[EM] Election-Methods Digest, Vol 235, Issue 30

[Michael Ossipoff]

On Tue, Feb 20, 2024 at 00:10 robert bristow-johnson <
rbj at audioimagination.com> wrote:

> The unavoidable question is how much should a voter is how much should I
> score or approve my sec ond-choice candidate?  If I score them too high (or
> Approve them), I reduce or lose my vote for my favorite.
>
> What if it turns out that the race was most competitive between my top two
> choices?  Then I'll regret Approving (or high scoring) my second-choice
> candidate. Especially if my less-preferred candidate wins
>
> But then what if it's a competitive race between my second-choice and the
> candidate I hate?  Then I'll regret *not* Approving (or low scoring) my
> second-choice candidate. Especially if my less-preferred candidate wins.
>
> […]
>
But then we have to at least think about whether to Approve our
> second-choice. We don't wanna waste our vote approving our second-choice
> candidate and hurting our first choice.  But we also don't wanna harm our
> second-choice to defeat the candidate we loathe.
>
So the problem is that you don’t know whether your objectively-optimal vote
(the one that would maximize your utility over many similar elections)
would be to approve or not approve your 2nd choice.

No problem. Neither do the other voters, so don’t worry about it.

A rank-balloting method is an automatic-machine that receives your
preference-ordering as input, & then outputs an answer.

It does everything for you.

Fine, but, as Arrow showed, & likewise does the incompatibility between
Participation & the Condorcet Criterion, every rank method is a rather
imperfect contraption.

Additionally, every method other than Approval can be criticized as
arbitrary.

“Why this one instead of some other one?”

Approval is the absolute minimal method that allows & counts multicandidate
merit-comparison-expression (y/n ratings, in this absolute minimal method)
for a multicandidate election. …an obvious requirement for democracy.

…& being absolute-minimal makes Approval the unique completely unarbitrary
method.

… in addition to conferring an elegance possessed by no other method.

Approval is the simple, reliable handtool (as opposed to an automatic
machine). Yes, it’s  a different kind of voting, different from rankings.

You might want an automatic-machine to do it all for you, but that comes at
a price (in addition to the already mentioned ones):

Harder to define, describe, explain, propose, enact, implement, administer,
& audit against error & count-fraud.

I claim that those big disadvantages, those big losses of Approval’s
advantages, are much too high a price to pay for having it all done for you
by an automatic-machine.

Much better to just trust voters to be able to use Approval well.

…& that’s even before mentioning that Approval’s Myerson-Weber equilibrium
is the voter-median, where the CW is.

Achieving the objectively-optimal vote (defined above) with a cardinal
method isn’t feasible. It would require guessing lots of
difficultly-guessed probabilities, & estimating difficultly-estimated
numerical utility-quantifications.

Forget it. Forget objectively-optimal voting. That isn’t how Approval, or
any cardinal method works.

Other than speaking of maximization of overall utility over many
hypothetical repetitions of similar elections—objectively-optimal
voting—probability & expectation are subjective.

They’re based on what information you have & use.

A shuffled & cut deck of cards is on the table. I take the top card, look
at it, without showing it to you, & put it in my pocket.

I ask you the probability that the new top card is the ace of spades.

For you it’s 1/52. For me, it’s either zero or 1/51.

There are many ways to vote, & many ways of choosing how to vote in
Approval.

…& any one of them maximizes your expectation based on the information you
have & use.

This post is already too long, so I won’t go over all of them. But I listed
lots of them in a recent post to the Voting Science Forum.

But one is to just approve everyone you like. You thereby maximize your
probability of the election of someone you like.

Approve whom you feel like approving, what feels like the best vote. Based
on what you know & feel, it is.

You get the idea.

Which feels like a worse risk: that your 2nd choice will take the win from
your 1st choice, or that your last choice will take the win from your 2nd
choice m. Go with your feeling, if that’s your best information.

Maybe you have a feel for who the 2 frontrunners are. Fine. If that feeling
your best
Information then go with it & approve down to the better frontrunner. (…but
don’t let the mass-media convince you that the 2 frontrunners are a greater
evil & a lesser evil.)

Better than expectation: Maybe you perceive an expected merit for the
outcome. …a feel for which candidates would be a pleasant surprise if
elected.Then you can approve those  above-expectation candidates, which
will raise your expectation based on the information consisting of that
feeling.

If you have information about the likely CW, & that’s your best
information, then approve down to hir.

I don’t claim to have listed here all the ways to choose how to vote in
Approval, but you get the idea.

Approval is a completely different kind of voting from ranking, inputting
your pairwise preference feelings into an automatic-machine that will do it
all for you.










----
> Election-Methods mailing list - see https://electorama.com/em for list
> info
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20240228/abe8890e/attachment.htm>