At first the only FBC/ABE method we knew of was MMPO. Then MDDTR. But both had
criticisms which, though I don't recognize their importance, could still be
used against them in an enactment campaign. Other than those, there are my
conditional methods, and Chris's ICT. ICT avoids the co-operation/defection
problem by completing Improved Condorcet with a top-count. But the fact that it
uses a pairwise count makes it vulnerable to the offensive strategies of burial
and offensive truncation. As I said, that's mitigated by the fact that those
strategies can ony help a more favorite candidate. But that strategic
consequence of pairwise-count is still there. As I said, the top-count is what
avoids the co-operation/defection problem. But, when AOC uses
conditionality-by-top-count, one of thekinds of conditionality that I've
suggested, it's making use of the same solution. Let me repeat how that simple
kind of conditionalitly works:Give top rating to the candidates you approve
unconditonally. Give middle rating to those you approve conditionally. They'll
get an approvalfrom you only if they get more top ratings than do your
top-rated candidates. I prefer conditionality by mutuality, but what I
described in the previous paragraph avoids the kind of calculations needed by
mutuality-conditionality, while retaining optionality by candidate. All of the
methods that I call "conditional methods" are simply and closely based on
Approval, thereby having Approval's avoidance of many strategy problems that
other methods have. And the conditionality gets rid of Approval's remaining
problem, the co-operation/defection problem. Ordinary Approval is good enough.
We've discussed various ways of dealing with ABE in ordinary Approval. Forest's
diplomatic solution. Or publicilzed principled refusal of a compromise. It's
been shown that Approval will soon home in on the voter median, avoiding an
unliked Myerson-Weber equilibrium. The conditional methods have been objected
to by someone, on the grounds that he doesn't like coalition, or doesn't like
conditionality. 1. As I said, "... then is it important to you that you not
help the people whose help you want?"2. Any method that elects C in the ABE, if
B voters don't co-operate, is, itself, conditional, even if not explicitly. The
simple, direct, obvious and transparent explicit conditionality of the
conditional methods, used with Approval, and Approval-basedmethods such as MTA,
MCA and ABucklin, avoids the strategy problems of other methods, without adding
one. Of course all ballots-only nonprobabilistic methods have strategy. The
strategy considerations that the conditional methods have are not problems, by
that term's meaning in discussions of other methods. I'm only comparing the
conditional methods to other ballots-only methods. And, as I've said, the
conditional methods have the advantage of being obvious and simple improvements
on Approval. So closely related to Approval are they that AOC, MTAOC, MCAOC and
AOCBucklin can be offered together as options in an Approval election. Options
areuncontrovesial, and difficult to criticize. Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
