[EM] Approval Voting and Long-term effects of voting systems

[Michael Ossipoff]

On Mon, Nov 21, 2016 at 2:24 AM, robert bristow-johnson <
rbj at audioimagination.com> wrote:

>
>
> there is some truth mixed in with this, but some hooey mixed in also
>

.then I'll reply to some of it.


>
> I don't see any reason why IRV results in the dominance of the two major
> parties.
>

...except that IRV's national track record is one of 2-party dominance.


> The problem that IRV suffers is that sometimes IRV does not elect the
> Condorcet Winner and then voters feel that the election failed when a
> candidate is elected despite that more voters marked their ballots that
> they preferred a specific different candidate.  Then this cascades into
> other   problems like spoiler and later harm..
>

That well characterizes what happened in Burlington, but IRV doesn't
violate Later-No-Harm.  ...or Later-No-Help either.

IRV is a good method. But don't advocate or adopt it unless you understand,
accept and want its problem, and its unique offering and tradeoffs. IRV
isn't for everyone.

In Burlington, you repealed it because you didn't like its elimination of a
CWv. Then why did you adopt it??? You didn't know that it could do that?
Did FairVote assure you that it would never happen?

As with any voting system, IRV's advantages come at a price. In particular,
freedom from chicken-dilemma always comes at a high price. As I've said,
IRV's price for CD isn't more than the "going rate" for the price of CD.

IRV's problem is its ability to eliminate a CWv and its failure of FBC. I
wouldn't accept a method that fails FBC, unless there's a good trade for it.

 I prefer the relative certainty & solidity of FBC & Approval voting, for
my own voting, and also to reassure lesser-of-2-evils overcompromisers. IRV
gives up FBC, for a completely different approach.

Other than its problem, the fact that the supporters of a CWs have no way
to protect hir from elimination, and that other voters, too, can't protect
hir other than by favoite-burial, IRV brings some big advantages, CD in
particular, along with MMC, & the combination of LNHa & LNHe.

That problem has some mitigation, as I've described in previous postings:

If the eliminated CWv's voters transfer to someone worse, that's because
they're closer to Worst than to Favorite. If so, then maybe you aren't
losing much if Worst wins instead of the CWs.

And if your candidate & party are big enough to eliminate the CWs, then
they'll be well-known (if we had honest media). The CWs's voters will know
about your candidate. If you like the CWs, then hir voters are likely to
like your candidate, and will transfer your way.

And if you _don't_ like the CWs, and especially if s/he's in your strong
bottom-set, then you won't like the results of any voting-system.

I don't advocate IRV, and I wouldn't include it in a proposal. But if
others like and want it, then that's fine, because it has its own approach
to good results. I merely suggest that they familiarize themselves with its
problem, and be sure that they're ok with it, before they advocate, support
and enact it.   So that there won't be another Burlington.


> Setting aside the Condorcet cycle, none of this would occur with
> Ranked-Choice Voting with a Condorcet-compliant tabulation of votes.
>
I wouldn't give up FBC, except in trade for avoiding chicken-dilemma.
Ordinary Condorcet fails FBC.

Daniel was right, with regard to Condorcet: Lesser-of-2-evils
overcompromisers (will thnk that they) have reason to rank some dismal
compromise alone in 1st place, over their favorite.

If you want Condorcet, then at least use Kevin Venzke's Improved Compromise
(IC), which meets FBC.

Kevin has introduced an IC//MMPO, which meets FBC & CDalong with an
improved kind of Condorcet winner-choice.

When you top-rank 2 candidates, you don't do so in order to have one of
them beat the other and give the election to Worst instead. You top-rank
them both because you want one of them to win. The IC beat-definition
counts an equal top-ranking of X & Y as a vote against either beating the
other (which could give the election to Worst).

It's your vote, and counting it to your advantage isn't unfair.

....or use the IC beat-definition with your Ranked-Pairs. Then you'll at
least get FBC ...but use wv instead of margins. Margins gives no protection
from burial, or even from innocent, nonstrategic truncation.

But, without protection from chicken-dilemma, a rank method brings zero
genuine improvement over Approval, and can only be justified by a
psychological need for rankings.

[Replying farther down] :


> > This may be more surprising to
> > people, but looking at the underlying cause, it seems we can make a
> rather
> > important simplifying argument about most voting systems regarding this
> > problem. I would assert that the underlying cause of this problem of
> > two-party dominance in any voting system is that it gives voters the
> > ability to rank or order at least one candidate higher than the rest.
> >
> > The reason this ability to rank candidates becomes a problem is the
> spoiler
> > effect, where voters will have a strong motivation to give their highest
> > rank to one of the leading candidates. If they don't, then they weaken
> the
> > chance of that candidate winning and therefore strengthen the chance of
> the
> > less preferred leading candidate. Because one of the leading candidates
> is
> > likely to win, all the rest of the rankings of non-leading candidates
> > hardly matter at all.
>
> Again, this is a demonstrated problem with IRV, but not so much with
> Condorcet (sans cycle).
>

If you prefer Jill, but you believe that Hillary is worth electing and
better than the Republican, and you're told, and believe, that the winner
must be either Hillary or the Republican, then your best strategy in
ordinary Condorcet is to rank Hillary alone in 1st place, over your
favorite.  ...because of Condorcet's failure of FBC.


> > doesn't join one of these two major parties is excluded.
>
> Just not true.  And little evidence to support the claim.
>
> > So once voters and candidates figure this out, any such voting system
> ends
> > up devolving into the dominance of two major parties that we get with
> > simple plurality voting.
>
> Again, a claim made with no justification.
>
As I said, IRV's longterm national track-record is one of 2-party
domination.


> > In fact, one might argue that plurality voting is
> > better just because it is simpler.
>
> THAT does happen.
>
> > But Approval Voting avoids this problem.
>
> No, it just creates more trouble.  It's Approval Voting that devolves into
> Plurality Voting once voters start wondering whether or not they should
> Approve their second-choice candidate (or whether they should Approve their
> third-choice).
>
Approve only your favorite if s/he's the only candidate you like.

Approve only your favorite if you insist on voting tactically (based on
predictive information) and if it looks as if your favorite has a win.

Is that a  problem to you?

Ii suggest that if you're a progressive, then there are a large set of
candidates whom  you so much prefer to the alternatives, that your best
strategy is to approve them all.


Voters don't want their second-choice to beat their first-choice.  How do
> they stop that from happening?  (By bullet-voting their first choice, and
> that devolves Approval into Plurality.)
>

You're worrying too much about voting _among_ the best candidates. If
there's a set of candidates who are qualitatively the best, incomparably
better than the rest, then your best strategy is to maximize the
probability that one of them will win. That doesn't include voting _among_
them. Just vote _for_ them.

In any case, I don't recommend tactical voting (voting based on predictive
information)...partly because the corporate mass-media's predictive
information is disinformational, in the service of the people who "own" the
mass media.

...but also partly because it's most to your advantage to just approve the
whole set of candidates that's qualitatively, incomparably, better than the
alternatives.   ...to maximize the probability that one of that strong
top-set will win.

(But until we have verifiable vote-counting, of course the voting system is
irrelevant, as is the voting.)





> > Equal-rank approval votes mean
> > voters don't get the option to express their preferred ranking, but
> because
> > of that, they aren't at all motivated to bias their ranking dishonestly.
>
> That's baloney.  They may very-well not Approve of their second-choice
> candidate (whom they do actually approve of) to protect their first-choice
> candidate.
>
If they strongly approve of their 2nd choice, but not their 3rd choice,
then they'll approve their 2nd choice. In our actual candidate lineup, the
merit differences among the progressives are negligible in comparison to
the big gulf of a merit difference between them and the Republocrats.

Don't worry about it so much. Just approve all of the best.



>
> > They only have to decide which candidates to approve, or where the
> cut-off
> > is between approval and disapproval.
>
> No. That is a falsehood.
>
> They have to consider whether or not their second-choice candidate has a
> chance of beating their first-choice.  And they have to consider whether or
> not their third-choice has a chance of beating either their second-choice
> or their first-choice.  And then they have to decide if that risk is worth
> it to Approve their second or third choices.
>
No, they don't have to do any of that. They don't have to try to vote
tactically. Partly because there's no reliable information on which to base
tactical voting, and partly because their best strategy is just to approve
all of the set that is incomparably, qualitatively better than the other
candidates.

>
>
>
> Robert says:
>


I think most of your gushing for Approval voting is not well-founded.  What
if the voter prefers Candidate A as their favorite, but suspects that
Candidate B and Candidate C are the leading candidates?  How should that
voter vote with an Approval ballot? Please answer that question with
specifics.

[endquote]

We'd need better "specifics" about the situation that you've incompletely
described. It depends on how much you like A better than B & C, and how
much you like A & B better than C.

You don't mention any set's qualitative or incomparable better-ness,and you
suggest that there's good predictive information that the winner will be B
or C.

So, if you don't have a strong top or bottom set, and the merit differences
between your successive choices are uniform, and if there's reliable
predictive information that the winner will be B or C, then of course
approving {A,B} is a good way to vote.

But in our actual political lineup, the media have people believing that
there are 2 parties that always must and will win. Say your favorite is
Jill Stein, and that the other 2 candidates are Hillary & Donald. You're
told that only Hillary &I Donald are well-liked by everyone enough so that
one of those two beloved and popular candidates is sure to win.

For one thing, you should find something fishy about such a claim, given
what is known about those 2 candidates being the most despised presidential
candidates in this country's history.

For another thing, the merit of Jill is so much greater than that of
Hillary or Donald that, no matter what your'e told or believe about
predictive information, your best strategy is to vote for the only
acceptable, honest and progressive, genuinely pro-people candidate.

That latter interpretation of your scenario is the one that you probably
intended, because it's the one that actually obtains in elections here.

Michael Ossipoff











Approval Voting (and also Score Voting) requires *more* voter strategy than
> does Ranked-Choice Voting decided in a Condorcet-compliant manner.  IRV has
> problems (like a spoiler).  Of course Borda and Bucklin have weird problems
> that also motivate tactical voting.  But when there is no cycle, Condorcet
> Voting expresses exactly the will of the majority voters over all
> combinations because it satisfies one simple to understand property:
>
> "If more voters mark their ballots that they prefer Candidate A over
> Candidate B than the number of voters who mark their ballots to the
> contrary, then Candidate B is not elected."
>
> I still don't know how or why anyone would argue with that democratic
> value.
>
>
>
> Sorry, you Approval advocates have *never* made your case persuasive.
>  (And that includes Warren, whom I have a lot of respect for.)
>
>
> --
>
> r b-j                  rbj at audioimagination.com
>
> "Imagination is more important than knowledge."
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20161121/6c9bef72/attachment-0001.htm>