On Apr 28, 2010, at 11:29 AM, Raph Frank wrote: > On Wed, Apr 28, 2010 at 4:52 PM, Jonathan Lundell <jlund...@pobox.com> wrote: >> This is, I think, a decent general solution to ordering a set of STV >> winners: re-count, with only the current winners eligible, for successively >> smaller numbers of seats. > > Yeah, it is reasonable. > > The fundamental problem is that if you use PR-STV to elect N > candidates from N+1 candidates, then one of the factions that was > represented ends up not represented at all. This isn't so big an > issue when N is large, but it becomes a larger problem as N gets > smaller. > > For example, if the voters were arranged as a circle, and each > candidate represents a 120 degree sector, then picking any 2 of them > is not ideal. > > Something like CPO-STV might help, but the problem seems fundamental.
Whether it's a problem depends on what the ordering is intended for. There's no guarantee in any group that you'll find a majority choice. It's just that successive counting is in some sense a defensible ordering, while order of election is IMO not. ---- Election-Methods mailing list - see http://electorama.com/em for list info