[EM] The worst about each system; Approval Preferential

[Jameson Quinn]

>
>> Correct strategy in APV when the two frontrunners are ideologically
>> distinct is to disapprove one and everybody worse, prefer the other and
>> everybody better, and approve everybody in between.
>>
>
> Eh? That forces favorite betrayal, doesn't it?


No.


> "Correct strategy" presumes that a particular goal is "correct." What is
> it, in this case? Presumed here is an assumption that ideology is the issue.
>


I give a more precise and general definition below for the shorthand
"ideology", which makes your objections moot. Perhaps you should finish a
paragraph before writing several in response to the first sentence.


>> If voluntary and honest. But one dishonest strategic expression can
>> "poison" a number of honest expressions. Moreover, even semi-honest strategy
>> creates two classes of voting power - strategic and nonstrategic - which
>> hurts legitimacy.
>>
>
> I would encourage all voters to vote strategically and honestly. In a
> Bucklin system there is no advantage to dishonesty that is worth the risk.
> I.e., from a game theory point of view, net zero or negative expected gain.
>

I happen to agree. But some voters might overestimate their capacity to
predict behavior, and the fact that they're wrong doesn't stop the harm they
do. In my system, where there is never more and almost always significantly
less payoff from dishonest strategy, this is less of a danger.

Anyway, you didn't respond to my critique of semi-honest strategy. In APV,
naive human nature strategy is closer to being the same as optimal strategy,
so the difference in voting power is less. This means more legitimacy.

This is important: the math hasn't been done, but my intuition here is that
> the rational strategic Bucklin ballot is a Range ballot with these
> restrictions: the candidates are divided into two sets, approved and
> non-approved, and range ratings are best as sincere ratings *within these
> sets*.


I vigorously disagree. The rational strategic Bucklin ballot is, to first
approximation, an approval ballot. In some cases (which I don't quite have a
handle on) it might be rational to move a single candidate down from maximal
to minimal approval, or to add a minimal approval to a single candidate who
would not have made the cut under approval. Intermediate approved rankings
are never rational if all voters are purely rational, though if there are
some honest voters, it may become rational to use intermediate approvals
occasionally. Unapproved rankings besides the bottom are used for
turkey-raising if at all.

>
>  That's why adding levers and knobs to your voting system is dangerous if
>> they can be used strategically with impunity.
>>
>
> I've seen no cogent example of this. In evaluating strategy, I'd encourage
> Mr. Quinn and anyone else to *start* with utilities.


I do.

Voters will bullet vote, commonly. It is a rational and sensible and
> *sincere* strategy, properly understood.


Absolutely true in many, but not all, cases. I'd guess that mostly it's
rational and sincere; sometimes it's rational and insincere; and rarely it's
irrational and sincere.

Mr. Quinn's proposal is indeed simpler,


Simpler; and better at harmonizing naive, individually optimal, and socially
optimal strategies. Both are important advantages.


>> No more bizarre than closed primaries, at the very worst.
>>
>
> One is getting desperate when one justifies a system as being no worse than
> a bizarre system.


A lower bound which is acceptable to most is not desperate. It is not the
average performance.

>
>
>   That is, a solid majority coalition might elect its more radical member,
>> not the centrist. "Solid majority" means that the median voter is a member
>> of that coalition, supporting the "radical" on that side over all other
>> candidates. Unlike closed primaries, if there's a majority but it's not
>> solid, the centrist from that side is probably elected.
>>
>
> In the example shown, there was a drastic difference between the winner and
> loser.


And it was an artificially-constructed example, one which both the naive and
the correct strategy (which are the same in this case) would tend to
discourage from ever happening.


> The loser was massively approved, the winner only barely. Bare approval in
> a situation like that is quite likely to be an anomaly. But without having a
> set of utilities to start with, we cannot judge a scenario outcome, not
> well, anyway. If the approvals represented sincere approval, the approval
> winner would *certainly* have been the best. Mr. Quinn is arguing that
> preferring the most-preferred candidate will, then encourage additional
> approvals. But then he discards those approvals in this scenario, making
> them useless. Tell me again, exactly why do we want to encourage people with
> a strong preference for first preference to add additional approvals? Beyond
> the natural encouragement of avoiding a runoff -- when "strong preference"
> means they'd rather have a runoff?


If their preference is strong enough, they are free to not add approvals.
But (lightly) encouraging behavior that avoids runoffs has obvious social
benefits even if one minority faction wants a runoff. If a majority wants a
runoff, they can easily vote so as to get one.


>
>  Personally, I don't see that as necessarily bad - think of it as a small
>> taste of time-series proportional representation. In other words, a bit of
>> diversity, instead of centrists winning always, could be healthy.
>>
>
> Sure. But if that's what we are doing, it would be much better for it to be
> explicit. In Alcoholics Anonymous, standard method for electing delegates to
> the World Service Conference is by two-thirds vote. If, after what is
> considered sufficient balloting, and voluntary withdrawals, no 2/3 consensus
> can be found, they determine the delegate by lot from the top two. It is
> explicitly done in order to provide a little additional representation,
> overall, for minority views. That's pretty good for a method that was, I
> think, devised in the 1950s.
>
>
>  ... instead of his 1, 0.75, and 0, it should be 1, 0.5, 0. But that isn't
>> used in this present statement of the method. It's simply Range analysis.
>>
>>
>> This is only for the nonbinding poll. People can set these numbers
>> explicitly, those were just defaults. Actually, the right default value for
>> the "approved" rank is the average of the value people explicitly write in
>> for that rank. I do suspect that people would be more likely to write in
>> values above 0.5 than below it, so I suspect that number will be closer to
>> 0.75 than to 0.5.
>>
>
> You have two approved ranks. Range ballots are always representative of a
> range of actual utilities. If we assume that voters set midrange as their
> approval cutoff, meaning the middle expected outcome, approval could mean
> above this level, or it could mean that the voter rounds off to this level.
> In other words, goes a little below the middle, as "close enough."
>
> I'd simply move to an explicit Range ballot if the jurisdiction were ready
> for that. It would still be analyzed as Bucklin for most purposes. I'd want
> to see as many ratings on the ballot as there are candidates, probably with
> rating of zero being implicit, not explicit. I won't argue this now, but
> there are several reasons for it, not the least of which that it's
> traditional. You mark candidates that you are, in some way, voting *for*. If
> we are going to use voting *against* as the model, we might want to use a
> +/1 scale. It's possible to set a different default than min-range, but I
> won't go there now.
>
> If we assume that voters are classifying candidates into three sets, i.e.,
> with traditional 3-rank Bucklin, and if we assume that the range of
> preferences involved are *equal*, so that the difference from rank to rank
> is equal, except for (original Bucklin) half the ballot range is
> "unapproved" We would get, for range analysis, the middle of each rank
> range:
>
> 0   -  0.25  /
> 0.25 - 0.5  / unapproved, average across both ranges 0.25 (1/4)
> 0.5 -  0.75 / approved, average 0.62 (5/8)
> 0,75 - 1.0  / preferred, average 0.83. (7/8)
>
> Subtracting 1/4 from each, we get ratings of 3/8 and 5/8
>
> Converting to a full vote, this is 0.6 and 1.0
>
> But we could also treat equal ranking top as different from bullet top.
>
> I'd prefer to keep range simpler. In reality what matters is the difference
> between range steps. Each step represents a certain strength of preference.
> There is roundoff error, to be sure. The higher the resolution, the less
> significant it is. The effect of simple Range counting is that the members
> of each rating class are treated as if they were all valued at the top of
> the range in the class.
>
>
You seem not to understand my proposal for a nonbinding poll.


>
>
>>
>>
>>  If not, then the two candidates which are most preferred against all
>> others (ie, the two Condorcet winners based on these simple ballots, or the
>> two most-preferred in case of a Condorcet tie) proceed to a runoff
>>
>>
>> Utility theory would not suggest his pair. Utility theory suggests the sum
>> of scores candidates. I only suggest including a Condorcet winner because of
>> conflict between utility theory and democratic majority theory. If a result
>> is to be based on "greater summed good," the majority should accept it.
>>
>>
>> Utility theory only works for a nonbinding poll.
>>
>
> Which is what is involved in determining candidates on a runoff ballot, if
> write-in voting is allowed. Write-in voting means that the result of the
> first poll is *not* binding unless a majority was found. A majority of
> voters, or even, in fact, a plurality in the runoff, can nullify it. And
> they have been known to do just that. Candidates not on the ballot can
> sometimes win an election.


Do you think that Hanabusa could win a runoff between Case and Djou?
Certainly not a plurality runoff, and I doubt it seriously for an approval
runoff.


>
>
>  If you're relying on it working, people will manipulate it, and that
>> manipulation will be arguably biased; that is, one side will arguably be
>> doing it more than the other. (Whether there truly is a bias or not doesn't
>> matter; the mere appearance of bias undermines legitimacy.)
>>
>
> What I have not seen is a manipulation scenario that makes sense *for the
> manipulators*, when risks are considered along with possible gain. As I've
> mentioned, though, it is possible that with full knowledge, any voting
> system can be "manipulated," but, if it is a utility-maximizing system, the
> "damage" is limited to the voting power of the manipulating faction. It's,
> in fact, questionable if it is damage at all, and if that faction directly
> uses its power, it would not be bothering with "turkey raising" as Mr. Quinn
> proposed.
>

Again, want to bet? You program your agents to vote your strategy, I'll
program mine to take them to the bank. All agents have random preferences,
except that each candidate is either "Jameson-style" or "Abd-style", with a
small bonus to the utility of all respective agents for that candidate. We
can do any reasonable set of pre-election imperfect information, though of
course zero information is by far the easiest to program for.

Otherwise, this is just hot air.


>
> I know that Bucklin *did* work. I have no reason to believe that it
> wouldn't continue to do so. The changes I'm suggesting simply improve it to
> reflect modern voting system theory, they are not drastic.
>
> The "manipulations" as designed would be somewhere between difficult to
> impossible. Mr. Quinn did not show an exact scenario that reflected how the
> method would actually work. Basically, it's possible that a faction of
> voters, speaking roughly, could improve results, from their perspective, *a
> little*, using a complex strategy that could backfire, easily. Many
> theorists have proposed voting strategies that were like this. In fact,
> serious voting strategy, that requires voters to vote with reversed
> preference, seems to be quite rare, that we would see it in nonpartisan
> elections seems vanishingly possible, to impossible. The strategist is
> asking voters to state on the ballot what they don't believe.
>
> As I've mentioned before, if my favorite candidate suggested that to me,
> the candidate would no longer be my favorite.


Ever heard of Limbaugh's "operation chaos"? It's not the candidates you
should worry about, but their loose-cannon allies.


> Thus promoting a turkey-raising strategy, in this context, would be very
> hazardous. And I doubt that voters would do it individually. This is a red
> herring.
>
> The basic error that Mr. Quinn makes is in assuming that bullet voting will
> be a natural and practically universal strategy.


Where on earth did I say that? In fact, I said several things which
contradict that: that Bucklin would encourage (strategically misguided)
extra ranks, and that correct strategy would find a CW in one round if
non-supporters cared about the distinction between the two frontrunners.


> Definitely, bullet voting will be common. It's common with IRV where it's
> allowed. It's a sensible, rational strategy for a voter without sufficient
> information to do more than vote for their favorite, and Lewis Carroll
> invented Asset Voting to enfranchise these voters with STV.
>
> Then, to the contrary, he posits some clever, conniving strategy


Oh, I get it, calling your system design a "trick" (later clarified to mean
not evil but unstable) is not fair, but calling mine a "clever, conniving
strategy" is OK.


> to manipulate who gets into the runoff so that results, it's believed, will
> improve. I.e, the faction believes that B, say, is more popular than their
> favorite A, but their favorite could beat C. Notice: B is the centrist. If
> they try to arrange for the runoff to be between A and C, they run the risk
> of electing C, the worst candidate from their perspective.
>


> Now, how, exactly, would they do this? To understand if they have a
> rational motivation for doing it, we'd need to posit sincere absolute
> utilities for this group. (To get voters to "lie," requires strong
> *absolute* utility difference).
>
> We know that they don't want to approve C, and Mr. Quinn is proposing that
> the problem is in the Condorcet or Range analysis used to determine
> candidates for the runoff, and to suppress the presence of B in the runoff.
> The only way for them to do this would involve these ratings:
>
> (using a Range 4 ballot, Condorcet analysis will be used.)
>
> A: 4
> B: 0
> C: 1 <- insincere, a sincere vote might be B, 1, C 0.
>
> We must posit these conditions:
>
> A majority of the electorate prefer B to A, so they fear B will beat A in
> the runoff. Right away, I'd be ashamed to be a member of a group that wanted
> to defeat the will of an actual majority, but, yes, some people don't think
> that way. (I'd prefer to put my effort into educating the majority, it's a
> long-term effort that can pay off, whereas a short term election victory,
> when the electorate isn't ready for it, can seriously backfire.)
>
> Enough voters prefer C that the extra votes of this manipulative faction
> could put C into the runoff, but keep A out.
>
> Under the conditions, we must assume that B might gain a majority of
> approvals in the primary. What's the expectation of the rest of the
> electorate? Is there any hope for A? Bucklin isn't vulnerable to center
> squeeze, with sincere votes. This faction apparently believes that it can
> beat C in a runoff. How solid is that idea? To be solid, very likely, we
> have first preference votes for B exceeding those for C. The centrists, the
> supporters of C, will likely be split in the runoff between A and C, or may
> not even vote, if it's an A:C runoff.
>
> So we have first preference leader, lets say, is A, B is second or third, C
> is second or third. But second preference? A is a frontrunner, and likely,
> lets' accept, to win a runoff between A and C. So look at the C voters. They
> fear the election of A, so ... they will be relatively highly motivated to
> add an approval for B. In three-rank Bucklin, they may vote C>.>B. (Some
> will vote C>B, and some just C.)
>
> There is a very good chance that B wins in the primary. Let's assume,
> though, that the faction using this strategy is more radical than C. This
> means that B voters will prefer C to A. So some B voters may also add
> approvals for C. Depending on the balance, if the A voters add additional
> approved rating for C, they risk a win for C, in several ways. If they use
> the elevated disapproved rating, they could bring C into the runoff, whereas
> C might have been excluded, but this would only be bringing in C as an
> additional candidate, not shoving B out. They lose, their strategy
> accomplishes nothing, and the voters who voted that way will have it in
> their mind that their party could only win by "cheating." Not the way to
> build a party!
>
> It is major parties, the leaders, that "cheat," when they can get away with
> it.
>
> Here is the scenario, under the possible rules I proposed, for the rating
> of 1 to cause C to win the primary.
>
> It happens if the approvals and ratings for top two show B and C both
> getting a majority. If the rules only trigger a runoff if the top approval
> getter is beaten pairwise by the runner-up, that preference expressed will
> be looked at: C>B. If C was top, and there are enough C>B votes, C wins.
> Oops!
>
> Suppose there is a majority for C, none for anyone else. It's possible that
> the unapproved votes would be moot, here. But let's assume that we still
> want to see if there is a condorcet winner. C>B might confirm the C victory,
> whereas B>C might cause a runoff.
>

The scenario is that A is first and C is second or third. Voters can be
pretty certain that C won't get a sincere first-round majority, and their
strategy doesn't help it get one.

>
> I do not see this as a strategy that can be used with "impunity." There are
> risks, both political and practical.
>
> The strategy would be visible in ballot analysis, by the way, because votes
> of C>B for this faction, as described, would be rather obvious as insincere.
> Think Nader>Bush>Gore in this system. Individually, sure, but a whole
> faction voting that way> I'd say, next election, even if they prevailed,
> they'd be, to use the technical term, screwed.
>
> The rest of the electorate would not be helpless. The terms of the problem
> require the A voters to reasonably expect they will beat C. If this is a
> fact, the C voters, as I mentioned, will be motivated to head this off, and
> they do it by voting *sincerely*. We know that the B/C combined faction is
> quite a bit larger than A, probably about double the size. It only takes
> enough multiple approvals to keep B in the running to defeat the strategy,
> and those are sensible, sincere approvals. Likewise, I've suggested, the B
> faction, even though reasonably confident, may see some risk from A, and
> they are (centrally) further from A, most likely, than they are from C. They
> may cast additional approvals for C. This situation could bring C up to
> majority approval, with the scenario described above, where the insincere
> ranking of C might even make C win.
>
> We have no evidence at all that anyone, in a system like this, would even
> try it. Probably by the time Condorcet analysis is added, we'd have good
> ballot data, so we could watch how elections go. Combined with other
> information, attempts to gain value by "turkey raising" would become
> visible, those voting patterns would show in the ballot analyses, whereas
> turkey raising in top two runoff is invisbile, except for publicity leaking.
>
> Let me list the ways I find my proposal better than yours, in order of
importance.

1. Simpler to explain.
2. Naive, rational, and socially-optimal semi-honest strategies are closer
to each other.
3. My system has stronger reasons not to use dishonest strategies
(turkey-raising).

You're arguing against reason 3 by saying your system's
anti-dishonest-strategy properties are strong enough. You may be right. I
still prefer to be doubly safe. But that's my weakest reason anyway.

What are your system's advantages? To me, it's a lot of rules - for nothing
that my system can't accomplish using a nonbinding poll and a write-in
runoff.

JQ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20100527/34ba6142/attachment-0004.htm>