[EM] IRV vs Plurality

[Abd ul-Rahman Lomax]

At 10:51 PM 1/14/2010, Jonathan Lundell wrote:
>On Jan 14, 2010, at 7:17 PM, Abd ul-Rahman Lomax wrote:
>
> > Again, as I mentioned, the Condorcet Criterion looks good, it's 
> "intuitively satisfying." Unfortunately, it depends on pure rank 
> order, neglecting preference strength.
>
>Just for the record: for many of us that's an advantage.

Sure. Those who don't trust voters and who imagine that Range voting 
is susceptible to damage from Strategic Voting (which is a misnomer 
because the term originally referred to votes which reversed 
preference order, and Range never encourages that, period), are going 
to think that allowing voters to present a maximally accurate 
preference profile is "harmful."

With no evidence that it *actually* is. All evidence I've seen that 
claims to show it has been based on preposterous assumptions. Such as 
the assumption that voters who voted 100% A and 99% B are somehow 
going to be screwed by a minority of voters who voted 0% and 100% B.

Pure ranked methods can't distinguish at all between a trivial or 
barely perceptible preference and a deep and important one. So the 
fix these people propose is: don't allow voters to express their true feelings!

However, rank order methods which allow equal ranking do provide a 
partial fix, but then the same people will scream that voters are 
again being STRATEGIC if, on the one hand, they use equal preference 
if they "really" have a preference, or, on the other, they don't 
equal rank if they "really" approve both candidates equally but don't 
equal rank them. Can't win with these people.

It was noticed long ago that there was no single "sincere" Approval 
vote, but rather a class of such votes, being all voting patterns 
that do not violate rank-order by reversing sincere rank. This is how 
the major disagreement arose between whether or not Approval was 
vulnerable to strategic voting, with Brams originally claiming that 
it was "strategy-free" and then others arguing that it was highly 
"vulnerable," which is assumed to be a bad thing. Approval "strategy" 
involves the voters estimating the need for compromise, and according 
to that estimation, they lower their approval cutoff. In other words, 
it's a compromise, and all compromises are "strategic" in the sense 
that if the voter knew nothing about the other voters and had any 
significant preference at all, the voter would simply vote the 
preference. But they are not strategic in that the vote, as writ, 
indicates a sincere division of the candidates into two groups: 
approved and not-approved. Or not approved-yet.

In approval voting as a repeated ballot method (an excellent 
application), voters in the first round would bullet vote, or only 
add approvals that were so close to the first preference that the 
voter prefers to get it over with than start out with a bullet vote.

In other words, approval voting in rounds results in a sliding down 
of approval cutoff (plus the possible introduction of new candidates 
in some repeated balloting methods, or the simplifying process of 
candidate withdrawals), which makes repeated ballot approval 
incorporate sincere personal utility estimates.

The point I was making was that there exist circumstances where, if 
we know the true situation with the voters, forget voting methods 
entirely!, we will not choose the Condorcet winner, and, this is very 
important:

Neither will the voters choose that, once they know. In some of the 
examples proposed, we can safely assume that the voters will 
unanimously approve the choice of other than what the initial 
majority preference was. All it takes is a weak majority preference 
and a strong minority one, within certain normal limits.

This is a conclusive proof, if you look at those situations, that the 
Condorcet winner can not only be suboptimal, it can be highly so. But 
this has nothing to do with "core support," which is a bogus 
criterion made up by FairVote to try to find some kind of silver 
lining in the massive cloud that surrounds IRV. Unfortunately, the 
silver lining doesn't exist for IRV outside of a certain situation: 
2-party system, no-hope third parties who can only spoil elections at 
most, and there IRV protects the major parties and largely screws the 
minor ones in the long term. Did you ever wonder why Australia ended 
up using IRV? Try looking at the history!

But, of course, FairVote has convinced quite a few minor party 
leaders that it will be Good For Them because it "allows their 
members to express their sincere preference." But that good is 
accomplished through polls, in fact, and is better handled by Fusion 
Voting anyway, which gives actual clout to a third party, which 
retains the ability to spoil elections if it decides that's better 
than tolerating the major party snubbing them.

Third party strength is also better measured by donations to the 
party, which might be better spent in other ways than useless 
campaigning. However, that campaigning expense isn't wasted if it 
brings the party platform to the attention of the people.

Note: Ralph Nader could have vastly improved the situation of the 
Green Party in 2000 if he'd been more flexible and really had the 
party's interest at heart. Simple: in the last days of the election, 
withdraw and urge his supporters to vote for the best major party 
candidate, and, to show support for the party, and for him, send a 
donation to the Green Party, whatever they could afford, no donation too small.

The Green party would have been in fat city, and it's highly likely 
that Gore would have been elected.

One thing that I keep telling people is that Plurality voting works 
fine if the people are actually organized with organizations (there 
can be many of them) that seek the welfare of their members instead 
of what becomes too common: organizations seek their own welfare, 
being the welfare of the inevitable oligarchy that arises, per the 
Iron Law of Oligarchy. Fix the organizational problem, for voluntary 
organizations, and you will actually fix the overall societal problem 
as a consequence.

In that fix, voting methods, as such, are almost an inconsequential 
detail. Asset and standard deliberative process for the electors thus 
chosen (not "elected!") is probably quite enough, and terminally 
simple, as well as creating hierarchical structure that would still 
allow the oligarchy to form, and function routinely in its positive 
role, while still being capable of restraining it when it loses 
contact with the original purpose.

And being outside of government, there is *nothing* stopping this 
from happening except inertia and despair. If it can't happen in 
large organizations yet, because of this, well, then, do it in small 
ones. Make the organizations as small as needed. It starts, really, 
with *two people.* A two-person organization doesn't need Asset, it's 
a fish bicycle, but if the two understand that, if the organization 
grows, it's going to need structure that will survive the challenges 
of scale, and, as more join, they lay down and set up the principles 
and critical structures, that's it. The seed crystal is formed that 
then catalyzes the next stage of implementation.

And when the people are organized, through organizations that they 
trust and have good reason to trust, it really doesn't matter much 
what voting system is in use for "public elections." Because the 
people can use almost any system, and, if the existing system is so 
bad that they can't use it, they will stop saluting it and set up 
something new. Dictators cannot stand up against effective mass 
organization of the people, so they have to try to stop that in order 
to retain their power, thus becoming more repressive and less 
tolerable, until the inevitable happens.

Oligarchies in modern democracies are highly unlikely to attempt to 
prevent FA/DP organization, they will only succeed in disrupting 
organizations that don't follow the FA principles (which I won't 
outline here beyond saying that Alcoholics Anonymous has been the 
model). The organization itself is not going to remove any member of 
the oligarchy from power, nor will it even, as an organization, 
express an opinion. But the members will be connected through 
structures that will allow them to find consensus on *any* topic that 
they *personally* decide is worth discussing, and this will be 
outside official FA structures, it will be through personal contact. 
It could be totally invisible, absent continuous monitoring of every 
person (or at least sampling of same) in every interaction, including 
personal face to face conversations. The DP structure of the 
organization then allows consensus to form and spread. The medium is 
the message!

Instead, the oligarchy, which usually isn't stupid, will join it and 
use it and serve it, and it will reward them appropriately. FA/DP is 
not about mob rule, at all. DP and the related Asset Voting are about 
creating a hybrid between representative and direct democracy, a 
synthesis, if you will, that preserves the best aspects of both. And 
it will be developed through experience accumulated on a small scale, 
where it's relatively harmless. Until it is time to do more.

So what do we do in the meantime? Whatever. Just keep connected, this 
one is worth watching. Join the Election Science Foundation, create 
the connection, it doesn't matter at first if you just go to web-only 
status, but make yourself available for contact by the moderators. 
Later, when it is set up, name a proxy, someone who will get to know 
you and whom you will get to know. The proxy doesn't vote for you, 
actually, but will presumably inform you if there is something that 
they proxy thinks you might want to know about, and will advise you, 
being, perhaps, more familiar with the business of the organization. 
You remain completely free, and your power won't be collected and 
used without your permission.

The Election Science Foundation isn't an FA, exactly, rather it's 
becoming what is called in FA-speak a "service corporation." Such 
corporations are legally independent, and are allowed to collect some 
reasonable funding and property. If they become *too* independent, 
the separation becomes more clear and the links will mostly be 
severed. FAs can fission and fuse, whatever maximizes the freedom of 
the members as well as their efficiency. Rather naturally, though, if 
they are using DP, it's trivial to maintain connections for the 
purpose of negotiating consensus across the nominal organizational boundaries.

I absolutely hope that FairVote activists will join, and, at least, 
leave behind cogent representatives able to negotiate or at least 
advocate on their behalf. ESF isn't against FairVote, even though 
quite a few members might be. Like me, for one, I think FairVote has 
done a lot of damage to the cause of improved voting systems, but 
others are totally free to disagree with that. And I'd like to keep 
the communication lines open and, in fact, to make our communication 
more efficient and more effective. Perhaps we can avoid useless 
contention, and focus on finding ways to find common benefit for 
ourselves and the public. Isn't that the goal of voting?