The ballot is a normal preference ballot.
The first thing that is needed is a list of all outcomes, like PAV.
Then, all outcomes are compared against each other pairwise. Each voters preferences are compared, and each voter gives a score to either outcome based on which candidates were ranked higher on their ballot. The score is assigned by determing which outcome has more people that were ranked higher on the voters ballot. If the outcome had 1 person ranked higher, that outcome gets a score of 1. If it had 2 ranked higher, it get 1 + 1/2 . If it had 3 ranked higher, it get 1 + 1/2 + 1/3, and so on.
The scores of all the voters given to each outcome are all added up to get the voters total "satisfaction" with the score, and the outcome with the highest satisfaction wins.
I haven't thought about how to apply completion methods yet. I don't really read up on the things that much.
Obviously, this method is going to require quite a bit of computational power. A prohibitive amount. But I believe this is probably the Proportional Condercot Method that comes closest to actual Condercet, and is therefore at least worthy of note. If only we could find a way to do these kinds of elections without the lengthy comparison of all outcomes method...
Get amazing travel prices for air and hotel in one click on Yahoo! FareChase
---- election-methods mailing list - see http://electorama.com/em for list info