[EM] Oops! Squeeze-effect.

[Raph Frank]

On Mon, Jan 21, 2013 at 5:05 PM, Michael Ossipoff
<email9648742 at gmail.com> wrote:
> Elimination would start at the extremes. Transfers would be sent
> inward, until the candidates adjacent to the CW would have collected
> all of those inward-transferred votes, enough to eliminate the CW.

It seems more realistic that the CW is the centrist and one of the 2
large parties wins.

> So it's safe to say that IRV isn't at all good at electing CWs or fair
> compromises.

Right, that is the problem.

IRV supports the 2 party system.  It is not clear if the effect is
weaker than with plurality.

It might be weaker, but it is still sufficient.

See Australia, they have IRV and a 2 party system.

The 2 party system forces 1 dimension to politics.  You can't selected
policy direction and low corruption (or even 2 have 2 dimensions for
policy).

> In that squeeze-effect scenario, of course the voters preferring
> candidates to one side, plus those preferring the CW, must add up to a
> majority. So there are two majorities, and the one that prevails will
> be the one on the side that the CW's voters prefer and transfer their
> votes to, when the CW is eliminated.

45) L > C > R
10) C > L > R
45) R > C > L

The center candidate loses.

The mutual majority is (C, L), but the C candidate is the CW.

> IRV is a special purpose method that should
> only be supported by befeficiaries of a mutural majority.

It benefits the 2 parties, since they will be the last 2 parties to be
eliminated.

Spoiler effects mean that voters can't pick the best candidate.  In
the L/C mutual majority that can just pick L/C as a bloc.

Anyway, "better" systems would not be an advantage for the current
parties, better to have a 50% chance of winning the election, than a
centre party wins most of the time.