[EM] Warren: MDDA, FBC, Approval, RV

[MIKE OSSIPOFF]

Warren--

You wrote:

MDDA fails "add top".  That is, if you add some identical honest votes 
ranking A top,
that can harm A (e.g. by creating a Condorcet winner [who is not A]
who then wins, whereas previously there was a Condorcet cycle and A was the 
winner
on approval counts).

I reply:

MDDA doesn't look for CWs, and doesn't only use Approval when there's a 
Condorcet cycle.

Here's MDDA's definition:

1. A candidate is disqualified if another candidate is ranked over him/her 
by a majority.

2. The winner is the un-disqualified candidate who is ranked on the most 
ballots.

[end of ordinary MDDA definition]

Yes, if a method met Condorcet's Criterion, it would have to fail FBC. But 
MDDA doesn't meet Condorcet's Criterion.

As for "add top", I'll take your word on that. I consider it important that 
people not have incentive to bury their favorite. So, whatever add top is, 
I'd rather have FBC.

I don't agree that add top failure is like FBC failure. If it doesn't give 
you incentive to vote someone ofver your favorite, then how much like FBC 
could it be?

Add top or Mono Add Top sounds like a Woodall re-naming of one of the 
criteria related to Participation. I call those "adverse results criteria". 
So far as I know, all rank methods other than Borda (and maybe a few of its 
close relatives) fail Particiption.

You continued:

Now this may not technically count as an FBC failure, because I daresay 
there
exists some way to rank A top and dishonestly order the remaining 
candidates,
which still leaves A the winner.   However, in practice, it may have a very 
similar
effect to FBC failure.

I reply:

How?

Yes, your lower rankings could hurt your favorite, but that doesn't make you 
not want to vote. Ranking your favorite 1st improves your expectation, 
because you have no way of knowing if the lower rankings will hurt the 
favorite. In fact, by increasing the defeats of everyone below Favorite, 
your complete ranking is, on the average, going to benefit Favorite. I'm not 
sure, but, on average, you probably help Favorite more than you hurt him 
when you rank the lower choices, because you're helping them have majorities 
against eachother. Yes, it can go the other way.

If helping Favorite is really the important thing, in ordinary MDDA, don't 
rank anyone else, so that you give Favorite an Approval difference over all 
the others. That's probably better than ranking them all, losing the 
Approval difference, and hoping to beat them by giving them majorities 
against eachother (but maybe erasing a majority defeat among them).

With Deluxe MDDA, you can rank them all, and still withold Approval votes 
from them, so Deluxe MDDA, in that way, gives you more of a strategy 
dilemma. Should you rank the lower choices? How should you rank them? 
Probably you'd try to rank them in reverse order of winnability, even though 
that risks the add-top failure scenario.

Now you see why I like power-truncation. Just power truncate everyone but 
Favorite. You're effectively voting all but one of the candidates over each 
one of the non-favorites.

Instead of just Favorite, I usually talk about a preferred set. As I said, 
power truncation makes MDDA as good as Approval, when it's an 
acceptable/unacceptable situation.

So sure, for the a/ua situation, it takes power truncation to make MDDA or 
Deluxe MDDA as good as Approval or RV. And I, and many progressives, 
consider the elections to be a/ua (though we don't all agree on what is 
acceptable).

So no argument from me--there's a good case for Approval or RV over MDDA, or 
any rank method.

Somtimes a rank method can have less strategy than Approval or RV. But when 
it's a/ua, the strategy of Approval or RV is as simple and obvious as it 
could be, and that isn't true of MDDA (or other rank methods) without power 
truncation.

So you have a good point: Maybe the rank methods are more trouble than 
they're worth.

So what's the appeal of MDDA? With RV, the lesser-of-2-evils progressives 
might give the Democrat a little less than Nader, thereby helping Nader 
compared to what would happen with Approval. With MDDA, though Approving 
Dean, they're voting the better candidates over him, thereby giving him (and 
the Republican) a majority defeat if they're numerous enough. So, in that 
way, MDDA could improve on Approval in the same way that RV does, but maybe 
more.

And when MDDA, in a subsequent initiative, eventually gets power-truncation, 
it matches Approval's and RV's a/ua simplicity, while still offering SFC for 
the sincere voters. So then MDDA begins looking better than Approval and RV.

Of course if voters are mostly sincere, with rank methods, as some believe, 
then MDDA's SFC compliance could give it advantage over Approval & RV, even 
without power-truncation.

By the way, there's a sense in which Approval or RV matches what the 
pairwise-count methods do:

If ranking is sincere, and more people prefer X to Y than vice-versa, then X 
pairwise-beats Y.

In Approval, if the number of voters who consider the X>Y preference to be 
important enough to make a point of voting it is greater than the number who 
consider the Y>X preference important enough to make a point of voting it, 
then X outpolls Y, and Y can't win.

So, in pairwise-count and in Approval, it's a very similar pairwise contest 
between X and Y. Which gets more preferences, versus which gets more 
emphatic preferences. With RV it's similar to Approval except that, if 
voting isn't all strategic, it could partly be a matter of which has a 
greater sum of preference strength over the other.

So one counts preferences between X and Y, and the other counts emphatic 
preferences between them. Is there really much difference?

In this reply I've made a better case for Approval & RV > pairwise-count 
than vice-versa.

Mike Ossipoff

_________________________________________________________________
Don’t just search. Find. Check out the new MSN Search! 
http://search.msn.click-url.com/go/onm00200636ave/direct/01/