[EM] Election-Day musings. Sainte-Lague is the best PR.

[MIKE OSSIPOFF]

Yes, election day was a long time ago, but I reside in a remote rural area, 
and I don't often get to a computer.

I already knew this, but it occurred to me again that, if we have to have 
representative government, it would be better if people were voting for a 
party. Voting for a party platform instead of for a personality, a hairdo, 
someone who acts presidential, etc.

In representative govt, that would be PR. I've talked here about how Direct 
Democracy with Delegable Proxy (a term I got from James) is the form of govt 
that I prefer. I believe that people would accept it long before they'd 
accept PR. Proxy is reinvented and re-proposed again and again by people new 
to the subject. I've observed that myself at a PR meeting. But the 
proportionalists stubbornly reject it.

As I've said all along, single-winner reform is the significant but winnable 
electoral reform, since it seeks only to do in a fairer way what we already 
do, without trying to drastically change the system of govt, often without 
even a constitutional amendment.

But if it has to be representative govt, and if we disregard PR's poorer 
winnability in the U.S., then PR has the big advantage that people vote for 
a party. Even with STV they can vote for a party if, as in Australia, they 
can mark their favorite party's recommended ranking, by making one mark 
above the line.

Obviously, given what we've been hearing about the reliabilty of computer 
vote counting :-( STV is at a great disadvantage, as the most 
computation-intensive method (with IRV as a close 2nd). STV and IRV are out, 
just because of their computational requirements and resulting count-fraud 
opportunities.

If you want to vote for candidates, what's wrong with the PR system whereby 
you get as many votes as there are seats available, and you can give them to 
any candidate you want to. The candidates are listed in party lists on the 
ballot. All the votes received by a party's candidates count together as 
that party's party list PR allocation votecount, for awarding seats to 
parties. Then, after the party seat allocation, each party gives its seats 
to its candidates starting with the ones who got the most votes.

There are a number of countries that use systems like that.

I still say that single-winner reform is a much more worthwhile thing to 
discuss than PR, due to SW reform's better winnability, and the fact that, 
no matter what the system is (single-member districts, Direct Democracy, PR 
parliament, etc.) it will still be necessary, at some point, to often choose 
one alternative from among several: Single-winner social chioce methods are 
the fundamantal mechanism of democracy.

Anyway, some states or municipalities might claim that it's illegal to 
mention parties on a municipal ballot. So a party list system could use 
slates instead of parties.  Slates are very much part of every municipal 
multiseat election.

But, if slate-list PR is still not winnable in munipal elections, then CV 
(cumulative Vote) would be better than STV, due to STV's computation 
intensiveness. The form of CV that I prefer is integer CV, where the voter 
has as many votes as there are seats to fill, and can give them has s/he 
chooses, concentrating them on one or more candidates if desired. In an 
N-seat election, the N candidates with the most votes win. I prefer that 
form of CV, because the whole point of proposing CV instead of STV is to 
avoid computation.

The Single-NonTransferable Vote (SNTV) is ok too. It places an additional 
organizational demand on voters: "If you're a Democrat, and your last 
initial is from A to J, then vote for Jones". So why mention SNTV? Because, 
if a handcount is desired (and believe me, a handcount is definiltely 
desired), SNTV greatly reduces the number of votes needing to be counted. 
Australia uses handcounting, so don't say it isn't feasible for national 
elections. We have more voters? Yes, and more counters too.

But CV is probably a lot more winnable than SNTV (or STV). CV was adopted in 
several municipalities during the 80s or 90s. CV was used in one of the 
states for legislature elections.

Which PR list method? Sainte-Lague.

What's the obvious allocation? Divide the total votes by the total seats, to 
get the correct vote-cost for a seat. Divide each party's votes by that 
"Hare quota", and round off to the nearest whole seat. The whole purpose of 
that is to use a uniform seat-price, and one that will give the desired 
number of seats. But, due to the vagaries of rounding-off, it might not 
actually give exactly the desired number of seats. What to do then? Well, 
what's the whole purpose of the Hare quota? I stated it a few sentences ago. 
When Hare doesn't achieve its own purpose, it's easy to find a quota that 
will: Raise or lower the Hare quota so that it will result in the desired 
number of seats.

There, you have the Sainte-Lague seat allocation.

Yes, SL is usually stated differently. But it's easy to show that the usual 
SL definition is just an implemtation of my above definition.

Say we start with a quota so high that no one gets a seat. Then we start 
lowering it, awarding one seat at a time. The next seat will go to the party 
for which votes/(current seats + .5) is the highest. That will be the next 
party to round up to the next whole seat-number, as we lower the quota.

Now, if we double those "current seats + .5 numbers, then we get a sequence 
of odd numbers. That's where the well-known SL odd numbers formula comes 
from.

It's often mistakenly said that SL favors small parties. Not so. SL is the 
unbiased monotonic allocation method.

I'll continue this in a subsequent posting.

Mike Ossipoff

_________________________________________________________________
Share your latest news with your friends with the Windows Live Spaces 
friends module. 
http://clk.atdmt.com/MSN/go/msnnkwsp0070000001msn/direct/01/?href=http://spaces.live.com/spacesapi.aspx?wx_action=create&wx_url=/friends.aspx&mk