[EM] About some best methods for a proposal, including Approval, IRV, MJ...

[Michael Ossipoff]

Approval/Pairwise-Count cobinations:

There might be merit differences between MDDA, Smith//Approval,
Pairwise-Sorted Approval, Brams' PAV, MAMPO, and other methods that combine
Approval and pairwise-count...but I don't know which is better.

Just to clarify, PAV differs from  Approval only in that if there are 2 or
more majority-approved candidates, and one of them majority-beats the
others, s/he wins.

It seems to me that PAV should regard any ranked candidate as approved, as
does MDDA.

MAMPO uses MMPO to choose among the majority-approved candidates.

No doubt there are other Approval/pairwise-count combinations as well.

PAV & Pairwise-Sorted Approval start with Approval, in contrast to MDDA &
MDDTR, which start with majority-defeats, using Approval only if
majority-disqualifications don't give an answer.

Well, as an Approvalist, I should like better the ones that start with
Approval.

I guess Pairwise-Sorted Approval chooses from the Smith-set, but it can't
be exactly equivalent to Smith//Approval can it? What are the most
important properties by which it differs from Smith//Approval?

Any other Approval/Pairwise-Count methods that should be considered for a
proposal?

Approval:

Approval's way of dealing with chicken-dilemma doesn't look bad at all,
when compared to the high cost of CD. ...and chicken dilemma is the only
thing in Approval that even comes close to being (sometimes) a "problem" in
Approval. (I'd call it a sometime-nuisance instead of a problem).

There's no reason to change my claim that Approval is the best, and that
improvements on Approval are illusory.

...especially given the fact that, when there's a choice between
equal-top-ranking your strong top-set, and ranking them, in order to choose
_among_ them, then if you choose to rank them, you increase the probability
of electing from your bottom-set.

Given that, Approval looks pretty good in comparison to all the other
proposals. I'd say, the only reason why I'd propose a rank method instead
of Approval is because (as I've been saying), organizations and activists
seem to want rankings, and some voters (overcompromisers & rival parties)
are likely to do better with rankings, which would soften their mistakes.

IRV:

There's not a thing that the CWs's voters can do to prevent hir
elimination, and there's nothing that you, as a voter in one wing, can do
about it either--short of favorite-burial. That's a problem for you if,
when s/he's eliminated, hir voters transfer in the other direction.

But if you're majority-favored, then IRV is perfect, the rank-balloting
ideal.

Alright, if you aren't majority-favored...then.you're just out of
luck--What can I say?

But there's mitigation:

1. If the CWs's voters transfer the other way it must be because the
candidate they're transferring to is closer to them than your candidate is.
Doesn't that mean that their candidate also is likely to be closer to the
one they're transferring to than to your candidate & you?

If so, then you aren't losing much when they transfer the other way. Those
2 candidates must be rather similar--similarly different from what you want.

2. If your candidate is big enough to eliminate the CWs, then surely your
candidate is well-known (assuming honest meda, which we don't have--but
don't get me started on that).

If you like the CWs, especially if you like hir as a strong-top-set
candidate, and if your candidate is well-known, and therefore well-known to
the CWs's voters, then won't they probably like your candidate too?

...and transfer in your direction?

...in which case, no problem.

"But...", you might say, "...it's a problem to the CWs's voters."

Sure, but the CWs is the least favorite candidate, and so the people to
whom s/he's favorite aren't many. So hir elimination isn't a problem to
many people. Well, it's a problem to the people in the other wing.

Ok then, so what if the CWs's voters like the opposite wing better than
yours, and transfer that way. Then you probably don't like the CWs a whole
lot either. If you don't like the CWs, if s/he's bottom-set to you, then
you've got a problem anyway, and no voting system is going to be very good
to you.

And, anyway, IRV's problem should be regarded in the context of the fact
that CD is costly anyway. It comes at a high price, with any CD method.

>From these considerations, I suggest that IRV isn't bad.

Kathy & Robert: Reconsider IRV's merits.

MJ:

Because ER Bucklin needs skipping of rankings to protect your top-rated
candidates from your middle-rated candidates (to give your top-rated
candidates to receive lower-choice vote-sharing before you give to your
middle-rated candidates), then the Bucklin version that I like is the one
that MJ is equivalent to.

So, because I like that version of Bucklin, MJ would be fine.

But offer it as ER Bucklin with skipping. Don't talk about median.

Bucklin is known to have been accepted by the people of at least 39 cities
in this country. Saying that people can rate as many candidates at the same
rating level as they want to isn't complicated. Adding that they can skip
rating-levels isn't complicated either.

So if you're going to propose MJ/ER Bucklin, then propose it as Bucklin,
because you know that people accepted Bucklin.

Michael Ossipoff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20161117/a87f8b59/attachment.htm>