[EM] Re: ATLO, defection, method/criterion model

[Kevin Venzke]

Mike,

 --- MIKE OSSIPOFF <nkklrp at hotmail.com> a écrit : 
> Kevin--
> You said:
> 
> I have a doubt about ATLO. Assuming that the C voters use ATLO, then on
> these (same) ballots:
> 
> 49 A
> 24 B (withheld C preference)
> 27 C|B
> 
> A is elected. But in IRV, it's always the case that A is elected here, since
> there is no ATLO. So it seems to me that in this scenario, IRV is more
> effective than WV+ATLO in encouraging B voters to give C a preference.
> 
> I reply:
> 
> That's odd, that you'd say that, since ATLO in wv is completely effective in 
> giving the B voters incentive to not truncate. You say that with IRV there's 
> no need for ATLO, because the co-operation-defection dilemma is gotten rid 
> of without ATLO? That's real great, but IRV has other deficiencies. Getting 
> rid of that particular problem without ATLO isn't important enough to 
> justify giving up all that must be given up when using IRV.

You seem to think I'm telling you to use IRV. What I'm saying is, if ATLO
uses the same incentive to address the defection problem that IRV uses, it's
useful to ask whether IRV does a good enough job at that. I don't believe it
does.

> You continue:
> 
> Of course WV+ATLO is better in that the C voters don't have to consider
> whether to order-reverse. But it seems to me that this should also make it
> easier for the B voters to persuade the C voters to vote as the B voters
> wish, compared to the same situation in IRV: B voters don't need to convince
> C voters to abandon C; only to not use ATLO.
> 
> I reply:
> 
> Ok, you're not serious. You're wating your time and mine with a practical 
> joke. You now claim that the more of a drastic a giveaway is needed as a 
> defensive strategy, the better, because it's more difficult to persuade 
> voters to use the strategy.

Where do you get "the better"? What I'm saying is: Suppose B is a lesser evil,
and hopes to persuade C voters to vote in such a way that the election won't
be thrown to A when B voters truncate. In IRV this means the C voters can't
vote for C. In WV+ATLO this means that the C voters can't use ATLO. I think
the C voters will find the latter proposal easier to swallow.

> So, then, for you, the more drastically 
> insincere a defensive strategy is, the better. Well, that's good, because it 
> means that we've found a fundmanetal basis for our disagreement on that. Now 
> that that has been accomplished, there's no need for further discussion 
> about that.  There's no need for us to discuss ATLO, my criteria, or wv any 
> further, because we've found that we have completely opposite goals, on a 
> fundamental matter. When the source of disagreement has been found, as it 
> now has been found, there's no need for further discussion.

Heh.

> You continue:
> 
> What do you think: Given the sincere preferences in this scenario, does
> IRV misbehave when the B voters truncate?
> 
> I reply:
> 
> I"m going to tell you something that you didn't know: Offensive truncation 
> isn't a problem in IRV. I couldn't get James to understand this, and maybe I 
> won't be able to get you to understand it either: Even though IRV doesn't 
> have a problem with offensive truncation, or offensive order-reversal, the 
> same undesirable results that offensive order-reversal and (much less often) 
> offensive truncation can cause in wv are present in IRV, _even when no one 
> is using any offensive strategy!_

I know what you mean when you say this (and so does James).

So you call it "offensive truncation" when, under WV, the B voters refrain
from giving C a preference? That sounds unnecessarily negative to me.

> You continue:
> 
> The approach I prefer towards the defection problem does not require failing
> votes-only Minimal Defense.
> 
> I reply:
> 
> You forgot to say what approach that is.

I say it below; I guess that will be in your second email.

> You continue:
> 
> My interpretation of election methods is that all of them have a ranked
> ballot component (allowing full, partial, and equal ranking).
> 
> I reply:
> 
> ...Excuse me...but they don't. All methods don't havae a ranked ballot 
> component.

I'm not going to argue with you about whether methods can be interpreted
as ranked ballot methods.

> You continue:
> 
> When a criterion
> says "if some voter votes in X manner," this refers to this component.
> 
> I reply:
> 
> Which rank component--the ballot's actual one, or the one that you're 
> assuming. 

The one I'm assuming.

> Or are those two rank components the same on rank ballots? Meaning 
> that you have different rules for rank ballots and other ballots.

When the physical ballot allows all preference orders, then they're the same,
because no change needs to be made to make the "assumed" ballot valid. I
don't agree that this is a different rule.

> You continue:
> 
> (This
> component isn't necessarily completed sincerely.) When the ballot format of
> the method restricts the input of the ranked component, or has additional
> components (such as an approval cutoff), it has to behave properly 
> (according
> to a given criterion) no matter how the unrestricted ranking may be changed
> to be accepted as a valid ballot.
> 
> I reply:
> 
> Come again??

As an example, take Minimal Defense again, in the form "if more than half of
the voters rank X over Y and Y over no one, then Y mustn't win." The ranking
done by voters in that sentence refers to the intended ballot. (Criteria do
not ever refer to the physical ballot in my interpretation. They only regard
the intended ballot, and the probability distribution resulting from the cast
physical ballots.)

If we check First Preference Plurality against this criterion, it fails.
Suppose we have these intended ballots:

49 A
24 B
27 C>B

Every voter expresses a strict first preference, so there is only one case to
check for the physical ballots: 49 A, 24 B, 27 C. A wins, contrary to the
criterion.

If we were using a method requiring a complete ranking, then we would have a
lot more cases to check, since the 49 A voters would each have to vote either
A>B>C or A>C>B, and similarly for the 24 B voters.

I asked you recently whether VFA satisfies SDSC, and you said no. I also
interpret that VFA doesn't satisfy votes-only Minimal Defense, based on the
above reasoning.

> You continue:
> 
> This is why I don't feel compelled to include Steve Eppley's requirement in
> votes-only Minimal Defense that all preference orders must be accepted
> 
> I reply:
> 
> No, you just say the same thing in a different wording. I'll get to that 
> soon, below.

And that's basically what I said, except that you cut that part off of your
quote for some reason.

> You falsely assume that Plurality is a rank method. 

This sentence makes no sense unless it is either "true" or "false" that
Plurality is a rank method. I'm not saying it's "true" that Plurality is a
ranked method. I'm saying I interpret all methods to be ranked methods.

I don't use this interpretation because it is "true," but because it's
*useful*.

> So, when you say that 
> Pluraity fails your votes-only Minimal Defense, what you really mean is that 
> imaginary-rankings Plurality fails it.

Now you're just being stubborn. My interpretation of Plurality doesn't change
it into a different method.

> Let me suggest a better way to do that:
> 
> The Universal Rankngs Fiction (URF):
> 
> With non-ranked methods, in addition to whatever ballot marks are actually 
> allowed and counted on the ballot, if the voter were also allowed to make 
> one and only one ranking on the ballot (even though it isn't counted and 
> even though it in fact isn't provided for or allowed on that method's 
> ballot), then the following criterion should be met:
> 
> [end of URF definition]

Well, I think this definition is confusing because it isn't clear that the
ranking might contain part of the counted ballot marks.

> URF is a statement that should be explicitly written at the beginning of 
> each of Kevin's criteria. Sure, Kevin, you could say that URF should be 
> understood to be at the beginning of all of your criteria,even if it isn't 
> actually written there. But the trouble with that is that some people are 
> going to find the criteria, and they'll have no way of knowing that you want 
> it understood that they begin with URF.

That's a valid point. Hopefully other people do not have a major problem
understanding me when I quote criteria.

> If URF is dealtwith in that way, as part of your criteria, instead of just 
> vaguely saying that it's your interpretation of the methods, then it has 
> validity. But that doesn't make it an elegant approach to criteria.
> 
> It's counterfactual, awkward, and roundabout.

I consider your "counterfactual" argument very weak.

I don't agree that it is "awkward"; it allows criteria to be written very
concisely. The "cast ballots" don't have to be referred to at all, and there
is no intelligent "translation" process from the intended ballot to the cast
ballot. It can be assumed to be arbitrary.

I don't understand why you say "roundabout."

> As I said above, when the criterion refers to rankings, URF is nothing but a 
> statement that nonrank methods fail the criterion. No different from what 
> Steve said.

Instead of "URF" I'm going to use the term "RI" for "ranked interpretation."

That isn't true. If it were true, then it wouldn't be possible to write many RI
criteria that Plurality passes. Yet Plurality satisfies the Plurality criterion,
Clone-Loser, Later-no-harm, Later-no-help, Participation, Mono-raise...

> Let's look at Approval & Plurality with URF votes-only Minimal Defense:

I agree that either with Steve's complete rule, or using RI, Approval and
Plurality both fail votes-only Minimal Defense.

> What would be your URF votes-only version of WDSC?

Well, I'm sure you won't let me define your criteria for you, and also I
don't use any version of WDSC, so I don't know.

However, since methods are defined by their results given cast ballots and
not sincere preferences, a votes-only version of WDSC must be possible, if
difficult to use.

> Notice that it gives special status and mention to a particular balloting 
> mode.

That's because it can encompass most other modes. It would be straight-forward
to adapt RI to also accept cardinal ratings. Perhaps it could be called "RRI"
for "ranked and rated interpretation." Then "RRI" criteria could be written
which would then be applicable to all methods.

> And, in your URF system, what would be your votes-only version of WDSC?

See above.

Kevin Venzke



	

	
		
__________________________________________________________________
Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour vos mails ! 
Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/