Lesser of two evils

[Mike Ossipoff]

Yes, the lesser-of-2-evils axiom that you posted seems to me
a good concise statement of what we'd ideally like, when we
say we don't want there to be any lesser-of-2-evils problem.

But it's a stronger criterion than LO2E-2, and isn't met by
any of the methods proposed here. I believe it may well
be possible to design more deluxe methods that could meet
that axiom you stated, and they'd have to be perfect
methods, with regards to the lesser-of-2-evils problem. It
might be possible to devise 1-balloting sw methods that
are so near perfect that they can be said to meet that
criterion, but their definitions would be more involved
than those of plain Condorcet & Smith//Condorcet.

Just one thing, though: You left out something that was in
your previous statement of that axiom: the part that specified
that we're talking about voters _in M_ who rank some alternative
z over x. Or maybe that part isn't necessary.

Anyway, when it's this late at night when I write, I have to
disclaim that anything I say could easily be incorrect at this
hour.


Mike


--