[Election-Methods] Elect the Compromise

[Abd ul-Rahman Lomax]

At 02:32 AM 8/28/2007, rob brown wrote:
>On 8/27/07, Abd ul-Rahman Lomax 
><<mailto:abd at lomaxdesign.com>abd at lomaxdesign.com> wrote:
>At 01:07 AM 8/27/2007, rob brown wrote:
> >On 8/26/07, Abd ul-Rahman Lomax
> >Social animals and eusocial animals are totally different
> >things.  Worker bees, by virtue of their anatomical design and their
> >behavior, will give their own lives for the sake of the hive (they
> >sting and die).
>
>Whereas humans won't? What planet do you live on? It's *instinctive.*
>
>
>Uh, yeah.  Everyone here is motivated solely for the good of the collective.

I've now read Mr. Brown's little document on his web site, and I know 
understand better where he is coming from. He is not listening, he 
has an idea fixed in his head about what we are trying to say, 
already, before we say it, and it is not what we think or how we think.

Look at what I said above, and his response. What I did was to point 
to a very common and very human response. Not everyone will do it, 
but when an airliner crashed into the Potomac river, bystanders 
jumped into the freezing water to rescue people. They rescued many, 
and one of them himself died. This kind of action happens over and 
over again. Not always, not everybody, but many. Does this mean that 
the person who jumped into the river and died was "motivated solely 
for the good of the collective?" No. He probably didn't think about 
the "good of the collective." He saw people who needed help and he jumped in.

Does this mean that we can just sit back and trust that people will 
spontaneously act to prevent the tragedy of the commons (which is 
what his document is about, previously mentioned in this thread)? *Of 
course not.* People are complex, we are mixtures of self-interest and 
community connection. We have complex selves; we identify who we are 
and thus "us" in complex ways; in some situations, "us" is only our 
immediate family, in others, it is our co-workers, in others, it is 
our ethnic group, in others, it is our town, in others, it is the 
entire human community, and, occasionally, it is the entire community of life.

These allegiences conflict with one another, often. So which one 
predominates at any time is not something we can predict with ease. 
However, some general principles do seem easy to apply. People will 
not generally make a great sacrifice for the common welfare unless 
the action is clearly likely to succeed, and even there such actions 
would be spotty. However, people will routinely, most of them, make a 
small sacrifice for a larger net community good; this is even more 
likely if the beneficiary can actually be seen, it is less likely if 
the benefit is diffuse and abstract.

>   In fact, I think its lovely how here, on this planet, no one 
> tries to game the system or find loopholes or manipulate things to 
> their advantage.  If they did, people would, say, do things like 
> take advantage of the openness and cheapness of the email protocol 
> to send unsolicited commercial messages to people.  But luckily no 
> one does that...

Isn't sarcasm in the service of arrogance and ignorance wonderful?

People will try to seek advantage. Advantage for what? For 
themselves. But *what is the self?* It is not a fixed thing, we have 
overlapping concepts of self. At one moment it might be my, myself, 
and I. At another, my children. At another, my town. At another, the 
members of a mailing list. At another, every person on the planet. 
Not all people experience all of these, but healthy people experience 
most, at least occasionally.

*We want people to seek advantage for themselves through the election 
method.* We do not want them to act "altrustically," for the election 
methods that we promote are seeking to discover the common welfare, 
by amalgamating individual welfare as expressed by the voter. If the 
voter thinks that the voter benefits by some manner of voting, *this 
is what we want them to do.* They are providing the information we 
seek, which is what they think best for themselves.

Mr. Brown apparently has no understanding of this at all. It's not 
that he actually disagrees, for you can't actually disagree with 
something you do not understand. He *thinks* he understands, he has 
us pegged as "those naive, wide-eyed people who believe in innate 
human goodness and lah-te-dah everything will be rosy and we will all 
live happily ever after since we all love each other and always act 
for the common good."

No. We don't always act that way. A naive approach to the tragedy of 
the commons is not going to resolve it. The little story Mr. Brown 
tells, though, is quite warped. I would guess from it that he has 
never lived in a healthy small town where people have a sense of 
ownership of the place. I have, though only recently. It is not what 
I expected. Until last year, I lived for a few years in a small New 
England Town Meeting town, with what remains of direct democracy in 
the U.S. Quite simply, those roads would be fixed. They aren't 
libertarians, but if they couldn't raise the taxes -- there are 
limitations on what you can do with taxes in Massachusetts -- they 
would find another way. The amount of money involved in the story is 
piddling, and townspeople would be ashamed to leave the roads doing 
such continual damage when they could fix it for pocket change.

But, of course, this town does tax its residents.

I'm *not* a Libertarian. I propose voluntary associations to 
ameliorate a lot of problems, and it is possible that some voluntary 
solutions could replace some coercive governmental solutions, but I 
consider this far from proven and would prefer to see it actually 
working before dumping existing governmental solutions. In my view, 
libertarians should wake up and smell the coffee and start to put 
libertarian solutions to work to prove the point. If it works, people 
will use it. There are plenty of examples where there is nothing 
stopping us but inertia, governmental interference is not stopping 
us, taxes are not stopping us, etc., etc.

Where libertarian solutions, in my opinion, are exactly appropriate 
is with voluntary organizations, peer associations, and most 
specifically those concerned with collective judgement. It turns out 
that there is a lot of quite old political theory to back this up. 
Judgement is contaminated by power, give direct power to a judging 
body, and it loses its ability to make free and unconstrained 
judgements, and thus it loses, at least to some degree, intelligence. 
Intelligence, to be maximized, requires freedom.

And it is intelligence that I'm looking to enhance, not some 
political position or specific cause. Aggregative methods are, in 
general, pretty bad at this, they are shortcuts that do not take 
advantage of the very powerful deliberative processes which developed 
over millenia, they are devices developed to deal with problems of 
scale and secret ballot, and the cost of the efficiency and 
protection they bring is far greater than we normally realize. But of 
the election methods, Range is the closest to what one would get from 
standard deliberative process. And it can be improved with certain 
hybridizations, most specifically by incorporating a runoff stage if 
there is an apparent failure of the majority to consent to a result.

But deliberative process is far better, for it has no restrictions on 
outcome, it is adaptive and preferences shift as information and 
arguments are shared. The only serious problem with it -- with an 
even somewhat informed assembly -- is scale. It becomes impossibly 
tedious and inefficient when the scale gets too large -- which really 
happens well before a few hundred people, though certain devices can 
save the day for a time. The U.S. House is way too large, but the 
committee system makes it somewhat functional.

However, contrary to the common wisdom among political scientists, 
not to mention the public, the problem of scale is soluble, and the 
solution has been staring us in the face for centuries, waiting for 
someone to notice. As it happens, four different people, in different 
parts of the world, came across and began writing about a solution 
over the last decade; this would be delegable proxy, which some have 
called liquid democracy, I coined the term fractal democracy, which 
some like, and the Swedes used the term "advisor" for the proxy, 
which expresses the outward communication function of the proxy. 
James Armytage-Green also calls it delegable proxy; I think he might 
have gotten that from me, but I'm not certain. And there are others 
beyond these. Warren Smith came up with an election method, especally 
good for Proportional Representation, that he called Asset Voting. 
It's delegable proxy in disguise. It turns an election process into a 
deliberative one, by a quite simple device; votes are tranferable by 
the one who received them.


>Seriously, what planet are YOU posting from?

Earth. Where we live in societies and are normally highly dependent 
on one another. We have complex social systems that are not as simple 
as Mr. Brown thinks. We think like those we live among, far more than 
we think differently, but this may not be visible to us unless we 
have an opportunity to experience something very different.

We are free-living, i.e., we have the capacity for independent action 
and we have a sense of self that is sometimes quite contracted and at 
other times is quite inclusive. All of this gives us great 
flexibility and power, but also we are vulnerable to what might be 
called diseases of our adaptations. We are rational, that is, capable 
of rational analysis, but we are also instinctive and intuitive, 
often to degrees that we don't recognize, because we rationalize why 
we do things, we tell ourselves stories that we believe.

Without meaning in our lives, most of us would die fairly quickly, 
whether by disease or suicide or "accident." And what supplies 
meaning? I'm not going to answer the question here, at least not 
fully, but it is relevant, and one very common form of meaning is in 
society and social rituals, as was realized by Durkheim long ago.


>Yes, it's not exactly like honeybees. We have a far higher level of
>independence, but we are still social animals, with, *normally*,
>great concern for others.
>
>
>The difference is that humans reproduce directly, hence 
>fundamentally different Darwinian pressures on humans vs. the 
>non-reproducing worker bees.

Mr. Brown is revealing himself, quite a lot more than he might think. 
Nobody has here argued that humans are honeybees, but Brown was 
*contrasting* humans with honeybees by asserting that honeybees would 
die for the colony, whereas humans .... wouldn't?

To Mr. Brown, apparently, "I regret that I have but one life to give 
for my country" is mere propaganda, or perhaps how a man facing 
execution rationalized the situation to make it tolerable for himself.

We consider such people heroes. Most of us don't do what that man 
did, but most of us aren't called to. People go off to war and risk 
their lives for what they believe is right, the defense of their 
country, and that is pretty common. (It might be mistaken, too, but 
that's not the point here. The point is that people *are* willing to 
do what the honeybee does, die to defend the society. But honeybees 
don't think about it, they just do it. We think, and we make choices, 
and that means that sometimes we would do this, and sometimes we 
would run and hide, and sometimes we would do something else.

>The whole conception is off. Adaptation is driven by survival of the
>genes, not by survival of the individual
>
>Yes I know all about Dawkins selfish gene model, which supports the 
>concept of eusociality quite well, thank you.

You're welcome. I was not referring to Dawkins, I was referring to the obvious.

>However, like Dawkins, I'm not big at all on group selection, as I 
>think it is an extremely weak force.  And everything you say seems 
>to only make sense from a group selection mindset.
>Hey, if you are so selfish, what in the
>world are you interested in election methods for? For personal gain?
>There are much easier ways to find personal gain!
>
>And I am not advocating selfishness, in any way.  If you 
>misunderstand this point, read the article I linked.  If you still 
>don't understand it, read it again.

I understood it, though I will, indeed, read it again.

>I am advocating a system that does not give an unfair advantage to 
>those who DO behave more selfishly.

Is there something wrong with being selfish? What?

If I hand you a ballot and tell you, if you mark this ballot this 
way, we will be moved toward *this*, and if you mark it that way, 
*that,* tell us what you want, and you ask for what you want, marking 
the ballot in the most effective way to get that, are you being "selfish"?

I'd say you were doing what was suggested you do, and what is natural 
for you to do. And a good method will take this information and use 
it to further the common welfare. Now, what is the best way to do that?

*How* would you answer this question? By guessing? By unproven theory? By what?

Here is what some "experts" do. They come up with a list of criteria 
that a good election method will satisfy. Of course, this is 
circular. What we are trying to do is to figure out what makes a good 
election method! So how do we know what are good criteria to use?

Anyway, they end up with this list of criteria, and then, of course, 
since they have realized that the election criteria, particularly for 
ranked methods, conflict with each other, they cannot all be 
satisfied, they try to figure out which are the *important* criteria 
and which ones can be sacrificed. Sometimes it seems that the best 
method is deemed the one that satisfies the most criteria, even 
though that is obviously flawed, since there is no merit in the 
*number* of criteria satisfied; and people can and do make up new 
criteria to make a method look better, such as Richie's Core Support 
criterion for IRV.

There is something that actually is a metric, and it is well known 
and understood in economic theory, and that is utility.

Utility is often not measurable, at least not directly, but the 
concept is still useful nevertheless. For starters, it is possible in 
some situations to determine real utilities accurately, and we can 
then study how different election methods function in those 
situations. In all this voter behavior must be considered. Utilities 
do not automatically determine votes; rather, the assumption goes, 
voters have some kind of internal utilities assigned to alternatives; 
utilities are a quantification of preference, plus absolute utilities 
have something tying them to a common standard

Mr. Brown imagines that we think that voters will simply give their 
"sincere utilities" and thus everyone will live happily ever after. 
We don't think that. We do say, sometimes, that "If people voted 
sincerely in Range, Range would be the perfect voting method." But 
this is dicta. We don't think that people will vote sincerely, and, 
in fact, in real elections, to vote "sincerely" is quite undefined, 
people need some algorithm to translate their internal utilities to 
actual Range votes, and there is not only one algorithm, no single 
standard for what is "sincere."

But there are simple ways to turn a set of preference into a set of 
Range votes, and, it happens, these ways also can maximize personal 
utility. They are "selfish." But Range works quite well with them. 
And you can't do better than them without simultaneously, in most 
situations, risking loss. A balancing loss.

Contrary to what is often asserted, such an optimal Range vote 
procedure does not produce only approval-style votes. It can include 
intermediate ratings, and there are good reasons for voters to use 
those ratings, they are safer in some cases and at least harmless in others.

>   My assumption is not that all people act selfishly, or even that 
> most people will.  And I especially don't think people SHOULD act selfishly.

I do. But I also think that they should have a healthy sense of self, 
which is not confined, necessarily, to what is inside their own skin. 
Fortunately, most of us have this. Those who do not are literally ill.

>But if you design a system that rewards those who are the most 
>selfish by giving them the most power, I think that is a very very bad thing.

I would not dream of suggesting such a system.

There are a number of problems here. Mr. Brown is identifying the 
voting of strong preference as "selfish." But what if the voter has a 
strong preference? Is it selfish, then, for him to vote that way?

In real life, people negotiate with each other, and preference 
strength information is exchanged. You tell me that this is important 
to you, and I will try to offer it to you, so that you will give me 
what is important to me. If we both have the same strong preferences, 
and a limited supply of this thing, then we have a zero-sum game, and 
not much can be done. But zero-sum games are, fortunately, rare. 
Usually there is some way to maximize our collective benefit, and 
humans, given time and opportunity, are quite good at finding it, 
particularly if they have learned how to do so.

Preference strength information is critical if you actually want to 
maximize voter satisfaction. So what information do you use? Ranked 
methods assume, essentially, equal preference strength between ranks, 
which is blatantly distorted. If individuals exaggerate, in Range 
Voting, they get what they automatically get in ranked methods: a 
major loss of preference strength information, it is simply missing 
from a ranked ballot.

Mr. Brown wants to jump over the true question. Because he thinks 
that we must have "sincere votes" in order for Range to work, and he 
has never really examined what that means, he then thinks that Range 
will *not* work if people "exaggerate," i.e., "vote selfishly."

But, in fact, a "selfish" vote is simply an expression of strong 
preference. If you don't want to express that voting power, why would 
you do it? If you are exaggerating, you might get what you claim to want!

(Where this hurts is if you bullet vote for your favorite, you are 
simultaneously claiming that there is no difference for you with 
respect to all the other candidates. That is the effect of 100% for 
one and 0% for everyone else. So if you miss your target, the system 
will take you at your word. You don't care about which one of the 
others it will be, and so you have no power in the choice between 
them. You voted them all zero, they are thus considered all worthless 
to you, and so your satisfaction is moot, unless the voters 
collectively accept your favorite.)

Range Voting *constrains* preference to a range of one full vote. 
Ranked methods allow you to, so to speak, have your cake and eat it 
too: you are claiming maximimum preference of A over B, maximum 
preference of B over C, maximum preference of C over D, and all of 
these are a *full* preference. And that is impossible as an 
expression of how satisfied you'd be with the outcome.

Far from encouraging exaggeration, Range Voting *constrains* it.

I'd challenge Mr. Brown to show otherwise. Experts have tried, he 
should know, and so far, it appears, what they can come up with can 
easily be shown to be not what is sometimes claimed. If I vote a weak 
preference, *of course* someone else's strong preference may outweigh 
it. But if my less preferred candidate wins, *and I voted 
accurately*, I shouldn't mind. I had, after all, a weak preference. 
Be my guest. Suit yourself, I'm happy with this.

If you have a strong preference and I have a weak one, do I really 
care if you are lying? I'm not going to allow you to impose something 
on me contrary to a strong preference of mine, simply because you 
claim to have a strong preference. But if my preference is weak, why 
should I object?

Sure, if some small loss is repeated over and over again, I'd not 
like it. But what happens then is that my preference gets stronger!

>And as best I can see, that is what range voting is.  It punishes 
>people for doing what feels like the most "moral" choice, which is 
>to express their preferences honestly.

Bullet voting is honest, if you vote for your favorite. Multiple 
approvals are honest, if you are willing to accept them all under 
current conditions.

Now, suppose that I vote 100% for A and 90% for B. And a B partisan 
votes 100% for B and 0 for A. If B is elected, I'm getting a very 
good outcome! Why should I even care if the B voter was 
"exaggerating?" I *wish* I had seen elections turn out with such high 
satisfaction, it's been rare!

But was the B voter exaggerating? What's the basis for that claim? 
There were other candidates in this election, B presumably also voted 
zero for them, he was saying, by his votes, that they were all 
worthless, maximally bad. But thre is no standard for determining any 
particular vote as "sincere."

There is a naive standard, which would suggest that, since no 
candidate is perfect, I should vote less than 100% for all of them, 
and since no candidate is the devil incarnate, I should vote more 
than 0% for all.

Thus this "sincere vote" would essentially throw away my voting 
power. It's based on a misunderstanding of what voting is. Voting is 
a response to a question of choice. It's obvious when there are two 
candidates: the question is A or B. Not, how do you rate A or B on 
some absolute scale? Votes are "choice strengths," and Range simply 
allows you more flexibility than Plurality or Approval. It is 
perfectly okay not to use that flexibility, and it is not dishonest 
or sincere. Range never rewards truly dishonest voting, the only kind 
that can exist in ranked methods: preference reversal. It is, 
essentially, stupid.

Range is asking you to make a choice. Between what? Well, the best 
answer is "between the frontrunners." Other candidates should not 
influence your vote for the frontrunners, who are the only real 
candidates, by definition, in real public elections, the others can't win.

(People can, of course, err in identifying the frontrunners, but 
major failure in regard to that is rare in two-party systems, by definition.)

So you set your Range based on the frontrunners and vote max for one 
and min for the other, or perhaps *slightly* different than that. And 
then you place all other candidates where they fall given these 
ratings, never reversing preference, but perhaps allowing equal 
rating in some cases where you *do* have a preference. And if you do 
this, nobody is going to take unfair advantage of you and your vote. 
Absolutely, if your preference is weak between A and B, then someone 
else who expresses a strong preference between A and B might prevail. 
But we just stated that you had a weak preference. *You don't care*!

>Once again, I don't have time to read the rest of your lengthy post.

It takes far less time to read than to write....

>   If you are writing to others, or to yourself, fine, but I'm not 
> going to read and respond to each of your novels.

Good. We will all be better off. Or, at least, I will and others 
will. Maybe even you will. I am not the only fount of knowledge!

>   I found Jobst's challenge interesting and thought provoking, but 
> I get the impression he is getting frustrated as well with your 
> overly wordy illogic.

Jobst's challenge was interesting. Some of his comments about it were 
less than cogent. He has no understanding of utility theory, yet he 
used "ratings" to define the problem. Ratings abstracted from utility 
are almost meaningless, you cannot determine a just outcome from them 
alone. And I believe that I showed that fairly clearly, but, of 
course, if a tree falls in the forest, and nobody is listening, does 
it make any sound?