[EM] MAMPO variations
Forest W Simmons
fsimmons at pcc.edu
Sat Mar 24 14:19:04 PDT 2007
Thanks, Kevin, I should have known that Woodall would have already
considered this in his systematic way.
BTW, the more I think about MAMPO, the more I like it.
How about three slot MAMPO?
How about this variation of Chris Benham's idea?:
For each candidate X, let EQF(X) be the number of ballots on which X is
ranked first or equal first.
Let MPO(X) be the maximum pairwise opposition to X.
Elect the candidate X for which the score EQF(X)-MPO(X) is greatest,
i.e. for which MPO(X)-EQF(X) is least.
This method kind of tacitly acknowledges up front that these FBC
compliant methods tend to encourage lots of Equal First rankings.
But combining this method with ER Bucklin or Range Bucklin would allow
voters simpler zero-info strategy:
With range ballots, let R be the highest rating at or above which some
candidate is rated on more than half of the ballots.
Now collapse all of the ratings above R to "Equal First" and apply the
Kevin V. wrote ...
>Incidentally, Woodall calls the level of "consent" for A over B the
>"doubly-augmented gross score" for A over B. He calls MMPO "MinDAGS."
>I believe what he calls "DminDAGS" is equivalent to the method you're
>talking about, in which case it should satisfy plurality, majority for
>solid coalitions, and clone independence, but not Condorcet or LNHarm.
More information about the Election-Methods