[Election-Methods] RELEASE: Instant Runoff Voting

[Aaron Armitage]

--- On Tue, 7/29/08, James Gilmour <jgilmour at globalnet.co.uk> wrote:

> From: James Gilmour <jgilmour at globalnet.co.uk>
> Subject: Re: [Election-Methods] RELEASE: Instant Runoff Voting
> To: election-methods at lists.electorama.com
> Date: Tuesday, July 29, 2008, 3:45 AM
> Aaron Armitage > Sent: Tuesday, July 29, 2008 1:11 AM
> > IRV and all 
> > other ranked choice systems ask for the same input
> from 
> > voters
> 
> This is where you make your first mistake.  IRV and other
> ranked choice voting system do not all ask for the same
> input from the
> voters.  IRV asks voters to mark preferences in the
> knowledge that those preferences will be used as
> contingency choices, so that a
> later preference can in no way affect the chance of
> election of an earlier preference.  Some other ranked
> choice voting systems, in
> a variety of different ways, make simultaneous use of all
> the preference information recorded on the ballot paper,
> such that the
> later preferences can affect the chances of election of the
> earlier preferences.  The voters know in advance which
> counting rules
> will be used in any particular election and modify their
> marking of preferences accordingly.  So the inputs are not
> the same.
> 
> 
> > and produce the same kind of output, namely a single 
> > winner. 
> 
> Here is your second mistake.  Both kinds of voting system
> do result in the election of a single winner, but the
> outcome (output) can
> be quite different in terms of what that winner represents.
>  In the case of IRV that winner is the contingency choice,
> with all the
> implications of that.  In Condorcet, the winner may be
> decided in a very different way from IRV and represent
> something very
> different in relation to the voters.  In a Borda count, the
> winner may represent some sort of compromise even when there
> is one
> candidate who has an absolute majority of the first
> preference votes.  So all these outputs are quite
> different.
> 

You claim, in short, that using the same inputs differently makes them different inputs, and that producing the same kind of outcome differently makes it a different outcome. But you began by contrasting IRV to "social
choice" methods like Condorcet and Borda, even though Condorcet and Borda
treat the inputs very differently and informed voters will presumably take
that into account when they make their rankings. Are Borda and Condorcet
also incommensurable, or just IRV?

In general, if we followed your logic we couldn't compare different ways
of doing the same thing, since if they handled their inputs the same and
their results represented the same thing -- that is, the same internal
process -- they wouldn't be different methods. A hybrid car uses its gas
differently and drives the wheels differently; can we directly compare
a hybrid's fuel efficiency to a straight internal combustion car's fuel
efficiency, or do we decide that because a non-hybrid isn't trying to be a
hybrid it can get away with being less efficient?

 


> I would reject Buckilin because it does not comply with
> "one person, one vote".
> 

Under any definition of "one person, on vote" that Bucklin fails, IRV also
fails. But that wouldn't be a proper definition anyway.

> I am VERY sympathetic to Condorcet and think the basic
> concept is "sellable" to the electors (presented
> as a "head-to-head
> tournament"), despite the inevitable opposition of
> most politicians, big business and the media moguls.  I
> foresee bigger problems
> in selling any of various cycle-breaking and tie-breaking
> solutions that have been proposed.  But the real problem
> with Condorcet is
> the weak Condorcet winner.  It is my judgement (based on
> long experience as a practical reformer, but only in the
> UK) that such an
> outcome would not be politically acceptable to the
> electorate in an election to public office.  Such a winner
> would, of course, be
> the real Condorcet winner, but that would not, of itself,
> make the result politically acceptable to real voters.
> 

Every scenario for this that I've seen assumes two strong candidates and
a third candidate who is only a placeholder; in other words, it implicitly
assumes that only serious candidates will be the Republican and the
Democrat, or the Conservative and the Labourite or Liberal Democrat, or
whatever the local equivalents are. But what happened to all the serious
admonitions to consider how the system will change people's behavior? In
a Condorcet election there's no reason there would only be two major
candidates, and a public thinking in terms of Condorcet would have a 
completely different understanding of what strong and weak candidates are.
In America we could dispense with primary elections, so that all the major
party candidates run directly in the general election; unless I'm mistaken
the nomination process in the UK isn't as open as ours is so it might take
a few more adjustments, but it's the same general idea.

> IRV has, of course, a corresponding "political"
> weakness, in that it can reject the candidate who might be
> everyone's second choice
> (the Condorcet winner).  But experience shows that the
> electors are prepared to accept that outcome.
> 

Do you mean experience talking to them in focus groups? I'm not sure that
will hold up in the event.

Take an example. Louisiana uses the same election system that France does,
and it malfunctioned the same way in both places; a crypto-fascist got
enough votes to make it to the runoff, produced a fair amount of panic, and
duly lost to an opponent whose only real selling point was being the only
alternative. In the Louisiana case, there were three major candidates:
Buddy Roemer, the then-incumbent who had switched parties while in office,
Edwin Edwards, a former governor who was known to be incompetant and a
criminal, and David Duke, a former Grand Wizard of the KKK who was claiming
not to be as racist as he'd been before. Polls showed Roemer beating
any of the other candidates heads up, so we know he was the CW. But he came in third, producing a runoff where the most famous slogan was "Vote
for the crook. It's important." The candidate who came in fourth was a 
conservative whose supporters would probably have split between Roemer
and Duke (Roemer was a conservative himself), and the others had too few
votes to change the relative positions. So under IRV the final contest is
probably Edwards vs. Duke. The best case scenario here is the same outcome,
but in the full knowledge that Roemer, the non-crook, had beaten Edwards.
Would the voters have accepted that? Should they have? The unknown here
is the dark question: without a chance to take a good hard look at the fact
that they were about to make either Duke or Edwards governor, how would the
Roemer supporters have broken? And what do you suppose would have happened
if it was Duke?