[EM] IRV / RCv advances

[Kristofer Munsterhjelm]

On 2018-07-14 02:04, Christopher Colosi wrote:
> In promoting RCV, you can't just credit it in cases where the outcome 
> would have been the same given plurality / first past the post.
> 
> I can't speak for the others in this list, but London Breed was the 
> front runner and would have won in a traditional FPTP/plurality 
> election.  She nearly lost due to RCV and effectively spoke out against 
> it when saying it was the method we are stuck with.  You can make the 
> claim that it was "thanks to RCV" due to the fact that SF previously 
> would have had a separate run off election between her and the next top 
> candidate (would have been Leno or Kim), which she could have lost, but 
> none of Breed's supporters are saying "thanks to RCV".  So it is hard to 
> argue that this is good press to anyone beyond those who were already 
> fans of RCV.  And that causes me to question all the others in the list.
> 
> Living in SF, I can say that many people who filled out ballots 
> correctly really didn't understand how it worked.  Education on this 
> stuff is tough.  The press didn't feel positive.

That's a problem with many ranked vote methods, and moving to 
multiwinner will make it worse still. I suppose voters in places like 
the NZ counties that use Meek just trust the method without entirely 
knowing how it works, but such trust would have to depend on results. In 
that respect, STV has it easier than IRV because it's more clear that it 
has an effect: a proportional council looks very different from a 
two-party-controlled one.

> As a side note on this specific election with regards to RCV, it made me 
> question if we should still have a separate runoff when less than some 
> amount (maybe 2%) separates the candidates after all rounds.  When after 
> RCV it was still so tight of an election, the community may have been 
> better served (even if the outcome was the same) to get a week or two of 
> campaigning and a chance to choose solely between these two candidates.

Warren makes a related argument that although you'd expect IRV to 
dominate top-two because IRV has many rounds and top-two has just two, 
in practice, the IRV winner tends to coincide with the Plurality winner 
more often than the top-two winner coincides with the Plurality winner: 
http://rangevoting.org/TTRvsIRVrevdata.html

Being a Condorcet fellow, I'd suggest a runoff between the IRV winner 
and whoever most voters prefer to the IRV winner, if that candidate 
either beats the IRV winner pairwise or is within, say 2% of doing so. 
But I don't suppose the election authorities would be all that 
interested in doing so if IRV has been marketed as a cost-saving measure 
("get the results of a runoff without having to actually hold a runoff!").

(I would prefer the runoff to be between the CW if there is one and he's 
not the IRV winner, otherwise as above, but that might just be too complex.)