[EM] "majority" == "majority approval" ?

[Abd ul-Rahman Lomax]

At 02:38 PM 1/3/2009, Brian Olson wrote:
>I would like to propose a colloquial, common english solution to the
>debate about what majority is. A "majority candidate" is one with
>"majority support" or "majority approval". I think this would fit with
>the spirit of what most people would think of when they hear about
>'electing a majority candidate'.

It begs the question. "Majority" of what? All the people? Would Mr. 
Olson them be proposing "absolute majority" as the standard meaning? 
Yet this is at total variance with common usage, we say that someone 
won an election by a majority when a majority of voters, generally 
defined as those who cast a formal ballot in the election, have voted 
for the winner in some way. It seems that most accept that this vote 
may be an alternative vote, it need not be a first-preference vote.

Australian law does consider that the majority created by mandatory 
full ranking is an "absolute majority," but this does mean only a 
majority of formal ballots. They do not call the quota in Optional 
Preferential Voting a "majority," unless they qualify the term to say 
"majority of" something.

You can find references on the web to OPV election results that note 
that the winner didn't get a majority of the votes. This would be 
impossible if majority meant what FairVote and Bouricius have claimed.

>We know how to do approval voting with a simple optimal zero-info
>strategy. So, you could measure based upon that. Alternately you could
>have the no-strategy 'do i approve of this candidate' voter.

It is not possible to determine "approval" from pure absolute 
zero-knowledge voter preferences, because our very notion of 
"approval" is contextual. An outcome that we will prefer, even leap 
for, under some circumstances, is one we will disapprove of under 
others. This indeterminacy is what led Arrow to disregard systems 
like Approval, not even considering them to be "voting systems."

Yet, in fact, that is what they are: "voting systems." Systems 
whereby a collective *choice* is determined by amalagamating 
*individual choices.* The error is in assuming that a vote is an 
expression of pure preference, and that voter choices *should* be 
made on the basis of zero-knowledge preferences. It was this 
assumption that leads, inexorably, to Arrow's Theorem, the paradox. 
Yet people make decisions all the time, including collective 
decisions, and preference strength and probability estimates are part 
of these. We save time by not even asking for our absolute first 
preference, where we think it effectively impossible that it would be 
collectively accepted. Which would you prefer, $100, $10,000, or $1 million?

Wouldn't it depend on what you thought you could get? You might vote 
No on all of these, if you think you could get more than a million, 
or you didn't like some other aspect of the "deal." If you'd be glad 
to get anything, you might vote Yes on all of them. (Suppose if there 
is no majority approval, there is no "deal"? You don't get anything. 
*Maybe* there would be further process, which might offer $1, $10, 
$50. Or which might offer $2 million, $4 million, $10 million.)

There is no single "sincere vote." Only if we allow votes to 
incorporate probabilities, as if each vote was a vote in a lottery, 
can we then even talk about sincere votes, without making hosts of 
unwarranted assumptions, truly, and then the votes are actually 
optimally strategized if sincere. They are sincere or *stupid.*

You sold your birthright for a mess of pottage?

Well, I was hungry, and it looked good.

>There's already an infuriating amount of false mandate thrown to a
>candidate who gets voted for by 1/6th of the US population (and how
>many actually _like_ the candidate and how many are just going for
>what 1/20th of the population voted for in the primary). I don't want
>IRV to add to that by mislabeling candidates as having majority
>approval when they do not.

I'm with you on that. We don't gain anything by pretending a majority 
vote. It's quite possible to argue that IRV is a reasonable single 
winner voting system, though I'd disagree strongly, given the 
options, but to pretend that it determines a majority when it clearly 
does not, when parliamentary authority agrees (specifically: Robert's 
Rules of Order) that it does not, when it's being compared with a 
system that, in the first ballot, *fails* if no majority is found, 
when with exactly the same votes, apply the IRV process to them, it 
would "find" a majority, is simply deception.

It is as if "majority" is simply some technical requirement that we 
"should have," regardless of the real meaning, and that we do a 
"good" thing by fabricating it by how we define words. Appearance 
over substance. It's actually what's wrong with politics, in general.