Margins, majority, strategy

Mike Ositoff ntk at
Fri Sep 18 18:25:23 PDT 1998

> Simple Approval voting is defective.  If number votes were being used, a
> Condorcet winner can lose using simple Approval votes.
> Simple Approval voting is defective because it fails to distinguish among
> choices using number votes.

So Approval is defective because it isn't rank-balloting? If, to
you, rank-balloting is absolutely essential, then, for you,
Approval won't do. Again, you have a right to whatever standards
you believe in, and they aren't wrong, though they differ from

But it isn't clear to me what you mean when you say that a method
is defective. Not as good as some other method? Of course. I've
always asserted that Approval isn't as good as the votes-against
Condorcet versions we've discussed. Or inadequate in some absolute
sense? That depends on what you want.

Look, Demorep, the best rank-balloting count rules, for all practical
purposes, guarantee that you won't have strategic reason to rank
a less-liked alternative equal to or over a more-liked one.

Approval guarantees that you'll never have strategic reason to
rank a less-liked alternative over a more-liked one.

Therefore I claim that Approval is fully half as good as
the best rank-methods, in regards to strategic guarantees.


But it seems to me that, when it comes to the actual
democratic benefit gotten from that guarantee, Approval is much
more than half as good as the best rank-methods.

I claim that, just having the guarantee that Approval offers would
bring tremendous benefit, with respect to voter freedom of expression,
ability of candidates to enter the race without fear of being
spoilers, etc. Adding the further benefit of rank-balloting would
be better still, and of course I'd rather have one of our best
rank-methods. But the democratic benefit of the simple & modest
Approval reform would be nearly as good.

> Simple Condorcet voting is defective because it fails to have a YES majority
> Approval-type vote on each choice.

You're using that word "defective" quite freely. What you mean is
that Condorcet without Y/N voting isn't acceptable to you.

There are many times when we have to vote between several alternatives.
Absolute Y/N votes aren't really meaningful in that kind of election
(though they'd be fun). It's a _relative_ choice.

> The range of choices are--
> desired       compromise      opposed
> The desired choices of minorities will lose using YES/NO votes.  The
> compromise choices of such minorities will move up using the number votes.
> Probably a compromise choice will be a Condorcet winner.
> I note again that laws get passed by YES majorities in legislative bodies --
> and not by YES minorities.   I fail to see why YES majorities should not be
> required in a single winner election method to elect executive and judicial
> officers.

Let's not use current practices as a justification.

> Mr. Ossipoff apparently thinks that voting YES on one or more compromise
> choices along with one's desired choice(s) is somehow bad.   I would suggest

As I said, when you add Y/N voting you add a strategy problem.
You complicate a method that otherwise has simple strategy, an
and you add strategy problem to methods that are otherwise
virtually strategy-free.

Further, it complicates the count rule, and can cause embarrassment
when it vetoes all of the candidates. I talked about that
in the letter to which you're replying.

> How many real public elections will have 3 or more choices getting YES
> majorities (such as Kennedy, Carter, Clinton or Eisenhower, Nixon, Reagan or
> Kennedy, Clinton, Reagan) with such 3 or more choices in a circular tie ?

If you like Nader better than Clinton, do you give Clinton
a strategy "Yes" because you need him? Or maybe a strategic
"No" because you believe that Nader can win, and you want to
eliminate a rival. We don't need problems like that.



More information about the Election-Methods mailing list