This would basically be a change to the last round for PR-STV. Once (N-1) seats have been filled, the last seat is picked using a condorcet method. All ballots would still have their adjusted weighting, so a ballot that was used to elect one of the already elected candidates would be given a lower weight.
This breaks later no-harm for filling the last seat, but potentially increases the quality. Ordinarily, the last seat is completely elected via IRV, and has the issues with that method. Since that is the case, it may be worth uneliminating all the candidates. However, that means that a vote could be deweighted because it was used to elect a candidate, even though it could have helped elect a preferred candidate in the last round. This method collapses to condorcet in the single seat case, but that is somewhat cheating. ---- Election-Methods mailing list - see http://electorama.com/em for list info
