[EM] Poll, preliminary ballots

[Richard, the VoteFair guy]

On 4/18/2024 12:32 PM, robert bristow-johnson wrote:
> 
>  > The payoff is that RCIPE would not have failed in Burlington and Alaska!
>>  That's huge.
> 
> Any Condorcet method would have that huge payoff.  What "huge" benefit 
> does RCIPE have that a simpler Condorcet method doesn't have?
> 
> Like Condorcet-Plurality or Condorcet-TTR?
> 
> Or MinMax?

Condorcet methods require explaining pairwise comparisons between every 
candidate and every other candidate.  That's a significant disadvantage 
for voter education.  And voter understanding.  And voter trust.

In contrast, I have never heard anyone claim that a pairwise losing 
candidate doesn't deserve to be eliminated.

Regarding the simplicity advantage, the RCIPE (Ranked Choice Including 
Pairwise Elimination) method only needs two simple sentences added to a 
well-written IRV law:

"Pairwise losing candidates are eliminated when they occur.  A pairwise 
losing candidate is a candidate who would lose every one-on-one contest 
against every other remaining candidate."

Robert, I know you strongly dislike the misrepresentations from the 
FairVote organization.  I too dislike those misrepresentations.  (I've 
been hearing them for the three decades since I got the VoteFair domain 
name.)

Yet I do not share your opinion that those misrepresentations justify 
your low opinion of IRV as being worse than plurality (if I recall your 
poll ballot correctly).

Specifically, unlike you/Robert, I see value in these aspects of IRV:

* Uses ranked choice ballots (like Condorcet methods, unlike score, 
unlike Star, unlike Approval)

* Eliminating one candidate at a time can work well if the eliminated 
candidate is really the least popular candidate during that counting round.

* When a counting round does not include a pairwise losing candidate, 
eliminating the IRV-based fewest-transferred-votes candidate is a 
reasonable cycle-resolution method.  In particular it's cloneproof, but 
without the complexity of the Schulze method.

So in my opinion, adding the elimination of pairwise losing candidates 
elevates the resulting method, RCIPE, to the top of my poll ballot.

It offers both math advantages and 
easy-for-voters-to-understand-and-trust advantages.

In contrast, most voters don't trust starting with the idea of creating 
a full pairwise counting matrix and then looking for a candidate who 
wins every one of their pairwise contests.  That's why Condorcet methods 
have been so easy for FairVote and other organizations to criticize and 
dismiss.  And why it's so difficult for us math-savvy experts to explain 
to non-math folks.

As I indicated in my poll ballot, RCIPE is "simple, almost Condorcet, 
almost cloneproof."

Richard Fobes
The VoteFair guy


On 4/18/2024 12:32 PM, robert bristow-johnson wrote:
 > ... [see above]
> 
> I'm not sure it's agreed or wisely assumed that this poll is about the 
> most preferable method for public, governmental election (where the 
> stakes are high and voter s have a partisan interest to maximize their 
> own political interests), but I'm sorta assuming that.
> 
> Then when choosing a method to promote, the actual correctness and 
> fairness of the method is paramount.
> 
> Then the *perception* of fairness of the method by the public and by 
> policy makers is important.  To satisfy that concern, the method needs 
> to be clear and understandable to the public and to policy makers.  This 
> is necessary to get legislation passed.
> 
> Hare RCV may have a simpler procedure, the Single Transferable Vote, but 
> it doesn't have a simple *principle* of ethic that it's actually 
> faithful to.  Its ethic is its procedure, the STV, essentially saying 
> "This is the right thing to do since this is what we're doing and this 
> is always we had done it.". But that's no justification.  It's circular.
> 
> Condorcet is based primarily on the principle of "One-Person-One-Vote", 
> that our votes must count equally.  No enfranchised voter deserves to 
> have their vote count less than some other enfranchised voter.
> 
> But that means Majority Rule must apply (I don't mean the "majority 
> criterion").  This means that, at the end of the day, if more voters 
> prefer A to B, then B is not elected.  If B is elected, then the fewer 
> voters who preferred B had cast votes that were more effective than 
> those larger number of voters preferring A.
> 
> IRV is not committed to that simple ethic, whereas Condorcet is 
> committed to it.  This is essentially the Condorcet criterion.
> 
> So then the secondary concern would be public perception as well as that 
> of policy makers.  These secondary concerns are the only concerns that 
> differentiate Condorcet methods.  It's about how to justify to people 
> why the CW is the right choice to elect.  And what rationale there 
> exists for the rare event electing someone when the CW does not exist.