[EM] Another PR method based on ranked ballots

[Steve Eppley]

On 20 Mar 2003 at 15:33, Alex Small wrote:
> Steve Eppley said:
>-snip-
> > So I prefer a PR system such as the following: Each voter ranks the
> > parties from most preferred to least preferred, then seats are
> > awarded to each party in proportion to the number of voters that
> > ranked it topmost, with the following exception:  The voters'
> > rankings would also be tallied by a good single-winner method (such
> > as MAM), the winning party according to that method would be awarded
> > enough extra seats to make it the largest party.
> 
> I see your point concerning simple PR, but simplicity has its virtues.

Those virtues should not be over-estimated, of course.  Children are 
capable of ranking things in order of preference, and the existence 
of multiple majorities when there are more than two candidates is 
fairly easy to understand once it's pointed out.

> Another way to check the power of "largest minority" extremist parties
> is bicameralism:  Elect one house of a legislature by PR (pick your
> favorite simple method), and the other by some suitable single-winner
> method from large single-member districts.  
-snip-

Again, there would be a trade-off between "simple" and "good" single-
winner methods.  Simple methods would have the problem we commonly 
see, where two big parties each nominate one candidate per office 
(and other candidates are sure losers) and these two don't 
necessarily converge toward the median. (I refer to their actual 
policies, not their cheap talk).  On the other hand, if Alex is 
proposing the "suitable" single-winner be a "good" one (Condorcet-
consistent etc.) then his proposal is not as simple as the one I 
offered.

On 21 Mar 2003 at 0:38, James Gilmour wrote:
-snip-
> Why rank the parties?  Why not rank the candidates?  Then 
> you could get PR of far more than just the parties.

If the legislature has a reasonably large number of seats, say 50 or 
more, there would be so many candidates that no voter could be asked 
to rank that many, nor even rank enough so society could learn their 
preferences regarding the ones whose platforms are among the best 
compromises.  

Australia had this problem using STV PR, so they provided a short-cut 
on their ballot that allows the voter to optionally  pick a party 
instead of ranking the candidates. My understanding is that most 
voters use that option, so they get a PR with far less than if they'd 
ranked the parties, not far more.  

Here's another PR system involving voters' rankings:  

   Let each voter rank some candidates.  

   All candidates are "elected" but do not have the same
   weight in the legislature.  Whenever the legislature
   votes, some may abstain, so set the weight of each 
   non-abstaining legislator to the number of voters who
   ranked her over every other non-abstaining legislator.

  (Note that each citizen may rank himself at the top of
  his ordering, so this would be a hybrid form of direct
  democracy that allows busy citizens the option of 
  being represented by their favorite proxy.)

James Gilmour asked Alex:
> How do you define "suitable"?
> All single-winner methods will distort the wishes of 
> the voters.  In my book, NO single-winner method 
> is "suitable" for this purpose.

When a legislature (PR or otherwise) votes to adopt a new policy or 
maintain the status quo, the wishes of many voters are not granted.  
That's the nature of compromise.

If the policies adopted by a legislature (PR or otherwise) are 
similar to the policies the citizens themselves would adopt if they 
had time to deliberate and vote directly on the issues, then the 
legislature is representative, in my book. 

Since PR legislatures will tend to adopt compromise policies, their 
policies would be similar to the policies of legislatures elected by 
a good single-winner method (which tends to select candidates who 
espouse reasonable compromises).

I think the big open question, when comparing PR to a legislature 
elected by a good single-winner method, is the nature of the vote 
trading that will occur.  A "vote trade" is when a legislator 
reverses her vote on some issue in exchange for another legislator's 
reversal in her favor on an issue she cares about more.  Vote trades 
can make policies more representative, if the citizens would make 
similar trades when voting directly.  But if legislators have 
conflicts of interest, such as the desire for re-election, their 
trades could make policies less representative.  Pork is a common 
example.  Another example is the Shas Party in Israel's PR gov't 
winning public funding for their religious priorities in trade for 
their support of the Likud Party.  Would the Israeli citizens make 
that trade if they could vote directly?  That's a difficult question 
to answer.

-- Steve Eppley