[EM] Methods that elect all seats by the same faction:

[Paul Kislanko]

Donald wrote:

>- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>"Paul Kislanko" <kislanko at airmail.net> wrote on the EM list:
>
>>A theoretical, though impractical way to handle this would be as follows:
>>
>>Have independent ballot-choices for place 1, place 2, etc. which each
allow
>>the ranking of all candidates. The voter may choose to rank A>B>C>D... in
>>every position, in which case their "vote" degenerates into some variation
>>of IRV, but this would allow the voter to provide more information if she
so
>>chooses. To count the votes:
>>
>>1. Determine the place-1 winner (by whatever method you like).
>>
>>2. Remove the place-1 winner from every position in all the place-2
ballots,
>>with lower-ranked candidates "inheriting" the vacated rankings and
calculate
>>the place-2 winner.
>>
>>Repeat step 2 (remove the place-2 winner from all the place 3 ballots,
etc.)
>>until all seats have been filled.
>>
>>There are circumstances where this approach works well, but I am not sure
>>what all the ramifications would be depending upon the method chosen to
pick
>>a winner at each step.
>>
>
>
>This method is not proportional.
>
>A simple majority of the voters will be able to elect every seat.
>
>Donald

The procedure I mentioned isn't proportional, but it isn't a "method'
either. It's just one way to analyze ballots to extract more information to
use to analyze whatever method is being used. By asking the voters to
indicate preferences separately for each place, you're just making them give
you more information. As I said, it's not at all practical to do that, and
most methods that analyze the separate ranked ballots would probably be
negatively affected - the requirement to provide more information also leads
to more opportunities for manipulation.

As an analytical tool for some methods this approach provides some insight
into how the method works. For others it doesn't tell us anything.

Paul