Winning Votes intuitive?

[MIKE OSSIPOFF]

Rob Legrand said:

49:Bush>Gore=Nader
24:Gore>Bush=Nader
27:Nader>Gore>Bush

which should be equivalent to

49:Bush>Gore>Nader
49:Bush>Nader>Gore
24:Gore>Bush>Nader
24:Gore>Nader>Bush
54:Nader>Gore>Bush

I reply:

Why do you believe that's equivalent? In one case the Bush voters
express no preference between Nader & Gore, and in the other case
they all vote preferences between Nader & Gore. It's naive to say
that half the Bush voters voting Nader over Gore, and half voting
Gore over Nader is somehow the same as all of them voting no preference
between Nader & Gore.

For the purpose of determining whether Gore pairbeats Nader, of
course those 2 scenarios are the same. But for a method that
is interested in finding out how many people vote A over B,
it makes no sense to say that those 2 scenarios are the same.

The idea, with wv, is to go by the number of people who prefer
the pairwinner to the pairloser. That's the number of pairvotes that
will be overruled if we drop that defeat. wv is the count that
notices pairwise majorities. That's why, in your example, wv, but
not margins, enforces the majority for Gore vs Bush.

Your modified example changed the rankings in a way that changes
facts that are important according to the standards on which wv
is based. For the purpose of wv's standards, then, your modified
example is not equivalent to the one that it's modified from.

Rob contines:

But if I were one of the Nader voters (*shudder*) and a
winning-votes method were being used, I'd vote Nader=Gore>Bush.

I reply:

Nonsense. Doing so would make sense only if you rate Gore more than
2/3 of the distance from Bush to Nader. I suspect that most Gore
voters wouldn't rate Gore anywhere near that high. I certainly don't.
My sincere voted ranking would be: Nader. In other words, N>G=B.\

You'd often be well-advised to vote that way in RP(margins) too.
As you may know, there's no nonprobabilisitic voting system that
doesn't sometimes give strategic incentive. That includes RP(m) too.

The difference is that SSD rewards you for N=G>B in a few special natural
circular ties, while RP(m) requires you to vote that way in order
to protect a sincere CW's win or to enforce majority rule. A majority
can lose badly because they didn't use that strategy in RP(m).

No question about it, it's an embarrassment and a nuisance for wv
methods that it can give that strategic incentive. But it's not
a what I refer to as a defensive strategic need.

When, as is always the case in our elections, there is a set of
unacceptable candidates, I'd always, in wv methods such as SSD,
CSSD or BeatpathWinner, vote all the acceptables in 1st place.
In Approval, of course, I'd vote for (only) the acceptables.

It isn't possible for nonprobabilistic methods to improve on
Approval as much as you might like.

But I wouldn't do that in SSD because I'm defensive-strategically forced to. 
Incentive isn't necessarily the same as defensive strategic need.

That strategy situation should remind you that there are no
entirely strategy-free nonprobabilistic methods.


Rob continues:

Winning-votes
methods may discourage truncation, which is nothing more then tied rankings 
at
the bottom, but they strongly encourage tied rankings at the top.

I reply:

margins strategically forces truncation & order-reversal, as
defensive strategies. wv methods  can sometimes reward insincere
equal ranking in some special natural circular ties. There's a
difference in the degree of need for that strategy.

Rob continues:

In fact, I'd
vote ties in the top half or so of my ballot, even if I had *zero* 
information
about the other voters' preferences!

I reply:

You might benefit from it, or doing so might worsen your utility
expectation in the election. It depends on how you rate the candidates.
Strategy calculation for rank methods tends to be complicated, and
you don't have a formula for when your strategy is beneficial to you.
But it can be said that it's good if, to a large degree, the
merit difference between your shared-1st-place candidates, and your
other candidates is much greater than the differences among those
2 sets of candidates.

Rob continues:

"Margins" methods don't encourage tied
rankings like that on average, although of course there will be situations 
that
reward insincere voting in any method.

I reply:

No, margins doesn't "encourage tied rankings like that". It
defensive-strategically forces them, something that wv doesn't do,
at least not unless offensive order-reversal is a likely threat.
Even then, that offensive strategy can be deterred by mere
defensive truncation, in wv methods, getting rid of defensive
strategic need for ranking a less-liked candidate equal to or over
a more liked candidate whom one ranks.


I don't know how this issue came up again. I thought we covered
it before.

Mike Ossipoff


_________________________________________________________________
Join the world’s largest e-mail service with MSN Hotmail. 
http://www.hotmail.com