[EM] IRV - Approval, Condorcet - Approval hybrids

[Kevin Venzke]

Chris,

 --- Chris Benham <chrisbenham at bigpond.com> a écrit : 
>> "David's method gives me a similar, simpler idea that would seem to be an
>> improvement over IRV.  The method would be:
>> 
>> 1. The voters rank the candidates they would be willing to support, and also
>> place an approval cutoff.  (Alternatively, all candidates ranked non-last
>> could be considered "approved.")
>> 2. While there is no (voted) majority favorite, eliminate the Approval loser.
>> 3. Elect the voted majority favorite.
> 
> CB:In step 1, why did  you specify that voters only rank the candidates "they would
>  be willing to support"? 

I suppose this was unnecessary to the description...

>What the hell is the "approval cutoff" for? Pardon my tone,
> but I have no sympathy for the stupid tendency to want to confuse ranking with rating.

I don't understand your objection...  How do you propose this system work without
an approval cutoff?

I'm not sure what you mean by confusing ranking with rating.  Approving all candidates
ranked non-last is just for convenience: the voter doesn't have to worry about the
cutoff, and he will probably truncate most candidates, anyway.  You don't have to
agree with me that this is a good option.

> I am strongly of the opinion that the default placement of the approval cutoff should
> either be between 1 and 2, or nowhere.In a healthy democracy (in which the voters are
> not apathetic and the nomination process is fair and open)it can reasonably be assumed
> that most voters consider their first preferences acceptable.There is no reason to
> assume that they find any other candidates acceptable, 

My general thought is that, if a method is such that ranking too many candidates
can be harmful (think Bucklin), perhaps it's not so bad if some status is automatically
given to all the candidates the voter deemed worth of placing in a ranking.  Otherwise
I agree there is no reason to make assumptions.

>and I see no reason why they
> shouldn't be allowed to rank candidates they find unacceptable.

Agreed, unless the method accomplishes something by preventing this.  That is, I
don't see an inherent advantage to preventing voters from ranking unacceptable
candidates.

> 
> You didn't mention how you think equal preferences should be handled. 
> I assume you share my opinion that they should not be split.

I think they should be treated as full votes.  It would be bad if a voter sunk both
favorite and compromise because he up-ranked compromise equal to favorite.  (Such a
method would fail weak FBC, I've little doubt.)

>In that case,what do you think should happen
> when there is more than one "voted majority favourite"? I guess you favour the MCA-like
> solution of just electing the candidate with the "biggest majority". I strongly think that
> if there are 2 majority favourites then they should runoff. 

Here's the difference as I see it: In MCA you have very limited ability to strictly
rank in a meaningful way.  In "Approval-Elimination Runoff," you are not as limited,
so there's not as much incentive to rank multiple candidates in first.  Thus, in my
view, multiple majorities will tend to be flukes, rather than common disconcerting
occurrences.  I'm just thinking out loud, though.

>But what if there are more than
> two? There are many plausible alternatives. 
> 
> Which leads me to another idea on how to count these ballots: a series of elimination runoffs
> between the 2 (remaining) candidates with the fewest approvals.
> This of course ensures that a CW will be elected if there is one, but does that mean that
> compliance with Participation has (therefore?) been sacrificed?

I believe so, because as with all cycle situations, there could be incentive to
try to ensure that your compromise is undefeated, even by your favorite.

But I'm not so sure that Approval-Elimination Runoff (what is described at the
top of this message) meets Participation.  I can sort of imagine one's vote preventing
a candidate from winning at a certain stage (by increasing the amount needed for
a majority), causing a less-liked candidate to win afterwards.

The fear of this possibility, I think, is what would cause people to up-rank
compromises, and possibly create multiple majorities.  But I don't see a good
way of fixing this without inviting cycle-related problems.

I might also add that I see it as an asset that "AER" does not require the
notion of "pairwise contests" (let alone "cycles") to be understood.


Kevin Venzke
stepjak at yahoo.fr


___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com