On Mar 26, 2012, at 2:44 PM, MIKE OSSIPOFF wrote:
Dave:
On Mar 24, 2012, at 3:49 PM, MIKE OSSIPOFF wrote:
Approval can't be improved upon, other than questionably and
doubtfully.
You wrote:
This is a bit much, considering that there are many competing
methods that offer various worthy capabilities.
Looking at the ABucklin that you mention:
Assuming that I wish to elect A, but want to have B considered ONLY
if I cannot get A elected:
. I cannot say this with Approval, where I must give equal
approval to every candidate I approve.
. With ABucklin I can give B a lower rank than I give A, to be
considered only if A's rank does not decide on a winner.
[endquote]
Yes, and I don't deny that Abucklin's improvement can be desirable.
In fact, if our voting system now were Approval with the
options that I've been describing, I probably would use the MTA
option, or, better, the MTAOC* option. The only reason why
I wouldn't use ABucklin or AOCBucklin is because there wouldn't, for
me, be many levels of candidate-merit. Under different
circumstances, I might use ABucklin or AOCBucklin.
*(We've been talking about how the conditional methods have a
ridiculous secondary defection strategy. I'd use the
MTAOC option anyway, because I don't think that people would use
that ridiculous, counter-intuitive, and potentially
disastrous defection strategy. So, while 1st-level defection is
discouraged,
there might well not be any 2nd level defection. Let me just add
that, because I only suggest AOC and MTAOC, etc., as _options_, the
appearance of complexity of the conditionality-implementation
software code isn't an acceptance problem, because
everyone will know that s/he needn't use it. An _option_ for
managing one's Approval voting power isn't a problem. Anyone's
voting power is his/her own, and if s/he chooses a complicated way
of managing it, that isn't anyone else's problem.)
What did you say?
On or before the 24th I wrote of ABucklin based on a partial
definition of it since I could not find anything complete and solid.
Here I read of what must be collections of methods:
. Deciding on implementing would require decisions on ballot
format and counting rules.
, To be a voter would require much of that - and if vague or
incomplete would properly inspire complaints.
It isn't that Approval can't be improved on at all. I'm just saying
that voting system reform advocates often have (in my perception)
an exaggerated impression of _how much_ Approval can be improved on.
For instance, though I like ABucklin, and it's one of my favorites,
it isn't perfect. Improvements and refinements of Approval
don't bring perfection. Maybe you rank one of the acceptable
candidates in 3rd place, because you want to distinguish between
the merit of the various acceptable candidates. But then, in the
count, someone gets a majority when ballots give to their 2nd
choices. A candidate unacceptable to you wins because you ranked
that acceptable candidate in 3rd place.
Or maybe the opposite could happen: You give 2nd place rankiing to
B, and 1st place ranking to A. No one gets a 1st place
majority, and so all the ballots, including yours, give to their 2nd
choice. B then gets a majority and wins. But A would
have gotten a majority in the next round. Or maybe A and B both got
a 2nd rank majority, but B got a bigger majority
than A did. A would have won if you hadn't ranked B. Of course that
can happen in Approval, and, in fact, of course
ABucklin makes it less likely. My point is merely that it's still
possible.
Yes, I know that ABucklin offers something that Approval doesn't
offer. I'm just saying that it doesn't _always_ prevent
accidentally giving the election away to a 2nd choice. And you can
regret not voting Approval-style. Probably some
improvement--I'd use the multi-level MTA or MTAOC--but not the
perfect improvement that some expect.
And, whether in Approval with options, or in Abucklin, the person
voting an Approval ballot has simpler strategy
(though he/she has to of course be willing to forgo the multi-level
nature of ABucklin or MTA).
Of course ABucklin adds MMC compliance, and I value that.
Bottom-line: Improvement, yes. Perfect or complete improvement, no.
I suggest offering improvements, such as the options of AOC,
ABucklin, AOCBucklin, MTA, MTAOC, etc., maybe delegation,
sometime after
the enactment of Approval. Especially if there's considerable talk
about wanting something fancier than ordinary Approval.
Which leaves me promoting Condorcet. It allows ranking but, unlike
ABucklin or IRV, all that a voter ranks gets counted. Further, any
voter able to match their desires to Plurality or Approval for a
particular election, can vote by those rules and have them counted
with the same power by Condorcet rules.
Dave Ketchum
Mike Ossipoff
Election-Methods mailing list - see http://electorama.com/em