There should be brief names for Bucklin (= whole simultaneous) and IRV(= whole).
I suggests ABucklin and AIRV ("A" for "Approval"), or BucklinA and IRVA.
UP is met by MMC-complying methods with the (= whole) specification. Methods
allowing equal ranking, with a whole vote
to everyone you rank at a particular position. Maybe that could be generalized
to say "with equal-ranked candidates having the full status of that
shared rank position, as if ranked there alone".
What I like about protection levels is that, if a majority agree with you about
who goes in one of the protection levels, then you have
a big assurance, and the method will enforce that distinction.
Say you're a progressive who feels that the Democrat candidates are worth
protecting against the (identical) Republicans.
(I'm going to sometimes abbreviate "progressive(s) as "P"; "Democrat(s) as "D";
and "Republicans" as "R").
So you can refuse to rank any Republicans (unless there's an exceptional one--I
won't name names), and equal-top-rank all of the progressives.
In DP, that means preferring only all the P, Approving the D, and none of the R.
In ABucklin or AIRV, as I said, you equal-top-rank the P, and rank the D in
whatever order you like, and you don't rank any R.
If a majority agree with you about the equal-top-ranking, or the truncation,
then a 3P method will enforce that distinction.
That's why I like 3P and UP.
UP enforces such a distinction at any rank-position.
Say you're a P who has no use for the D. You equal rank a few of the _best_ P.
In ABucklin or AIRV, you rank the other P in any order you like. And of
course you refuse to rank any Republocrats ({R,D}).
In DP, you prefer the best D, and approve the rest, but none of the R.
I like the power that 3P and UP give to voters who are in a majority.
Of course we have an antagonistic electorate, maybe (probably?) so antagonistic
that there are no mutual majorities,
no solid coalitions. Then 3P and UP won't offer anything.
Certainly I don't expect P and D to be in a mutual majority, though they most
surely add up to a majority. Though the D is (at least a little)
better than R, to the P, that isn't mutual: If the policies, humaneness and
honesty-record of the D are really your favorite, then R will be your 2nd
choice,
since they're nearly identical, in terms of their policies, humaneness and
honesty-records. So, no mutual majority there.
Regrettably, I don't know if there's any mutual majority among the P either.
They're at least as antagonistic among eachother as they
are toward non-P. But there could and should be mutual majority among the P.
MMC, 3P, and UP _are_ of value.
What if there are no mutual majorities, solid coalitions?
The ABE non-failure of DP and AIRV might do much to mitigate the lack of mutual
majority, as will the SDSC-compliance of
DP and ABucklin.
With DP and ABucklin, if a majority all rank even one same candidate, and
there's a candidate-set that none of them rank, no one they don't
rank can win. There should be a criterion about that. Maybe it's covered by
another criterion. It sounds almost like it comes under MMC (but not
literally). But
doesn't AIRV meet MMC? And does AIRV have the property stated in this
paragraph's 1st sentence? I don't think it has it (but tell me if I'm
wrong).
If that property in the previous paragraph's 1st sentence isn't covered by
another criterion, I'll call it "1 candidate mutuality" (1CM or OCM)
AIRV probably doesn't meet 1CM, and I don't know if its LNH-compliance can make
up for that. Bucklin fails in the ABE. That leaves DP.
Since DP meets 3P, I don't think UP is as important as ABE non-failure. So
doesn't DP emerge as the best method, by these considerations?
CD, LNH, SDSC, and 1CM can mitigate the lack of mutual majority.
When I say "CD", I'm referring to the tentative replacement for CD that I
posted yesterday. My original CD, the one mentioning the
sincere CW (SCW), I'll now call "CCD", for "Condorcet CD".
I'd like to add one more premise stipulation to CD:
The A voters are more numerous than the B voters.
[end of added CD premise stipulation]
That makes for a more demanding requirement for CD taking effect, which would
improve it, from the perspective of anyone who
isn't sure that they like CD.
My advocacy of SFC was based conclusions about
offensive-order-reversal-deterrence based on 3-candidate examples.
If order-reversal is well-deterred, then SFC's premise is likely to be
effectively met (not enough order-reversal to affect the outcome).
But is order-reversal well-deterred when there are more candidates?
Say D is SCW. The P & D voters vote D over R, giving R a majority defeat. But
the R voters rank, over D, some candidate W, who is
terribly majority-beaten, making a large majority against D too. Now the R
needn't worry that there insincerely upranked candidate W will
win.
Without reversal-deterrence, SFC loses its reassuringness.
Other criteria, some of then named above, remain as assurances when there is no
mutual majority.
Is SDSC a little too demanding?
How about a WSDSC that requires only that no one need to rank someone equal to
or over their favorite, in order
to assure that Y won't win?
Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
