Party List P.S.

[Mike Ositoff]

When I suggested several single-winner methods to use, where
they're needed for the list choosing & ordering procedure I
described (to choose the candidate for the number 1 place in the
list, and whenever not all of the N-1 candidates that you've already
given list positions to are among the N winners of the current
N-candidate STV count).

But I left one out: Plurality With Withdrawals. 

Steve defined it on ER, but I don't know if he defined it on
EM, so let me re-state the definition here:

The candidate at the top of each voter's list receives a Plurality
vote from that voter, and the candidate with the most Plurality
votes wins. The pairwise totals from the voters' rankings are
available to the candidates, and the they're also informed of the
winner of the Plurality count. Any one or more of the candidates
may withdraw from the election, and call for another election.
Of course if any one or more withdraw after a result is announced,
then that still calls for a re-count in which those candidates
are removed from the rankings (so if one of them previously topped
your ranking, now your next choice does).

Candidates can withdraw any time a result is announced, and so
the process of withdrawals & new count is repeated till no one
else wants to withdraw.

***

This method, which is a simple & obvious remedy for Plurality's
problem, might be the most easily proposed of the best single-winner
methods. It would save you having to define the other best EM methods
and save you from having to explain methods that are new to
people, and save you from having to deal with people who propose
different count rules.

So my suggestion would be incomplete without suggesting
Plurality With Withdrawals as one of the best methods from EM.

***

Of course if there's strong insistance on another count rule,
such as IRO, then that other rule could also be used with
the candidate withdrawal option.

***

I don't believe that it's advisable for people here to be 
suggesting that you change your country's election system. Of
course we should try to answer your question about choosing &
ordering the list, but I don't beleive that we should be telling
you what changes should be made in your national election system,
because we aren't there. You said you want to propose some changes
in it, and you're of course the one who's in a position to know
just what needs improvement, and what would likely work better.
Plainly, the system you have is working fine (but if there's some
way in which it isn't, that's something that you, not anyone
in the U.S., would know about).

If you're inclined to propose STV, in an open list election, to
determine which of a party's candidates get the seats that that
party wins, that seems like an ideal way to do it.

If you want to propose that STV be the national PR system, then
you don't want the small districts that are usually used with
STV. If it were used, it should be used with the whole country
as one district, as is the case with the current system. If
the current system is working well, then caution & hesitation
are advisable when considering replacing it with STV, a method
with a new & different set of problems and possible public 
objections. If I were there, I wouldn't consider proposing
such a major change to a system that already works well.

***

But I've suggested STV as part of my suggested procedure for
choosing & ordering the list, and there's been some discussion
about the choice of quota. And so, even though I don't ordinarily
discuss PR anymore, I'd like to comment:

The advantage of the Hare quota (total votes/total seats) is
that it's the optimally proportional quota, and the unbiased
quota. It's even-handed with regard to the size of factions
in your party.

The Droop quota slightly favors large factions, giving them
somewhat more than their share. But its advantage is that it
guarantees that a faction with a majority of the voters will
automatically get a majority of the seats. 

It, of course, achieves that guarantee by being biased in favor
of the large. With the Hare quota, a faction could be a little
over a majority of the voters and not get quite a majority of the
seats, or it could be a little under a majority of the votes, and
still get a majority of the seats. In other words, the natural
inevitable random deviations from proportionality could go either
way. Droop's bias for the large makes sure that a party with
a voter majority gets a seat majority.

If your party has a well-defined majority faction, then they'd
likely insist on the Droop quota, and would have the power to
get it. Maybe that's why Droop is used so much. No problem. It's
perfectly ok. Just not as ideally unbiased & proportional as Hare.

I've personally felt that PR is about inclusion & proportionality
rather than majorities, though I consider majority rule important
for single-winner methods (as most people do). So I like the Hare
quota for STV, and that's what we proposed when we tried an
STV initiative in the town where I live (it didn't get enough
signatures to get on the ballot--that's what usually happens to
initiatives for electoral reform here). People not in the majority
faction needn't feel guilty or undemocratic for favoring Hare, which
doesn't guarantee majority control, because there's no law that
says that minorities have to work for the majority. And even
minorities that are part of the majority faction have the right
to defect from that majority if they feel that Hare is better for
their subfaction.

Anyway, this is the last time I'll get on the subject of PR.
My suggestion for the party lists uses STV, but suggesting the
use of STV needn't involve talking about ways of doing STV.

***

I feel that my suggestion for your party list is the best
that can be done for proportionality in that situation.

Another possibility, of course, would be to simply repeat the
use of a single-winner method. Use it to pick the number 1 candidate.
Then use it among everyone but him, to pick the number 2 candidate.
Etc. But that wouldn't be PR, wouldn't be an attempt or approximation
to proportionality between the composition of your party's fractie
and the composition of its voters. I believe that my suggestion
combining STV with the use of a good single-winner method is
the best that can be done toward that goal, in that situation.

Mike Ossipoff