[EM] Comments on MJ discussion

[Michael Ossipoff]

My favorite method, Symmetrical ICT, fails IIAC. I don't claim that
IIAC failure is an unacceptably, unforgivably deplorable thing.

IIAC compliance is desirable, but there can be more desirable
properties chosen instead. Approval's and Score's IAC compliance
counts for them as an advantage.

If you lose IIAC, only lose it in order to get something more important.

I don't think that IIAC-failure is as much of an outrage as
Mono-Add-Unique-Top failure. ICT and Symmetrical ICT fail that one
too, but that's ok because it's traded for CD and Condorcet
Compliance. Though traditional unimproved Condorcet meets its form of
the Condorcet Criterion, its failure of FBC and CD make unlikely the
sincere rankings with which Condorcet Compliance can mean something.

Advocates of traditional unimproved Condorcet claim that, with it,
ranking will be sincere. But I suggest that we shouldn't advocate
methods whose advocacy depends on an assumption that people will vote
sincerely when it's clearly contrary to their best interest (based on
their perceptions, assumptions, beliefs and information, whether
correct or incorrect).

But it still makes more sense to pass IIAC than to fail it, and that
counts as an advantage for Approval and Score. The fact that Smith has
been deleted from the ballots shouldn't change the relative standing
of Jones and Doyle (even if you'd change your vote in a subsequent
election without Smith). If it does, then the contraption (rank
method) is doing something idiosyncratic and unintended, as
complicated contraptions can sometimes do. It reminds us that our rank
methods _are_ contraptions, and aren't acting ideally as we'd most
like.

Mike Ossipoff