/stv.html> &
     <http://archive.rte.ie/election97/pr.html>



>
>
> Elisabeth