[EM] Does Top Two Approval fail the Favorite Betrayal Criterion [?]

[Abd ul-Rahman Lomax]

At 02:46 AM 6/13/2013, Kristofer Munsterhjelm wrote:

>Finally, I'd like to say that I do understand that reality is a lot 
>less neat. What Abd says about differences in turnout in the first 
>and second rounds of a runoff means that criteria are not as useful 
>as for single-round methods because the votes in the different 
>rounds would change.
>
>One could even argue that if they don't, there's no reason to add a 
>runoff to an advanced method, and the only reason for Plurality to 
>have a runoff is to patch problems in Plurality itself. I have seen 
>reasoning of this sort from some IRV advocates who both say "top-two 
>runoff is also nonmonotonic, so don't go around saying TTR is better 
>than IRV" and "IRV is better than TTR in every way because it's 
>clearly better than the contingent vote".

"IRV advocates," in general, do not value an actual majority choice, 
and have left centuries of democratic traditions entirely behind.

Plurality in nonpartisan elections is based, my suspicion, on a 
*simulation*. That is, the leader with a plurality will *usually* go 
on to win a repeated election seeking a majority, and if not being 
the best, is at least the second best, perhaps good enough.

Partisan elections are then based on an expectation that most people 
will be divided into two major parties.

Voting systems activists have, in general, demonized Plurality, when 
it, *in context and usually* works much better than might be 
expected. However, it has obvious breakdowns when there is a 
three-party system and partisan elections.

We know this about runoff voting: the election is a "comeback 
election" about a third of the time. I base this on a FairVote study. 
What should actually be done, I don't recall if they did it, is to 
look at the *margins* in those elections.

IRV, *in nonpartisan elections*, almost never is a comeback election, 
and the exception I saw was with a *very close* margin. I don't 
recall clearly, but there might have been an additional example, 
which was an election where the ethnic identity of the candidate was 
clear from the name, and thus this was a *different kind* of 
*partisan* election. Ethnically partisan.

(For those who don't know, a "partisan" election permits party 
affiliation of the candidates to be on the ballot. Voters may thus 
choose candidates based on that, which is a more stable preference, 
and more predictable, having little or nothing to do with the 
candidates themselves. I.e., we might assume that a Green Voter, in 
an IRV election, might then rank the Democrat second. The problem in 
Burlington was that the wider second choice was *eliminated* before 
the Republican, and many Republicans did rank the Democrat second, 
because the alternative was the Progressive candidate. Hence Bucklin, 
in that case, could easily have elected the Democrat *with a 
majority.* Someone tell Burlington that they can have their majority, 
probably, they do not need to be content with 40%, and their present 
system can do quite the same thing as IRV. It didn't in the last 
election, back to TTR/40%, because no Progressive candidate ran! 
FairVote has been strangely silent.)

IRV activists generally will not mention that TTR produces different 
results than IRV. However, they hint at it by claiming that runoff 
elections are "less representative," completely ignoring that many 
runoffs are held with the general election. Further, "less 
representative" results are commonly accepted in democratic 
organizations, as long as every qualified voter had the right to 
vote. They are *not* worse choices, in general. Rather, those who 
vote represent those who care enough to vote -- it's an effort! -- 
and thus results are probably *more generally acceptable.*

When a *major choice* is presented in a runoff election, people turn 
out to vote in droves, even if it's a special election. Lizard vs. 
Wizard, and a similar French election, a boring centrist vs a 
right-wing extremist.

This is the problem with FBC: it is interpreted *strictly*. I.e, *if 
there is any possible scenario where a voter might have an incentive 
to "betray the favorite," and no matter how small the incentive, the 
method "fails," the criterion becomes functionally useless. There are 
methods which obviously and routinely fail FBC. There are methods 
which don't, period, or at least no scenario has been proposed.

Now, Range does not fail FBC. However, Range has a problem. It can 
declare a winner when the majority of voters oppose that result. By 
the nature of range, that opposition is not at full strength, 
obviously, but it is a *lack of consent* to the result. Further, 
because of normalization error, the Range result, even with votes 
considered "fully sincere," can fail to be *actual utility 
optimizing.* Hence, Smith found that Top-Two runoff Range had better 
BR results than ordinary Range, with voters who vote with ordinary 
strategy. I.e, they normalize. They might also bullet vote or expand 
ranges, but, in general, the application of voting strategy to voting 
systems has been fairly naive.

Nevertheless, repeated elections represent iterated consideration and 
can be expected, in generaly, to be superior to instant amalgamation, 
we *must* make this decision NOW. That is the operation of the 
amygdala, the lizard brain, it's been called, and it's a necessary 
human response. I.e., to avoid being eaten. There is no time for 
iterated consideration, there is action based on the *immediate 
state* of *immediate responses.* No back and forth, no discussion, no 
presentation of an option and weighing, back and forth. Warren has 
claimed that Range Voting is used in nature, he often cites honey 
bees, but he hasn't stated the half of it. We use Range Voting, 
internally, all the time, as a "preponderance of reactions" based on 
stored experiences and instinct.

So ... I can expect that *any voting system* will improve with 
iterated consideration. The paradoxes and flaws of voting systems can 
readily be seen as resulting from the demand for an immediate result. 
People sense this, and runoff voting is popular, which is 
particularly significant because of the cost.

So, if any runoff system fails FBC, and I think Kristofer may be 
correct on that, this indicates that FBC is a criterion inimical to 
making optimized choices! It's like Later-no-Harm, which was reported 
to have made the referee "ill" when he read it, in the publication of 
the Woodall paper that introduced it. I could see why.

Translated to human negotiation process, this is the equivalent to 
"I'm not going to allow or express anything other than my favorite, 
until or unless my favorite has been eliminated and is out of the 
question." This is antisocial behavior, actually pathological.

"Okay, THIS is my favorite, but I see that others have different 
opinions. Let's try to find what will be mutually acceptable. First 
of all, what are the possibilities? I could see accepting this one or 
that one, even though I do prefer my favorite and will stand for that."

Someone who will stubbornly stick only to "My Favorite!" will delay 
the process. Sure, if it become clear that the group will not accept 
the favorite, they might then change.

But this is what I've actually seen in a multiround system in an 
organization that valued unity and that was highly functional.

Strong majority, maybe 70%, preferred the status quo, and a few of 
them had said, "Over my dead body" would there be a change. This had 
been the status quo for many years, maybe even thirty or forty, 
probably longer than any member still alive.

But there were some who were offended by the status quo, for reasons 
that could be labelled "politically correct." In fact, though, the 
reasons were personal. The status quo was the use of the Our Father 
prayer to close meetings. When that had been introduced, it was very 
widely acceptable, and few members were non-Christian, even though 
the organization believed that it was not sectarian, and that was an ideal.

So when the proposal was made to change the closing prayer, it 
aroused strong responses. A member of the meeting, experienced, 
suggested that this become the topic of a special meeting, which was 
the regular meeting devoted to this question. That was accepted by 
consensus. Nobody wanted to push an immediate vote.

So when the meeting came, someone had forgotten to tell the scheduled 
speaker.... and so he showed up.

In any case, one of the first steps taken was to take a poll on a 
series of other options, starting with the status quo. Yes, 70% of 
members found that acceptable. However, notice: 30% of members did 
*not* find it acceptable, and this is an organization which has, 
explicitly, for more than fifty years, valued consensus.

There was another option where all persons present, but one, said it 
was acceptable.

When that poll was complete, there was a motion to adopt the new 
closing "affirmation," which is what that option actually was. And 
that motion passed with *unanimity.*

So, this was an approval poll followed by a yes-no question. The 
result was organizational unity. The majority preference *failed* to 
prevail. And for very good reason.

Suppose this had been an actual multiple-choice deterministic poll. 
With voters who want to "protect their favorite from lower-ranked choices."

The result? Probably, continuation of the status quo. Possibly a few 
members who would leave in disgust.