[EM] Yet Another PR method

[Blake Cretney]

The following is my proposal for a PR based electoral system.

First, let me point out that this is not actually my personal
preferred system.  I prefer a closed-list PR with a vanishingly low
threshold.  I consider the following method to be inferior in a number
of ways.  However, I think it is much closer to common opinion both
among reformers and the general public.  Essentially, this is a
variant on open-list PR as used in Finland.


Ballot:

>From the voters point of view, the following changes will be evident.
 Ballots have more names, including multiple candidates from each
party.  As well, each district (constituency/riding) is larger and has
more members representing it.  I suggest districts should have two,
three, or four members.

Each candidate is paired on the ballot with his or her party, with
the exception of independents.  Each vote is therefore either a vote
for a candidate-party combination, or a vote for an independent. 
Voters will only vote for one candidate, as they do in FPP today.


Parties:

In general, votes are counted for both candidate and party.  However,
two tricks are used to handle special cases.  First, an independent is
considered to be a party of one.  Second, some parties will drop
bellow a specified threshold (for example 5%).  They will not simply
be discarded, but instead they will be fractured as follows.  A
fractured party has its support in different districts tallied as if
they went to different parties.  That is, the Illiberal Party in
Cabbageshire might be tallied as if it was an entirely separate party
from the Illiberal Party in Lettucehaven.

This threshold has no effect on independents, since they by
definition only run in a single district.  However, parties with
small, spread out, national support will likely be reduced or
excluded.


Tallying:

Firstly, each party (with the adjustments above), is given a total
national vote.  Then, candidates are assigned seats using the St.
Lague formula.  What this means is that seats are handed out one by
one to whichever party currently has the highest score for

Number of votes/[(number of seats already won) + .5]

Once a seat is assigned to a party, it must be assigned to a
district.  The district is the one with the highest score for

Votes for party in that district/[(number of seats for the party in
the district) + .5]

Finally, the candidate elected is the candidate in the appropriate
party and district, with the highest number of personal votes.

Eventually, districts will start to fill up.  That is, they will have
all their seats allocated.  When that happens, they are no longer
eligible for seat assignment.

Sometimes, a party will have all its members for a given district
elected (before the district runs out of seats).  When this occurs,
any votes above

(total votes in district / number of seats in district) * number of
seats allocated to party

are subtracted from the party's national total.  As a result, parties
will likely run enough candidates that this will only very rarely be
necessary.


Effect:

This method strictly enforces proportional representation at the
national level.  At the district level, proportional representation is
also largely adhered to.  District size and number of seats can vary
as desired, with no Gerrymandering effects in national party
percentages, except that the fewer seats per district, the greater the
effect of the threshold.

>From the point of view of the ballot, the election is so similar to
FPP that some uninformed voters may not even be aware of the change.

The strange provision about subtracting votes from the national total
when a party gets all its members elected in a district is to prevent
a peculiar result.  Otherwise, a party could intentionally run few
candidates in some districts, knowing that its underrepresentation in
these districts would have to be made up elsewhere.  The result would
be that districts would be saddled with unproportional results. 
Instead of allowing parties to abuse the national proportionality
guarantee, they are punished if they don't run enough candidates to
receive their support.

My rationale in advocating this method is that it is proportional,
but in other ways is as close to FPP as possible.  History has taught
us that anti-PR campaigns are usually better funded than pro-PR
campaigns.  The result is that many of their criticisms will go
unanswered.  It therefore makes sense to reduce the differences
between the proposed PR method and FPP, so that the pro-PR campaign
can be more focussed.  For example, I don't think geographic
representation is a good idea, and I believe that pro-geographic
arguments can be effectively countered, but realistically, this
argument will never be made.  The same can be said for preferential
ballots and closed party lists.


Perceived Advantages:

Proportional Representation
    The method is extremely proportional at a national level.  The
only other ways to achieve such a high degree of proportionality would
be to use a top-up party list (as MMP) or have only a single district.
 Because it retains proportionality even with small districts, it
would be possible to have as few as 2 candidates per district.

Ballot similar to FPP

Geographic Representation
    Actually I think this is a disadvantage, but some people insist
on it.  Districts are larger though.

Votes for candidates
    One of the most common arguments against closed-list is that the
parties are deciding what candidates get elected.  In this method,
each candidate is clearly and separately responsible to the voters. 
The voters are able to elect any subset of the candidates running,
restricted only by the number of seats allocated to each district. 
There is no ability for a party to place an unpopular candidate at the
top of their list.

Thresholds
    Thresholds are implemented, but in a way that does not adversely
affect independents, or parties with strong regional support.


Disadvantages and Oddities:

If a district has only one seat, it is possible for a candidate with
a majority of the votes to be defeated.  In general, it is necessary
to keep districts large enough that the results will not be perceived
as going against local majority wishes.  The legislation should
specify a minimum and maximum number of seats for a district.  I
suggest between 2 and 4, inclusive.

Disparity in voter turnout can create a situation in which a district
gets unproportional representation.  For example, if there are two
districts, each with 4 seats

         Red  Blue
Dist 1   40   20
Dist 2   80   160
Total    120  180
%        40%  60%

The problem here is that because district 1 has an equal number of
seats to district 2, but far fewer votes, it will be necessary for
Blue to be over-represented in each district, in order to be fairly
represented over all.  Although I don't recommend it, this perceived
problem could be eliminated by scaling districts according to their
seat allocation before the votes are tallied.

It is possible to argue that in some cases the order in which
candidates in the same party and district are elected is almost
arbitrary.  If there is one well known popular candidate, the other
candidates may be ordered by a tiny fraction of the vote.  However, in
FPP the vast majority of candidates are elected based on accidents of
geography by people who cannot differentiate them from their party. 
Perhaps the ability to pick and choose candidates in this method would
stir voters to become better educated about their local candidates ;) 
In practice this problem is greatly reduced by having fewer candidates
per district.

Often countries like to give small regions extra representation.  If
a country has an island, for example, that the government decides must
have a single representative, despite its population, some
modifications have to be made.  The simplest is to give the island its
seat, to be decided by FPP, and tally the election as completely
separate from the election as a whole.  A small number of low
population areas could retain FPP.  If this is to be done a lot, it
would be better to use a formula to scale the votes.

The method is fairly complicated.  Simplicity is a good thing, but I
think its necessity is exaggerated.  FPP isn't a simple system.  The
drawing of district boundaries is critical to FPP, but few people have
any idea how this is done.  People understand the system well enough
to know how to vote, and to have some idea of what's going on.  Most
people seem to think that that is enough.  As well, the media seems to
uses "complicated" almost as a synonym for "new", so I don't think it
is realistic to expect the pro-PR side to get much credit for
proposing a simpler method.  

In fact, if you ignore the details, the method is fairly simple.  It
gives each party seats in proportion to their votes, and uses a
formula to give close to proportional results in each district. 
Candidates with higher number of personal votes are elected before
other members of the same party, in the same district, with lower
numbers of votes.

Any comments?

---

Blake Cretney