[EM] Why the concept of "sincere" votes in Range is flawed.

[Abd ul-Rahman Lomax]

At 05:26 PM 1/8/2009, Juho Laatu wrote:
>This is a reply to an old mail. (I didn't cover this well enough earlier)
>
>--- On Fri, 5/12/08, Abd ul-Rahman Lomax <abd at lomaxdesign.com> wrote:
>
> > >  One
> > > could say that any placement of the approval
> > > cutoff is sincere as long as it respects the
> > > sincere preference order of the voter. But one
> > > could also require that the approval cutoff should
> > > be placed on some "main gap".
> >
> > Why? I don't see any reason why that particular
> > position is better than any other. It is simply *easier*.
>
>If one's ratings are A9 B3 C2 then the
>"big gap" is between A and B, and it is
>very natural to the voter to approve
>only A. Doing something else might be a
>result of strategic thoughts.

The whole concept of strategic voting is flawed when applied to 
Range. Voters place vote strength where they think it will do the 
most good -- if they think. Some don't. Approval is essentially, as 
Brams claimed, "strategy-free,"  in the old meaning, and the only way 
that it was at all possible to call it vulnerable was that critics 
claimed that there was some absolute "approval" relation between a 
voter and a candidate.

Consider the example given; on some absolute scale, let's assume, we 
have A9 B3 C2. It seems to make sense that a sincere vote would be to 
bullet vote for A. But what if A is essentially an irrelevant 
candidate, the voter knows that A cannot win? Great candidate, far 
better than the frontrunners B and C. Given that B and C are, in the 
estimation of the voter, the only realistic candidates, then, the 
voter votes for both A and B. "Sincerely?"

What does sincerity have to do with it? It's a *choice,* a *vote*. We 
choose things based on our expectations, which include probabilities.

On the other hand, suppose that the ratings are A9 B8, C2. If all 
three candidates are viable, in the estimation of the voter, then 
voting for A and B makes sense, because the loss of utility from B 
winning is small compared to the loss if C wins. But if C is not a 
viable candidate, a looney, perhaps, the voter is confident that the 
vote for C won't even make 1%, *C is irrelevant.* What we really have 
is a two-candidate election between A and B, and the voter votes 
sincerely for the favorite.

It's claimed by some that Approval doesn't meet IIA. It does if the 
voters treat irrelevant alternatives as such! And if they don't, they 
aren't irrelevant in the ordinary meaning.


> > There is no absolute approval cutoff

In other words, it's not possible to define an Approval vote as 
"sincere" or "insincere," unless it reverses preference, which is 
insincere, and which Approval does not reward; at best it is moot.

>There are no "absolute" and measurable
>opinions. Much depends on what the voters
>think they are supposed to do.

An odd view, in my opinion. What the voters are "supposed to do"? Who 
is the sovereign who sets these requirements?

Here is what I'm supposed to do, as a voter: exercise my choice. 
Hopefully, it's an informed one, which includes understanding the 
voting system and how it works, the context, and the likely 
consequences of my action.

How I vote is *entirely* up to me. It's secret ballot, we can 
presume. I can take a pen and scribble on the ballot. I can vote for 
every candidate or none. I can vote for one or many. I can write what 
I like on the ballot, and nobody can do anything to me, there is no 
law against silly votes or nasty votes or, especially, selfish votes.

I can vote in a manner that I think will overall benefit society, or 
I can vote in a manner that will benefit me personally, hang the rest of it.

What happens to the ballot depends on the rules, but those rules do 
not constrain me, unless I want my vote to be effective, in which 
case I need to know them. I can use them as I choose.

In Open Voting, i.e., Approval, I can vote independently for or 
against each candidate, there is no connecting constraint. (Perhaps I 
vote for the candidate by marking a box on the ballot for that 
candidate, I vote against the candidate by default.) Essentially, I 
am putting a weight in the box of no candidates, one candidate, or 
more than one. The candidate with the most weight wins. It is an 
action, and the consequences of the action do not have any connection 
with my intention: if I have a "sincere" intention, or a "strategic" 
intention, it doesn't matter.

The big objection that someone like Saari makes to Approval is that 
it's indeterminate: you cannot just take a preference ranking of the 
candidates and predict, simply from that, how voters will vote. But 
we can, nevertheless, make some predictions, and Saari totally missed 
this, he imagines that voters will vote according to an insane "mean 
utility" strategy, which, of course, is highly vulnerable to clones 
and irrelevant alternatives. Now, if the "mean utility" is adjusted 
by probabilities, that's another story. But that's not what Saari did.

What voters will really do, in substantial numbers, is bullet vote. 
They will pick their favorite and leave it at that. And, in fact, in 
most elections, this works just fine! There are exceptions, and those 
who would need to add additional approvals generally know who they 
are. It gets difficult only in quite rare circumstances.

It's unpredictable. Quite like intelligence in marginal circumstances.

>  They may
>think that they should mark all candidates
>that they accept for the position, or they
>may be just told that they should mark
>candidates that they want to promote in
>the given competitive situation (based on
>the available poll information, expected
>winning probabilities etc.).

We tend to overemphasize "poll information." Every voter is a sample 
of the population, and the voter generally knows where this sample fits.

What's been really interesting in studying nonpartisan elections with 
IRV is how the supporters of candidate A appear to be, with respect 
to their preferences for B and C, quite like the overall population. 
If the overall population prefers B to C, A supporters will, with 
similar percentages, prefer B to C. I must say that this was totally 
unexpected; but the key is "nonpartisan." When partisan loyalties and 
prejudices are involved, it is quite different.

The Burlington election, in spite of having party affiliations on the 
ballot, was pretty much like this. I haven't looked at the lower 
choices of the Democrat and the Progressive, but the voters for the 
Republican candidate, in spite of the relative political positions of 
the parties, were about evenly split between the Democrat and the 
Progressive, with a slight majority of them voting for the 
Progressive candidate, i.e., more to the left. The Democrat was 
considered "Republican light" by the Progressives. She had been a 
Republican, previously. She was, in the legislature, a moderate 
Democrat, according to one source.

Burlington is not a very large town; it's even a bit surprising that 
the mayoral election there is partisan, many towns that size don't 
have party affiliations on the ballot.