Who is wronged in Kevin's MMPO bad-example?
-------------------------------------------
Yesterday I asked how bad C can be, in that example, if nearly all the A voters
are indifferent between B and C,
and the only one not indifferent prefers C to B.
I'd like to additionally ask who is wronged in that example. Someone who is
indifferent between the winner and the
other top candidate? Hardly.
Surely the "wrongness" of a result must be judged by whether or not someone is
wronged by it.
Kevin's MMPO bad-example, MDDTR's Mono-Add-Plump failure, are
Plurality-prejudice aesthetic matters.
.....................................................................................................
MCA protection of top-rated from middle-rated:
----------------------------------------------
When I addressed that matter yesterday, I should have emphasized that,
typically, with a 1-dimensional
political spectrum (which we pretty much have), if you hope that someone will
win who is no worse, from your
point of view, than X, then you must hope that the voters preferring X to those
you like less are a majority.
If there's no indifference, a Condorcet candidate has a preference majority
over all the other candidates.
So, typically, majority is part of winning.
If one of your top-rated candidates can win, then most likely s/he has majority
support against those s/he can win
against. So, the requirement for a majority of top ratings for a 1st-stage win
isn't an unreasonable or unattainable
requiement. Majority is typically needed for winning anyway.
That's the value of 3P, UP, 1CM and SDSC.
......................................................................................................
3-Slot SFC:
---------------------------------
I'd like a criterion name for the SFC-like benefit available in a 3-slot
method such as MDDTR or 3-Slot MMPO. So I'm here defining 3-Slot SFC.
The 3-Slot SFC that I've found isn't very elegant, because part of its premise
stipulates a
3-slot method. That's very inelegant, when a criterion stipulates a type of
method. Of course
that's the only way that the inelegant votes-only criteria can keep Plurality
and Approval from
meeting those criteria.
So, this 3-slot SFC shares the inelegance of the votes-only criteria so beloved
among academics.
3-Slot SFC (SFC3):
If the method is 3-slot, and if no one falsifies a preference, and if there is
a Condorcet Candidate
(sincere CW), and if Y has a majority of the voters voting someone over hir,
then Y shouldn't win.
[end of 3-Slot SFC definition]
Plurality and Approval neither pass nor fail SFC3, because that criterion is
only for evaluating
3-slot methods.
................................................................................................
The 100, 15, 0 utility example:
___________________________________
I spoke yesterday about MCA in the following example. Now I'd like to comment
on other methods
with that example.
In 1995, there was discussion of an example like this:
(The numbers are utilities as judged by you)
A 100
B 15
C 0
Of course with 0-info Approval you vote only for A.
No one knows the mathematical stragegy of MCA, but, if C is completely
unacceptable, then C
almost surely is too, and doesn't get a middle rating.
What about MMPO and MDDTR?
With those methods, which meet Later-No-Harm (LNHa), there's no question about
whether to give
B middle or bottom rating. You give B middle rating. You have no reason not to.
You're free to use
B to help defeat the worse C.
So, along with the Approval bad-example (ABE), this 100,15,0 example shows the
big strategic
improvement that comes with MDDTR and MMPO.
Of course with any method, including MDDTR and MMPO, it can be necessary to
vote a compromise
equal to your favorite(s) in order to keep someone worse from winning.
But, as I said, if C is completely unacceptable, then surely a candidate with
only 15, our of 100, will be
likewise completely unacceptable, and shouldn't get a top rating.
MDDTR and MMPO ease your strategic dilemma in this example, as they do in the
Approval bad-example.
...But they can be "controversial", because they can be aesthetically
criticized for not looking enough like
Plurality in their results.
As I said, I'd like a good method to be enacted, and achieving that is more
important than _how_ good the
method is. MTA and MCA bring great improvement over Plurality. They meet FBC,
and have the great majority-
rule guarantees of 3P and 1CM. I would probably, then, just propose MCA or MTA,
rather than risk the
distraction and confusion that opponents could cause, regarding MMPO's and
MDDTR's greater departures from
Plurality. But it's still worth talking to people about MMPO and MDDTR--their
big strategy improvements,
and the not-valid criticisms of them.
Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
