[EM] Voting systems theory and proportional representationvssimple representation. (Abd ul-Rahman Lomax)

[Abd ul-Rahman Lomax]

At 06:56 PM 3/20/2010, Kathy Dopp wrote:
>I have not had enough time to study this in depth but would personally
>support this method only if it were counted using a Condorcet-like
>method and thus avoids all the flaws such as nonmonotonicity, and
>unequal treatment of voters' that STV exhibits.  I don't know what the
>best method would be to count these, but this system sounds good if it
>were monotonic and equitable, therefore STV counting methods would not
>work, but I don't claim to know the best method to use to ensure
>approximate proportional representation that is simple enough to count
>to make it easily audited for accuracy and is fair to all voters and
>monotonic.
>
>http://electionmathematics.org/em-audits/US/PEAuditSamplingMethods.pdf


We are so accustomed to the problems of single-winner elections that 
we fail to notice that multiwinner elections, where accurate 
representation is the goal, operate under almost completely different 
criteria, at least for the bulk of the representatives.

Single Transferable Vote used for proportional representation isn't 
an ordinary election. Why does every voter only get one vote? Think 
about it, please! In Plurality-at-large, often used for multiwinner 
here, every voter gets as many votes as there are seats to be filled. 
How does that work?

In a well-run STV election, with enough seats and not way too many 
candidates, seats start to be assigned before there are any 
eliminations. All these seats are clearly appropriate! Every one is 
given to a candidate who was preferred by a quota of voters.

I have not described how candidate proxy would work in an STV 
election, and I don't like candidate list, precisely because the 
rigidity requires circumstances where there might be monotonicity and 
other failures. I just think that candidate list is better than party list.

In candidate proxy, otherwise known as Asset Voting, there would 
really not be any eliminations. Rather, there would just be the 
creation of seats by the assemblage of a quota of votes. If the quota 
is V/N, the Hare quota, what can happen is that there are unassigned 
seats, which means there are unused votes. Any time those holding 
those votes can assemble a quota, a new seat is created. It's a 
deliberative process, negotiation.

STV is, however, much better for PR than it is single-winner. The 
problems arise with the last elections, and the very last one is, in 
fact, just an IRV election. Candidate list would allow completion 
without ballot exhaustion, and good voting strategy by the candidates 
would really prevent most problems. Vote for someone who uses bad 
strategy? Well, you voted for the person to sit in the Assembly or 
whatever, that would be even worse, surely!

Please understand this: for proportional representation, it is a goal 
that is not utterly ridiculous that every seat is elected 
unanimously. The PR Method assembles the coalitions that do that.

Candidate and party list STV, with as many seats per district as 
possible, would be better than any other method currently in use for 
public elections. Condorcet methods don't apply to multiwinner, not 
on the principle of preferred or chosen representation, which is not 
about "contest."