[EM] reply to Heitzig criticzing range voting

[Abd ul-Rahman Lomax]

At 08:49 PM 8/29/2005, Warren Smith wrote:

> >Do you suggest the election system should rather declare one of the
>candidates which are not approved by anyone the winner than to demand a
>new election because of lack of approved candidates. (I certainly don't
>agree to that.)
>
>--yes I do.  The job of a single-winner election system is to 
>produce a single winner.
>It is not to say "I refuse to do my job"  and then the voters try 
>again.  Assuming
>the voters produce the same votes for the same candidates (and why 
>should they not?
>can you produce examples from history when they have not?) the same 
>thing then
>happens ad infinitum.

First of all, there is an obvious assumption here: that the 
candidates for the second election would be the same as for the 
first. When run-off elections are held, I've never heard of it being 
the same election de novo. And even if it were, the conditions would 
surely be different, i.e., more time for arguments, and candidates 
and those who support them might well tailor their arguments based on 
the first election. While this would be no guarantee that the 
election would not again fail, if it is at all a decent method, 
failure is not likely.

So the run-off might be, as it normally is, for a subset of the 
candidates for the first. If a run-off is between two candidates, and 
no other votes are allowed, there will be a majority winner (if 
abstentions are not counted). A runoff will also bring voters out who 
did not vote the first time, especially if the election was close.... 
So a complete election method would include defining what happens if 
a runoff is necessary. Top two? Approval winner, Condorcet winner, if 
any, Range winner? And, of course, the runoff might have different 
rules, designed to guarantee a winner.

Now, as to the argument that it is "the job" of the system to produce 
a winner, the system is a cold and unintelligent machine. It does not 
have a job, it is not a responsible being, it is merely a mechanism, 
and artificial intelligence has in no way come to the point where we 
could "give it a task" and expect it to complete it well under all 
circumstances, unless we have designed it perfectly, something which 
is possible with simple tasks, and rarely with complex ones.

This is why the best election methods, in my view, incorporate 
deliberative process. Of those listed on the EM wiki, I only have 
noticed Mr. Smith's Asset Voting and the delegable proxy election 
process I contributed to the wiki. But, of course, elections in many 
bodies involve some deliberative process....

Alcoholics Anonymous has an interesting process by which delegates 
were (are?) elected. If no candidate gets a 2/3 majority, they 
continue voting. I think they might drop the lowest candidate. I 
don't know if they use Approval, they might. (My guess is that AA 
members would be averse to discarding ballots, I don't know if the 
ballots are written or not, I've never witnessed an AA election). If, 
after a certain number of ballots, no candidate has reached the 
required majority, the delegate is chosen by lot from among the top two.

It's interesting because it provides a certain level of 
representation for what might be a minority faction....

(AA is the model Free Association, and the Conference to which the 
delegates go is mostly an advisory body; the legal authority is in 
Alcoholics World Services, Inc., a board-controlled nonprofit. As I 
recall, board members *are* elected by the Conference, but only a 
certain number are alcoholics, the bylaws require a certain number of 
non-alcoholics -- it used to be a majority, but that was changed. And 
I don't know if the Board has to accept Conference-elected delegates, 
though certainly they routinely do. The true power that the members 
of AA would be to withhold the regular contributions on which AA 
World Services depends, since it is prohibited from building or 
maintaining more than a small reserve. And I'm not aware of any 
serious controversies between the Conference and the Board. But I 
might not necessarily know....)

> > 1. Can you explain to me the difference
> >between assigning 64 or 65 points to the middle candidate?
> > 2...there is no such thing as Util(A) or Util(B).
>
>--1. Why should anybody have to explain it?  Why should you have
>to understand it?  Why should anybody have to feel they understand it?

Well, perhaps if we are considering working for it?

>2. There is such a thing.  I have explained how to try to define it 
>in terms of
>money, or brain chemicals, or neuron events.  The latter (or even the former)
>may not be easily measureable. (And I did not claim it was.)  But 
>what we definitely
>know is a false claim is that there "is no such thing."

What Warren is calling a "thing" may actually be more than one thing. 
It hasn't been defined.

>3. Furthermore, some Condorcet advocates - including, I think, you recently,
>go too far in trying to deny utility, and thus cause grevious harm 
>to humanity.

I think that utility is a useful concept, but calculating utility is 
a hazardous enterprise. Sometimes it is better than nothing, 
sometimes it can be misleading. There are irrational factors....

>[...]
> >So you suggest that when candidate A gives $200000 to 1 voter and
>nothing to the other 99 voters, but candidate B gives $1000 to each of
>the 100 voters, then candidate A should be considered best for society.
>
>--YES!!  (at least, if utility=money.)
>
> >That's strange, isn't it?
>
>--NO!!    And in fact the very fact that money is fungible  (I 
>assume we both are
>allowed to ignore, or have already factored in, inflation...) makes it quite
>clear A is better - if it were non-monetary utility this would be less clear.

The example is backwards and might as well have been designed to 
confuse. In a rational use of money as a measure of utility, it would 
be the voters paying the society as a whole for the privilege of 
electing a candidate. So candidate A bribes 1 voter. What do the 
voters do? Presumably they make an offer stating the utility of the 
election of their preferred candidate. So the voter A, wanting to 
make a buck (literally)  makes an offer of $100,000 to elect 
candidate A. The B voters also offer $100,000 plus whatever personal 
value they assign to the election. Given the average tax burden of 
citizens, it might be well over $1,000 each. Candidate B was merely 
defraying their expense.

The B candidate wins. But let's presume that the value to B voters is 
$1,000 each, and the value to the A voter was 0, but he decides to 
turn over (i.e., bid) the entire amount, $200,000. So A wins. But the 
$200,000 is distributed to all the voters, giving each of them 
(almost) $2,000. Since this is worth more to them than preventing the 
election of A, they have all profited by the generous contribution of A....

If you are going to consider a money-value election process, at least 
make it a sensible one! Value in such an election is not merely a 
*stated* value, but an actual bid.

Warren, you have work to do. I suggest you stop wasting your time in 
pursuing arguments over details....