[EM] Replies to Craig Carey and David Catchpole

[Blake Cretney]

Donald E Davison wrote:

> Craig: If you have in mind some little sample elections that
test methods'
> proportionality then perhaps you could give them.
> 
> Donald: Yes, I do have `some little sample election' in mind,
and I will
> give it.
>      Suppose an association of 120 persons is having its
election for board
> of directors. There are ten candidates running to fill five
seats. The
> election method is STV with random transfer of ballots.
>      Further suppose that the voters know all the ten
candidates well
> enough to rank all ten, and that all voters do rank all ten. I
know this is
> ideal, but it removes the influence of exhausted votes, because
there will
> be no exhausted ballots. I am willing to compare Hare and Droop
without
> exhausted votes, but in a real election the Droop will have
more exhausted
> ballots because there will be more transferring of ballots. The
number of
> possible exhausted ballots is depended directly on the number
of ballots
> transferred.
>      Anyway, if this election is conducted using Hare Quota,
the quota will
> be 24 votes, and the results will be five candidates elected
with 24 votes
> each.
>      This is an ideal PR election. Every vote ended up on one
or another of
> the winning candidates. Every elected member received the same
number of
> votes, a number which is an equal portion of the total votes.
>      Now, if this election is conducted using Droop Quota, the
quota will
> be 21 votes, and the results will be five candidates elected
with 21 votes
> each.
>      But, this is not an ideal election. Not all votes ended up
on one or
> another of the winning candidates, 15 were wasted. While every
elected
> member did receive the same number of votes, that number is not
an equal
> portion of the total votes.
>      If we were to regard the ideal election as a 100, then
this Droop STV
> election example would be 88, (105/120).
> 
> Craig: How much better is Hare than Droop?
> 
> Donald: In this example, Hare is `exactly' 14.3% better than
Droop (15/105).
> 

Let me suggest a different way of looking at standards for
electoral methods.

It seems to me that there are two types of standards:  functional
and procedural.  A functional standard is one that judges a
method by its results.  It is a standard that says "in situation
X, candidate Y must be elected."  All the widely accepted
criteria are functional in nature.

A procedural standard goes beyond what a method does, to how it
does it.  Simplicity is a procedural standard, and a useful one. 
A standard like, "Exhausted votes should be minimized," is
clearly procedural because it is defined based on the procedure
of vote transference.  It is not possible, as far as I can tell,
to determine the number of exhausted votes based solely on the
outcome from a given set of ballots.  You must know the procedure
to be used.

There is a big problem with procedural standards.  Except in
cases like simplicity and computability, we should be concerned
only about the results a method gives, not how it gets them.  If
a method is superior because of results, then a functional
standard could be found to argue this.  If a method's advocates
cannot find a functional standard by which a method is superior,
they are not judging the method by its results, and we have no
reason to believe their favourite method gives results that are
superior.

Your (Donald's) entire line of reasoning involving wasted
votes,proper quotas, and where transferred ballots end up, involves
nothing but procedural standards.  To me, it means nothing more
than that you find the Hare procedure  more aesthetically
pleasing than the Droop.

> 
> Craig: What has PR got to do with STV?

STV will tend to give parties representation in the
legislature approaching their proportion of support in the
electorate.  This increases with the number of representatives
from each district.  People who want proportional representation
usually hope that by giving each faction of the public
representation proportional to its numbers in the electorate, the
legislature will act more like the public would if it was
feasible for them to thoughtfully represent themselves.  They
will find that STV approaches this goal.  It certainly does
better than plurality.

Opponents of PR generally hope to give a fairly uniform group or
individual total power (as limited by such things as judicial
review and a limited term).  They expect that such a government
will be more effective, and more likely to behave responsibly. 
They will tend to find STV inadequate, at least if districts have
enough candidates, and no party has majority support.

-
Blake Cretney