[EM] WV, Margins wrap-up

[MIKE OSSIPOFF]

WV vs margins:

All along, I've been referring people to the electionmethods website
for the defensive strategy criteria that justify my method arguments.
I might as well restate it here: http://www.electionmethods.org

Two people have pointed out that, though margins has strategy
problems that wv doesn't have, it's also true that wv has possible
stratgegy temptations that margins doesn't have.

This isn't surprising. Gibbard & Satterthwaite demonstrated that
, with any method, there can be situations in which someone can
gain from strategy. So it's a matter of choosing which strategy
problems you want, and which you most want to avoid.

For instance, comparing wv's strategy situation with that of margins,
which causes a lesser-of-2-evils problem? Which causes a gross
majority rule violation? Answer: Margins.

Additionally, wv's strategic temptation to sincerely rank 2nd choice
equal to favorite only brings benefit in a natural circular tie.
Margins' strategy problems can result in truncation stealing the
election from a sincere CW.

When, in wv, you incerely rank unequally your equally-preferred
lower choices, you're falsifying preferences.

But falsifying a preference in that way is like half of an order-reversal. 
Both wv & margins have order-reversal as an offensive
strategy. Rob wanted a method that deters order-reversal better than
Margins does. Ok, the wv methods do. With the wv methods, simple
defensive truncation is all it takes to make offensive order-reversal
backfire badly. If offensive order-reversal ever began to be used,
people would soon know about it, and if some were sophisticated enough
to use it, others would be sophisticated enough to deter it.
It would be impossible to organize a jurisdiction-wide order-reversal
without others hearing about it. And using counterstrategy.
Offensive order-reversers would usually regret it very much. After
a few tries, they'd quit using it.

I claim that offensive order-reversal won't be used on a scale
sufficient to change the election result.

Thwarting order-reversal in Margins requires order-reversal. Thwarting
truncation requires at least truncation. With wv, truncation thwarts
offensive order-reversal, and no one has to do anything to thwart
offensive truncation, because it can't steal the election from
a sincere CW or a member of the sincere Smith set, under the
plausible conditions described in SFC & GSFC.

The bottom-line of these strategy problems is: What kind of
strategy problems do they cause for you if someone else truncates
or order-reverses? Because truncation & order-reversal are the
only things that can threaten a CW in pairwise methods.

Ok, truncation causes no problem in wv, but it causes a problem
in Margins. In Margins, but not in wv, offensive order-reversal
requires order-reversal as a defense.

The strategy _needs_ that Margins imposes are the problem. The
unimportant strategy temptations that wv has aren't a problem.

In the poll whose balloting commences tomorrow, if you designate
a margins method, and if some people truncate, as I and many others
did in the previous poll, then there's a good chance that your
Margins method will, as a result, miss the sincere CW and
place your final Approval cutoff point lower than it should,
giving the election away to something that you like less than that
sincere CW.

Likewise, Margins can miss by giving one of your final Approval votes
to something outside the sincere Smith set, and less liked by you
than some member of that set, giving away the election to that less-liked 
alternative.

Maybe you like Margins better for aesthetic reasons, because it
looks more symmetrical, but I suggest that, when you designate
a method for Voter's Choice, you consider which method is going to
miss and give away the election.

And then ask yourself if you'd want, for public elections, a
method that you don't want to designate in Voter's Choice.

Mike Ossipoff








_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com