[EM] RE: [Condorcet] A "Condorcet" by any other name stillsmells as sweet?

[Abd ulRahman Lomax]

At 10:53 PM 9/30/2005, Paul Kislanko wrote:
>All of this notwithstanding, no one has ever explained to me how including
>an extraneous "+" is different from


>my_preferred>all_others_could_live_with>>those I think would save me some
>problems by having a heart attack.

It's not different. Why would you think it is?

>I repeat, if I can say A+=B> then you shold be able to infer that from that
>me saying A>B>>(all others).

Yes. My point exactly, this is A+PW as I defined it. Mr. Simmons has 
a different proposal.

I think it has been missed, A+ is a ranked ballot, but ranking is 
truncated to two expressed ranks, and the  Favorite rank is not used 
to determine the winner, it is used for other [very important] purposes.

A+PW is probably the minimum Condorcet method. It has two expressed 
ranks. The winner is the Condorcet winner; Condorcet cycles are 
resolved, I'd suggest, by choosing the Approval winner from the cycle.

>The "plus" added to a ranked ballot is unnecessary, but it could turn an
>approval ballot into a semi-ranked ballot. If that is desirable, just use a
>ranked ballot to begin with.

I'll repeat this: the main objection to Approval is the inability to 
simultaneously vote strategically (i.e., for the favorite among the 
front-runners) and preferentially (for one's favorite over the 
front-runners). Adding an extra slot to Approval allows this. If the 
extra slot is not used for determining the winner -- which would 
still suffer from the strategic problem -- then it still has other 
important purposes.

And if the extra slot is used in determining the winner, A+PW is a 
Condorcet method, the simplest of all possible Condorcet methods. (It 
has three ranks. Fewer than three ranks, you have Approval. Disallow 
overvoting, you have Plurality.)

The more I look at it -- I've been doing some doodling with election 
scenarios -- the more I've become convinced that A+PW is ideal for a 
public proposal at this time.

It adds very little to the ballot. Essentially, any voting system can 
handle and report votes on items with two options. Adding extra ranks 
complicates the ballot and complicates counting. We may argue that 
the benefit is worth the expense, but it is hard enough to get 
Approval implemented, even though probably 90% of the election reform 
goal is accomplished with Approval. What is the main obstacle to 
Approval? The inability to specify a favorite. A+ accomplishes that, 
while leaving counting *very* simple.

A+PW uses the same ballot as A+, but the rank information is actually 
used as part of the election method. I'd say that this would 
accomplish 98% of the reform goals. With more ranks, there is 
improvement, to be sure, but at a cost.

If A+ is implemented, it is easy to go to A+PW, and A+PW might be 
better from the start, if it is politically possible. If A+PW is 
implemented, and a need appears for more ranks, again, not at all 
difficult to add them.

What is crucial is that the Approval cutoff be visible. A+PW does 
this, by essentially dividing candidates into three categories: 
Favorite, Preferred, and [Not Preferred]. A simple ranked system does 
not do it. You can't tell from a fully-ranked ballot where the 
approval cutoff actually is. So if you want to use full ranking, if 
you've got the political capital to pay for the ballot complexity, 
then it would be highly recommended that Approval Cutoff be specified 
on the ballot. Otherwise the meaning of a ranking is not clear. 
Favorite is obvious from a ranked ballot. What is not obvious, unless 
a means is used to make it clear, is what candidates were Approved, 
that is, the voter is expressing an opinion that the outcome of the 
election is reasonably satisfactory if any one of the Approved candidates wins.

(I've seen one object to the use of Preferred. As I'm using it, 
Preferred means "Preferred over all unmarked candidates." It's 
accurate. But the naming is not the core of what I'm suggesting, and 
the common name in election-speak for Preferred is Approved. But I've 
seen much more cogent objections to Approved, on the basis that 
voting "Approved" may not accurately represent the opinion of the 
voter, who is merely indicating a preference, not actually approving 
the candidate, about whom the voter may feel that he has to wash 
after voting "Approved." I can say that I'd prefer Genghis Khan to 
Adolf Hitler (difficult choice), but I'd have real trouble if I had 
to mark "Approved" after the name.