[EM] Approval for many candidate, non-partisan, multi-seat elections

[James Gilmour]

Rob Lanphier Sent: Monday, August 15, 2005 9:46 AM
> As you saw earlier, some people on Wikipedia are considering 
> a switch to Bloc voting (plurality-at-large) for a multi-seat 
> election, away from Approval.

Whatever the merits or de-merits of multi-seat Approval, I am sure Bloc Voting (multiple-X) will be worse.  You don't
need organised slates of candidates or organised groups of voters ("parties") for Bloc Voting to produce distorted
results.  BV will give all the seats to the largest minority of voters who vote for a similar selection of candidates,
leaving the other large minorities (who may constitute a majority of the voters) with no voice.

> There's a good argument to be made against STV in this 
> context.  This isn't about finding a diversity of opinions 
> (proportionality)...they seem to want a cluster at the median.

Do you really want a cluster at the median - a bunch of clones?  Who are "they" who want this?  Do they speak accurately
for the electorate?  It seems odd to choose the members of the Arbitration Committee by election if the purpose is not
to ensure that the committee is representative of its electorate, or at least, representative of those electors who
exercise their right to vote.
 
> So is Condorcet (specifically Schulze(wv)) the best 
> recommendation to make here, with each winning candidate 
> popped off the top of the stack until the five seats are 
> filled?  Or is there arguments to be made that some 
> proportionality is always a good thing, and therefore a 
> proportional system should be used?

I don't know how multi-seat Condorcet would work out in practice, but maybe (NB maybe) multi-seat IRV might give us some
insight.  How often, in real elections, would the Condorcet winners be different from the IRV winners?  If they would
usually be the same, the comparison might be valid and useful.  Multi-seat IRV does ensure that every elected member has
the support of 51% (round figures) of the votes (then in play, etc, etc).  But that could well leave 49% of those who
voted with none of their preferred candidates elected.  Those elected would form a cohesive group, but they would not be
clustered around the median of the voters' wishes.  Would multi-seat Condorcet have the same majoritarian defect?

The objective in this election may not be to secure maximum representation of the full diversity of opinions among the
voters, but STV-PR might well give a much more "centrist" balance than any single-winner method applied to a multi-seat
election.  But if there are significant divergent views among the community on whose behalf this Committee will work,
why should that diversity not be reflected in the Committee's composition?
James Gilmour