[EM] Proportional Representation Systems I'd Support

[Kristofer Munsterhjelm]

Kathy Dopp wrote:
> On Thu, Mar 25, 2010 at 3:01 PM,
> <election-methods-request at lists.electorama.com> wrote:
>> Send Election-Methods mailing list submissions to
>>        election-methods at lists.electorama.com
>>
>> From: Kristofer Munsterhjelm <km-elmet at broadpark.no>

>> The general scheme would then be: party list with some majoritarian
>> method determining the internal composition. Each party gets as many
>> seats as the party list method says they're entitled to, then the
>> majoritarian method allocates candidates within those "blocs".
> 
> Now that sounds like a GREAT method if:
> 1. the voter can vote for one party
> 2. the voter can rank members within that party (say up to 3 members
> to keep the ballot doable)
> 3. the voters' consensus ranking of members within that party is
> decided using the Condorcet method.

How about a ballot like this:

	Party name

# Party candidate A                    # 1 #   #
# Party candidate B                    # 2 #   #
# Party candidate C                    # 3 #   #
# Party candidate D                    # 4 #   #
# Party candidate E                    # 5 #   #

where the official ordering is in the first numeric column, and the 
voter may number the candidates (in the second numeric column) to 
override the default order? E.g. someone who wants D > B > A and doesn't 
care about C or E would write:

	Party name

# Party candidate A                    # 1 # 3 #
# Party candidate B                    # 2 # 2 #
# Party candidate C                    # 3 #   #
# Party candidate D                    # 4 # 1 #
# Party candidate E                    # 5 #   #

Thus the voter can rank as many or as few as he wants. If it's closer to 
closed list PR, the ranking would be completed in the default order; if 
it's closer to open list, putting in any numbers in the second numeric 
column means it counts as a (possibly truncated) Condorcet vote.

The "near-closed" version would complete the vote to give D > B > A > C 
 > E. The "near-open" version would just submit the vote as D > B > A.

I'd imagine that unless regulations limit them, parties would submit 
lists with as many candidates as there are seats in the district in 
question, so that if by some miracle a single party gets unanimous 
support, it would have enough candidates to fill all the seats. Thus, 
for something patterned after Ireland (5-seat constituencies), a complex 
ballot might not be needed.