Margins, majority, strategy

Mike Ositoff ntk at
Mon Sep 14 23:05:24 PDT 1998


When I posted 2 Margins truncation bad-examples the other day,
I didn't accompany them with any comment, and so I'd like
to point out a few things about them.

In both examples, the A voters, by truncating, succeeded in
gaining the election of A, by defeating a Condorcet winner,
in violation of the expressed wishes of a majority.

In example 1, Margins, by electing A, elected the only
candidate with a majority against him.

That sort of thing is predictable for a method that ignores
majority wishes by not scoring by how many people rank the
defeater over the defeated.

To defend against that result, the C voters would have to vote
B equal to C, insincerely voting a less-liked alternative equal
to a more-liked one. A form of drastic defensive strategy.


Example 3:

100 voters.

 44  28  28
  A   B   C
  C       B

The A voters are using order-reversal against B. It succeeds,
using Margins, but is thwarted, in the votes-against versions,
by the B voters declining to list a 2nd choice.

True, a truly strategy-free method wouldn't even require
strategic truncation. But at least the votes-against methods
don't require voting a less-liked alternative equal to or
over a more-liked one.

What would voters have to do in Margins to thwart the order-reversal?

If the C voters ranked B equal to C, that wouldn't do it. A
would still win. To prevent the success of the order-reversal,
the A voters would have to rank B _over_ their favorite. They'd
have to insincerely rank B in 1st place, and put their favorite,
C, down to 2nd place.

That's what we don't like about Plurality (FPTP). How much of
a reform is it that retains that problem?


Unlike the Margins method, Approval never requires the defensive
strategy of voting a less-liked alternative over a more-liked
one. Unlike Margins, Approval never gives reason to not vote
one's favorite in 1st place. Margins isn't as good as Approval.
Somewhat better than IRO, but not as good as Approval.

The same can be said for Copeland. In fact, as I've said,
only a very few of the very best rank-methods are as good
as Approval. The Condorcet versions discussed on this list,
using votes-against are the only methods that are better than
Approval, in regards to degree of defensive strategy required.



More information about the Election-Methods mailing list