[EM] range voting mind-bogglingly stupid? (Reply to Rob Brown re voter strategy)

[Warren Smith]

>rob brown:
>"Deluded" is certainly a word that comes to mind regarding the suggestion
that people will, in significant numbers, choose to reduce the strength of
their vote to some non-zero value below the maximum possible strength they
are allowed.

>I don't even know how to make this argument, it just seems common sense.  It
seems about as useful an option as having, on your tax return, an option to
reduce the amount of your refund.  I suppose its ok to have the option, but
if you actually think it will make a significant impact, I'm saying
"deluded".

--let me make two replies.
First of all, mathematically,
it simply is not always the case that strategic range voting is the same thing
as approval voting, see examples in
   http://math.temple.edu/~wds/crv/RVstrat1.html
In other words, even for a voter with full knowledge of the utilities
of all the candidates, casting an approval style vote may be a bad idea strategically
and it can be UNIQUELY BEST for the voter to cast a range-style vote involving
one or more scores intermediate between full-approval and full-disapproval.

However, this mathematical point probably is not important in large real elections;
in them I suspect the optimal approval-style range vote, in fact,
as strategically good as the optimal range vote, up to negligibly small
utility sacrifice.

Second.  Rob Brown relies on his alleged knowledge of human psychology to argue 
people will not range vote, they will approval-vote.  Therefore, he argues, it is
silly to allow them the option of range-voting.  But in fact, so far
there have been at least 3 people (WDS, Jan Kok, and Lomax) here who
have expressed support for RV and/or stated that they have intentionally
gone with a weak vote, i.e. in Lomax's case intentionally abstaining from
voting on an issue he felt he had little knowledge about.  (I have done the same,
e.g. in every judgeship election I have ever voted on so far.)  That sample
suggests the fraction of people who will behave that way IS "significant."
And I have conducted an actual exit poll using over 100 real voters.  The fraction who voted
approval style was (with high confidence) below 25%.  So, we conclude from
this that Rob Brown's intuition is experimentally incorrect.

The question is: should we dismiss high-confidence experimental facts, figuring that
Rob Brown's psychological understanding trumps any amount of reality? 
I recommend another course.

So:  Why is Rob Brown so much smarter than over 75% of the US voter
population?  It is an interesting question.  I speculate that a large part of the answer
is that in range voting, your strategic incentive to dishonestly exaggerate
about the candidates can be very small if the candidate in question is not a "front runner."
And I mean VERY small.  Not 0.00001.  More like
0.0000000000000000000000000000000000000000000000000000000000000000000000000000000000001.
In that case, the natural human desire to be honest often trumps the
minuscule desire to get more power, causing the experimentlaly observed behavior
that approval-style ballots are rare.

On the other hand, one's desire to exaggerate to full-approval or full-not is
larger in the case of front runners because then it really is 0.00001.  
Our data supports this hypothesis:
voters experimentally often gave very high scores to Kerry and zero to Bush or
vice versa, but intermediate scores were common for third party candidates.
It was much more rare for a ballot to give a high score to a third-party guy,
and an intermediate score to the top major.  So there surely was dishonest
strategy going on, but not in a candidate-symmetric manner.

Let us make a similar analogy to Rob Brown's example of an option
on a tax return to allow voters to contribute more taxes than they owe (and there
actually is such an option) - he said this would be rarely used.
Well actually, when it comes to range voting for non-major candidates,
a better analogy is: having an option on your tax return where
you get to contribute $1 more than you owe, but that magically has the effect
of contributing $9999999999999999999999999999999999999999999999999 to support
your country.  If such an option were available and possible (which it isn't)
I suggest to Rob Brown, that it actually *would* be commonly used.  That is
exactly the sort of effect range voting gives you.

The result of this all is an experimental effect called the "nursery edfect",
which is very important.
   http://math.temple.edu/~wds/crv/NurseryEffect.html
It is this effect which enables range voting to have a decent chance to actually get 
enough dedicated support to actually get enacted.   Rob Brown, if he supports, say,
Approval Voting or Condorcet Voting, will not have the Nursery effect working for him.
He will therefore not be able to get a dedicated large group of
suporters from third parties.  That is because  (and now I rely on MY commmon-sense
knowledge of human psychology) those third parties will not be interested in supporting
a voting system that experimentally gives them an order of maginitude fewer votes
than range voting does.  My common-sense knowledge also suggests to me that if 
Rob Cannot attract the support even of third parties (those most moticted to get
voting reform) then he cannot get his voting system enacted.

So if Rob wants an imporved voting system, I suggest he support Range Voting.
Go to the CRV web site
  http://math.temple.edu/~wds/crv
and click "join" (and/or join the RV bulletin board) to get started.
wds