[EM] Improved Instant Pairwise Elimination

Richard Lung voting at ukscientists.com
Sat Jul 10 09:12:21 PDT 2021


Dear All,

As there is only one truth, a valid voting method must have the same count for an election as an exclusion/elimination (symmetrical count requirement) otherwise one of the counts must be false (more or less). 
(The basic dysfunction of MMP is that it has votes for two contradictory election counts, so one of them must be wrong -- indeed both are!) 
Generally, voting methods have one election count and a buttress of an elimination count, to keep the election count going to completion. The elimination generally is not even considered as an equal count, in its own right, to the election count. Elections are generally just that, (uninomial) election counts. 
Symmetrical election and exclusion counts are a binomial count. It took me a life-time to arrive at that (when over 50). It took another 14 years to fully develop FAB STV. (Afterwards, I worked out how to do a 2-D voting system with a complex number count. I consider myself retired.)

FAB STV Is monotonic. Shifting about the preferences won't give perverse results. And so it is not manipulable by strategic voting.

Regards,
Richard Lung.




On 9 Jul 2021, at 10:18 pm, Kristofer Munsterhjelm <km_elmet at t-online.de> wrote:

> On 09.07.2021 23:02, Susan Simmons wrote:
> 
> Very good point, Richard, these loser-elimination methods are just
> stop-gap methods that might be a stepping stone to something better ...
> we shall see.

Sometimes it's the only way we know how to get certain properties (like
Condorcet and DMTBR), but I suspect that just means that the theory is
lacking.

The manipulability integer programs suggest that a monotone
strategy-resistant method exists. But I have no idea how to actually
construct one, because the strong monotonicity requirements for
something like Benham are way too strong - as are the DMTBR ones.

-km


More information about the Election-Methods mailing list