[EM] Kristofer: MMPO objections

[Kristofer Munsterhjelm]

On 12/29/2011 10:49 PM, MIKE OSSIPOFF wrote:

> Kristofer continues:
>
> You say
> that it is non-valid from your own point ofview  <http://www.textsrv.com/click?v=VVM6MTQwMTY6MTIxOnZpZXc6ZmIzNmIxM2E4NTZhY2M2YjczNGZjZWNhZjA1NDJjNGU6ei0xMDMyLTEwMzY4Omxpc3RzLmVsZWN0b3JhbWEuY29t>, where you think the
> tradeoff made by MMPO is worth it.
>
> [endquote]
>
> Kindly don't quote me saying something that I didn't say.
>
> I told why the criticism is nonvalid. See my paragraph above.

Okay. I was inferring from what I thought you were saying.

(By the way, what's going on with your mail client? It inserts links 
into www.textsrv.com for common words, e.g. "view" above. I've seen it 
insert links to electorama, too.)

>> Others (like me) may have different
>> opinions.
>
> [endquote]
>
> If you're saying you have an opinion that someone is seriously wronged by the election of
> someone over whom no one prefers anyone other than their favorite, then can you tell us
> in what way that seriously wrongs someone, or are you merely expressing a personal opion
> (as, of course, is your right)?

There are really two parts to what I'm saying:

1. Having to restrict myself to considering only single voters instead 
of a "class" seems to be just as arbitrary, if not more so, than to 
consider Plurality as valid.
2. Therefore, I do not think that you can, on one hand, say that Kevin's 
MMPO example is non-valid, period (because it relies on a "class" rather 
than a single group of voters), and on the other, dismiss the Plurality 
criterion (or the mono-add-plump criterion for another method that 
doesn't pass it) from the reasoning that it's arbitrary opinion and not 
an objective standard.

For that matter, I would go further. I would say there's plenty of 
precedent to consider "classes" rather than only single groups of 
voters. I've already covered the Plurality criterion, but consider the 
mutual majority criterion. You could say that no single group has been 
wronged when a method fails to elect from a set that a (otherwise 
disparate) majority ranks first, but not in the same order, above 
everybody else. You could also say that no single voting group has been 
wronged when a method fails to elect from the Smith set (or possibly 
even when it fails to elect the Condorcet winner, because the majorities 
preferring the CW to other candidates may differ depending on the non-CW 
candidate in question). Yet these are all valid criteria. If anything, 
it's your restriction that an error must "seriously wrong someone", a 
single particular voter, or a group preferring the same candidate at 
top, that seems arbitrary.

Voting is a collective process. Why shouldn't a collective process be 
subject to standards that consider properties distributed over broad 
groups as well as those regarding individuals or narrow groups?

> Kristofer continued:
>
> Saying that you're willing to make the tradeoff is quite a
> different thing than saying that the criticism (e.g. MMPO's Plurality
> failure) is inherently invalid.
>
> [endquote]
>
> Quite so. I'm saying both things.
>
> The criticism is nonvalid, for the simple reason that I stated.

>
> I'd conceded that, if you want to combine the A and B plumpers as plaintiffs
> in a class-action lawsuit, then it could be argued that, when considered together,
> they're _slightly_ wronged.

Slightly? In Kevin's bad-example, the "class" has a decision go against 
them, even though the members of the class number ten thousand and the 
non-members number in the single digits. In fact, you can make the 
"class"'s size as large as you want. You could have seven billion in the 
"class" and the example would still work.

It doesn't matter how slight the preference violation is (not that we 
can infer strength of preference from the example in the first place), 
because you can multiply it by as large a number (the size of the 
"class") as you want.

> But the dependence on that prestidigitation doesn't inspire any confidence in the
> "wrongedness.  ...when you can't show that the A voters themselves, as a group, or
> the B voters themselves, as another group, are wronged by the election of someone
> to whom they don't prefer anyone other than their favorite.

So you would thus throw away mutual majority, or Smith?

> All voting system standards and criteria come down to a matter of opinion about
> what is important and what isn't. And what is more important than something else.
>
> I suggest that it's important to get rid of the worse strategy problems.
>
> Maybe you don't agree with me about that. Fine. There's no reason why we should
> all agree in our purposes and goals. You aren't wrong.
>
> What (I suggest) is "nonvalid" is your claim that the A plumpers or the B plumpers are wronged
> badly enough to give up important method-properties. --But, as I said, that's a
> subjective judgement, and maybe you have very different notions about what is important.
> You aren't wrong. By your own, different, personal purposes, your criticism could be
> valid.

I didn't claim that you could only look at the A-plumpers or B-plumpers 
separately. I think that the "class action suit", as you put it, is more 
than just sleight-of-hand. What I am claiming is that the "class" or the 
broad group or what you want to call it - that they are wronged as a 
whole. That was the point of my cake analogy.

If you think that criteria that can't be applied to each single voter 
individually to be invalid, that's your opinion, and I can't stop you. I 
can try to show you why I think differently, but I can't go beyond that.

Saying "nonvalid" right out just seems to imply more than a mere 
difference of points of view about what tradeoffs to make.