[EM] alternate MMP method

[James Green-Armytage]

Dear election methods fans,

	At the end of my last message about MMP (by the way, I'm sorry for
posting that twice), I wrote: "Personally, I prefer parallel systems to
MMP systems, simply because MMP can't exist without assuming some kind of
party unity and interchangability." But as I wrote it, I started wondering
if it was true. Not sure, I added "At least that is the case in any
conception of MMP that I've ever heard of."
	But I started wondering whether a true MMP system could be developed
based on STV rather than party list. Here is the idea that I came up with.
Not surprisingly, perhaps, it is based on CPO-STV.

________

	For example, imagine for the sake of simplicity a 100 seat legislature
with 55 district seats and 45 adjustment seats.
	Some number of candidates would be running for the 100 seats. Of those
candidates, each candidate would have the option of running for one (and
only one) of the district seats, assuming that they met whatever residency
requirements you might want to impose, etc.
	Voters would rank the candidates on a single ballot (equal rankings
allowed).
	Tally step one: With the electorate divided into the 55 separate
districts, each one only considering the set of candidates running for
that particular district seat, the Condorcet winner (or the winner based
on a Condorcet completion method) would be selected for each district
seat. These 55 local winners would be guaranteed seats in the legislature.
	Tally step two: A CPO-STV tally would be held considering all of the
candidates running, in competition for all 100 seats. However, only those
outcomes containing all of the 55 local winners would be considered. The
winning outcome would be the final result of the election.

_________

	This idea only occurred to me yesterday, and I don't really know how well
it would work. I would not prefer it to straight CPO-STV, but it might
come in handy if there happens to be a situation where people would like
to have a non-party-based semi-PR system but demand single-member
districts. Clearly, it is not as fully proportional as pure CPO-STV, but I
imagine that it should be more proportional that a parallel system
combining CPO-STV and Condorcet.
	I'm not sure if a system with similar benefits can be devised without the
use of comparison of pairs of outcomes.

	Anyway, let me know if you have any criticisms of this idea, or
suggestions as to how to improve it.

-- James Green-Armytage