---------------------------- Original Message ----------------------------

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

From: "Daniel LaLiberte" daniel.laliberte@gmail.com

Date: Tue, December 6, 2016 1:21 am

To: "Kristofer Munsterhjelm" km_elmet@t-online.de

Cc: "EM" election-methods@lists.electorama.com

...

�
i think we all agree with Arrow that every system has flaws.
...

no.
�

and with RCV, there is no strategy "in your face".
who is your favorite? �who is your contingent candidate? �who is your third choice? �who are the candidates you would never want elected? �this preference translates directly into
sincere rankings. and that cannot be said about Approval or Score voting, by their very nature.
and unless there's a Condorcet cycle, there's really no strategizing if the ranked votes are tabulated according to a Condorcet compliant method. �the election is decided exactly the same as
if the CW was up against any other candidate in a one-on-one race with a simple-majority vote and one-person-one-vote equal weighting on the votes. �it's because IRV didn't deliver on that promise for the Prog-hating GOP (who found out that simply because they ranked their favorite #1 they
caused the election of their most hated candidate) that people start considering if strategy would help their political interests.
but with RCV, there is no strategy "in your face" in the voting booth. �only a strategy that confronts you with a bad tabulation method such as
Borda or IRV or Bucklin.
with RCV decided according to a simple rule*** that nobody disagrees with, voters know what to do to express their political interest, and if they can have confidence that the tabulation method won't screw them, there is no esoteric need for strategizing.
�eventually the election will be decided between the two candidates with the broadest support and their preference for one of the two over the other will be treated with the same weight as with the preference of any other voter. �when there is no cycle, they can be confident that they are
treated equally with other voters and that there vote will not be misinterpreted by
***the simple rule that nobody disagrees with is: "If more voters mark their ballots that they prefer Candidate A over Candidate B than voters who mark to the contrary, then Candidate B is not
elected."
the problem with Approval is that it doesn't get enough information from voters. �in a multi-candidate race, there is no way for a single voter to express that they prefer Candidate X over Candidate Y for every {X,Y} pair combinations from the set {A, B, C, D ...}.
�either the voter has to bump up B to be as good as A, in order to make sure that B is preferred over C or the voter has to bump B down to be as bad as C in order to make sure that A is preferred over B. �that's a burden placed on the voter. �
"what to do! what to do! oh me
oh my! oh me oh my! what to do!"
and, despite the loquacity, neither you nor Michael can adequately address that concern. �you cannot simply answer "Should the voter Approve their 2nd choice or not?" �But I can simply answer ":How shall the voter rank their 2nd
choice?" �I have a simple answer and you do not have a simple answer for the question put to you.
but, to avoid the burden of strategizing being placed on the voter, you must be able to answer that question simply: Should the voter Approve their 2nd choice? �Their 3rd choice?
�Yes or no?

another goofy thing about Burlington Vermont (and surrounds) is that our State Senate district is the largest legislative district, in terms of number of at large seat holders, in the United States. �the Chittenden senate district elects 6 state senators where
everyone, regardless of party, is running against each other. �a party is allowed up to 6 nominees on the ballot. �only the Dems get 6 candidates on the ballot. �voters are allowed up to 6 votes. �in the past, because of one reliable GOP winner, it was like musical chairs for the
6 Dems, because one of them would lose. �relatively few voters voted all 6 allowed votes because there were specific favorites nearly every voter wanted to see elected. �most people voted for one or two.
we have virtually the same as Approval Voting in the 6-winner Chittenden senate
district. �everyone voting for less than 6 is voting tactically and is voting exactly as they would if it were Approval Voting. and some who are voting for the maximum are voting tactically. �and it's a burden. �and it's because it's a multiwinner race (6 top vote getters win) that
makes it a big clusterfuck.

--
r b-j � � � � � � � � �rbj@audioimagination.com
"Imagination is more important than knowledge."

On Sat, Dec 3, 2016 at 9:09 PM, Michael Ossipoff email9648742@gmail.com
wrote:

Thanks Michael.  I recognized some similarities, but Kristofer is certainly
giving more agreeable, reasonable arguments, and he appears to agree
regarding the value of Approval Voting.  The contention, if there is any,
is about when strategizing is required, and whether it matters.

I agree with him that there is this one bit of strategy required when using
Approval, which is deciding where to draw the line between approval and
disapproval that is possibly affected by what other voters will do, and it
is interesting to understand exactly why this happens.  But I also agree
with you that, when it is not so clear how you might want to strategize
with Approval Voting, where exactly you might draw the line, it doesn't
matter much either.  So there is some strategy required at least sometimes,
but fortunately, it happens mostly when it doesn't matter very much.

But whether this is true depends on some other factors, such as what
happens in earlier stages of the election where it is determined which
candidates will run in this election.  If Approval Voting is used in closed
primary elections, then this will amplify the distinctions between the
nominated candidates for each of the parties, and thus the general election
will be more contentious rather than homing in on the median very quickly.
I'll continue that discussion in another message replying to Kristofer.
(Or maybe we should continue here.)

The popular assumption is that it's essential to try to elect the best

I think it is reasonable to expect that everyone should try to elect the
best candidate, but it is not reasonable to expect that everyone can be
satisfied, and I don't think anyone has that assumption.

How much we care about trying to get the best candidate also depends on how
critical it is that we get a half-way descent candidate.  When most of the
candidates are reasonably good, and one of them is likely to win, then it
doesn't matter much.  When there is a large range of quality and a
freakishly high number of bad candidates, then we need to be concerned, but
then the difference is also fairly clear, and Approval Voting will work
very well in that case too.

I like to remind people that, very often, "Good enough is better than
best."  That is, a voting system (or a candidate) that is "good enough" may
very likely better than one that is "best".

"Approve about half" is a good enough, easy to remember guideline.  It
would seem to maximize your impact as well.

Whether "about half" is good enough does depend where the frontrunners are
in each voter's ordering of candidates, but given that the frontrunners are
likely to be close to the median across all voters anyway, then they will
likely be positioned near the median of most voters' ordering.

I'm not so worried about electing the best.  I would worry about electing a
much worse candidate in a surprising upset.  Elections really ought to be
much more boring, but not enough to put us to sleep.

I'm still planning to reply to a couple of your earlier messages with a
couple more comments.

--
Daniel LaLiberte
daniel.laliberte@gmail.com

Hi Daniel--

This reply to the December 6 posting will be in a number of parts. I'll
skip around, answering first the things that are most readily answerable.

On Tue, Dec 6, 2016 at 3:37 PM, Daniel LaLiberte <
daniel.laliberte@gmail.com> wrote:

Tactical voting, based on predictive information isn't possible unless
there's reliable predictive information. When Noam Chomsky so
self-assuredly told us that voting for Hillary was our only strategy,
Chomsky was being a chump, even if an astoundingly arrogant one.

Well he was right to say that voting for Hillary was strategic
voting--really bad strategic voting.

I'd suggest that voting in Approval should never be based on predictive
information...at least not in this social world.

One thing that I like about Score is that, when people moderate their
strategy with sincerity, that softens any strategic errors that they're
making.

Rankings could confer similar benefit.

But it has to be realized that ranking-methods don't really get rid of
strategy concerns. Pairwise-count methods seem to offer the most, and seem
to be the only methods that can meet FBC and avoid the chicken-dilemma
problem, but they're subject to the offensive strategies of burial, and
often truncation.

I'm for offering the best rank methods, preferably the
FBC-&-no-chicken-dilemma ones. But, other than IRV's popularity due to
heavily-funded promotion, I'd bet that the relative complexity of
rank-methods would spoil their chances of acceptability.

I recommend against that. I say, "Vote optimistically. Approve only  the
very best. Give the people some credit for being able to elect one of them."

Yes, I'm talking about a different kind of voting and a "strategy" of
optimism.
.

Sure, and with the more morally-diverse candidate line-up, there will
definitely be a strong bottom-set.

What needs to be emphasized in this discussion is that, as I was saying,
the 99%, with our current lineups, always have strong top & bottom sets.
That makes it really obvious how to vote. Just approve (only) all of your
strong top-set.

Any talk of situations that aren't like that, is talk about voting after
honest elections & media have been in place for a while,  in some utoptian
future in which your strong bottom-set has withdrawn from the elections
because they're unwinnable.

So, that's a good answer to someone who worries about strategy in Approval.

Trying to say it in a positive way: We won't have to worry about that any
time soon.

More later,
Michael Ossipoff

Daniel--

You wrote:

Sure, not everyone can be satisfied by the election-result.

But I wouldn't say that everyone should try to elect the best individual
candidate that they can get. What if trying to elect the best candidate
that you can get, choosing among your strong top-set, increases the
probability of electing from your strong bottom-set. Is that good?

By the definition of the strong top *& bottom sets, electing from the
strong top-set instead of from the strong bottom-set, is incomparably more
important than the matter of which strong top-set candidate you elect.

Instead of electing a progressive candidate who is honest, and wants the
same improvements that you (& the rest of the 99%) want, you elect a
corrupt, bought sociopathic Republocrat?  Increase the probability of that
in order to choose among the progressives (instead of just supporting them
all, to make it more likely that one of them will win?

MDDA & MDDAsc protect your ranked & approved candidates as well as Approval
does, while also choosing among them (by influencing majority pairwise
defeats among them). So it seems you aren't losing anything in that regard
with MDDA & MDDAsc, compared to Approval.

But, when you're voting in MDDA or MDDAsc, you can improve Pt, the
probability of electing from your strong top-set (which we can assume are
your ranked & approved candidates): You can rank them all at top.

If you have a majority who prefer your strong top-set to the others, but
some voters similar you approve lower than you do, resulting in some of the
worse candidates having maybe more approvals than your strong top-set do,
the preferrers of those worse candidates couldduse truncation or burial to
make everyone majority beaten, and then win because of their higher
approval total.

But if you've ranked your entire top-set at top (and maybe a few similar
voters do), then you have the best chance of preventing the truncation or
burial to make everyone majority-beaten. If lots of people rank someone at
top, how can any smaller-than-majority group make them majority-beaten? You
& lots of other people aren't ranking anyone over that person.

So, by equal top-ranking all of your strong top-set, instead of ranking
them in order of preference, you're lowering the probability of electing
from your strong bottom-set.

So often that's the choice: Choose among your strong top-set instead of
maximally supporting them all, and you thereby increase the probability of
electing from your strong bottom-set.

Of course. Then you don't have a winnable strong bottom-set.

You can try to maximize your expectation, among those all-good candidates,
in any of various ways.

...and that's how it is now.

That's when Approval voting is at its easiest. You definitely have a strong
bottom-set. Approve (only) all of your strong top-set.

More next time.

Michael Ossipoff

(all replies in this message are above this point)

---------------------------- Original Message ----------------------------

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

From: "Michael Ossipoff" email9648742@gmail.com

Date: Wed, December 7, 2016 10:18 pm

To: "Daniel LaLiberte" daniel.laliberte@gmail.com

Cc: "EM" election-methods@lists.electorama.com

we call those folks "losers" or the voters who voted for the losing candidate. �this is, of course, unavoidable.
but, under normal conditions (and i would say a cycle is not normal), we can satisfy every
voter that their sincere vote served their own political interests (at least how they evaluate their own political interests) as best as could be expected. �Rank Choice Voting, decided with a Condorcet-compliant method (and no cycle) can do that, but neither Approval nor Score voting
can.
This is because, if the voters favorite candidate loses to their second favorite, voters will wonder if they should've approved their second choice. �Or if both their fav and contingency candidates lost, but they didn't vote for (approve) their contingent candidate (out of fear of
screwing their fav) will they regret not approving their second choice?
yet with Condocet, if your candidate loses, it is because you know that some other candidate was preferred over your candidate by more people with the same power of vote you have (that is "one person one
vote"). �you may not like that your candidate lost, but you won't be thinking the election was stolen or a screwup of the tabulation method (like IRV in Burlington 2009) elected another candidate that was less preferred than yours or that a spoiler robbed your candidate of victory.
�you will be dissatisfied being a loser, but you can accept that your side lost fair-and-square. �with Approval, you just never know. �and that is because, similarly to Plurality with the single-mark ballot, not enough information is solicited from voters.
�

I wouldn't say that everyone should try to elect the best individual

gee, isn't that normally called the tactic of "compromise"? �it's the burden of tactical voting and, according to Duverger (and common sense), it's why we have a dominant 2-party system in the U.S. �it
turns out that "the best candidate that you can get" will be one of the two major party candidates.
�
as disagreeable as i am, Mike (Kristopher is much more agreeable than i am), being unable to answer the fundamental challenge i put to you is not synonymous with me being unreasonable. �like the DNC after Nov 8, you need to turn inward a little to focus blame.
so by not
saying "that everyone should try to elect the best individual candidate that they can get", you're essentially saying that voters should expect to vote tactically. �and, unreasonable me is saying that voters should not be expected to vote tactically to best express and serve their
political interests. �they should be able to walk into the voting booth and express their political interests without fear that the voting system will misunderstand their political interests and act, on their behalf, contrary to their political interests.
BTW, i don't read farther than
the first 20 or 25% of your posts. �you might want to learn the meaning of "concise". �(and "concise" doesn't mean alphabet-soup nor does it mean relying on jargon. jargon-jockeying is not particularly persuasive.)
�

--
r b-j � � � � � � � � �rbj@audioimagination.com
"Imagination is more important than knowledge."

When I used the example of MDDA & MDDAsc, to illustrate that it's better to
equal-top-rank your strong top-set, rather than choosing among them by
ranking them in order of preference--That wasn't intended as criticism of
MDDA & MDDAsc.

The fact of it being better to equal-rank the set that is important to you
instead of choosing among them is true with other rank methods too.

In particular, it's true of Condorcet & Bucklin. It would be true of IRV
too, if IRV allowed equal-ranking.

I don't have proof that it's, in principle, a property of all
ranking-methods, but I don't know of an exception.

One fairly obvious thing that can be said for MDDA & MDDAsc is that your
protection for your strong top-set, even when ranking them (and no one
else) in order of preference, and approving them all (as is the default),
is as good as your protection of them in Approval, when you approve only
lthem.

A majority doing so in approval will elect one of them.

A majority doing so in MDDA or MDDAsc will give a majority-disqualification
to everyone else. And if preferrers of one of your strong bottom-set try
burial or truncation, and if they thereby manage to make everyone
majority-disqualified, then someone in your strong set will win the
Approval count.

That suggests that MDDA & MDDAsc let you choose among your strong top-set,
and still protect them from your strong bottom-set just as well as Approval
would have let you. That's an improvement over Approval.

Of course an additional improvement is that MDDA & MDDAsc give you an easy,
convenient, & reliable way to avoid chicken-dilemma (by denying approval to
the candidate of the distrusted faction.

It's just that MDDA & MDDAsc allow you to further enhance the protection of
your strong top-set, by top-ranking them all. If a majority do that,then it
would be quite impossible for buriers or truncators to majority-disqualify
them. Of course if any significant number of voters similar to you top-rank
those candidates, that makes it much more difficult, or impossible, for
buriers or triuncators to majority-disqualify them.

If you use the chicken-dilemma defense of denying approval to the candidate
of the distrusted faction, and that candidate is someone whom you top-rank,
then you're still protecting hir from burial & truncation, for the reason
described above.

If the candidate to whom you deny approval is someone you rank below top,
then that is no longer true. If the method is MDDA, that candidate still
has the full truncation-proofness protection that any ranked candidate has.
If the method is MDDAsc, that is no longer guaranteed. But, if
Mono-Add-Plump is necessary for public acceptance, then the cutting-loose
of that middle-ranked candidate of the distrusted faction is a regrettable
but justifiable action resulting from reasons that that faction has given
you for defection-deterrence.

Likewise, though MDDA protects your middle-ranked candidates from
truncation by eachother's factions, that protection isn't essential,
because reliably choosing among your strong top-set isn't the important
thing.

In MDDAsc, you're still fully protecting your top-ranked candidates against
everyone else, and you're still fully protecting all of your rannked &
approved candidates against your unranked, unapproved candidates. That's
what's important.

MDDA & MDDAsc are the rank methods that best deliver the benefits that are
available from ranking-methods.

Now, to resum my reply:

Exactly. Eecting one that is good enough is much more important than
reducing the probability of doing so, by trying to choose among the ones
that are good enough.

[Replying farther down] :

Yes, you're voting between the maximum number of candidate-pairs.

[Replying farther down] :

Then, in Approval, approve all of your strong top-set.

With honest elections and honest media, elections wouldn't be boring,
because you'd be choosing among various versions of the very best. The
choice among them, the discussion regarding their different approaches to
the best policies and directions, would be anything but boring.

What's boring is when the media keep claiming your choice is between two
criminallyi-corrupt, bought candidates, and when people believe it.

Michael Ossipoff

---------------------------- Original Message ----------------------------

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

From: "Michael Ossipoff" email9648742@gmail.com

Date: Thu, December 8, 2016 6:35 pm

To: "Daniel LaLiberte" daniel.laliberte@gmail.com

Cc: "EM" election-methods@lists.electorama.com

if IRV allowed equal-ranking, who of the equal-ranked 2nd choice are promoted to the 1st choice (because that's the only choice that gets counted in IRV) when the 1st choice is eliminated? �unless you split votes in half, there is no way to do it.
�and try selling the concept of a 1/2 vote to legislators or voters considering it? �in Burlington Vermont, the ballot instructions were clear that assigning the same rank to two candidates would spoil your ballot.
�

property of all

lemme see, you're claiming that, without an exception that you know of, in any ranked method it is an advisable strategy to equally rank all candidates that you "sorta like"? �you think voters that are free to unequally
rank them (but not to have to score them) would prefer to follow your tactical advise in their particular circumstances that you have no idea of? �you can discern that a voter's best political interest would be served by that?
and then you think it's little burden to the RCV voter to
actually grok that decision in the voting booth: "Lessee, I really like Andy over Bob, but I can live with either and since Michael Ossipoff recommends equally-ranking the top set, I'll rank Bob equal to Andy, just to make sure that this will more effectively defeat Curtis whom I don't want
elected?"
you think a voter would prefer that to simply ranking Andy > Bob > Curtis?
the main reason we want to use other systems than Plurality is to reduce the burden of tactical voting on voters and let them express their political interest with little worry about shooting
themselves in the foot and voter regret. �tactical voting is not pleasant. �and another reason is to reduce the effect of voting strategy by campaigns or interest groups trying to swing elections by persuading a quantity of voters to vote a certain way.
i don't see how this advice
is attractive to regular voters, who want to be heard, who don't want surprizes because of irregularities. �why would they be motivated to do that? �why would they be motivated to even try to remember such advice and its detail?

--
r b-j � � � � � � � � �rbj@audioimagination.com
"Imagination is more important than knowledge."

On 12/9/2016 10:05 AM, Michael Ossipoff wrote:

Mike,

If  IRV allows equal-ranking, it should definitely be the "fractional"
version (so that in every round each ballot gives a single vote or
fractions of a vote
that sum to 1).

If instead a ballot gives a full vote (approval style) to each candidate
it highest ranks then the method is much more vulnerable to Pushover and
fails
Mutual Dominant Third.  I gave an example of that a while (probably
years) ago.

Here is an example of Push-over that is also a failure of Unburiable
Mutual Dominant Third:

45: A=C  (sincere is A>B)
35: B>A
20: C>B

B is the sincere MDT candidate, and so with normal IRV is strategically
invulnerable, but under  ER-IRV(Whole) loses to A.

The fractional version also makes Push-over strategising a bit easier,
so for IRV and Benham I'm opposed to allowing above-bottom
equal-ranking.

But if that is insisted on I suggest that to decide which candidate is
next eliminated we first order the candidates by the fractional method,
and then have ballots that equal-highest rank more than one candidate
give a full vote to the one of those that is highest in the order (and
nothing to any other candidate).  Then eliminate the candidate with the
fewest votes.

So in the above example the initial order is  C 40.5  >  B 35 >  A
22.5.  The 45 A=C ballots give a full vote to C, giving C 65 > B 35 > A
0.  We eliminate
A and then B and C wins.  (There are better examples where the device
does some good).

You wrote "I don't have proof that it's, in principle, a property of
all ranking-methods, but I don't know of an exception."

It's true of  Winning Votes and Bucklin and any of the proposed methods
that meet FBC, but I don't see how it is of ER-IRV(fractional).

Or MinMax Margins or  ER-Benham(fractional)  or even Smith//Approval
(implicit).

In the latter case obviously the voter should truncate hir bottom-set,
but couldn't it be the case that if the voter strictly ranks hir top set
then one
of them will be the CW while the most approved candidate is in hir
bottom set but if the voter equal top-ranks hir top set there will be a
top-cycle
that includes the most approved candidate?

Chris Benham

On 12/9/2016 10:05 AM, Michael Ossipoff wrote:

On Mon, Dec 12, 2016 at 12:52 AM, C.Benham cbenham@adam.com.au wrote:

Hi Chris--

The only reason why, some years ago, I suggested ER-IRV(whole) was because
I thought it met FBC. When Kevin pointed out that it doesn't, I no longer
have advocated that method.

Ordinary IRV, Benham & Woodall are fine, as-is.

I've discussed the fact that they have a problem, but that the problem has
mitigations. IRV's worst problem has been the dishonest promotion of it,
with false guarantees substituted for the disclosures and warning that it
needs.

IRV would be fine, if, before enacting it, people understood & accepted its
problem, and weren't going to favorite-bury,  or repeal it as soon as it
eliminates a CWv.

So, if, in some jurisdiction, IRV is the method that's popular as a
proposal, my message to its advocates would just be to promote it honestly,
and to not advocate or enact it unless they understand and accept how it
works, what it can do, its unique character & tradeoff.

Well, this is progressive partisanship, but one reason why i wouldn't want
IRV now is that it appears that, here, under current conditions, IRV's
problem, even if mitigated in general, would work against progressives if
it happened.
But I suspect that, with honest media, and well-informed voters, IRV's
problem's mitigations would make its problem a lot less important.

Benham & Woodall elect a CWv when there is one, and that's an improvement.
They're quite vulnerable to burial or truncation, but the worst that those
can accomplish is to give IRV results. IRV at worst, Condorcet at best.

But the trouble is that IRV has always been dishonestly promoted without
the necessary disclosure, which was why it was angrily repealed in
Burlington, by people rightfully angry about having been deceived by
promoters.
.
Also:  Though, with public ballot-imaging, any voting-system is secure
against count-fraud, we don't currently have that (maybe never will) , and
any effort to catch or prevent count-fraud is doubtful, then it's important
to not make successful count-fraud any easier than it has to be.

That means not using a voting system that isn't precinct-summable. That's
the other big reason to not enact IRV or its derivatives, under current
conditions.

You wrote:

You wrote "I don't have proof that it's, in principle, a property of all
ranking-methods, but I don't know of an exception."

It's true of  Winning Votes and Bucklin and any of the proposed methods
that meet FBC, but I don't see how it is of  ER-IRV(fractional).

[endquote]

Maybe. I hadn't considered fractional IRV or Margins Condorcet when making
that statement. Admittedly I was only looking at a few kinds of methods,
and so my statement was too much of a sweeping-statement.

You wrote:

Or MinMax Margins or  ER-Benham(fractional)  or even  Smith//Approval
(implicit).

[endquote]

Smith//Approval?

What if there's member, X, of your strong top-set that is only in the voted
Smith-set because, if you equal-top-rank, it ties or beats a certain
member, Y, of the Smith set who is in your strong top-set.

But now you rank your strong top-set sincerely, and now Y beats X. Now X is
beaten by everyone in the voted Smith-set. X had been the winner, because
of its high approval-total, but now some other voted Smith-set winner, Z,
with next highest approval, wins, and Z is in your strong bottom-set.

So are you sure that equal-top-ranking all of your strong top-set isn't the
best strategy in Smith//Approval?

But I don't advocate Smith//Approval, because I feel that, if you're going
to use a ranking-method, it should meet FBC  & at least have a convenient
way to avoid chicken-dilemma.

(IRV is an exception, regarding FBC--Under the right conditions it could be
acceptable, but of course it should be selfishly opposed by people who'd be
on the wrong end of its problem).

You wrote:

In the latter case obviously the voter should truncate hir bottom-set, but
couldn't it be the case that if the voter strictly ranks hir top set then
one
of them will be the CW while the most approved candidate is in hir bottom
set but if the voter equal top-ranks hir top set there will be a top-cycle
that includes the most approved candidate?

[endquote]

True. Like all un-improved Condorcet methods, Smith//Approval fails FBC,
and equal top ranking your strong top-set could elect a member of your
strong bottom-set.

But isn't it true that the voter ordinarily wouldn't know which strong-set
member s/he should strategically rank over the others?  ...which strong
top-set member shouldn't be moved to top?

But that wouldn't usually be unpredictable, and so wouldn't it still be
best toequal-top-rank.one's whole strong top-set?

Michael Ossipoff

Chris Benham

On 12/9/2016 10:05 AM, Michael Ossipoff wrote:

On 12/06/2016 09:37 PM, Daniel LaLiberte wrote:

I did actually write a reply to your previous post, but I thought it'd
be a bit cumbersome to continue our discussion in two different threads,
so I was intending on folding it all into one thread once I get a reply
to my other post. But since what I wrote is relevant to Approval, I'll
quote it here:

=====

IRV is Plurality with instant run-off. The analogous method for Borda is
called either Nanson or Baldwin. In Nanson, you do Borda, then eliminate
the candidates with below-average score, then rinse and repeat until you
have a winner. In Baldwin, you eliminate the loser one at a time.

Both Nanson and Baldwin are Condorcet methods, because a Condorcet
winner always has above mean Borda score. But like IRV, neither is
monotone. They're also relatively vulnerable to burial and neither
Nanson nor Baldwin is cloneproof. I'd take the Borda-IRV methods over
plain IRV anyday, but for Condorcet, there are better options.

IRV fixes some of the problems with Plurality. It's kind of an example
of a patch idea:

• Plurality has a problem (vote-splitting where very small parties draw
  enough votes to make the aligned major party lose)
• How can we fix that?
• We add a patch where the small parties are eliminated first.
• Then the right major party wins!

It's true, but such patches are usually inadequate because they only
deal with the problem currently present. So IRV works well as long as
there are only two major parties. But when there are three, IRV no
longer knows which is the minor party and behaves erratically (cf
Burlington 2009).

In the past, I've heard some Approval supporters go "well, you know the
parties you absolutely do like and the parties you absolutely don't, so
just approve the former and don't approve the latter". But this is
another kind of patch. It works as long as that's true, but then fails
when it's no longer the case. In a Burlington scenario, both IRV and
Approval would have done well as long as there had been only two major
parties. But when there're three on the line, IRV becomes confused and
the chicken problem appears in Approval.

=====

So I don't find the counter that you can just "approve of the candidates
you like and not approve of the people you don't like" very persuasive.
This assumes that there will always be a clear delineation between the
candidates the voters like and the candidates the voters don't like.

So suppose that there will always be such a clear delineation. Then most
systems (like Condorcet) that support equal rank and reduces to Approval
when everybody equal-ranks will work: just equal rank the acceptables
and don't rank the rest. In particular, because such a preference type
is singlepeaked, there is automatically a Condorcet winner.

On the other hand, suppose that there won't always be such a clear
delineation. Then Approval will start running into trouble and/or put a
burden on the voters when you get into a fuzzy territory. In a
Burlington scenario, Approval could misjudge -- like IRV, except it
would misjudge in favor of a centrist rather than in favor of an
extremist, to the extent that the voters are risk averse and approve
more candidates than they would under optimal strategy.

So my point is that in "easy to call" elections where all the voters
have a sharp delineation between who they like and not, most reasonable
methods get it right. The difference is that ranked methods don't force
you to vote as if that were the case when there is no such sharp
delineation. On the other hand, in the scenarios where it's unclear what
society's preference is, like when there are multiple real contenders,
methods like Condorcet can have trouble. That's right, but so will
Approval.

The trouble manifests differently: in a Condorcet method, it permits
strategic voting if the voters know what they're doing. In Approval, it
increases the burden for honest voters (and the risk of a wrong result
if the honest voters don't pay attention).

A RangeVoting page on IRV says something like this:

'Q: Do you support IRV over Condorcet since Condorcet has problems with
strategy?
'A: Condorcet has problems with strategy. But IRV has problems without
strategy! So no sale!'

and I would say something similar:

'Q: Do you support Approval over Condorcet since Condorcet has problems
with strategy?
A: Condorcet has problems with strategy. But Approval burdens the voters
even without strategy! So no sale!'

And if Approval doesn't burden the honest voters, then neither do the
ranked methods.

Again, this kind of loops back to my concept of "manual DSV". Approval
can't do the impossible even though it seems to to do so, because
[Approval + whatever you do in your head to quantize the ballot] could
be considered just another voting method. If you do a fair comparison
between Approval and other methods by explicitly stating an algorithm
that does the quantization step for you, then the resulting quantized
Approval method is susceptible to strategy like every other.

Consider a method like, say, MAM. This method passes the Smith
criterion, which means that it'll elect from the set of candidates where
all of them would win a one-on-one runoff with any of the candidates
outside the set. Another way of saying this is that it'll be uncertain
of whom to elect (in the absence of strategy) only among those who are
close to the median anyway. So it picks someone close to the median
without the honest voters having to accommodate the restrictions of the
system.

Now, you could say that Approval is more resistant to strategy. It is,
because you have to play the strategy game no matter whether you're a
honest voter or not. If the voters can absorb the noise they need to
absorb in order to get Approval to behave nicely, then they can absorb
the noise that would result from strategy campaigns as well -- as long
as those strategy campaigns don't poison the polls, etc.