[Election-Methods] Simple two candidate election

[rob brown]

On Dec 21, 2007 7:41 AM, Jan Kok <jan.kok.5y at gmail.com> wrote:

> Hi Rob,
>
> I congratulate you for focusing on some questions that underlie the
> endless debates about which is the "best", "fairest", etc. voting
> method.
>

Cool.  glad to see someone gets where I'm coming from. :)


> To restate the questions my own way: What do we _mean_ by "best",
> "fairest", "most democratic", etc. Is there some standard (criterion,
> figure of merit) that we can all agree upon for evaluating and
> comparing voting methods?
>
> It appears that, so far, there is no widely agreed-upon definition for
> "best", fairest", "most democratic", etc. that can be used for
> comparing voting methods. Nor is there some "gold standard" that we
> can all agree upon for comparing methods. I had thought there were
> only a couple commonly held standards, but I see from this thread that
> there are perhaps four or more:


Agree.


> The Majority Criterion: roughly speaking, the majority of voters get their
> way.
>

Personal opinion: I don't find this one at all compelling. (note that I find
majority works fine in a two person election, but I don't think that
generalizes to elections with more than two candidates. In fact I think it
is next to meaningless in some cases ....an example being my "vote for a
number" scenario, where there are essentially an infinite number of
candidates)


> Social Utility: pick the winner that maximally benefits society, or
> gives maximal overall voter satisfaction, etc.
>

Let's make sure we are in agreement that this is "short term satisfaction
with the results alone", and doesn't bring into it all the other (intangible
and possibly long term) factors that can satisfy people, such as feeling of
fairness, lack of conflict between feeling like you are "cheating" or being
a "sucker", etc.

My discussions with Clay got ridiculous because he would include these
intangible things under social utility so that he could say "it is a fact
that social utility is the 'correct' thing to judge a voting method on", and
then he would not include them when he did his "proofs".  In other words, he
changed the definition (dramatically) to suit him.

In any case, I think social utility -- in the restricted first meaning -- is
one of many criteria, but not the only one and probably not the most
important one.  But that's my opinion.


> Equal voting power: all voters have equal influence over the election.
>

This is, to me, a fundamental concept of democracy and one of the top
criteria.  It directly contrasts with "social utility", since it by nature
has to ignore strength of preference.

Minimize or eliminate the need or temptation for voters to vote
> strategically.


This is an extremely important one to me as well.  It overlaps a bit with
the "equal voting power" one (if strategic voting provides benefit, that
means non-strategic voters have less influence over the outcome than
strategic ones).  But they are ultimately separate issues.

I think there are other important criteria as well.  My biggest problem with
plurality is its tendency to cause parties to form.  I think this
polarization, which is of course a long term effect, is one of the most
destructive things in society.

I think a system that tends to elect centrist candidates is better than ones
that elect candidates that bring out strong feelings of a particular
faction.  All the moreso in places (Iraq comes to mind) that are civil war
prone.  I would argue that the differences between north and south in the US
could have been resolved peacefully, but for the polarization caused by
plurality (in fact the election of Lincoln, who was hated in the south, was
the specific reason the south seceded...but it went way beyond that)

Again, this overlaps with the strategic one, but they are not one and the
same.

Instead, I'll say a few words to promote my view that Social Utility /
> Overall Voter Satisfaction / "Maximum Net Tangible Utility" (I like it
> :-) is the "best" standard.
>
> Preface: I don't think there is a way to "prove" that any of these
> standards are "correct" or "best", using mathematical or philosophical
> arguments.


That's good to hear.  I agree.  Clay does not agree, and demands they are
proven, which drives me insane when trying to discuss these things with
him.  I have had the same experience with Warren and Abd.


> I have NOT chosen voting methods
> first, then chosen criteria that justified my choice of voting method.
>

I think that is a good policy.


> Ok, here are a couple of scenarios:
>
> (Inspired by Clay Shentrup.) Consider an election where the choices are:
> A. Spend $10 billion to find a cure for the common cold.
> B. Spend $10 billion to find a cure for multiple sclerosis.
> (Assume equal high probability of success for either choice.)
> Presumably a lot more people suffer from colds than from MS, but MS is
> devastating, while colds are generally just a nuisance. So, if voters
> are selfish and shortsighted (and "honest" about expressing their
> preferences), option A might win with a simple majority vote.


Not trying to be difficult, but actually I happen to be of the opinion that
A should maybe win.  Maybe if I ever encountered someone with MS I would
change my mind, but I don't think its unreasonable to think that more
suffering (and lost productivity) is caused by colds.

But....I get your point.


> But,
> let's assume that option B gives better social utility. Wouldn't that
> be the better outcome? A method such as fine-grained Range Voting,
> which allows voters to express preference strength, would have a
> better chance of making the "right" choice, if voters who are not
> afflicted with MS and not worried about getting MS vote "honestly"
> their weak preference for option A. On the other hand, if the A voters
> strategically exaggerate their preference for A, then A would win.
> That's no worse than the majority rule result.


But no better.  I do see what you are getting at.  However, I think in the
absense of some sort of weighted voting (where people can save up their
voting points to "spend" on the issues they care about), democracy simply
fails to achieve the outcome you want for such decisions, and Range voting
does not solve this, it only distracts from solving it.  Representative
democracy works better at dealing with this sort of thing, because a
candidate who wants to spend more on MS research would tend to get the votes
of those who really care about MS, which takes into account strength of
preference (since the candidates presumably have many other things on their
platform that voters will balance).

Regardless, I think range voting would work fine for such elections, on one
condition:  that humans were eusocial animals like bees are.  But we're
not.  Pretending we are doesn't solve the problem, it just distracts from
finding a solution

(see http://karmatics.com/docs/collective-self-interest-fallacy.html , which
has a similar situation where a solution which doesn't account for
tendencies toward Nash equilibria distracts from finding a solution which
actually works).

Here's another, similar scenario. The choices are:
> A. All people pay income tax in proportion to their income.
> B. People whose last names begin with A through F pay twice as much as
> they would in option A, and the rest pay $1 less than they would in
> option A.



> A simple majority vote would choose B.


Well, again, I think representative democracy is the way we address such
things, not direct voting on individual issues, which simply breaks down on
things like that, Range voting or not.  If a candidate put such a thing on
his platform, every single A - F voter would show up at the polls, and vote
his ass out. And then they'd probably stop by his house on the way home and
burn it to the ground.

G - F voters wouldn't care as much, and would tend to concentrate on other
issues on the candidates' platforms.

Also, remember in such calculations of utility, the "tangible utility" is
not the only factor.  For the winners, it might be.  For the losers, in
addition to the financial hit of having to pay more taxes, they have the
resentment of being treated unfairly. And this would not be matched by an
equal positive utility on the part of the people who profited by this
scheme.  Humans have a deep seated sense of fairness, that simplistic
"tangible utility" calculations tend to ignore.

(incidentally, even monkeys have this sense of fairness which goes beyond
tangible utility: http://www.primates.com/monkeys/fairness.html )


> Range Voting
> with honest voters is more likely to choose A. But if the G-Z voters
> strategically exaggerate their preference for B, then the result is
> the same as majority vote - no worse.


I have a hard time thinking that people who would vote for something that
unfair, would then "de-weight" their preferences just to be nice (really,
think about that).

More likely, they'd think "plan B's f*cked up" (recognizing that it would
cause a huge amount of anger and societal strife that doesn't make their $1
savings worthwhile) and vote for A.

I understand what you are getting at, but whether you use your extreme
example or a more realistic one, you have the same problem:
you are assuming that people will be selfish in which they vote for, while
at the same time altruistically de-weighting that vote.  That doesn't
compute for me.

Would voters vote "honestly" with intermediate values, thus
> voluntarily "weakening" their vote under Range Voting?
>
> Warren D. Smith's 2004 exit poll study showed that a surprisingly high
> fraction of the respondents (about 75% if I recall) voted with other
> than max and min values.


Sure, they might well do so in the first few elections.  Until their
candidate loses, and they realize that some on the other side had more
"voting power" because they voted with max values.  They would feel
suckered.  Then they'll say "screw this being nice crap" and from then on be
strategic voters.

I highly doubt Warren tested this long term effect.  Or even thought much
about it.  In Warren's world, if you can't assign a hard number to it, it
doesn't exist.

The fact is, a system that expects people to vote with "less than full
strength" is about as far from a Nash equilibrium as you can get.  Does he
really think that, over time, it isn't going to gradually move toward that
equilibrium?  That people will put up with being suckered time after time?
He might want to talk with some of my friends who voted for Nader in 2000
(they will say "Ok, no more 'voting my conscience'. Strategy from here on
out").

To be blunt, Warren doesn't seem to have much concept of human nature, in my
opinion.  The simple fact that he argues (
http://rangevoting.org/ApisMellifera.html ) that range voting works for
eusocial animals (whose Darwinian interest is the colony rather than the
individual), so therefore it should work for people....tells me that he's
been looking at numbers long enough that he's lost his ability to do common
sense.


> I personally have voted on bylaws or platform issues at political
> conventions, where I wanted to cast a weak vote rather than a strong
> yes or no vote. The reason was that I had only a weak opinion, and
> would have preferred to let those with strong opinions have their way
> - but if nearly everyone had only weak opinions, then I might as well
> register my opinion and participate in the decision.
>

I can appreciate that, but i think people's likeliness to "de-weight" their
vote decreases as the number of voters gets higher.  A single voters vote
already has microscopic effect on the outcome.  Will people really want to
reduce that further?

-rob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20071221/be2fd8b6/attachment-0003.htm>