[EM] Rob: MDDA, 10/15/05

[MIKE OSSIPOFF]

Rob--

You wrote:

I agree with Warren that a system that doesn't break down in the face of
full rankings is highly desirable, for many of the reasons that he
cites.

I reply:

I assume you're implying that MDDA breaks down in the fact of full rankings.

If everyone but me were going to vote full rankings in MDDA, I'd vote only 
for Nader, and so Nader would automatically win the approval count. Unless 
Nader has a majority defeat and someone else doesn't, Nader wins. The 
full-ranking electorate have given me the power to choose the president.

For everyone to rank fully in MDDA would be unstable, because any one voter 
could improve his/her expectation by bullet-voting, assuming that we don't 
have information about who is likely to be majority-disqualified.

Approval, too, would "break down" if everyone voted for all the candidates. 
I recommend against doing so. As Kevin pointed out, it makes no sense 
whatsoever to rank all the candidates in MDDA. Don't rank your last choice. 
Then you're voting everyone over him/her in Approval. You're casting an 
Approval vote that says who is the worst candidate. As Kevin said, if 
everyone did that, a tie would be as unlikely as it is now in ordinary 
Plurality.

So let's lay to rest this notion of MDDA having a problem with everyone 
ranking all the candidates.

You continued:

I think his argument could be reworded slightly to be less
controversial.  It's not that there's a universal or near-universal
desire to provide full rankings.

I reply:

That's for sure. I've just voted in an Internet presidential poll with 15 
candidates. Ranking all of them was a bit of work. Even more so in the EM 
presidential poll with 46 candidates. Few people will want to bother ranking 
all the candidates when there are more than a few.

You continued:

A system that violates later-no-harm
therefore has a serious political liability in that regard.

I reply:

That's a strong statement, and it requires justification. Who says that 
rightness requires that people should have no incentive to not rank 
everyone?

You continued:

My personal belief is that satisfying later-no-harm (or at least
minimizing violation to rare instances) is highly desirable.

I reply:

All criteria are desirable. It's a question of choosing which ones one 
considers most important, especially when it isn't possible to comply with 
all of them. To you it's very important to encourage people to rank all the 
candidates, with no incentive to do otherwise. To me it's more important to 
assure people that there's absolutely no reason to bury their favorite. I 
don't say that your standards are wrong. But remember that others' standards 
aren't wrong because they differ from yours. There aren't right or wrong 
standards. What's wrong is the notion that there are.

And, by the way, MDDA's SFC compliance gives some protection against LNH 
failures, for people who rank the candidates they don't like.  Its premise 
conditions aren't at all implausible.

But you're right that it would usually be a better idea to not rank the ones 
you don't like. Sure, I could help to give Bush a majority defeat from 
Kerry, but why should I do that when I can just vote an Approval difference 
against both of them by ranking neither.

But MDDA's SFC guarantee means that it doesn't do so much harm when the 
giveaway progressives insist on ranking Kerry. They're failing to vote an 
Approval difference between Kerry and their favorite, but they're still 
contributing toward a possible majorilty defeat for Kerry. If you think 
that's unrealistic that a majority would rank Nader over Kerry, consider 
that Nader is usually the voted CW in nearly all presidential 
rank-ballotingl polls.

Sure, Condorcet(wv) encourages ranking all the candidates, unless offensive 
order-reversal is expected. But I found it a bit yucky to rank candidates I 
don't like. I have no objection whatsoever to a method in which it's 
strategically better to not rank disliked candidates.

You continued:

I
personally believe that getting people to think deeply about compromises
is how we get to a more civil state in politics.  If people can be
jostled out of their comfort zone and consider the relative merits of
candidates who they might at first blush consider "evil", then perhaps
we'll truly get less evil candidates than the current batch who foment
partisan rancor.

I reply:

I disagree. Why is it better if people consider the relative merits of Kerry 
and Bush, or McKain and Hillary Clinton, etc.? To talk of their relative 
merits implies that there is a significant merit difference. Why waste time, 
debasing oneself with concern about the merit-difference between meritless 
candidates?

That won't give us less evil candidates, but if we continue trying to find 
how one is better than another, then it tends to perpetuate their undeserved 
success.

Discard the undeserving candidates. In Condorcet(wv), if you don't expect 
offensive order-reversal, go ahead and strategically rank them in reverse 
order of winnability, if you can hold your nose tight enough. That will 
decrease the chance that one will win.

Blake and others have complained about that incentive and used it to argue 
for margins over wv. But margins has much bigger strategy problems--major 
defensive strategy need, to protect majority rule or the CW, as opposed to 
relatively minor strategy incentive.

You continued:

Here's a related set of questions I've been meaning to ask:

1.  Are the Later No Harm (LNH) criterion and the Sincere Favorite
Criterion (SFC) mutually incompatible?

I reply:

SFC stands for Strategy-Free Criterion. Doesn't MMPO meet both of those 
criteria?

You continued:

2.  Are LNH and the Favorite Betrayal Criterion (FBC) mutually
incompatible?

I reply:

Doesn't MMPO meet both?

You continued:

3.  Are LNH, SFC and FBC mutually incompatible?

I reply:

Doesn't MMPO meet all of those?

But regrettably, MMPO has a big problem, as Kevin pointed out.

Mike Ossipoff

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