James--
As you said, S/WPO apparently has no pure-merit disadvantage with respect to AERLO & ATLO, and S/WPO combines AERLO &I ATLO in a generalization. Of course, in principle, it's always better to give more choices to the voter, though, as you suggested, S/WPO might not be as practical a public proposal.
Having AERLO & ATLO as two separate options seems more likely to be understandable to the public, as opposed to giving them a bigger and more complicated generalization that can be used as AERLO or ATLO.
You said:
I see two subtly different possible ways to count weak preferences: (1) The direction of pairwise defeats are determined by both strong and weak preferences, and the strength of pairwise defeats are determined only by strong preferences.
I comment:
Of coures that's using S/WPO as a simplification of Cardinal Pairwise.
(2) If there is a cycle, ballots are simply changed so that >> becomes >, and > becomes =, and then an ordinary winning votes tally proceeds from there.
That's the part that implements AERLO & ATLO, but is more general too. More complicated for the voter. I suggest AERLO and ATLO as ways of protecting the above-line candidates, in two different ways. One could apply AERLO & ATLO on the same ranking, of course. I'd expect that it would make the most sense to apply them both at the same point, to protect the same best-set of candidates, but of course each could be applied anywhere.
I have some reservation about S/WPO having both of your two meanings for the ">" and ">>", because the voter might just want to use one of those two options at a particular place.
If one is going to have a generalization like that, then why not go the rest of the way with it, and give the voter the option of which, or both, of 1) and 2) s/he intends to use at a particular ranking point.
One other thing. It's true that the mark of (otherwise) successful offensive order-reversal is an all-majority-defeats Smith set. So if AERLO or ATLO's only purpose were to thwart or deter offensive order-reversal, then it should only take effect if the cycle is a majority cycle.
But I'd use AERLO for another purpose, other than just anti-order-reversal: I'd use it to protect the best-set candidates even in the event of a natural non-all-majorities cycle.
So, if it had to be either yes or no, for the majority requirement, I'd prefer to leave out the majority requirement for AERLO & ATLO taking effect.
So, for instance, I'd word AERLO:
The voter may mark a line in his/her ranking, to indicate that if none of his/her above-line candidates wins, and if a circular tie contains above-line and below-line candidates, then that voter wants to promote all his/her above-line candidates to 1st place, and repeat the count.
[end of AERLO definition]
The ATLO definition is the same, except that it drops the below-line candidates instead of promoting the above-line candidates.
So, if specifying that that circular tie be an all-majorities circular tie has to be in or out of AERLO and ATLO, I'd leave that requirement out. Better to have AERLO useful in non-all-majority natural circular ties, even if that means having it take effect when it isn't needed for deterring offensive order-reversal.
Of course the ideal would be to give the voter the option of whether he wants the cycle referred to in the AERLO & ATLO definitions to have to be a majority cycle. Then, if the voter is only interested in order-reversal-deterrence or thwarting, and doesn't want to use it otherwise, then s/he could specify on the ballot that only majority cycles count.
But that's a complication, and I'd suggest, at first at least, just leaving the majority-cycle requirement out, and defining AERLO & ATLO as I defined them above in this message.
Mike Ossipoff
_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
---- Election-methods mailing list - see http://electorama.com/em for list info
