Condorcet vs. 5

[Steve Eppley]

DEMOREP1 wrote:
>The six more common methods rate as follows, in my opinion, in
>their relative ability to get a single winner favored by a majority.
>rankings= candidates are ranked 1, 2, 3, etc.
>
>1. Plurality (no rankings)
>2. Top 2 go to runoff (no rankings)
>3. Instant run off (rankings)
>4. Condorcet (rankings)
>5. Limited approval voting (rankings)
>6. Plain Approval voting (no rankings- all candidates are in effect 
>     ranked 1)

Is method 5 common enough to be included as a choice?

I'll assume that 5 and 6 are versions of Approval which disqualify 
all the candidates if more voters disapprove each than approve each,
since this principle is what you've been writing about so much.

If methods 5 and 6 can disqualify everyone, it's not clear
whether they're relatively better at getting a single winner favored
by a majority.  They might get no winner favored by a majority...

Is your rank order best to worst, which is the norm for rankings?
It looks like it's upside down.

Here are my rankings on this "majority approval standard" of the 
methods which I consider sufficiently common:
1. Condorcet+NOTR
2. Smith-Condorcet
3. Condorcet
4. MostApproved
5. Most Approved of the Approved
6. Runoff, Instant Runoff
8. Plurality

The reason I think Smith-Condorcet and plain Condorcet are usually
better than Approval at electing a majority-approved candidate, even
though they don't explicitly let the voters express approval, is
because Approval has strategy dilemmas for the candidates and the 
voters which can distort the results.

The reason I think Runoff is about the same as Instant Runoff is
because in a 3-candidate Runoff election the voters have no dilemma--
they can freely vote for their true favorite.  But they can have 
the LOE dilemma in a 3-candidate IR election.  In other respects IR 
is better than Runoff, and I'm not sure which of these considerations
is most important.

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)