[EM] minimizing voter despair

[MIKE OSSIPOFF]

>This is difficult.  Numerically representing your level of satisfaction 
>from
>a number of different outcomes might be "honest" from your point of view,
>but without a standard that is followed by all of the voters, the honesty
>doesn't mean anything.

As Bramms & Fishburn use the term, and as I use the term, the only
insincere Approval vote is one that reverses a preference ordering.

>
>Say there is a polity of sincere voters, who would not ever dream of voting
>strategically.  This polity votes for their president using Approval 
>Voting.
>So far so good.  However, in our election for president out of ten possible
>candidates, one group of voters vote for all those candidates that are
>strongly liked (between one and four, becuase even this group has different
>standards of what "strongly like" means), one group of voters vote for all
>candidates that are reasonably acceptable (between three and seven), and 
>one
>group of voters vote for all candidates that are not strongly disliked
>(between five and nine).

That sounds a lot like strategy. Why vote for candidates just because
you don't strongly dislike them, or because they are reasonably
acceptable. If you're not voting strategically, and there are some
candidates whom you really like, doesn't nonstrategic voting mean that
you vote only for them? I don't think a nonstrategic voter would
vote for someone mediocre when there are excellent candidates, and he
certainly wouldn't do "hold-you-nose" voting.

Sure, we can't objectively measure & predict what excellence means to
different voters, but it seems to me to mean enough by itself: How
good does a candidate have to be before a certain voter rates him
excellent? He has to be excellent, for that voter.


>This does not include the voters that vote for
>exactly half of the running candidates every time

That's another strategy, the one used when no probability information
is available, and the voter is voting according to his ranking of
the candidates rather than his ratings of them. So again, that isn't
nonstrategic voting.

, or have invented some
>other system of voting.  All these voters are voting sincerely

By the most accepted meaning of the term, they're voting sincerely
if they don't reverse a preference. And, if they're voting for
mediocre candidates, candidates whom they don't dislike too badly,
when there are excellent ones in that Approval election, then they
aren't voting nonstrategically. Only strategy makes you vote for
mediocrity when there's excellence in the election.


, but because
>of a lack of common standard, the results could get quite arbitrary.

Approval is often accused of arbitrariness for that reason, but
the nonstrategic voter votes for whom he really likes a lot, and
the strategic voter votes so as to maximize his expectation. I shouldn't
repeat those expectation-maximizing strategies now, because I
discussed them in a recent letter. The result isn't arbitrary. It's
the result of those people voting in the way that seems best to them.
Different people have different reasons for voting for the number
of candidates that they vote for in Approval, but they're all good
reasons. Let's not assume that they draw a number from a hat.
For example, if there are 2 candidates whom you expect to be the
frontrunners, then you'll vote for whichever of those you like better,
and also for everyone whom you like more. That's pretty much the
same as saying that you'll vote for the candidate you'd vote for in
Plurality, and for everyone who is better. That's what people will do
if we use Approval for public elections. Of course I'm talking about
strategic voting now, because that's what people will do. But
nonstrategic voters voting for the excellent candidate(s) aren't
being arbitrary either.

I believe that Approval would work just fine for nonstrategizing voters.

Cardinal Ratings, with completely nonstrategizing voters, people
rating the candidates sincerely to the best of their ability, would
do somewhat better, by social utility. If I were a member of that
kind of a society, I'd suggest Cardinal Ratings as the voting system.

Strategy doesn't spoil Cardinal Ratings; it just turns it into Approval,
which is perfectly ok. Of course, since voting _will_ be strategic
in our political elections, then we gain nothing by starting with CR.
Why not just start win Approval instead, unless CR is more winnable.
When I say that, I like to emphasize that Condorcet would be my
favorite, but if Approval or CR is more attainable then I'd be glad to
get that.

Mike Ossipoff



The
>problem is worse with 'higher resolution methods', where one can really 
>only
>be said to be voting sincerely if one casts their vote relative to the
>preferences of other voters.
>
>Say a voter is relatively unnaffected no matter which candidate gets in.  
>In
>order for a higher res system to really work with sincere voting, this 
>voter
>is expected to know that other voters are more affected than him by the
>choice of candidate, he should keep all of his scores / reported utility
>outcomes on a similar level.  This is unrealistic, but more importantly,
>undemocratic (as is our ideal society voting with approval), because
>everyone's votes do not count the same - they are not equal.  While I'm not
>one to jump up and down about how fantastic democracy is in its own right,
>it would be safe to say that, as a general rule, it is useful to maintain
>political equality amoung citizens, not least because we have no way of
>safely evaluating the differences in utility outcomes between individuals,
>so the best default is to assume that they're all the same (ie assume that 
>a
>person having their first choice elected is as utility positive as any 
>other
>person having their first choice elected).
>
>Approval is an acceptable voting system because it minimises the inequality
>of voters, especially given that the strategies available to the voters are
>simple to understand, and balance out the inequality even more.  Higher
>Resolution methods carry all of the disadvantages of approval, without any
>of the mitigations.
>

_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com