[EM] Does IRV elect "majority winners?"

[Abd ul-Rahman Lomax]

At 06:52 PM 1/5/2009, Juho Laatu wrote:

>In IRV one should of course also
>clearly state what kind of a majority
>(or non-majority) decision this is.
>It was maybe "a majority decision
>between the last remaining candidates".
>This is at least one nice way to hide
>the fact that some of the already
>eliminated candidates could have been
>a stronger "majority favourite" in
>some way (e.g. a Condorcet winner).


Actual election rules do specify adequately, ordinarily. (There are 
some rules I've seen which don't, which then leads to 
misunderstandings and possibly to lawsuits.) In Australia, with 
mandatory ranking, the quota is an absolute majority of valid 
ballots. And ballots without full ranking are spoiled, so they don't count.

(The majority found, then, is a coerced one. Bad, bad idea, without 
value but an illusory one, a claim of a majority result.)

For Optional Preferential Voting, the quota is a vote from a majority 
of ballots containing votes for continuing candidates, or some such 
language. They specify what "majority" means.

However, in the U.S., the *campaigns* for IRV don't explain the 
difference. This is so consistent that I conclude that it is planned. 
We know that FairVote has held conferences for activists, teaching 
them how to argue for IRV, I know that because I know people who have 
attended them, and who later realized that they'd been hoodwinked. 
I'm sure that these activists must know about the problem, and I'm 
sure that they are trained to deflect the issue.

They do *not* want to see exposure of the problem, for they are 
mostly arguing for IRV in contexts where there is a method which 
*does* require a true majority to complete with the first ballot. 
Then, in that context, comes the IRV activist -- or someone duped by 
them -- who states, "The winner will still be required to gain a 
majority of the votes."

They know how voters will read this. They also know that most voters 
won't think sufficiently about this, on their own, to realize the 
implications. The voters will think that they are getting their 
majority cake, and freedom from runoffs at the same time!

But they aren't getting that majority cake, they are getting a 
mathematical trick. The statement above, about the "requirement" is 
misleading in two ways. "Majority of the votes" is incorrect, in 
context, because, in fact, when it happens, a "majority of the votes" 
cast in the election were for other than the IRV winner. Secondly, a 
"requirement" implies some standard which must be met by a candidate, 
when, in fact, the method guarantees the result, it isn't something earned.

It's certainly not realistic on a large scale, but imagine an 
election where every voter (out of 10 voters) bullet votes, for a 
different candidate, out of nine, Except that two voters vote for the 
same candidate. That candidate wins in any Plurality method, but 
nobody would claim, sensibly, that it was by a majority.

Comes IRV, with those same votes, and it finds, in the second round, 
a "majority." Same votes. Is that a "majority?" Of course it isn't! 
It's a plurality.

IRV, if no runoff or further process is held, is a Plurality method. 
And, quite unexpectedly for me, it almost always, in nonpartisan 
elections and very often with partisan ones, produces the same result 
as Plurality.

Majority failure is common with it. Burlington is an example. 
Majority failure doesn't mean that the wrong candidate has been 
chosen; it is simply a symptom that *can*, under some circumstances 
mean that. In Burlington, Kiss seems to have been, from various kinds 
of vote analysis, clearly the best winner and highly likely to win a 
runoff against any of the other candidates.

This is normal for IRV and it is, in fact, normal for Plurality. The 
problem is the exceptions. How often do they occur. My estimate is 
one out of ten elections, which is based on "comeback elections" with 
Top Two Runoff. More work needs to be done, but that is a rough 
expression of the frequency of comeback elections. It is also about 
one out of three runoffs. (roughly two out of three TTR elections, 
then, find a majority in the first round, and this is clearly a true 
majority. But so is the same majority under Plurality.)