[Election-Methods] a story for solstice

CLAY SHENTRUP clay at electopia.org
Fri Dec 28 20:09:22 PST 2007

Regarding Rob's response to Jan..

You make an analogy about being able to lock up your things, which if
we complete the analogy to voting methods could only be taken to mean
"prevent your personal utility from being taken by strategic voters".
Of course, all deterministic voting methods have this problem.  Warren
Smith's social utility efficiency calculations take strategic
exaggeration into account, and find Range Voting to be comparatively
robust against this problem.

> there is no way in Range voting to prevent others from defecting by voting Approval style.

So with 100% strategic voters, Range Voting is equivalent to Approval
Voting, which is still superior to feasible alternatives.  And it's
obvious that a substantial number of people will _not_ vote
strategically with Range Voting, leading to an improvement in net
social utility compared to Approval Voting.  This is discussed in more
detail here:

You point out that there are externalities that these simulations do
not account for, including long-term effects.  I agree.  Many believe
that Range Voting will have additional effects such as:

* breaking up bipartism
* decreasing the importance of money in elections
* increasing voter turnout

That last point is based on a look at some scientific research into
factors affecting voter turnout:

You speculate that some aspects of Range Voting would cause indirect
effects which would lead to a long-term negative impact on society.
But it seems clear to me that, beyond being purely speculative, your
underlying logic is invalid.  Take for instance your hypothesis that
RV would cause voters to want to stay home, to avoid feeling the
"conflict" of showing up to vote, and having to decide between a
sub-optimal honest vote, or a guilt-inducing optimal insincere vote.
For voters who care more about maximal effectiveness than honesty
(regardless of any associated guilt), they will show up to vote
strategically, which doesn't cause as much harm with RV as with most
other methods.  For those who want to prefer honesty, they can simply
vote honestly.  I fear your implication is that they would feel an
uncomfortable sense of regret by casting a sub-optimal vote, and that
by staying at home, they could avoid that regret.  But by staying at
home, they'd be casting an even _less_ optimal vote!  This example is
emblematic of what I see to be your typically self-refuting rhetoric.

>To drive the point home: the one and only way to defect in the analogy is to
>take stuff from others' apartments, which is the analog of voting Approval
>style under Range.

Or to bury candidates with Condorcet or IRV or any rank-order method.
Strategic behavior is a strong point for Range Voting, as its
superiority over alternatives (e.g. Condorcet, IRV, Borda) _increases_
with a more strategic electorate.

> In both cases (Range voting and semi-communal housing complex), some people
> will FEEL dishonest if they act in a certain way.

Then they're free to vote honestly.  And those "naive" honest voters
will be better off in Range Voting, than they'd be as strategic voters
with most other voting methods.  So your argument doesn't make sense.

> A range ballot, by virtue of not being an Approval ballot,
> implies to many that you "should" vote sincerely rather than strategically.

A ballot with Nader on it rather than just the two actual contenders
implies that  you "should" vote for Nader, if he's your favorite.  So
should we remove third party candidates when they have no chance, to
prevent voters from foolishly voting honestly?  If people want to vote
Approval-style with Range Voting, that's fine.  They can do it.
There's no catastrophe except in your imagination.

>RE: Representational democracy ..
>Strength of preference DOES apply, but it does so in a way
>that is not asking people to de-weight their preferences out of the kindness
>of their hearts. (if you prioritize, say, welfare more strongly that other
>issues, you do so at the expense of your ability to express your opinion
>strongly on other issues)

You're wrong.  Range Voting has an element of revealed preference in
it, aside from any altruism.  For instance, if A B and C are
approximately equally likely to win, and you like B more than the
average of A and C, then you should "approve" A and B, otherwise only
A.  You are strategically "forced" into admitting something about your
compariaive support for B, and there's no way you can out-game the
system to hide that.  It's like I've pointed out to you many times,
voters can strategically exaggerate in _any_ deterministic voting
method; the issue is how severe the resultant problems are.  Range
Voting mitigates the effects of strategy, period.

And should someone choose to vote sincerely with RV, either out of
ignorance or the kindness of their hearts, that is a good thing, not a
bad thing.  It increases the net/social utility.

>here is always the possibility of considering two options to be
>equal, in which forcing a preference is not appropriate.

Well, no.  Once you hypothetically consider X and Y to be "equal",
then in an instant, as soon as even one little molecule in your brain
changes, you then destroy that equilibrium.  That is, if I have two
objects that have the same mass, then in an instant, that will change,
as soon as one of them, say, sheds an electron.

>But if a voter has a preference, my idea of a "fair" voting system
should expect them to
>express it as strongly as possible.

"Fair" is ultimately a very nebulous concept.  It stems from a
traditional belief in equality, which doesn't translate to social
choice - since there's no way to equally distribute candidates like
slices of cake.  I can't get one slice of Ron Paul while you get one
slice of Barack Obama.  Elections leave use faced with an assortment
of heterogeneous options like "give Brian 2 slices of cake, and Jane
1.3 pieces of cake", or "give Brian 1 piece of cake, and Jane 3 pieces
of cake".  The utilitarian philosophy essentially boils down to "pick
the option which gives the greatest number of pieces of cake"
(breaking the analogy to encompass the fact that cake has diminishing
marginal value, and utility does not).

The crux of your entire framework is to not accept that inability to
evenly distribute candidates like cake, and so rather than accepting
this and seeking to pick the cake-maximizing option, you instead keep
focused on the even distribution model, and try to impose it upon a
fundamentally different area - the election input rather than the
result.  But the result is the whole _point_ of the election.

Doctors certainly understand the basic premise that option X is not
necessarily better than Y, just because X pleases more people than Y -
intensity of preference also plays a part.  If a doctor can either
treat 2 patients with scrapes, or 1 with a broken arm, we know which
one most any person will consider the more sensible option.  Triage
works in order to do the most good for the most people.  The "best
outcome" is not necessarily majoritarian.  And if traditional common
sense isn't enough to persuade you, then there's the basic reductio ad
absurdum indictment of the majoritarian principle.

That's a proof, not an opinion.  It rules out majoritarianism
(Condorcet efficiency) as the correct way of measuring the quality of
a voting method.  That leaves plenty of other options, but I do not
know of any that can compare to additive utility (maximizing the sums
of individual utilities).  Many others have been discussed, and all
have internal contradictions or obvious failings.

But whether or not you accept the additive social utility function, it
is clear that the "equal power" concept fails on multiple levels, both
from traditional ethical standards, and axiomatic logic.  Harsanyi
made what is perhaps the best argument of all when he noted that it is
in the individual's best interest, when uncertain of his identity, to
maximize the net utility of society.  His "expected value" is that net
utility, divided by the number of voters.  So even if you cannot agree
from the perspective of a benevolent god, you can at least agree from
the perspective of a self-interested voter who has no problem seeing
other voters better off as well.

Clay Shentrup
clay at electopia.org

More information about the Election-Methods mailing list