[Election-Methods] Elect the Compromise

[Abd ul-Rahman Lomax]

At 02:59 AM 8/26/2007, rob brown wrote:
>Vote trading generally means the ballots can't be secret, so 
>elections would be inherently corruptible by anyone with money.

This is commonly assumed. But it probably is not true. First of all, 
the ballots don't have to be personally identified, all that is 
necessary is that the winner be known. If I'm agreeing to pay $X if A 
wins, I'm paying for a result, not for a vote. Who do I pay it to?

Well, remember, I only have to pay if I get what I want. Do I really 
care? But how much moneh would it take to pay off all the voters who, 
perhaps, register for the payment prior to the election. (To collect, 
they would have to be on the list of those who actually voted.) 
Normally, a lot of money.

What is offensive about vote-buying is the secrecy of it, and that 
only some people get paid when all suffer the outcome.

In the situations where these payments make sense, though, it would 
not be a large amount of money *per voter*. In any case, I'll leave 
it to someone else to work out the details. Suffice it to say that it 
is *not* plutocracy.

Too much money is involved!

>   Not good.  And I wouldn't think it would be ok given your problem 
> description, which is for a single election.

It's okay if the voters accept it! And if they don't, they don't get 
paid! (Even if they accept the offer and the candidate doesn't win 
that someone was willing to compensate for, there is no payment. But 
probably no loss, either.)

Essentially, the given situation was one where all voters would agree 
that C was a good compromise. (I argued that we could not exactly 
tell this from the ratings, but, with certain assumptions, this would 
be the case.)

The B voters gain very substantial value from the election of C, 
whereas the A voters lose only a little. Thus it makes sense for the 
B voters to compensate the A voters, the majority, for giving up 
their majority rights, to accept a choice of lesser value to them. 
And it would make sense for the A voters to accept the compensation 
that shifted the *net* utility to them such that C becomes their 
optimum choice.

There are other forms of utility shift that could be employed. C 
could make certain promises, could agree to include, as an example, A 
in his government. A, as the representative of the A faction, could 
agree on their behalf to accept some compromise which not only 
improves overall social utility -- which ultimately benefits everyone 
if maximization is general policy -- but also at least preserves 
utility for the majority. (Asset Voting could have this kind of result).

>   But....I suppose if we were able to talk about mulitple 
> elections, where a voter can earn "credit" for compromising which 
> can be spent in later elections, you could build that into the 
> system in a way that doesn't require losing the secretness, and 
> would solve this problem nicely.

Perhaps. It could get very complicated very quickly, and it results 
in the past binding the present, a problem we already have in many 
ways. If the voter has "credit," by the way, how do we know to apply 
the credit? Talk about complicated!

>Obviously range voting would solve this problem perfectly, if only 
>humans were eusocial animals -- most of us being sterile 
>worker-people, whose only Darwinian interest was the good of the collective.

We are far more that than seems to be realized. Not sterile, to be 
sure. But very much social animals. Most of us, most of the time.

>   Sadly, we're not, so range voting is (in my opinion)  best left 
> to bees and the like. ( 
> <http://rangevoting.org/ApisMellifera.html>http://rangevoting.org/ApisMellifera.html)

The argument here is that, because Range Voting is a perfect solution 
if people vote "sincerely," which is conveniently left undefined, it 
is therefore a bad solution if they vote "selfishly."

There is no basis for this claim. Simulations don't confirm it.

It's ironic, really. Range Voting *to some degree, not completely,* 
collapses to Approval Voting, under certain conditions with allegedly 
"selfish" voters. This is a feature, not a bug! Approval Voting is 
quite a good method!

And it is quite easy to improve Range in ways that should encourage 
more use of intermediate votes, which does, quite likely, improve 
performance. But even without these little touches, and with 
"strategic voting," which in Range only means what I call truncating 
the ratings, not preference reversal, Range still does very well. By 
the only real performance measures we have, so far: the simulations.

(Election Criteria are generally *not* performance measures, they 
don't generate "measures", rather each one is either satisfied or 
not. And evaluating methods by looking at a list of criteria and 
seeing how many are satisfied is obviously quite defective, for some 
criteria are, I think we would agree, more important than others. How 
do we know which ones are most important? It's quite subjective. 
However, there is a criterion that could be called the Social Utility 
Criterion. How come we never see that in lists of Election Criteria? 
It is really the only one that is associated with a *measure*.)