[EM] primary elections

[Daniel Bishop]

Alex Small wrote:

> Bart-
>  
> Good points.  Advancing the top N would be a bad idea if the top N 
> were determined by Approval totals.  OTOH, if the top N were 
> determined by first-choice vote totals then parties (especially small 
> parties or compromise parties) that split their vote would be excluded 
> from the final round.

That would definitely have the advantage of making it difficult for a 
single party to have all N of the final candidates, but it introduces 
the problem of vote-splitting.

Is there an election method for which the first- and second-place 
winners tend to have different views, but without vote splitting?  
Green-Armytage's "proportional ordering" system might work, but because 
of the large number of candidates, it might be infeasible.

Perhaps candidates could be ranked by a combination of Plurality and 
Approval votes.

> <>And if we went by the original proposal in this thread and advanced 
> anybody with over 1% approval then, leaving aside strategic objections 
> for a moment, the field would probably be pretty crowded so the 
> primary wouldn't really serve its purpose of narrowing the field. 

This could be remedied by setting a higher threshold.  But how high?

> So separate partisan primaries seem the way to go for partisan races.

The problem with partisan primaries is that it's the partisans who vote 
in them. 

> But for local races there is a real dilemma when local races are 
> non-partisan.  One could argue for simply using a good single-winner 
> method that can handle more than 2 candidates and be done with it, and 
> not have a primary.  However, many people see value in winnowing down 
> the field to allow for more focused debate before making a final choice.
>  
> If this is to be done, how do we justify to voters the notion of using 
> anything other than top 2 runoff?  It would be hard to justify 
> bringing 3 or more cnadidates into the final round.  Sure, we could 
> all think of good reasons that satisfy us, but will they satisfy the 
> average voter?

If that voter went to Texas A&M University, you could bring up the 
example of the 2003 Student Body President election.

Lucas Cheatham '03 was popular among the students (but hated by school 
administrators), and had received the endorsement of our newspaper.  
(Btw, I voted for him.)  His political bloc won a majority of the 
Student Senate, and a poll at a popular online forum projected that 
Cheatham would win the presidency.  But in the first round of the 
six-candidate election, the results were:

Burke: 2665 (22.6%)
Josefy: 2573 (21.8%)
Cheatham: 2472 (21.0%)
Pfluger: 1949 (16.5%)
Carlton: 1518 (12.9%)
Brown: 600 (5.1%)

The election was almost a 3-way tie, but only Burke and Josefy could be 
in the runoffs, despite the fact that the majority had voted for other 
candidates.  Furthermore, it was perceived that if Brown had not run, 
most of his 600 votes would have gone to Cheatham, because of his 
similar views.

This created a controversy about the runoff system, but unfortunately it 
hasn't been changed to this day.

(The bug in the online voting software that conveniently mostly affected 
older students might have had an effect as well, but that's off-topic.)