[EM] Three rounds

[Raph Frank]

On Thu, Nov 13, 2008 at 10:11 PM, Juho Laatu <juho4880 at yahoo.co.uk> wrote:
> This party list based method actually allowed the party supporters not to be rock solid supporters of the party. Those 7 A2>B>A1 voters were able to indicate that they preferred B to A1. And their favourite still won.

Hmm, it is IRV except but it uses a different elimination order.  A
candidate from the smallest party is eliminated first (and presumably
the candidate from the party with the fewest votes if the party has
more than 1 vote).

I assume that the party totals are re-calculated after each
elimination?  That means that the largest party doesn't automatically
win (as their candidates would be eliminated last)