Juho wrote:
They also said:
> Condorcet-Schultze med 30% kvotering
= Condorcet-Schultze with 30% quotas
I just wonder if that adds something to the basic Schulze method.
That means that 30% of the people on the list has to have some property.
I'm not sure what that property is, but I guess it refers to matters of
gender - at least 30% of each.
I don't see any mention of how they're going to achieve this. If I were
to make something like that, I would probably divide the result into
ordered lists, based on the property (e.g. gender), then go
straightforward down along the social ordering until either group
decreases below the quota. At that point, pick the highest ranked from
the list of the minority in question until that group is no longer below
the quota.
One could probably devise even more complex solutions involving global
optimization, particularly for methods that return a cardinal social
"ordering" (a ratings ballot as output). For such an ordering, one could
simply phrase it as an integer programming problem: maximize the total
rating, subject to that one may not pick more than the number of seats
and that no group may have less than 30% of the seats. I would be
surprised if that's how they do it, though, in particular since the
Schulze method doesn't return a cardinal social ordering.
----
Election-Methods mailing list - see http://electorama.com/em for list info
