[EM] Election-Methods Digest, Vol 96, Issue 22

[Juho Laatu]

On 13.6.2012, at 11.39, Nicholas Buckner wrote:

> Actually, on a weird second thought, wouldn't a method that refused to
> identify a winner in a three-way tie (Condorcet paradox) be compatible
> with both?

Election methods that are partial in the sense that they don't always find winners are also a good research area. Some criteria talk about probabilities, e.g. "doing ... shall not increase the probability of ...". These criteria would be valid for the partial methods if we define the winning probability of all the candidates to be 0 in the case that the method does not provide any results. In some other society "no result" could mean that the old elected representative will continue. In this case the probability is 1 for that candidate/representative and 0 for other candidates. "No result" could also lead to a lottery. Or maybe we will just use partial criteria that say "If the method gives a result, then ...". Yet one more approach would be to arrange new elections until we have a winner.

Juho