[EM] Best IRV Tweak?

[Kristofer Munsterhjelm]

On 7/20/21 5:41 AM, Susan Simmons wrote:
> Run IRV as usual eliminating one-by-one the candidates with least first 
> rank support among the remaining candidates UNTIL only the winner 
> survives OR the one on the chopping block is found to dominate every 
> other remaining candidate by the covering relation.
> 
> If I were an IRV supporter with a firm belief that low first rank 
> support was the best single indicator of inferior support, it might be 
> hard to convince me that a runoff loser should be promoted to election 
> winner solely on the basis that she was unbeaten pairwise among the 
> remaining candidates.

I've always found FairVote's enthusiasm for "Core Support" to be 
somewhat disingenuous. If the core support criterion is an absolute 
requirement, then nobody with zero first preferences should be elected. 
However, IRV can elect someone with a single first preference if the 
transfers are right.

So zero first preferences is bad but one is good? If core support is the 
end-all be-all, then surely Plurality must be a better contender.

It also clearly fails IIA: augment an election by having the voters vote 
for themselves in first place and now no candidate has more than a 
single first preference.

If core support is to be taken seriously, it must mean something 
different from "first preferences". But it shouldn't be a tailor-made 
criterion that says "whatever IRV does", either.

-km