[EM] "we only get one shot" (Re: RCV Challenge)

[Kristofer Munsterhjelm]

On 05.01.2022 04:35, Forest Simmons wrote:
> Kristofer and All,
> 
> What I get from this is to use a Coombs/STV hybrid. Perhaps mostly
> Coombs for Elimination and STV quotas for election ... something like that.

It's a perhaps surprising but still nice observation that it doesn't
matter how you eliminate candidates in STV as far as the DPC is
concerned: as long as you do it one at a time, the criterion still holds
because the election condition only looks at first preferences.

I like the idea of using Ranked Pairs to determine the elimination
order. Because RP passes LIIA, this gets rid of a major source of chaos
in STV/IRV because the eliminations don't change the social order. Only
the election and surplus distribution steps do that.

It may be a mar on the perfect proportionality of STV that Ranked Pairs
has a centrist bias (since it's a single-winner election method), and
thus RP-STV would have some element of centrist/consensus bias. On the
other hand, you could cast this as a feature rather than a bug: there's
certainly no lack of polarization in the current political world, so
perhaps a bit of centrist/consensus bias is not so bad.


As for Coombs STV, that's certainly one possible interpretation of
binomial STV. But if binomial STV *is* Coombs STV, then it definitely
isn't monotone, because in the single-winner case it's just regular old
Coombs.

-km