Brief Replies to Tobin's Letter

Mike Ossipoff dfb at
Tue Jul 9 00:26:30 PDT 1996

This is just a brief note, to reply to some of the
things in Tobin's letter. Tobin brings up a number of
new issues, arguments & suggestions, and I'll reply 
tomorrow to anything I might miss in this brief reply

With Condorcet's anti-order-reversal strategy, the object isn't
to punish order-reversal, but rather to deter it. It sounds like
the same thing, but it isn't. Because if it's successfully deterred,
it won't have to be punished, because it won't happen--precisely
because it obviously would be punished if it did take place.

It's true that the Clinton voters don't have a strategyk
against order-reversal that ensures that Clinton will win
even if order-reversal happens. But that isn't a fair
thing to say, because no method has a strategy that would
ensure victory no matter what under conditions like that.
Certainly not IRO or Copeland or Regular Champion--or
any method.


You mentioned that Smith//Random doesn't tempt order-reversal
because it only has a 1/3 chance of successs in a 3-candidate

But the thing is that order-reversal isn't the real problem
anyway in pairwise methods: The real problem is trunctation.
Order-reversal on a scale sufficient to change the election
result is improbable in a public election, especially considering
that the intended victims would be sure to hear about it in
advance, and would be ready to deter it! And especially since
it's so well deterred in Condorcet.

And in every rank-balloting election that I've conducted or
participated in, there's been much sincere truncation. That's
the problem. Smith//Random does nothing about that. In
contradistinction, plain Condorcet & Smith//Condorcet
never let truncation gain election for someone over whom
a majority have ranked the Condorceet winner. Truncation
isn't a problem in Condorcet, but it's a big problem in
every other pairwise method.

You said that there's only a 1/3 chance that the order-reversers'
candidate would win in the random choice of Smith//Random. But
there's also only a 1/3 chance that the order-reversal would
result in the election of someone whom the order-reversers like
less than the candidate from whom they're trying to steal victory.

Since there's not a thing that anyone can do to thwart the
order-reeversal in Smith//Random, except for the Nader voters
to rank  Clinton in 1st place, or for the Clinton voters to
use a probabilistic strategy to make sure that no one knows
which extreme will beat the other, and since there's only a 1/3
chance that order-reversal in Smith//Random will backfire in
a 3-candidate race, I say that order-reversal is less well-deterred
in Smith//Random than in Condorcet.


Mike Ossipoff


More information about the Election-Methods mailing list