[EM] Why I think IRV isn't a serious alternative 2

[Juho Laatu]

--- On Wed, 24/12/08, James Gilmour <jgilmour at globalnet.co.uk> wrote:

> I do not think you have to be anywhere near the zero
> first-preferences Condorcet winner scenario to be in the
> sphere of "politically
> unacceptable".  I am quite certain that the 5% FP CW
> would also be politically unacceptable, and that there would
> political chaos in
> the government in consequence.  The forces opposed to real
> reform of the voting system (big party politicians, big
> money, media
> moguls, to name a few) would ensure that there was chaos,
> and the electors would have an intuitive reaction against a
> weak Condorcet
> winner so they would go along with the demands to go back
> to "the good old ways".

> ... the Condorcet voting system will never get off the ground
> so long as a 5% FP Condorcet winner is a realistic scenario,
> as it is when
> the current (pre-reform) political system is so dominated
> by two big political parties.

The question is if methods that may
regularly elect a 5% first place support
Condorcet winner can be politically
acceptable.

One reason supporting this approach is
that most single-winner methods are
designed to always elect compromise
winners. (Some methods like random ballot
are an exception since they give all
candidates a proportional probability to
become elected.)

Using single-winner methods to implement
multi-winner elections is a weird
starting point in the first place. This
approach works for two-party systems,
although PR of those two parties will not
be provided.

If one uses a compromise / best winner
seeking single-winner method like
Condorcet for multi-winner elections
(using single-seat districts) it is in
principle possible that all the districts
will elect a 5% FP support candidate. In
the worst case there are the two old major
parties with close to 50% support and then
one or few compromise candidates in the
middle.

The proportionality of this single-winner
single-seat district based Condorcet for
multi-winner elections may thus be quite
biased. The same applies to all similar
misuse of single-winner methods.

What is the fix then? One approach is to
use a single-winner methods that do not
aim at electing the best (compromise)
winner in each case. Random ballot would
be one. We would get quite decent PR this
way. But the random nature of the method
is maybe nor what people want.

Another approach is to use IRV or some
other method that favours the large
parties. No proper proportionality
provided but this approach is close to
the current plurality based approach in
many two-party countries. This approach
may thus be acceptable in wo-party
countries (but probably not elsewhere).

A third approach would be to implement
some PR method. Typically this means use
of multi-winner districts (although not
mandatory since one can do this in
principle also with single-seat or
few-seat districts).

One can interpret this as one argument in
favour of IRV-like methods that will
to some extent maintain the dominance of
the old large parties, or as a warning
against trying to achieve PR by using
single-winner methods for multi-winner
elections.

- - -

Since I mentioned option of having PR
and "few-candidate districts" here is
also one sketch of such a method.

Each district has two seats. Votes for
each party are first counted at national
level and the number of seats will be
allocated to them proportionally.

At the second phase seats are allocated
in the districts. The district that has
strongest support of some single party
gets the first seat. A quota of votes is
deducted from its votes. Next the second
strongest claim will be handled. Claims
that would exceed the two seats per
district limit or the national level
allocation of seats to each party will be
skipped. The process continues until all
seats have been allocated.

One can expect that each two-seat
district got at least one representative
that the voters clearly wanted. The
second seat will in some cases go to some
small party that didn't get as much votes
in this district as some other party did.
This violation of "local proportionality"
is needed to maintain the "national
proportionality" and the "two-seat
district approach".

(Mixed member systems would be another
approach.)

Juho