[EM] three tier approval

[Craig Layton]

Just a note: I'll be posting from this address now, but it's still the same
old Craig.

Roy wrote:


>Your idea is very similar to Dyadic Approval, described in:
>http://groups.yahoo.com/group/election-methods-list/message/7157


Dyadic Approval will give a much better result, but it is complex to vote in
and very difficult to design a ballot for.

>If I were to distill both ideas to their essence, it is to take CR
>scores and use them in the pairwise contest matrix, but with
>particular limitations on the allowable CR scores.
>
>Has that particular idea been proposed? Would it ameliorate the
>strategy problems of CR?

Although I haven't tested this, I imagine that in a clear two-horse race,
three tier approval would collapse into the strategy of strongly approving
of your favourite of the two frontrunners and everyone you like better,
strongly dissapproving of your least favourite of the two frontrunners and
everyone you like worse, and approving of everyone between the two
frontrunners.  This would allow some room for a good compromise (Condorcet)
candidate to win whom would otherwise be overlooked by Approval.

In the vast majority of scenarios, there should be an insentive to put at
least one candidate in each category.

Craig