[EM] Demorep's Runoff Ideas

[Forest Simmons]

In recent postings Demorep has suggested some sequential elimination
methods based on Demorep style ballots (ranked preference ballots with
Yes/No tolerability markings).

These elimination methods (unlike Demorep's ACMA) are summable only in
data summaries that grow exponentially in size as the candidates increase
in number.

So in an election with much more than a dozen candidates the individual
ballots would have to be referenced repeatedly during the calculations, as
in IRV elections.

In some situations this may not be a problem.

Other than this problem I think the suggestions are very good for
elections in which the voters have the patience to rank the candidates as
well as give the Yes/No information.

However, I still think the five slot grade ballot (on which each voter
grades each candidate) would be easier for the voters to fill out than the
ranked preference ballots.

Although the five slot ballots don't allow complete rankings, the voters'
choices of which rankings to NOT collapse reflect intensity of preference
that is not revealed in a full ranking.

The five slot ballot with Yes/No column added could be adapted to any of
Demorep's methods (so far).

But, if we're going to allow non-effectively-summable elimination methods,
my favorite public proposal would be Instant Approval Runoff via Five Slot
Ballots:

At each stage each ballot is construed as approving all candidates above
that ballot's average grade of the remaining candidates.

The approvals are tallied for each candidate over all of the ballots. 

Each candidate with less than average approval is eliminated before the
next stage. 

[end of description of Instant Approval Runoff]


Forest