[EM] Unranked IRV versus Approval - divergent winners exist!]

[Martin Harper]

[fwd: missent]

Tom Ruen wrote:

> My point is a runoff process works to identify strong candidates and
so this
> should be a minimum reform over plurality.

Agreed. Plurality is sick, and IRV would be a good deal better. If it's
a choice
between IRV or Plurality I'd prefer IRV - but I'd prefer Condorcet or
Approval
more.

> Methods like Approval can work too, and more simply perhaps, but it is
still
> not very helpful in my opinion when there are 3 strong candidates. It
is
> still a game of chicken.

Just like IRV, as you said yourself.

At least in approval we won't crash and elect the second-worst candidate
- we'll
elect a compromise candidate who has broad support. Plus you don't need
to lie
to play the game.

> About random ballots, here's a fun PR system for gamblers.

> Give every voter
> 100 points to distribute among all their choices as they like. Then
add all
> the points and pick a winner randomly in proportional to votes
obtained.
> This throws majority rule out the window as no candidate is guaranteed

> victory without 100% of the vote.

PR system? It looks like it's a single winner system to me - where's the

proportionality? It is also identical to Random Ballot in results -
though
conceptually somewhat different, and harder to implement. The best
strategy is
to give all your points to your most preferred candidate.