[EM] MDDA: No extra truncation incentive. No order-reversal problem.

[MIKE OSSIPOFF]

Rob--

You expressed concern that MDDA would give more truncation incentive than 
Condorcet. It need't.

Sure, MDDA counts Approval, but it also counts pairwise, and has SFC 
compliance to enforce pairwise majorities. That means that you can forget 
about Approval, and rank sincerely. So the strategy is simple, just as in 
Condorcet: Rank down to the CW. And, as with Condorcet, if it's uncertain 
who the CW is, you can overshoot. In fact you can overshoot more safely in 
MDDA than in Condorcet, for reasons that I'll soon get to.

In fact, to be on the safe side, you could rank all the way down to your 2nd 
to last choice.

Say you rank all the way down to your 2nd to last choice. And say that the 
supporters of that 2nd to last choice, and other voters in that spectrum 
region, refuse to rank the CW. Approval-wise, you've been had. So what? The 
supporters of those candidates who, for you, are past the CW, can't do 
anything about the fact that their candidates have majorities against them. 
They can't win, no matter what they do.

About offensive order-reversal, with regard to MDDA, I mentioned the other 
day that it's deterred because to do so would give an Approval vote to the 
insincerely-raised candidate, a lower choice. That's true.

And it's deterred for another reason: Again, their offensive order-reversal 
can't change the fact that their candidate has a majority against him. In 
Condorcet, they can make their candidate win if they they make a big enough 
majority defeat against the CW that their own candidate is the least-beaten. 
That doesn't work in MDDA, because every majority defeat disqualifes.

But what they would accomplish, with their offensive order-reversal, would 
be to cause their insincerely-raised last choice to not have a majority 
against him, and not be disqualified. They could only elect their last 
choice by the offensive order-reversal.

Example:

Nader, Dean, and Cheney

Sincere preferences:

40: CDN
25: DCN
35: NDC

Voted rankings:

40: CN
25: DC
35: ND

Condorcet:

Defeat strengths:

Dean: 75
Cheney: 60
Nader: 65

Cheney wins. The offensive order-reversal succeeded.

MDDA:

All 3 candidates have majorilty defeats, so no one is majority-disqualifed.

Approval scores:

Cheney: 65
Dean: 60
Nader: 75

Nader wins. The offensive order-reversal elected the reversers' last choice.

Every nonprobabilistic method has a strategy problem of some kind. What kind 
does MDDA have?

If there isn't defensive truncation incentive, or offensive truncation 
strategy, and if offensive order-reversal doesn't work, what problem can 
there be?

Majority strength circular ties. But they'll be rare. Effectively, MDDA 
doesn't have any strategy problem.

Sure, there could be the offensive strategy of creating, via offensive 
order-reversal, a majority-strength cycle among one's opponents. But 
engineering a majority-strength circular tie is even more difficult than the 
organizing that would be needed to do sufficiently large scale offensive 
order-reversal in Condorcet(wv).

As I said, MDDA doesn't have a strategy problem.

The strategy: Just vote at least down to the CW. If it's uncertain who's the 
CW, then overshoot, to be on the safe side. Maybe rank down to your 2nd to 
last choice.

SFC guarantees to a sincere-voting majority that no one worse than the CW 
will win, if falsification doesn't occur on a scale sufficient to change the 
outcome. But in MDDA, as compared to Condorcet, that falsification must be 
of a much more difficult kind, the kind that seeks to create a 
majority-strength cycle.

So, Warren, that answers your question about MDDA's strategy. It's just 
simple sincere ranking.

Comparison between MDDA and DMC:

1. MDDA has a much briefer definition and a much simpler rule.
2. MDDA meets SFC, and fails Condorcet's Criterion, while DMC meets 
Condorcet's Criterion (lexocographically) and fails SFC.

I"ve told why SFC gives a more meaningful guarantee than Condorcet's 
Criterion does.

3. MDDA meets FBC. DMC fails FBC.

4. Offensive order-reversal is much more succeed-able and safe in DMC than 
in MDDA.

Here's why #4 is true:

DMC improves on Condorcet(wv), when it comes to offensive order-reversal 
(but not when it comes to SFC), because the creation of the artificial cycle 
can't improve the reversers' candidate's chance in the circular tie 
solution, by Approval. In fact, they're giving an Approval to the 
insincerely-raised candidate. But that could be cancelled out if the victim 
voters have ranked all the way down to the reversers' candidate. So, in DMC, 
as in Condorcet(wv) it's necessary to vote short rankings, to protect 
against offensive order-reversal.

That isn't true in MDDA. In MDDA, the offensive order-reversal can't get rid 
of the reversers' candidates' disqualifying majority defeat. But it does 
reliably get rid of the disqualifying majority defeat of the 
insincerely-raised candidate.  Unless the reversers manage to engineer a 
majority-strength cycle among their prospective victim candidates.

So, in MDDA, but not in DMC or wv, the offensive order-reversal reliably, 
directly gives the win to their last-choice insincerely-raised candidate.

MDDA is the simplest and the best, under present conditions.

As I said, there could be future conditions where there is no LO2E problem, 
and little need for FBC. I've always said that offensive order-reversal 
isn't really a problem in wv Condorcet, and in the future electorate I'm 
talking about in this paragraph, it could be even less likely. Then, the 
added deluxe array of majority-enforcing criterion-compliances that SSD has 
could make it the best method, in that future improved electorate, under 
those better conditions.

Mike Ossipoff

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/