[EM] Fine-grained option

[LAYTON Craig]

Jeandré,

you wrote (in part):

>A "binding none of the above" (BNOTA) option, and space to comment is
required for all
>referenda. If BNOTA wins, the most popular comments will be incorporated in
the revote.

The referendum is an interesting and important form of public vote/election
that doesn't often receive much attention.

>If a single winning option is required, the option with 50%+ of the votes
wins; failing
>a 50%+ winner, a revote must be taken including comments; if again no
option wins 50%+
>there is no winner/action-taken.

I guess the main issue is that sometimes you have to have a winner.  In the
case of your referenda model, there are three outcomes from the entire
process; the proposed change is enacted (yes voters win) some other change
is enacted (BNOTA wins) or there is no change (no wins).  As a result, NO
can win with much less than 50% of the vote.  I guess the best solution is
multiple option referenda, although I don't think most constitutions allow
this.

Single option referenda are subject to a basic spoiler problem.  Here's the
requisite example from Australia (again); the rebublic referendum was
largely defeated because of debate among republicans (as in, people who
don't support the constitutional monarchy, not people who voted for Bush
jnr) over whether or not to have the President directly elected.  The model
proposed was for an indirectly elected President, so the direct election
voters voted no, even though they'd quite likely prefer an indirectly
elected President to the current model.  Opinion polls indicate that a
significant majority of Australians want a republic.

>//Should there be different systems for 1 winner vs. multiple winner
referenda?

Multiple winner elections should probably use a different system, although
I'm not sure what the benifit is of using a multiple winner vote for
anything other than electing a group of representatives.  Unless, perhaps,
there is a complex interrelationship between the options, and you write
counting rules specifically for that election - this might be the go if some
options are mutually exclusive.  Although, the simple (if somewhat less
accurate) alternative would be to hold a series of votes, or build in a
simple elimination proceedure in a single vote count.

>If they vote for several options, with an X, those options each get an
equal percentage
>of the vote: 1 ÷ number of votes (NOV), e.g.: 4 selections means all 4
options get 0.25
>of a vote.

This system (sometimes called cumulative) is one of the best multi-winner
methods where there are no parties or factions involved.  I think STV
(single tranferrable vote) and Proportional Approval (invented by Forest
Simmons of this list) are better, but they're also a darn sight more
complicated to count.

Craig