Forest--
You wrote:
I was discouraged when Markus and Steve pointed out to me that MMPO failed
Clone Winner,
I reply:
It seems to me that the way that MMPO or PC fails ICC, MMC, or GSFC, is not
a failure that will be likely or frequent.
Copeland fails ICC in a much more flagrant and problematic way. And Copeland
is, or at least used to be, popular with authors. Anyone who has ever
praised Copeland is in no position to criticize MMPO for ICC failure.
You continued:
...since Clone Winner compliance is important for avoiding the spoiler
problem.
ICC is about a special case in which some candidates are so similar that
their merit differences among themselves are considered by everyone to be
negligible compared to other merit differences. Of course it would be
desirable if no faction could gain or lose advantage by running such a set
of candidates. But it's about a special case.
But the defensive strategy criteria deal with the spoiler problem, as well
as the lesser-of-2-evils problem, in its more general aspect.
And MMPO, in addition to meeting FBC, meets SFC and WDSC. And, with AERLO,
MMPO meets SDSC and Strong FBC.
That's a lot of lesser-of-2-evils protection.
As I said, initially plain MMPO could be proposed. AERLO could be proposed
in a subsequent initiative, allowing the first MMPO proposal to be simpler
and more proposable.
As you know, MMPO can be introduced to people very briefly:
Voters rank as many or as few candidates as they want to. Equal ranking is
allowed.
Elect the candidate over whom fewest people prefer someone else.
[end of suggested public MMPO introduction]
Sure, ideally it should be "voters" instead of "people"; and it should be
"vote" or "rank" instead of "prefer". But something more precise and
explicitly complete could follow, or be available. For this introduction, I
wanted to say it how it would sound briefest, clearest and most compelling.
You continued:
As you know, the problem is that an MMPO winner with a max pairwise
opposition of, say 40 percent, could be replaced by a cycle in which each
member was opposed pairwise by some other cycle member on more than 60
percent of the ballots.
I reply:
Sure. Same with PC. But I don't consider that to be something that will be
frequent or likely.
You continued:
In a case like this it seems natural to collapse the cycle, and then when
the MMPO winner turns out to be the collapsed cycle, find the MMPO winner
within the (reconstituted) cycle.
I reply:
Before BeatpathWinner/CSSD, Ranked-Pairs, or SSD were posted to EM, or at
least before I understood their advantages, I wanted to try to make PC
(easily shown to) comply with SDSC, and I tried something that I called the
"subcycle rule", which involved collapsing cycles. It had problems. Maybe
cycles can be collapsed in a way that doesen't have the problems that the
subcycle rules had.
When you refer to collapsing cycles, you mean the way that it's done in the
sprucing-up procedure, right?
You continued:
Spruced Up MMPO would be too drastic. Perhaps leaving off the first stage
of the spruce up process, and proceeding directly to the "beat clone
collapse" stage would do the trick.
Sprucing up is complicated, isn't it? Too complicated for a public proposal?
I don't want to ask you to write a long description of its advantages, if
you don't have a lot of time to, but, just summarizing briefly, what's the
nature of it's gain?
Could you tell me where I could find a complete definition of the
sprucing-up procedure, one that includes definitions of the terms used in
the definition, such as the names of special sets of candidates, etc.?
Mike Ossipoff
_________________________________________________________________
Don�t just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/
----
Election-methods mailing list - see http://electorama.com/em for list info
