Mike Ositoff ntk at
Mon Oct 12 21:06:58 PDT 1998

Sorry, I made an incorrect statement in a recent posting

I said that Runoff differs from IRO in not being able to
fail to elect a Condorcet winner that is everyone's exclusive
1st or 2nd choice.

No, Runoff can do that too, as can IRO. Though IRO is the
only rank-balloting method to be able to fail in that way,
it isn't the only method (of any kind) that can do it.

But my other statements regarding Runoff vs IRO are accurate.

By the way, when I pointed out that IRO, unlike Runoff, can
fail to elect a CW who has a plurality, someone could say:
"Sure, but FPP (the Plurality method) won't do that either--
does that mean that _it's_ better? In that respect, sure, 
but the methods that I advocate are significantly better
than Plurality in ways that many consider important, ways
that have been much discussed on these lists.

But I'd like to add that in Approval, if the CW has a plurality,
and more extreme voters don't vote for him, he still has a
plurality, and wins unless another candidate has his vote total
increased by other voters sharing. If the CW is a non-centrist
who's CW, then his votes, & those farther out, must add up
to a majority. Sure, if they all voted for a more centrist
compromise that they believed they needed, that could give
the election away. But I've talked about how such a giveaway
would require a very big mis-estimate, believing that the
opposite side has a plurality, when actually their own candidate
has a majority. Especially implausible if, in keeping with
normally distributed voters, the middle candidate isn't 
especially small.

And, in the pairwise count methods that solve circular ties
by counting votes-against in pairwise comparisons, the
only thing that can take victory away from a CW is the
devious & risky offensive strategy of order-reversal.

Mike Ossipoff

More information about the Election-Methods mailing list