[EM] Maine RCV

[Gustav Thorzen]

On Thu, 30 Apr 2026 19:59:15 +1000
Etjon Basha via Election-Methods <election-methods at lists.electorama.com> wrote:

> Good evening gentlemen,
> 
> This whole situation led me to think of how might someone dead-set on IRV
> get around this, which would obviously be through runoff voting.
> 
> Which in turn made me think of runoff itself, and how it's an inefficient
> way to parse a crowded field of candidates in round one. Once you get to
> round two, there's no games left to play, and a contest between two parties
> only is at least immune to on-the-day voter strategy.
> 
> Of course, if we let more candidates than two survive the first round, we
> would suffer from potentially fewer failures on round 1, but potentially
> more on round 2.
> 
> So, which is the optimal way to minimise the total strategizing across two
> rounds?
> 
> Three candidates? However many top-voted collectively clear half the votes
> cast? Any other ideas? Or is "top two" likely the best compromise anyway?

3 rounds can get greatly improved strategic properties in common real life scenarios.
Round 1 keep top 3,
Round 2 keep top 2,
Round 3 elect top 1.

This solves the today common problem of two dominant parties/candidates
with one of them having a serious competitor,
because both dominants and the competitor would make it to round 2,
where the dominant party with a competitor would in practice
only compete with their competitor for a spot in round 3,
which gets all the properties of only two candidates.

The votesplitting caused by the competitor would never
(in relevant scenarios) be an advantage to the second
dominant without a competitor of their own,

Someone would have to figure out what (any?) changes to the numerous
strategic equilibria of round 1 and their implications,
especially what happens if both dominant parties/candidates get their own
serious competitor for round 1.

Can't really do better with only two rounds then keeping top 2,
but if we decide on some fixed number of rounds R (2 or more),
then keeping top R in round 1 and reducing it by one every round
is more or less optimal as a system.

We could start with some number of keep top remaining half,
which would still bring a meaning full strategic improvement
for every such round before keep top R then countdown
eliminating one party/candidate per round.
But I think R round are acceptable then there is
more benefit to go directly to the keep top R
and eliminating one per round countdown.
(The most I have heard acceptable are 3 rounds.)

Gustav