[EM] Copeland's criteria

[Abd ul-Rahman Lomax]

At 01:52 AM 9/12/2005, Rob Lanphier wrote:
>As Abd alluded to in at least one email, it's possible to have a revised
>version of Copeland that works differently.  For example, it could be
>possible to not credit a candidate with a victory if they don't receive
>majority support (called "Copeland Majority" for purposes of this mail).
>We'll say one point for a win, no points for a loss, tie, or
>"draw" (where neither candidate gets a majority).

There are election methods which produce raw data. How that data is 
interpreted may or may not be part of the method. For example, a 
method may produce a tie. How the tie is resolved may be independent 
of the method. Whether or not there is an actual runoff, or if a coin 
is tossed, the method is still the same method.

Before an election method is chosen, there is a fundamental question 
which must be answered, which I have mentioned many times. I'll 
approach the question by giving my own opinion about the answer, 
recognizing that there are certainly arguments for a different 
opinion (and if the status quo is an opinion, it certainly has more 
adherents today than my opinion).

At what point do we consider a government democratic? I would say 
that the minimum standard is that a majority of citizens *consent* to 
it. Now, a democratic government might make undemocratic decisions, 
because of rigidities in its structure (perhaps inevitable, though I 
think the problem could be ameliorated substantially); however, then 
we might ask at what point do we consider a *decision* democratic? 
Again, the answer is, for me, the same: a decision is democratic when 
a majority of citizens consent to it.

(I am not addressing the issue of the definition of citizenship. The 
Greeks excluded slaves and women, if I'm correct; still, we can call 
Greek democracy "democratic" *with respect to its citizens*, though 
not with the population as a whole.)

So, for me, the implementation of an election is not democratic if a 
majority of *voters* have not consented to the election result. Now, 
there may be a mechanism which still produces a winner even when the 
winner is short of a majority consent. It could still be called 
"democratic" if the majority *consents to the mechanism.*

Such a consent may vary from time to time and election to election, 
so simply passing a law providing for an election resolution, by 
whatever method short of having measured a majority consent is not 
sufficient. My grandfather's generation might have thought, for 
example, that it was fine to elect a plurality candidate, and may 
have consented to that; but this does not mean that *I* do. Conditions change.

There is, of course, an alternate view, which is that it is more 
important that the office be filled without the delay and expense and 
inconvenience of a runoff election. Frankly, I find this a dangerous 
view, for the potential damage from a poor election result can be 
enormous, compared to which the expense and inconvenience are trivial.

So I generally favor election methods which are designed such as to 
encourage some kind of consensus. Approval and Range (with Approval 
cutoffs, not always part of Range methods) are two such methods. For 
a method which would truly maximize consent, I'd suggest Fractional 
Approval Asset Voting (or any form of Asset Voting, though FAAV 
shares the ballot simplicity of Approval and the more complex aspect, 
the vote negotiations, only kick in if there is a majority failure).

(In Asset Voting, where there is a majority failure, the votes are 
refigured as fractional approval, where, unlike standard Approval, if 
a person has voted for two, each gets one-half vote. Then those who 
have received the votes may reassign them to create a majority 
winner. The fractional aspect is necessary, in this process, to 
maintain one-person, one-vote.)

However, standard Approval Voting, without the Asset feature, could 
certainly  approach this, and I see no cogent argument that standard 
Approval would be *worse* than what we already have in plurality. 
Indeed, it does not guarantee a majority winner; however, it would 
*probably* produce such more clearly. It eliminates, as a minimum, 
the spoiler effect, where one gets a minority winner because the true 
majority candidate's vote has been split by a clone or by another 
candidate attracting part of that vote as a preference, but not as a 
preference over the ultimate winner. It is this kind of result, all 
too common, which is the worst feature of plurality; it should 
actually be considered intolerable, for it is truly minority rule.

Other methods can achieve similar results, though sometimes with 
problems. For example, strategic voting in IRV can result in burial 
of a consensus candidate, once again producing a minority winner.

As to the issue of majority failure, what if, with each election, 
there was a box to check, or, better, two boxes? I'll give the two:

() I consent to the result of this election even if there is no 
majority winner.

() I prefer the inconvenience and expense of a runoff to the 
acceptance of a minority winner.

(Of course, the actual wording would be a bit more terse.... but I 
think you get the idea.)

It would be more efficient if there were one box to check. There are 
two ways to do it, obviously, and it is not obvious to me which is 
best, though it is obvious that they could produce different results....