[EM] Re et al Chicken and Egg

[Kristofer Munsterhjelm]

David L Wetzell wrote:

> Me playing the sucker for punishment yet again with Kristofer...,

I'm not sure who the sucker is, actually!

>
>> On Wed, Dec 21, 2011 at 6:18 AM, Kristofer Munsterhjelm
>> <km_elmet at lavabit.com <mailto:km_elmet at lavabit.com>> wrote:
>>
>>>     On 12/21/2011 05:10 AM, David L Wetzell wrote:
>>>
>>>         Happy Holidays, I reply to RBJ, Ted Stern, Dave Ketchum and
>>>         Kristofer M
>>>         below.
>>>

>>>>             DK:But when marketers lie and get caught, potential
>>>>             customers get
>>>>             suspicious as to future marketing.
>>>>
>>>>
>>>      dlw: To simplify is not to lie.
>>>
>>>
>>    "IRV finds the majority winner". Given that IRV doesn't even do that
>>    by its own standards when it's limited to three ranks, is that a lie
>>    or simplification?
>
>
> It finds the majority winner more often than FPTP and always improves
> the percent of votes allocated to the winner over a comparable FPTP
> election.

That's not what FV says. FV says "finds the majority winner", period. 
When someone says "read my lips: no new taxes", he doesn't mean "no new 
taxes except property taxes". He means "no new taxes". If he then 
weasels his way out of it (as politicians tend to do) after introducing 
property taxes, the customers (voters) will get suspicious.

IRV only finds the majority winner to begin with if you define majority 
winner as "non-Condorcet loser" the way FairVote does. Even by the 
FairVote definition of "non-Condorcet loser", if it's limited to three 
ranks, or if enough voters fail to fill in all the ranks, IRV can't do 
what FairVote claims it does.

Thus, FairVote is "simplifying" (in a manner that to many customers will 
come off as lying -- see my "no new taxes" comparison) twofold. First, 
by redefining what a majority winner means. Second, by stating, even by 
their own definition, that IRV can always do what it only sometimes can 
do (bait and switch?).

(On a side note, I think it would be possible to construct a theoretical 
scenario where FPTP picks a winner that has less pairwise opposition 
than the IRV winner given these ballots. So "always improves the percent 
of votes allocated" is not strictly true, either. But you could probably 
dismiss that as a contrivance, so I won't.)

>>     KM: "IRV solves the spoiler problem". Given that IRV exhibits center
>>     squeeze, which by its nature involves spoilers, is that a lie or
>>     simplification?
>
>
> dlw: It's only a problem if the two major parties are not aligned around
> the center.  This is what tends to be the case in the US historically.
>   It's what would have become the case in Burlington VT if IRV had been
> retained.

Again, that's not what FairVote said. They didn't say "solves the 
spoiler problem as long as you vote strategically and thus reduce the 
influence of spoilers yourself, err...". They said "solves the spoiler 
problem". The Burlington voters found out that this wasn't true. They 
then repealed IRV. Who can fault them?

Or try this. Let's say I was going to market Plurality. I then claimed 
"Plurality always elects a Condorcet non-loser". Someone later comes up 
and tells me "hey! this isn't true when you have more than two 
parties!". I simply say "oh, but if you had let the system continue for 
a few more elections, the voters would have understood not to vote 
against the lesser evil, and so they'd all vote for one of the two major 
parties and they would win - hence Plurality would always elect the 
Condorcet non-loser". Would you be impressed by such an argument?

>>     KM:"In IRV, voters just have to vote honestly". Given the Burlington
>>     example, is that a lie or simplification?
>
>
> dlw: Once again, the standard case in the US, the audience of FairVote,
> is for there to be 2 major parties strategically positioned around the
> de facto center.  You cut the legs to what would have been an
> increasingly defunct GOP in Burlington, VT and it becomes 100% true.

It's a rather fragile balance. If the parties go too close to the center 
(de facto or not), they get center-squeezed. If the parties go too far 
away, they won't capture enough votes nor be at the center (de facto or 
not). How can you know that the fragile balance is exactly at where you 
want your de facto center to be, unless you define the latter in terms 
of the former?

Furthermore, that's besides the point. Of course you can make a 
statement true if you wave away all the counterexamples (like 
Burlington). But nowhere did FairVote say "voters have to vote honestly 
except when IRV seems to give significantly more choice than Plurality, 
in which case they then have to mop up IRV's problems themselves and 
vote strategically". They said "vote honestly", period.

>>     KM:Where do you draw the line between separating lie and
>>     simplification? Do you just fit the line or curve so that all the
>>     counter-IRV objections get clustered on the disingenuous side of the
>>     line and the IRV marketing arguments get clustered on the "oh,
>>     merely marketing simplifications" side of the line?
>
>
> dlw: Context.  And FairVote gets the benefit of the doubt in my book
> because they're the ones doing the spade-work of actually pushing for
> electoral reforms with US voters with very limited frames of reference
> for electoral reform.

Context is a really fuzzy description, good for moving a lot of 
goalposts. If you are going to be that kind to FairVote (out of context) 
and less kind to everybody else, that's your choice, but don't be 
surprised when the voters find FairVote's claims to be lies or 
politician-like "non-lie lies".

I don't think you need a "sophisticated frame of reference" to 
understand that "safe to vote however you want" and "safe to vote 
however you want except when there's real choice" is not the same thing.

>>     The uncertainty within the parameters that you call p_irv and x_irv,
>>     and the uncertainty within the truth that the party dynamics will
>>     behave the way you do.
>
>
> dlw: There's not much uncertainty about P_IRV being considerably bigger
> than P_Other, at least in the near future in the US.  As for X_IRV, I
> can only keep adding args that |Xirv-Xoth| is not-so-great in real life
> and that the absence of a clear-cut single alternative alternative to
> FPTP makes it so that lower PIRV does not raise POth proportionately.

Yes. P_irv is currently greater than P_other. That is too bad. I think 
P_irv is actually more like an economic bubble, because of the 
unappetizing properties of P_irv. If we do nothing, then P_irv will at 
some point burst. (That's what Jameson's wager is about). If we *do* 
start the race, we may overtake P_irv later and then we'll have avoided 
the risk of IRV blowing up in our faces.

As for |Xirv - Xoth| being not-so-great, you've really only showed that 
p(not(Xoth >> Xirv) | (DLW complex ecology comes to pass)). You've been 
somewhat circular in that respect. First you've said that the difference 
between other methods and IRV isn't so great because in your system, 
there will only be two major parties anyway. Then you've argued that we 
should stick with your "DLW complex ecology" because you can't get 
anything better with other single-office rules (because the difference 
between others and IRV is not so great) anyway. So you reach the 
conclusion that your system is the best if your system is the best.

Beyond that, my arguments have been twofold:

First, that (DLW complex ecology comes to pass) is rather arbitrary, 
i.e. that "your particular arrangement of parties into local and more 
widespread organizations, the contesting of two-party status by smaller 
parties under a duopoly, etc" will happen and is desirable. Here, you 
tend to throw "US isn't Norway" counters at me, even though other people 
on this list have stated they think we should move beyond two-party rule 
too.

Second, that even granted [dynamic two party rule], which is somewhat 
less arbitrary than your complex ecology, Xoth is much greater than 
Xirv. I've used Yee, but you don't think that ports to reality, not even 
the border arguments. I've pointed at center squeeze, but you think the 
de facto center is somehow within IRV's safe area but not Plurality's 
safe area. I've pointed at that other methods do better at avoiding 
embarrassing outcomes than IRV, but you say that it takes scheming 
Plurality campaigners to point out those outcomes and so they aren't 
important either. And so on...

> As for party dynamics, quasi-proportional PR in "more local" elections
> isn't rocket science, neither are the fact that it'd prevent either
> major party from dominating a state's politics, which is the basis for
> the perceived tendency among politicos in the US for a party to get a
> permanent majority.  If you take that prize away, you increase the
> incentives for whichever two major parties are on top to cooperate.

As long as your PR is PR enough, and as long as IRV won't pull too much. 
You're having a real balancing act on your hands, and your system 
requires such. Let's just dispense with that and go with something that 
will work without having to be fine-tuned.

And to get your PR by the means of that momentum you so like, you'll 
have to have FairVote agree to promote your system. You'll have to have 
them introduce PR soon enough that the IRV-without-PR doesn't give 
enough power to the major parties that said major parties can block PR. 
Remember, you didn't oppose Australia as an example of IRV tending 
towards uncontested two-party domination - you only said that your 
arrangement of PR could counterbalance it. If IRV gets there first, PR 
won't be around to do the counterbalancing, and then the big two will be 
even stronger than in Australia.

> dlw: I agree.  I think the root diff is our assessment of the badness of
> a two-party dominated system.  I believe from my country's history that
> it's not so bad if handicapped by the use of multi-winner elections in
> "more local" elections and if the duopoly is contested by minor parties,
> as was the historical practice in my country.   Because of this
> assessment, I don't mind the fact that IRV doesn't always get the
> Condorcet Winner when the two major parties fail to align themselves
> around the de facto center, as was the case in Burlington VT.  I see
> such as a way to incentivize the two major parties to realign themselves.

But IRV also fails (and you've said it did behave erroneously in 
Burlington) when parties go *too close* to the center. That's not a 
failure to approach the center, except by IRV's standard. And by IRV's 
standard, IRV is the best. Of course it is. That is tautologous.

I want to point out that "the badness of a two-party dominated system" 
was only one of the tines of my argument. I tried to show you that even 
if you want nothing more than contested two-party rule, IRV won't give 
it to you. No amount of reference to Europe will help that.

>>    If we had the same priors, i.e. the same judgement of either the
>>    likely range of these parameters or of the data from which we make
>>    the conclusion about the likely range of these parameters, we would
>>    be in agreement. We're not, so we don't.

> But the burden of proof is still on your side, not mine, due to how
> Pirv>>Poth.

"Pirv >> Poth" regards perception. I have argued that it's wrong. I have 
further argued, even "on your turf", as it were, that the other methods 
are better. So if you claim the burden of proof is on my side, I say 
that it has shifted, and that your counters of "well, within my complex 
ecology, IRV is good enough" isn't... good enough. Nor is "well, I'm 
middle-brow so I don't have to take that into account". Brow level tells 
us nothing about what really is the case.

>>>   PR does help 3rd parties(we agree right?) and American forms of PR
>>>   would tend to help LTPs since there'd be fewer seats in plenty of
>>>   "more local" "super-districts", this tends not to encourage
>>>   nation-wide 3rd parties.
>>
>>
>>     KM:PR does help third parties; or rather, we can say that systems
>>     that give political minorities representation help the organizations
>>     that represent those minorities. If your American PR does just that,
>>     then it will help third parties.
>
>
> dlw:By sort of helping third parties, we would give minorities more exit
> threat and thereby also more voice within the major parties.

If it's fine-tuned correctly. Again, I think that it would be better to 
just go to something that doesn't require that kind of fine-tuning. 
Instead of basing your construction upon an idea of how parties are 
going to act with respect to each other, look at the numerous examples 
from practice. Use real PR, not quasi-PR and know that, like in 
countries with STV, you *will* get minority representation within the 
body that uses it. By all means, use 5-seat STV, but please don't bias 
it by adding a majority component unless you compensate for it elsewhere 
like MMP does.

>>     KM:Of course, it is possible to twist PR so that it is no longer PR.
>>     Using a divisor rule with an enormous large-party bias would be one
>>     way of doing it, for instance. Having a very low number of seats per
>>     district is another way, as that imposes an effective threshold
>>     below which a group gets no representation.
>
>
> dlw: The barrier will be higher, but there'd be other positive effects
> without nailing proportionality.  But this is why I very much want 3
> seat LR Hare to get third party state reps elected.  I can live with a
> 3-5 seat STV with a Droop Quota for US congressional elections as a
> matter of realpolitik, but I think it's important to give third parties
> a constructive role to play in the US's democracy, which means helping
> them get representation.

Ordinary 3-seat gives a 25% threshold, which is rather large in my mind. 
5 is better.

As for realpolitik, I'd say this: if the powers that be had absolute 
power, there would be no discussion of electoral reform at all. So we're 
bound to upset the status quo somewhat just by advocating for a change 
that allocates power away from those who already have it. You could 
argue further that those who have power will oppose such reallocation to 
the extent that it is effective. They'll humor you if they think your 
change won't make a difference. Don't exert yourself just to water down 
your reform - you might concede too much.

In another post, you claimed RBJ was "defeatist". Well, isn't it 
defeatist to hobble proportionality just so the gatekeepers will be 
happy? Isn't it to hedge all your hopes on a model that has to be 
fine-tuned to achieve balance between plurality and domination, where 
the introduction of the two parts has to be done in sequence, when your 
tuning could be wrong or your timing off and you could end up with 
further uncontested two-party rule? The proportional representation 
leagues of the 20th century didn't have to hedge anything. They pushed 
for STV without any qualifiers, they got STV, and they got minority 
representation within the bodies that used it. They didn't say "oh, we 
can only hope that the powers-that-be are wrong and that our fine-tuning 
appears to give them continued power whereas it really gives a contested 
two-party rule". They said "let's get PR. Let's throw the machines out", 
and they *did* get PR, and they *did* throw the machines out. Only 
really dirty red-baiting tricks got the machines back in.

> dlw: It's based on my country.  Small parties don't need to build up
> enuf momentum to become minor parties who can challenge the major
> parties. The US is far more politically heterogenous than say Norway.
>   So even if our system tends to be dominated by 2 major parties, the
> "right" two parties will differ by state and that will enable minor
> parties who are able to be among the top two in at least one state to
> challenge the major parties.

If you get the balance right. How do you know that different states will 
have different major parties? Perhaps the influence of minor parties 
will be limited to poking at the big two, like what you mentioned below 
about Illinois and Minnesota. You should have something from which you 
can infer that "if we use IRV + PR, then at least in one state, a 
national third party will become one of the big two there".

Let's look at it differently. At one end of things you have the 
influence given to third parties by Plurality. Basically, this is when 
third parties say to the closest major party: "go in our direction or we 
pull voters to us, which will split the vote, which will get the other 
major party into office". That's clearly not enough to make third 
parties a counterbalance to money. At the other end of things you have 
true multipartyism. Disregard European vs US here, and just consider 
that multipartyism would solve the problem (otherwise there would be no 
such thing as a coalition).

Then there are two questions you have to ask. First: how do you know 
that your mix will pull far enough towards the diversity side that 
you'll get different local major parties? I suppose you'd use Illinois 
as your answer here. Did third parties come to lasting power in 
Illinois, or did they just "poke the sides of the major parties"? How 
about Wisconsin? Their Progressive Party doesn't seem to have lasted.

Second: why settle for this amount of influence with the tradeoffs you 
have to make? Don't voters deserve to be able to vote honestly most of 
the time? Don't third parties deserve to get their opinions heard beyond 
just as a check on the major parties? We know that from throughout the 
world, multipartyism *does* work. Not just in Europe, not just in a 
homogenous Norway, but throughout the world.
To this, you would probably answer "momentum and realpolitik". I've 
already replied to the realpolitik. As for momentum, it's not too late. 
If the advanced methods really are better, let's show that they are. If 
FairVote wants to champion IRV, ultimately we can't stop them, but it'd 
be better if we could just move to something better, all of us. 
Furthermore, p_IRV may be a bubble, as I've hinted. If so, the right 
thing is not to inflate the bubble further - the Plurality supporters 
have plenty of pins by which to prick it. Instead, we should choose 
something more durable. Neither the voters nor our opponents will be so 
forgiving as to say "it's their first try, let's be gentle".

>     KM:So, in the absence of all of that, the only thing we have is
>     theory. Well, theory and your Illinois example. I think that it's a
>     very risky assumption to put the Illinois example against Yee,
>     criteria, etc. and then say that the Illinois example wins.
>     Furthermore, I think it's silly to take that risk when the
>     opportunity of reform presents itself: better do it right and have
>     something that *will* work, than rely on limited examples and push
>     for something that only might.
>
>
> dlw: Has the Condorcet winner been used much for political elections?
>   If IRV examples are scarce then other purportedly better single-winner
> examples are even more scarce...   IL wins because it's empirical.  It
> lasted for a century and was only ended by a very misleading referendum
> campaign during a time of great antipathy against state legislators.

The Condorcet *winner*, sure. Even FairVote makes a point to that IRV 
picks the CW more often than does Plurality - which is part of why I'm 
so mystified that they don't want to add just a single little tweak to 
make IRV fully Condorcet. (They then point out "core support", but this 
seems more like a justification than a reason - it's easy to make 
elections where the CW has greater "core support" than the IRV winner. 
But I digress.)

Where Condorcet methods have been used, and some places where IRV has 
been used, too, the Condorcet winner has been elected. It's hard to say 
how often Plurality (or top two) has picked the Condorcet winner since 
you can't easily infer the voters' opinions from their preferences 
there. The Range Voting page gives some elections where it appears that 
the top-two second round winner was the right one but the first-round 
(Plurality) one wasn't: http://rangevoting.org/FunnyElections.html , but 
beyond that, it's hard to tell.

You probably meant a Condorcet *method* rather than a Condorcet 
*winner*, though. Off the bat, let's mention Marquette, Michigan, which 
used Nanson's method for local elections. I don't know for how long they 
did, but they did use it. Then we can continue by listing political 
organizations from the Schulze method Wikipedia article:

"Pirate Party of Australia
Pirate Party of Austria
Pirate Party of Brazil
Pirate Party of France
Pirate Party of Germany
Pirate Party of New Zealand
Pirate Party of Sweden
Pirate Party of Switzerland"

And before you wave those facts away by saying the Pirate Parties are so 
tiny they don't count, keep in mind that the Pirate Party of Sweden have 
representatives in the Parliament of the European Union, and that Pirate 
Party of Germany members hold seats in the House of Representatives of 
Berlin.

Those are the most clear cut ones. If you want to include elections 
within organizations involved in politics in a more indirect manner, you 
can also include the Free Software Foundation of Europe, as well as the 
Free Hardware Foundation of Italy. The Free State Project also used a 
Condorcet method (Minimax) to determine which state to focus on.

If you want to consider internal elections in organizations in general, 
then there are more.

>> KM:You still haven't given me any numbers. If we're going to resolve
>> anything, we'd have to find some kind of agreement as to what data
>> would be accepted.
>
> dlw: Case studies.  I'd like lots of case studies. There are (at
> least) three kinds of lies: lies, damned lies and voter utilities...
> I distrust rational choice models.  I do not expect hard numbers for
> Xs and find analytical args over which rule has the biggest X to be
> circuitous and not fruitful for raising the P of some lucky election
> rule...

P is (honestly) raised by advocating for a rule. P is lowered by 
pointing out problems with a rule. Analytical arguments take two forms: 
either that you show that generally, the outcome will be better; or that 
you show that there won't be problematic cases. The former boosts P 
directly, the latter shows that the current P is based on fact and thus 
makes the method more resistant to backsliding.

Raising P by closing your eyes and say "I do believe in momentum" will, 
at best, get you through with a method that's so-so when you could have 
got something that's much better. At worst, it'll inflate a political 
bubble that'll burst - and when it does, the fallout will hurt all 
electoral reformers.

As for case studies, I don't know how detailed a study you need. If you 
need multiple elections with the observation of dynamics that follows, 
you're not going to have many examples. Your best hope there would be to 
look at the organizations I have mentioned that use Condorcet internally 
- or for that matter, the UN's use of Approval. I haven't pierced the 
organizational veil enough to know what kind of dynamics have arose from 
the use of those, but I do know that they are still using these methods. 
Some in the Swedish Pirate Party have argued that the Condorcet method 
they use (Schulze) is so complex as to be technocratic rather than 
democratic, but despite this, the method is still in use.

If you lower your bar a little, there is the Orsay exit poll. Balinski & 
Laraki got 74% of the voters in Orsay, France, to vote Majority 
Judgement-style in an exit poll. The results showed a different winner, 
and the "spoiler" (Le Pen) got decisively placed last in comparison to 
his significant (if not winner-level) support in the official election. 
Furthermore, as I have stated before, the voters acted as if there were 
more serious candidates - there was less of a difference between the 
results than in Plurality - and the winner (Bayrou wins) was consistent 
with the results of larger-scale pairwise polls (of the form "who would 
you vote for in a hypothetical second round", where the polls showed 
Bayrou would beat Sarkozy, the actual top-two winner). So I think it's 
pretty clear that MJ improved upon top-two here. Sarkozy won in the 
top-two and would likely also win under IRV - although you could fudge 
the data to make any of the top three win under IRV.

You can see the details of that poll and analysis at 
http://rangevoting.org/OrsayTable.html .

Other French 2007 studies have also been made, using different voting 
systems, and you can look at 
http://rangevoting.org/French2007studies.html for the data. Bayrou wins 
in all of them except the official top-two count and IRV. There is a 
caveat: the voters in the ranked ballot poll was told the method would 
be IRV - the ballots were then run through other methods by Warren. 
Bayrou, the CW, came in third in the IRV order.

In 2002, Laslier and Straeten ran an Approval exit poll in Loiret and in 
parts of Orsay. 77.6% of the voters from these voting stations 
participated. Again, results were more even than with Plurality, even 
given that the Plurality round was just the first round of TTR (which 
would tend to make for more even results than a "have to vote for the 
lesser of two evils" one-round Plurality count). Warren has some data at 
http://rangevoting.org/FrenchStudy.html , and that page also links to 
the papers themselves (where you can read details of the experimental 
setup and so on).

There have also been Approval studies in Germany. See 
http://rangevoting.org/GermanApprovalStudies.html . Yet again, the 
results were a lot more even than under Plurality; people acted as if 
there were more choices.

But I can hear you already: "those are just the EU-friendly nations 
being used to multipartyism. Here in America, we have two parties! We're 
used to that, it's our way!". That's one way to do the "let's make the 
criteria so strict nothing passes muster so my method wins by default" 
dance. But even if you do, wander over here:

http://rangevoting.org/PsEl04.html . Another exit poll, this time in the 
United States, in 2004. Under both Range and Approval voting, the voters 
acted as if the third parties were more serious choices than under 
Plurality. In the Approval context, you can pretty much see Bush holding 
the conservative/Republican half (less a little), Gore and Nader holding 
the liberal half (less a little), and a bunch of small parties taking 
the rest. In Range, it's the same, but the small parties are no longer 
sliver-small.

So that doesn't convince you, either? Polls of the sort "how much do you 
like the candidates" regularly show a more even, more multiway race than 
do Plurality elections in the US. See 
http://rangevoting.org/RangePolls.html , which I've mentioned before.

In summary, voters seem to show that they aren't inherently two-party. 
When they vote, they vote in a way that shows there's a contest not 
among just the top two, but that the contest also includes parties that 
are usually considered "third" or "minor". And we know from the methods 
that they don't suffer center squeeze, so they can also make good use of 
the additional information without the voters having to twist themselves 
into an "I only like two parties" shape (as you want them to do under IRV).

If the results I've given for some reason aren't valid, then you still 
won't get a worse result with the advanced methods - but if you stick 
with IRV and the people really want more than two real choices (as seems 
likely by interpreting the above), then you've got a problem.

>> KM: Say, for instance, that I run the Bayesian regret calculations,
>> and that I decide to limit myself to four candidates (even though I
>> think there should be more), so that you can't dispute that aspect.
>> Say, further, that I get a result that the best Condorcet rule
>> improves upon IRV about 10% as much as IRV improves upon Plurality.
>> Then you could easily say "see, Condorcet isn't worth it". On the
>> other hand, if I got a result that the best Condorcet rule improves
>> upon IRV 10x as much as IRV improves upon Plurality, then you could
>> also claim "yeah, but that's a static simulated result under
>> conditions that aren't realistic, so in reality X_other - X_irv is
>> still small".
>
> dlw: I'd say it'd be impressive regardless, but I get your point...
>
> I would then add: I think 4 candidates is more realistic, given the
> tacit equitable distributional assumptions used to generate the lies,
> uhr voter preferences used for Bayesian Regret calculations. I also
> think that the X of Condorcet would be lowered more than the X or IRV
> by the bounded rationality of voter, since there'd be more GI and
> more GO and it makes sense that the top ranking would have more
> signal and less noise for many voters. Furthermore, the hybridization
> of IRV and Approval Voting would raise the BR and lower the gap...BR
> analysis has shown IRV to do its best relatively when there are only
> 3 candidates and that is the context in which the hybrid would use
> IRV.

In other words, you think the comparison wouldn't be right if it went 
against you. If you can rewrite the rules of the game after the results 
come in, then you can't really blame me for not wanting to play :-)

By the way, I already countered your claim of GIGO. Olson's graphs show 
IRV does worse with noise than does Condorcet, and the Condorcet cycle 
argument actually goes in favor of Condorcet when people truncate. 
Remember: if everybody truncated beyond their first choice, a Condorcet 
cycle would be impossible. If everybody fully-ranked, a Condorcet cycle 
is possible. Now what do you think will happen as you decrease the 
average number of submitted ranks? More or fewer Condorcet cycles?

(Also, empirically speaking, Google uses local Kemenization as part of 
their PageRank algorithm to attenuate the influence of the noise 
injected by SEO spammers. This procedure brings the results closer to a 
Kemeny (Condorcet method) outcome.)

Your use of hybridization is good compared to unadorned IRV. You only 
have to go one step further - from AV3/IRV3 to AV2/IRV2. Have you got 
FairVote to go even to AV3/IRV3? If so, why can't they - or you - go 
just one more step?

> Just as I think it's safe to say that the X of Approval and Score
> Voting would be lowered more by relaxing the cardinal utility
> assumption implicit in most Bayesian Regret calculations.  Let Xij be
> the initial cardinal utility of voter i for candidate j.  Let Si be
> generated from a log-standard normal distribution.  Let all the Xijs
> be transformed to become Yij=(10^(1-Si) * Xij^Si).    Then, decide
> who to vote for based on the Yijs, while assessing the results based
> on the Xijs.
>
> The bottom line: Bayesian Regret is a heuristic and hence the proofs
> gotta be in the pudding.

Hence my question as to what kind of pudding you would accept. For 
instance, if I did the above transformation and Range (or MJ) got 100% 
relative improvement wrt IRV as IRV wrt Plurality, would you change your 
mind, or would you just move the goalposts?

>> KM: After the fact, it would be simple for either of us to readjust
>> the rules of the game, as it were, so that we get off free. If the
>> Bayesian regret heuristic is going to solve anything, it must have
>> power, and it doesn't have power if we can just step around the
>> result no matter what it might be.
>
> dlw: It's one of those fuzzy things that can be persuasive without
> being defnitive in my book.  My views on election reform have been
> changed from my debates with Dale Sheldon Hess and also Broken
> Ladder/Clay Shentrup).  I'm more open to other election rules than
> IRV in the long run because of their work.  But as you know, in the
> long run, we're all dead...
>
>> KM:I suppose, then, that what I'm really saying is this: you discard
>> theoretical points by saying theory isn't practice,
>
> dlw: I don't discard or disregard theoretical points, I express
> diffidence towards theoretical points, based on an ethos that's
> somewhere between a "critical realism" and "instrumentalism" that I
> learned in my studies of institutional economics.

They seem rather tuned to let you get off with your points. Robust 
center-squeeze weirdness in Yee diagrams (that hold up even when the 
center shifts) doesn't matter to you, for instance. GIGO working in 
favor of Condorcet? You don't comment on it, but you do when you think 
it works against Condorcet and in favor of IRV.

And let's look at the GIGO thing further. You say "I think that voters 
will be noisy below the first ranks or just plain truncate". If I had 
said that, and said "so Condorcet is better", you would say that I don't 
have any data by which to back that up and so it's only theory. Yet you 
make that claim against Condorcet: "I think voters are noisy". 
Micronumerosity goes both ways, but you seem to press it a lot more when 
it's in your favor to do so.

>> KM: that you're middle-brow so it doesn't matter anyway,
>
> dlw: Since as a middle-brow, I believe theory is essentially a crutch
> for coping with a complicated, messy reality, not a precision laser
> for pinpointing the right election rule.

As I said earlier, your brow level doesn't affect reality. I have also 
given real world examples above.

> KM: and even *if* they showed the other rules are better, they don't
> show the other rules are *that much* better. You discard what little
> practical (experimental) data we have by saying that it's
> inapplicable (AU) or that the conclusion was just because of
> interference from scheming Plurality advocates (Burlington).
>
> dlw: Micronumerosity sucks.  As also does the problem of historical
> specificity.  That these problems are relevant to the matter at hand
> is not a matter of opinion, in my opinion.

The problem of micronumerosity is itself real - when you have few 
samples, you have a greater margin of error. However, I don't think 
you've been entirely fair here. You use a model of your own devising 
(what I've called the "DLW ecology") based on a few examples that are 
close to what you envision yet not quite there (Illinois, Wisconsin, 
neither of which use IRV) to state that more numerous examples either do 
not apply (Australia) or aren't in line with what you want anyway 
(examples of true multipartyism, top-two runoffs, etc). You do not say 
anything about what the voters really want, and you omit examples of 
multiparty rule in the US without IRV (locally, New York under STV) but 
press historical specificity when you'd otherwise have to weaken your 
position (multiparty rule in Europe and NZ). Then, after you've 
sanitized the field of samples that could count against your position, 
you state that there aren't enough instances of real world data to get 
you to change your mind.

Sure, you can always say there aren't enough examples if you raise the 
bar high enough, and you have a different bar for what's scarce when 
it's your data (Illinois) than when it's ours (Burlington). You can do 
these things, play with standards and levels of proof unequally. I can't 
stop you -- but playing like that isn't exactly conducive to agreement.

As for historical specificity, you could argue this way if different 
societies used different electoral methods, and consistently so. If 
Europe overwhelmingly used PR (without DLW ecology type 
counterbalancing) while Asia overwhelmingly used SNTV while America 
overwhelmingly used parallel voting, you'd have a point. However, let's 
take a (by no means exhaustive) list of nations and territories using 
proportional representation for their legislative bodies:

Algeria, Aruba, Benin, Brazil, Burkina Faso, Cambodia, Cape Verde, 
Colombia, Costa Rica, Cyprus, Dominican Republic, El Salvador, 
Guatemala, Guinea-Bissau, Guyana, Iraq, Israel, Lesotho, Mozambique, New 
Caledonia, New Zealand, Niger, Paraguay, Peru, Rwanda, Saint Martin, São 
Tomé and Príncipe, South Africa, Sri Lanka, Suriname, Turkey, and Uruguay.

To argue for historical specificity, i.e. that "the US is different", it 
seems to me that you'd have to argue that the United States is different 
in some way that affects the best choice of electoral method, and that 
it is on one side of a partition (call it the "needs managed PR" 
side)where all the places above are on the other side ("plain PR / 
multipartyism works" side). What history does Saint Martin have in 
common with Israel? Are you saying Cyprus is historically more similar 
to Sri Lanka than the United States is to New Zealand? The dividing line 
that places all the PR nations on one side and the US on the other 
seems... arbitrary, almost like it's fitted in advance to make the 
conclusion you want to have.

>> KM: At that point, very little remains. Thus I ask: what would it
>> take to change your mind? What demonstration, what experiment would
>> give you the data needed? What sort of argument would meet your
>> "middle-brow, true test of IRV" standards? If your answer is
>> "nothing", then we're done and this is just text on a screen.
>
> dlw: Push for multi-winner elections in the US.  Trust that third
> parties will enable an expansion of electoral experimentation,
> facilitated by smart electoral analysts like the many people on this
> list, to break the impasse. Recognize that in lieu of convincing
> empirical data over the Xs of electoral alternatives, it's logical to
> focus more on the Ps in the short run and strategically support IRV,
> or decline to strongly critique IRV, as a clear improvement over FPTP
> for the US's two-party dominated system that can be marketed to the
> US public.

That isn't a standard for an argument or experiment. That's *your 
strategy*, what you want in the first place! I'd ask again "What would 
make you rate IRV lower than you do?", but I think you've answered so 
before (case studies). I've replied to that above, and I hope you have a 
response to it. I do think that provides "convincing empirical data over 
the Xs of electoral alternatives", or close enough that inference can 
get us the rest of the way.

>> KM: (Incidentally, I didn't see you reply to the 36% backsliding rate
>> for IRV. Were all of those due to scheming Plurality advocates?)
>
> dlw: As with the Alternative Vote in UK, when the third party
> benefactor of an electoral reform loses popularity, it makes the
> election rule less popular.  If only we could assess election rules
> while wearing Rawlsian masks that make us abstract from whether
> they'd help our self-interests in the short-run...

What third party benefactor? Your system will only have two parties, 
after all, and a bunch of minors that may pull the parties in their 
direction, but only indirectly.

But since you didn't object to the 36%, I assume you agree there's a 
significant backsliding rate for IRV. You only disagree about the 
reason. Is that right?

> dlw: AU uses IRV in "More local" elections where it is less likely
> to help due to de facto segregation by characteristics correlated
> with political preferences.  It uses PR in "less local" elections
> where it is less needed.  And so yes, it's not generalizable. I
> consider IRV to be reliable improvement on FPTP and two-round
> elections, moreso when coupled with the strategic use of PR that
> takes the edge off of how IRV does not tend to end effective 2-party
> domination.

Then we're back to runoffs again. Do you agree that IRV countries tend 
towards uncontested two-party domination in the bodies that use IRV, 
whereas top-two countries don't, all other things equal? If so, how can 
the IRV component be better than top-two as part of your system, yet 
worse than top-two by its own?

Disregard PR. PR is an addition, and can be an addition to either 
top-two or IRV. That's why I say "all other things equal".

>> KM: Would IRV + PR be better than Plurality + PR? If you've
>> constructed all of this from the Illinois example, which did use
>> Plurality, why IRV? Or is IRV just an expedient, something one has to
>> swallow to get the whole FairVote package, PR and all, through?
>
> dlw: I support FairVote as the de facto leader of election reform in
> the US.  They deserve that status for their spade work on
> communicating election reform concepts to US voters ignorant of the
> electoral debate.  But yes, I believe IRV+PR > FPP+PR >> FPP.  At
> issue is how much better, I don't know.  I have been critical of
> FairVote in the past because they seemed to put too much of their
> political capital on IRV, as still is the case with the independent
> FairVoteMN in my home-state.  I am glad that's going to change for
> FairVote.  I want to do whatever I can to get that ball rolling in my
> country, including spending time on this list-serve arguing that such
> is more important than trying to replace IRV with another election
> rule.

If you can get FairVote to focus more on PR and less on IRV, more power 
to you. I hope FairVote will move in that direction, ideally to the 
point they throw IRV away. You'd like to have PR and I'd like to have PR 
too. If you're concerned there will be vote splitting, try to make 
FairVote a dedicated PR organization. I think PR would lead to fewer 
complaints than IRV, too -- many seats solve a lot of ills.

Don't throw that agreement over PR out just because FairVote currently 
feels like chaining it to IRV. There's even less of a reason for you to 
hold on to IRV at the cost of others' support for PR if you, as you say, 
don't know how much better IRV+PR is than FPP+PR. Then we can do the 
single-winner thing in parallel -- or do it later.

>>> dlw: I recommend that you not push for the use of IRV in Norway.  I
>>> also hope that you and others do not stymie the coupling of
>>> IRV(hopefully in hybrid form) + Am. Forms of PR that is emerging as
>>> what progressives/centrists/activists are going to be rallying around
>>> in the US.  When smart people like you and others here state
>>> unequivocally that Xoth>>Xirv it lowers Pirv without increasing
>>> Poth.
>>>
>>>
>> KM: I hope that the momentum (to what degree it exists) can be turned
>> in the right direction. Condorcet methods are being used in
>> organizations (as well as a political party) right now. The United
>> Nations uses Approval.
>
> dlw: In orgs with fewer better informed voters and more serious
> options, condorcet methods would be clearly superior to IRV.

You mention "more serious options". Since you like focusing on dynamics, 
I ask this: if you have a method that won't reliably deal with multi-way 
contests, how are serious options going to arise in the first place? You 
can't just say that "IRV is better because there's less choice" when IRV 
makes it hard for more choice to actually appear, or for voters to take 
those increased, more serious choices into account without risking a 
backfire. And I think that the exit polls, as well as the Burlington 
result, shows that the potential for serious choices *does* exist. Don't 
deny the voters the ability to make use of them - don't teach the voters 
that they better stick with the major party because otherwise the 
center-squeeze will hit them in the face.

You also mention "better informed voters". Why are the voters going to 
bother becoming better informed if the method they're using boils down 
to an election between Major Party X and Major Party Y (under 
Plurality), or "Major Party X, Major Party Y, or scare either candidate 
by voting for a third candidate - but not if there's a real danger of 
center squeeze" (under IRV)? If you want better informed voters before 
you'd think an advanced method is good enough, and the voters won't take 
the effort to become better informed within the context of a method 
where their increased information won't make a difference, then the 
electorate will never become informed enough for an advanced method to 
be worth it. To want to become informed, they'd need a responsive 
method, but they can't get the responsive method before they're informed!

(I incidentally think they are already informed enough. The ratings and 
Approval polls show opinions beyond "major party good, challengers 
bad!". Turnout rates, which are also about the effort the voters feel 
like making, are also higher in PR democracies.)

> Likewise with umpteen variations of proposals on the docket in the
> UN, Approval likely would be better than IRV.  The trend is for
> experimentation away from FPTP.

Please consider the data before you construct an explanation. The UN use 
of Approval is not for "every umpteenth variation". It is for the 
election of Secretary General. Last election, there were six candidates 
for Secretary General. That's not umpteen.

Back to what I was saying. Do you think organizational momentum could be 
used to advocate advanced methods elsewhere?

>> KM: Furthermore, I would not say that my statements that X_other >>
>> X_irv "lowers" P_irv as much as that it brings P_irv in line with its
>> true value, should it persist. That is, I think IRV has several
>> unappetizing properties that, when discovered, will have people leave
>> it; and it's better they see those aspects now than later, so that
>> they don't pull so much of the general idea of electoral reform down
>> with them when they *do* see the bad parts of IRV.
>
> dlw: Tell that to the nearly 50% of Burlington VT voters who wanted
> to keep IRV.  P_irv is a short-run concept.  I don't care what it's
> true value is in the long run, after all, I'm middle-brow on such
> things...

So you're middle-brow with respect to the truth as well? To me it seems 
that you're saying that "it doesn't matter to me if I inflate P_irv over 
the level it would have if people were to get a really good long-term 
look at it; all that does matter is that P_irv gets high enough, right 
now, that we can push it through before the voters figure out what's 
going on". That's not exactly voter-friendly.

> KM: True, claiming that X_other >> X_irv might not do much to
> P_other, except perhaps by making it less likely that IRV's flaws
> will taint the other methods. Raising P_other is a separate concern.
> It can be done by people signing the declaration, or when the parties
> and organizations currently using other methods serve as momentum of
> their own.
>
> dlw: So you recognize that the old saw-horse of divide and conquer
> might be at work to stymie election reform in the US???

I don't think that the scheming status-quo-ists are deliberately playing 
IRV against Condorcet etc. The latter is below their radar. There might 
be some implicit vote splitting when we lower P_IRV, but we should not 
*only* do that. We should also raise P_Other by (but not limited to) the 
means I've already given. Furthermore, if the decision of what sort of 
reform to actually make use of is done in a good manner (like in New 
Zealand or by a commission in ways that have been mentioned before), 
then the implicit vote splitting is minimal. Do recall that NZ had a 
two-stage question setup, and commissions don't have to limit themselves 
to Plurality-style thinking.

>>> dlw: Ethically, the burden of proof is on those who oppose the
>>> working consensus proposal for reform to show that their preferred
>>> approach is considerably better. I believe I have been holding to
>>> this ethical principle in my emails on this list.
>>
>>
>> KM:You have given proof based on your assumptions. You have then
>> argued, based on those assumptions, that the counters were not
>> applicable (such as that the theory is too theoretical or doesn't
>> impact enough).
>
> dlw: I have given plausible tweaks to rational choice assumptions
> that serve to argue that |X_irv - X_oth|  < P_irv - P_oth and that
> the strategic use of PR in the US is of even greater importance than
> purportedly improving upon IRV.  My args have been deconstructive
> such that what matters in teh short run are the short-run Ps, not the
> Xs or the long-run Ps.

We're not in it for the short run. This isn't a sprint to get across the 
finish line before the voters go "hey, what's up with these failure 
modes?". It isn't a race where we put all our eggs in the 
"untried-in-practice DLW ecology + IRV + PR can work" basket. I don't 
think the preconditions you build "X_other - X_irv is small" upon hold, 
and I don't think we have to limit ourselves to the DLW ecology. If we 
pick a good method and it gives only the DLW ecology, all we lost is 
time; if it gives us more than the DLW ecology, all the better!

And I say again: if you think PR is even more important, go do your PR. 
Just don't contaminate it with IRV, and we can all be happy! If you 
think we're interfering, try to get FairVote to focus on PR - the PR 
that you after allwant - and then leave others to single-winner reform. 
Your Illinois example, if you trust it, should show that PR + Plurality 
can work - and you seem to think, yourself, that single-winner isn't 
that important.

>> On 12/21/2011 05:10 AM, David L Wetzell wrote:
>
>>> Let me add RBJ that I really do appreciate your comments in response
>>> to Kathy Dopp.  I would add that if the GOP/Prog Haters cd go back in
>>> time to the '09 election then IRV would have worked better because
>>> more of them would have voted strategically for the Dem candidate as
>>> their first ranked choice.  So I'd say 2009 was a learning
>>> election....  and I have no problems whatsoever with some still
>>> having to vote strategically.  I see this as a consequence of how IRV
>>> retains a tendency for there to be 2 major parties.  What it does is
>>> makes it so those 2 major parties are more dynamically drawn to be
>>> centered around the shifting de facto center.
>
>> KM: Also, let me see if I got this right. You're saying that instead
>> of having potentially direct multiparty rule by the use of an
>> advanced method (or at least, no worse a rule than under IRV), you
>> want to have two-party rule. You want to use a method that behaves
>> strangely in certain situations, and you want the voters to take up
>> the burden of making it behave properly by having them vote
>> strategically.
>
> dlw: I accept that 2 party domination is inevitable in the US or that
> it's much, much easier to get election reforms in a two-party
> dominated system when they do not challenge 2 party domination.  What
> matters is that the 2 parties are given incentives to become more
> dynamic and that outsiders/dissenters are given voice and
> opportunities to move the center around which the two major parties
> center themselves.
>
>> KM: So instead of a rule where people could vote mostly-honestly and
>> could possibly get multipartyism directly, you want something where
>> people still have to vote strategically and won't get anything more
>> than (possibly contested) two-party rule even when they do. And why?
>> Because all the momentum's with IRV?
>
> dlw: Because even if I thought multi-party> two-party, I don't think
> it's >> two-party and as a matter of political cultural change, it's
> best to pick one's battles.  In my view, the critical problem in the
> US is it's tendency to single-party rule, not whether there's a
> tendency to two-party domination.  The latter can be contested more
> robustly and be more just if the two parties are more dynamic.

Focus. Once more, with emphasis:

Let's say I am on Let's Make a Deal. Behind door number one, Monty has a 
voting method where people have to vote strategically whenever a third 
party gets strong enough. Behind door number two, Monty has a voting 
method that is like the one behind door number one, except that the 
voters can vote honestly *also* when there is a strong third party.

Please tell me: why should I, as a Burlington voter, open door number 
one instead of door number two? Why should I, the voter, have to bend 
over backwards to accomodate the failure modes of IRV just because 
FairVote happened to stumble across IRV before they stumbled across 
Condorcet?

Note that this thrust of the argument has nothing to do with 
multipartyism. I only brought in multipartyism because the voters might 
prefer it to two-party rule and an advanced method might bring it about. 
If they do, and if it does, that's another advantage to door number two. 
If they don't, well, then that doesn't *hurt* door number two. That's 
why I said "possibly" and "potentially". Even if I lower my aim to your 
standard, the fact is that, in Burlington type scenarios, if the voters 
had picked door number two, they *could have voted honestly*.

The voters thought they could vote honestly with the method behind door 
number one. They then found out that the method (IRV) malfunctioned when 
they did, contrary to FairVote's lies, err, "simplifications" about that 
you can always vote honestly. If the voters then act as if there was a 
goat, not a car, behind door number one, you can't wave it away by 
saying "well, if only they had voted strategically...". That's the 
*whole point*. With IRV, the voters have to do in their minds what the 
method can't do itself, and thus vote strategically.

You've admitted as much. You said that IRV exhibited center squeeze in 
Burlington. You said that the voters would have to vote strategically to 
get around that center squeeze - at least until the Republican party 
makes its exit. So that is not contested by you.

Furthermore, we *know* that Condorcet would have given the right result 
in Burlington where IRV did not - unless, for some weird reason, people 
would vote honestly under IRV but not Condorcet. Therefore, we know that 
door one (IRV) required strategy and door two (Condorcet) didn't.

Why should the voters be burdened with strategy? How can you demand of 
the voters - whom the system is supposed to serve - that they make the 
right strategic contortions to fit your system? The system should fit 
the voters, not vice versa. All the expediency arguments in the world 
won't make the voters less angry when they find out that because they 
didn't compromise for "Liberal" instead of voting honestly for 
"Conservative", "More Liberal" won.

And if they can get not just freedom from strategy under a realistic 
scenario, but also direct multipartyism, that's icing on the cake.

I suppose you could argue that your DLW ecology is better than 
multiparty democracy to a greater degree than being able to vote 
honestly is better than having to vote strategically; or you could claim 
that Burlington won't ever happen again while, on the other hand, 
momentum is everything... however you put it, it seems that you would 
have to state that "momentum + strategic voting + DLW ecology at best" 
is preferable to "less current momentum + honest voting + DLW ecology, 
possibly direct multipartyism". Don't expect the voters to agree with 
you if you sacrifice their confidence in being able to vote honestly for 
simple expediency.

>> KM: Doesn't that sound a bit bizarre? "I know of a voting method
>> where you could vote honestly, but because FairVote got to IRV first,
>> you'll just have to pinch your nose and vote strategically under IRV
>> instead". Do you think that argument would go over well with the
>> voters?
>
> dlw: More like, "There are an infinite number of election rules.
> Many claims have been made that such and such an election rule is the
> best.  I make no such claim.  Neither do I claim that an even playing
> field across all parties will end our problems.  What matters most is
> that we subvert the cut-throat competition between our two biggest
> parties and make them both give more voice to more people on more
> issues or else face their replacement by other parties.  IRV and PR
> are well tested ways to do this.  They'll make our democracy into a
> melding pot that balances the need for hierarchy and equality,
> continuity and change."

Well-tested? IRV+PR has been "well tested" in Australia -- but that 
doesn't count. Where has your particular IRV+PR configuration been "well 
tested"?

Besides, that's only a side part of that argument. The voters still have 
to use strategy.

>> KM: Why should we expect the voters to jump through hoops when
>> electoral reform is supposed to remove the need to jump through
>> hoops? I could understand your tradeoff if all that jumping through
>> hoops gives you something (e.g. multipartyism) the other rules don't,
>> but IRV doesn't even give you that.
>
> dlw: It sets it up so that both major parties must adapt, listen to
> dissenters, so as to reposition themselves around the true, moving
> center.  When we become less ideologically stodgy, escaping the
> tailspin we got caught in in recent decades for a number of reasons,
> we won't need an EU-style multi-party system.  We'll do it our
> way...

Why should we expect *the voters* to jump through hoops? The voters.

And I'm sure your description of multipartyism as "EU-style" will come 
as a surprise to the people of New Zealand, or of all the other nations 
I've given earlier. Multiparty democracy is not an European thing alone.