[EM] Reply to Rob regarding RCV

[Kristofer Munsterhjelm]

On 9/24/23 23:55, Michael Ossipoff wrote:
> Sure, I said that I prefer the Pairwise-Count Condorcet-Criterion 
> methods, because they’re the ones that get rid of the 
> Lesser-of-Two-Evils problem (LO2E).
> 
> As I said, RCV’s disadvantage is that its merit & workability depend 
> strongly on the character of the electorate, & on the candidate-lineup.
> 
> …& yes, I’m talking about voters who won’t make the big giveaway compromise.
> 
> It’s a philosophical question: Is giveaway incentive a problem when the 
> electorate aren’t interested in giving it away?
> 
> It isn’t a problem to them, nor, in that case, to me.

My point is that the giveaway incentive can indicate a deeper problem 
that degrades the quality of the outcomes.

Suppose voters who don't want to compromise vote where to relocate their 
capital. They all vote honestly, none of them compromises. They use IRV. 
As a result, the capital is placed in the most populous subregion of the 
most populous region, instead of the place that minimizes the sum of 
distances to the voters. Even though the voters honestly listed their 
candidate sites in order of closest first.

You could say that the compromise incentive is separate from the 
propensity to elect the strongest wing of the strongest wing 
(recursively). But it's this dynamic that produces the compromise 
incentive in IRV, so they are connected.

> As I said, an electorate who have just enacted RCV by referendum didn’t 
> do so because they want to vote some one whom they don’t like over their 
> favorite. They want rankings because they want to rank sincerely. They will.
> 
> No problem.

Why doesn't this argument work for a hypothetical ranked Plurality?

"An electorate that has enacted a variant of Plurality where everybody 
ranks the candidates in order of preference and then the candidate with 
the most first preferences wins ... didn't do so because they want to 
vote someone they don't like over their favorite. They want rankings 
because they want to rank sincerely. They will. No problem."

Just like IRV, Plurality has absolutely no burial incentive. Just like 
IRV, Plurality has a rather large compromise incentive. But that 
shouldn't matter if the voters have decided they're not going to 
compromise, right?

Something seems off. "Ranked Plurality" would be a non-starter; nobody 
would go for it.

-km