[EM] Dual Dropping method and 'Preference Approval' ballot ideas

[Elisabeth Varin/Stephane Rouillon]

>> I'm honestly unsure if the final ranking of the candidates other than

>> the  winners, and the defeats that need to be overturned to produce
>> such a  ranking, are at all relevant.  It seems like this is more of
a
>> multi-winner  issue.
>
>
>You're right.  Excessive worry about the rankings of the losers is one
>motivation behind Donald Saari's work, at least from what I've seen in
his
>book _Chaotic Elections_.  This excessive worry (and a disproportionate

>view of the importance of symmetry) leads Saari to recommend the Borda
>Count.  Most of us would rather not go down that road....

Considering only overturned pairs with the winner if the winner
selection is
the only output you expect is reasonable.

However, if the complete ranking matters, like for multiple winners,
or when using a Condorcet ordering to make a round elimination process,
then all overturns are important.

It makes sense to adapt the error measure to best fit the expected goal.

Steph.

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em