[EM] Sainte-Lague vs d'Hondt for party list PR

[Michael Ossipoff]

On Tue, Jul 10, 2012 at 5:23 AM, Raph Frank <raphfrk at gmail.com> wrote:

It might be possible to tweak it by having more micro-parties, but the more
accurate it needs to be, the harder for it to happen in practice.

[endquote]

Yes, difficult and risky.

You continued:

It was really just illustrating the point that Sainte-Lague could favour
very small parties.

[endquote]

But, when it happens without being the rssult of successful splitting
strategy, it could just as easily _disfavor_ small parties. Maybe they're
all just barely short of a round-up point.

But, as a strategy possibility, and to the extent that it's a feasible
strategy, where the small parties can control whether they're favored or
disfavored, it's a strategy possibility that favors small parties. If the
problem is found to exist even with a first denominator of 2 (instead of
1), then maybe Largest-Remainder should be considered. But I don't think
it's even a problem in countries where the first denominator is 1.4  With a
first denominator of 2, the small parties, at best, at great difficult and
risk, could only increase their s/q by a factor of up to 4/3, if the
strategy is particularly daring, and successful. (as opposed to possibly
being able to almost double it, with a first denominator of 1). It doesn't
seem to be a problem.

You continued:

Even just using 0.7 as the first divisor instead of 0.5 might eliminate it
in practice

[endquote]

Evidently it does, because that's what they do in the Scandinavian
countries that use Sainte-Lague.   ...A first divisor of .7,  when they're
done as I was doing them would mean a first divisor of 1.4 when those
divisors are doubled to make integers (the odd numbers formula).

By the way, the doubling, to make the odd numbers formula, had especially
much practicality in pre-calculator days, because it reduced the number of
digits in the denominators.

You said:

, but that is bias against smaller parties which has the same effect as
d'Hondt, giving voters an incentive to vote for larger parties.

[endquote]

It only disadvantages small parties for the first seat. d'Hondt is much
worse, because it disadvantages them everywhere.

With Sainte-Lague, with a first divisor of 1.4 or 2, yes someone might want
to vote for an at least slightly larger compromise party--one that at least
is big enough to be sure to get a seat. (Unless the sort of transfers that
you spoke of are available, which would make it safe to initially give your
vote to your small favorite party).

In d'Hondt it's much worse, because, even if you vote for a small party
that's sure to get a seat, you still aren't being counted as fairly as if
you'd given the same vote to a larger compromise party. Vote for a party
that gets fewer seats per vote, and your vote doesn't count for as much.
Everyone would have incentive, need, to vote for a big party, to maximize
the weight of their vote.  That isn't so in Sainte-Lague.



> 2. d'Hondt can strategically force people to vote for a compromise party
>> instead of their favorite, in order to maximize their weight in parliament.
>>
>> What other solutions are there? Largest-Remainder. In your example, In
>> LR, the large party immediately gets 29 quota seats, and then the first
>> remainder seat. The small parties get the rest of the remainder seats.
>>
>
> Another possibility is Largest-Remainder, but with STV transfers allowed.
>

[endquote]

With the transfers, either one-only, or the whole STV ranking with
transfers, for ballots that voted for eliminated candidates, Sainte-Lague
with a first denominator of 2 would be fair, as you pointed out. But if
even the ability to strategically, by splitting, multiply their s/q by up
to 4/3 turned out to be problematic, then it would be time to go to
Largest-Remainder. The bottom-end elimination-transfers would be a good
thing with SL or LR. When reading about PR,I always wondered why party list
PR doesn't have that.

You said:


> The ranking could even be set by the parties.  This would mean no lost
> votes, but still be very simple (voters just pick one party).
>

Yes. It would keep the balloting simple. And it would keep the count easy
too, since the transfers wouldn't have to be made ballot-by-ballot. ...but
only party by party. Anyone could do the count at home, given the party
vote totals. That's one thing that I like about party list PR--The count is
so easy that anyone can do it at home, given the party-vote-totals, instead
of it requiring a large team of counters doing transfers.  Anything that
simplifies the count, and reduces the number of people involved, and the
number of operations involved, and makes computer-counting unnecessary,
makes it a lot easier to make the count secure against count-fraud.

STV advocates, and advocates of rank methods for single-winner elections,
take note.  How about Approval for single-winner elections, and party-list
for PR, to make the count easy and secure against count-fraud.

Japan used to use the Single-Nontransferable-Vote (SNTV). It, too, has a
simple and easy count, though it lets people vote for individual
candidates. It requires some sort of organization, agreement or
instructions among a party's voters, but that needn't be a prohibitive
problem. They used it for a long time, so it must have worked fine.

Likewise, the various open list systems have the simple and easy count of
party-list PR, while still letting people vote for candidates, to determine
which candidates will occupy the seats won by a party.

Given the count fraud problem, STV, and rank-methods for
single-winner elections, wouldn't be the most prudent.course.

Even national level pure STV, but by party, would be possible.  There would
be no "elected" party.   In each round the party with the lowest remainder
would be "eliminated" and have its number of seats "locked" by rounding
down.  There could then be surplus transfers.

[endquote]

Deluxe. Maybe, with those surplus transfers, and then eliminations at the
bottom end, that "party-STV" is the most deluxe kind of party-list system.
I think it would be great.

And of course, since that STV is by party, and not by candidate, the count
would still be quick and easy, and would be easily and quickly verifiable
by everyone, at home, if the rankings were made by the parties, so that
transfers wouldn't have to be done ballot-by-ballot.

Maybe, voting could be open-list, and, additionally, the voter could
indicate where s/he wants the party's votes to transfer to, in the event of
surplus or elimination.

The party getting the most transfer transfer marks on the ballots received
by party X would be the party that party X transfers to in the event of
surplus or elimination.

In that way, the voters would still be fully in control.

And you could make it a little more deluxe, by letting the voter give
transfer marks to as many parties s/he wants to, when marking hir ballot
giving hir vote to party X. So it would amount to an Approval count, to
determine where that party would transfer to in the event of surplus or
elimination.

You said:

I think a transfer system of some kind would make people much more
confident to vote for their favourite party.

[endquote]

Certainly, regardless of the expected votegetting ability of their favorite
party. One could show support for a party whether or not it can win. One
could maybe even help it win, where one might not have otherwise wanted to
risk voting for it.

You said:

 Even if on average the system is unbiased, voters wouldn't want to risk
giving a majority to a different group.  Transfers would more consistently
give a majority of the seats to a majority of the voters (assuming it is a
solid coalition).

[endquote]

Sure, the arguments for STV would apply. And having a parliament in which
are seated more parties that are someone's _favorites_ would make for an
especially progressive, interesting, dynamic and more truly representative
parliament.

Mike Ossipoff



> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120712/87387121/attachment-0004.htm>