[EM] More non-altruistic attacks on IRV usage.

[Kristofer Munsterhjelm]

We're still hitting the same disagreements. I say "look at the others", 
you say "this time it'll be different", I say "Condorcet >> IRV", you 
say marketing differences are great while in practice, there's no 
difference between Condorcet and IRV large enough to make a difference.

Thus, let me do some asking, because we're not getting anywhere. 
Consider in your mind: what kind of data could I show that would change 
your mind about whether IRV is stronger in the hegemonic direction than 
PR is in the enabling-contesting-parties direction?

Furthermore: On what do you base that reality is:

0 Plurality
0.7 IRV
0.72 Condorcet

rather than:

0 Plurality
0.25 IRV
0.72 Condorcet?

You keep saying that X_Condorcet - X_IRV is small. Is that just a 
belief, or do you have something on which to support it?

Some replies below.

David L Wetzell wrote:
>>    KM:If the cost of campaigning is high enough that only the two major
>>    parties can play the game, then money (what you call $peech) will
>>    still have serious influence. 
> 
> 
> dlw:My understanding/political theory is that $peech is inevitable and 
> all modern democracies are unstable mixtures of popular democracy and 
> kleptocracy/plutocracy.  To bolster the former, we must accept the 
> inevitability of the latter.  This is part of why I accept a two-party 
> dominated system and seek to balance the use of single-seat/multi-seat 
> elections and am an anti-perfectionist on the details of getting the 
> best single/multi-seat election.  Deep down, I am skeptical of whether a 
> multi-party system improves things that much or would do so in my country.

The mode of corruption you have in the US, where monied interests openly 
give the organizations that participate in the political process power 
by which to be seen, seems to be a particular thing to the US itself. To 
my knowledge, there's no Canadian OpenSecrets, nor, for that matter, a 
New Zealand one.

In more corrupt nations, corruption usually happens within the context 
of the state itself, and on all levels: you might be stopped by a police 
officer who wants some money to not claim your car has a problem -- or 
parties might pay the electoral commission not to rig the vote as heavily.

In other first-world nations, the parties may be given money, but such 
is usually tightly regulated. Thus, the corruption is less overt - 
corporations may collude to fix prices in a state bid, for instance, or 
try to convince appointed officials or mid-level bureaucrats that it's 
better if they do it "their" way.

Now, you could say that just supports your conclusion: if the US is 
different, then multiparty won't help it where it helped other nations. 
But you could also turn this the other way, and say that the difference 
between US and the other nations is that the US has two effective 
parties both by EFNPP as measured by seats and votes - i.e. that the 
reinforcing process of Plurality has gone so far that people are 
resigned to two parties alone. If so, to reverse the corruption, you 
should let other parties but the big two grow -- and other parties but 
the big two be seen as having a chance.

> dlw: It is counterbalanced by the fact that in a system with more 
> competitive elections, intere$t$ would need to hedge between the two 
> major parties and consequently accept a lower, more variable return on 
> their $peech and  that there'd be turnover wrt which two parties are the 
> major parties so it'd be a contested duopoly.  

A hedge among ten is better at that than a hedge among two.

> It is also counterbalanced by political cultural ways to move the 
> political center.

Political ways that will be hampered because other parties on the ascent 
to meaningful opposition to the big two will have to do a tightrope walk 
between appealing to the center (get more votes) and not appealing to 
the center (or they'll be center squeezed).

> dlw:Burlington's two major parties would not be the same as the two 
> nat'l major parties.  Republicans would vote Democrat in Burlington 
> mayoral elections.   This would add to the ferment of the system as a 
> whole being a contested duopoly or contribute to  a shift to a new 
> duopoly between Prog-Dems and Dem-Pubs.

Is it really worth the marketing advantage to burden people with having 
to vote strategically, or the parties to have to keep that in mind when 
they position themselves?

> dlw:It should be emphasized here that more "more local" elections would 
> tend to be multi-winner/PR.  This permits LTPs to win seats without 
> having to move too much close to the de facto center.  This gives them 
> the chance to move the center and/or possibly center squeeze the two 
> local major parties in single-winner elections.  I agree this could get 
> complicated, but I believe that the potential to center-squeeze is what 
> makes the center tend to become more dynamic.  And the unpredictability 
> is not unlike a similar undpredictability due to the nonexistence of a 
> Condorcet winner when there are 3 strong parties.  Plus, 
> unpredictability can force intere$t$ to hedge further, naturally 
> checking its influence and facilitating learning.

Center-squeeze is, in itself, not a good thing. When it happens, it 
means that although more voters are closer to a different 
candidate/party (the center), your method picked the winner it picked. 
In Bristow-Johnson's terms: it means you picked X even though the votes 
could have told you there's some other Y who a majority preferred to X.

And if you say the votes say no such thing, note that that cuts against 
IRV too. If noise makes it hard to infer true intent from the ballots, 
well, no election method can be a genie, some are better at it than 
others (as by Brian Olson's graphs), and we should at least have the 
method do well when there *isn't* noise.

>>    KM: (It might well be that the nature of IRV, plus cost of
>>    campaigning means there could only be two national-level parties. I
>>    don't think cost of campaigning alone would force there to be only
>>    two national-level parties - e.g. France - but the answer to that
>>    question isn't critical to what I wrote above. I'm saying that even
>>    if we assume what you're saying, you get into trouble on a more
>>    local level.)
> 
> 
> dlw:  My point is that more 3-5 seat forms of PR in "more local" would 
> remedy this problem.  I think there'd be a process of creative 
> destruction that tends to make for a more dynamic center that makes the 
> two major parties have to adapt a lot more.

If PR pushes harder than IRV pulls.

>> KM:I base my low confidence of PR's capacity to pull stronger towards 
>> competition than IRV does towards consolidation in that IRV pulls 
>> stronger wherever it's been tried. You say they aren't applicable. You 
>> may have that opinion, but then there's little I can show that will help.
> 
> dlw: Yeah, but AU uses PR where it reduces, not increases, the number of 
> competitive elections in the senate and IRV where it does little 
> value-added, due to de facto segregation in "more local" elections.  

IRV also seems to be keeping the same two parties major in most of the 
states, so the national influence bends local elections, too -- unless 
the local elections, too, are limited by IRV (i.e. from the bottom up 
rather than the top down).

> dlw: I think that the nature of the state as involving the use of the 
> monopoly on legit uses of violence tends to make multi-party systems 
> become two coalitions of parties, which ends up having similar 
> dysfunctions as two major parties.  So my prejudice is that it's less of 
> a diff than people claim and in both cases, serious changes require 
> political cultural changes apart from the electoral process.

I don't see that. The coalitions shift and change here, as they do in 
continental Europe. I have already stated why I think coalition politics 
are also more transparent than majoritarian ones, and why the voters are 
given greater choice when they can decide upon the relative sizes of the 
wings.

Or to put it differently: if coalitions were major parties, there would 
never be minority governments. Yet there are. If coalition members 
coordinated so much as to be like the wings of the same organization, 
all of the indecisiveness objections to PR would vanish. Yet people 
argue that PR can be indecisive, and setting up a government that all 
coalition members like requires serious negotiation.

Given the above, on what do you base that coalitions aren't different 
enough from two-parties to make a difference?

>> KM:Well, Burlington just confirms things. The simulations say IRV can 
>> fail to pick the CW, and may squeeze the center out, and the less minor 
>> the minor parties are, the worse it gets. As Burlington agrees with the 
>> simulations, that doesn't count in IRV's favor.
> 
> dlw: IRV does not always pick the CW.  But that's not so important so 
> long as the de facto center trends in the right direction due to 
> elections.

Do you then agree that if Burlington confirms what the simulations say, 
we can include both Burlington and the simulations; and thus that we can 
conclude that IRV exhibits center squeeze and can fail to pick the CW?

If so, we just have to find out how bad the artifacts of IRV are, not 
whether there are artifacts at all.

> KM:So why would IRV improve things enough over Plurality? That verdict, 
> too, has to come from somewhere.
> 
> dlw: more votes get counted in the final round than with FPTP.  Thus, 
> the de facto center is closer to the true center and third party 
> candidates can speak out their dissents and force the major party 
> candidates to take them seriously.  Why not look at the total number of 
> cities that have adopted IRV and see what a small fraction have had 
> buyer's remorse?

Sure. I'll take the Wikipedia list. If you have others, feel free to add 
them. (If any other list members have others, feel free to add *them*, 
because I don't know that you'll go looking for those that have had 
buyer's remorse.)

Let's first define the criterion as that IRV was used in the place in 
question, then at some later time, the place returned to their old 
method. Furthermore, we'll only consider those places where there have 
been actual IRV elections, not where IRV is on the books but hasn't been 
used; and we'll also consider runoff-IRV hybrids, where the system 
implements both TTR and IRV, as being too different to apply.

I'll mark the places that are still IRV as -, those where there was a 
backslide as #, and those that don't count as /.

Then we have:

(Historical use:)
# 1974, Ann Arbor, Michigan. Adopted in 1974. Repealed in 1976.

(Since 2002:)
- San Francisco, CA: Still uses IRV3.
/ Basalt, CO: IRV on the books. Hasn't been used yet.
/ Ferndale, Michigan. Wikipedia says "pending implementation". If this 
is -, feel free to correct.
- Berkeley, California. It has been used once, in November 2010.
# Burlington, Vermont. Nuff said.
- Minneapolis, Minnesota. Used once, in 2009.
# Pierce County, WA. Repealed in November, 2009.
- Takoma Park, Maryland. Tried in 2007 and 2007, but both in contests 
with only two candidates. Also tried in 2011, with one city council race 
having three candidates.
- Oakland, California. Used in 2010 elections.
- Hendersonville, North Carolina.
/ Kinston, North Carolina. Not enough candidates to go to IRV.
# Cary, North Carolina. Cary participated in the pilot, then chose TTR.
# Aspen, Colorado. Passed IRV in 2007, voters rejected measure to 
maintain IRV in 2009 and approved a binding amendment to return to TTR 
in 2010.
/ Sarasota, Florida. "Implementation is contingent on conditions that 
were not met as of 2010."
/ Santa Fe, New Mexico. "or as soon thereafter when equipment and 
software for tabulating the votes... are available at a reasonable 
price". The WP article doesn't say there was an actual election there, 
so I'm inferring it hasn't been, yet.
/ Memphis, Tennessee. I can't find any IRV election results, and what 
election results I did find from post-2008 elections are all plurality: 
see e.g. http://commercialappealapps.com/election/total/
- Telluride, Colorado. Passed in 2008, used in 2011 where the Plurality 
winner won.
- Saint Paul, Minnesota. Adopted in 2009, used in 2011.
/ San Leandro, California. This one could be counted either as # 
(changed from IRV to "either IRV or top two") or as - (did use IRV once, 
and one of the second-round options after amendment is to use IRV). Thus 
I'm not counting it for either.
- Portland, Maine. Adopted in 2010, used in 2011.

That's 5/14. 36% "buyer's remorse" - more than one in three - is hardly 
anything to be proud of. More ardent IRV opponents might argue that the 
number is lower, because races where the Plurality winner won shouldn't 
count -- but I've disregarded that so the numbers won't look fishy.

(If you say some of these don't count because "dishonest" campaigners 
were behind them, please tell me how to tell that kind of true Scotsman 
from a false one, in advance.)

>>>   dlw: 2.  AU does use IRV/PR in the opposite from ideal mix if the
>>>   goal is to increase the number of competitive elections.   3. WRT
>>>   France, we disagreed on matters of taxonomy.  I classified their top
>>>   two as a hybrid.  You classified it as a winner-take-all and used it
>>>   to show how IRV has been improved upon and could be improved upon
>>>   further.  
> 
> 
> KM: Let me try your pragmatism for a minute. You say that our 
> disagreement about top-two is taxonomy. Why should taxonomy matter, 
> though? If I have a "tacs"-type voting method, and an "intar"-type 
> voting method, both elect winners to single positions, and the voters 
> know both, but the difference is that the "intar" method produces a 
> greater diversity of winners than does the "tacs" method, then why not 
> use the "intar" method?
> 
> dlw: Because diversity isn't the only criterion that matters.  If there 
> are mults, tacs and intars, then a mix of mults and intacs (used in 
> different sorts of elections) might prove preferable over only using 
> hybrid intars....

Diversity is the most visible way to know more than the big two have any 
chance. The same argument goes for "going beyond uncontested two-party 
rule". TTR has shown to do so, even when there's no PR on the scene. IRV 
hasn't. Even if your "it doesn't apply here" objection holds, that just 
moves IRV from "won't" to "unknown" unless you have data showing IRV to 
dislodge the uncontested two parties.

Also, I don't think you have to disregard mults simply because you're 
using intars. If you pick TTR and don't get PR, that's good, because you 
know there's a chance you'll dislodge the uncontested two-party rule 
even so. If you pick TTR, get it, *and* get PR, all the better!

>> KM: I say top-two (TTR) is known, since runoff elections are used many 
>> places in the US. FairVote tends to market IRV as "runoffs without the 
>> runoff" there, and as my list showed, that particular form of marketing 
>> carries with it a risk of backfiring.
> 
> dlw:sure, but what sort of marketing strategy that simplifies the 
> options for low-info voters does not risk back-firing?  

Every pitch that bends the truth has some risk, yes; but FairVote bends 
the truth considerably on what is supposed to be the main advantages of 
IRV. IRV "lets you vote without considering spoilers" -- except when the 
spoilers become brave enough and large enough. IRV is a "runoff without 
a runoff and always elects the majority winner" -- but that's just not 
true when there are truncated ballots. "IRV *is* ranked choice voting", 
except not.

That's rather different than, for instance, not going into the details 
of a cycle resolution rule unless voters ask, or than saying of 
Condorcet "this will elect the winner that would win a top-two against 
every candidate, if he exists" when the dynamic/sequential game nature 
of TTR doesn't ensure that.

> dlw: If we could go back in time and push hard for TTR instead of IRV, 
> would it have worked (better)?  Maybe.  
> Let me reframe my statement:  "IRV leads to two-party domination within 
> IRV seats, but so likely do all single-member single-stage rules" and 
> "it may very well be a matter of political culture the preference 
> between two major parties and two major coalitions of parties.  If 
> IRV3/AV3 retains the former then that does not imply that it's the wrong 
> election rule alternative to FPTP for a country habituated to two-major 
> party rule.  "

Even if the goal is contested two-party rule, the diversity given by TTR 
serves as a lower bound that will tell you "we'll at least get a 
contested two-party rule". We may argue European PR vs US PR, but the 
bound is there, and it's that bound I want to draw attention to.

If it's true that all single-member single-stage rules lead to two-party 
domination, then you should advocate neither for IRV nor Condorcet, but 
for something similar to a runoff. If we advocate for IRV, IRV gets 
passed, and all single-member single-stage rules lead to two-party 
domination, then we have two-party domination. Same with Condorcet. But 
if it's just IRV that has this "feature", then IRV will still lead to 
two-party domination and Condorcet won't.

So if your statement is as you said above, focus on PR. Leave the 
single-winner for later. Disagreement about what single-winner method to 
use can only hurt PR in the sense that if FairVote's IRV gets 
discredited, they'll pull STV with them. Linking IRV and STV is 
FairVote's error, not ours.

>> KM: Condorcet regards not just a single very precise result, but a whole 
>> class of them. Therefore, it is resistant to perturbation, so I don't 
>> think there's "much ado about nothing".
> 
> dlw: Condorcet presumes (a majority of) voters have done their homework 
> and properly ranked an indefinite number of candidates.  When voters do 
> not do so and rank a subset, it more often leads to cases where there is 
> no Condorcet Winner.  

If a voter truncates after say, voting three candidates, he is still 
voting those three above all the others. The only way this can lead to a 
cycle is if enough people refrain from ranking both A and B, and A is 
in, say, an A > B > C > D > A cycle.

To put it in another way: truncation can't lead to a cycle where there 
wasn't a cycle to begin with.

So that means that those who did rank the votes expressed a circular 
preference. Too few voters that truncated stated any opinion about these 
candidates to alter the cycle.

I think that such a scenario would be unlikely to creep up on said 
voters even if they *did* only rank a small subset of the candidates. 
For the cycle to be a problem for an advanced Condorcet method, it must 
be a top cycle. I.e. if A beats everybody else, it doesn't matter if 
there's a B > C > D > B cycle. So the candidates in the cycle have to be 
somewhat well known, while the truncators either rank them equal, 
perpetuate the cycle (i.e. disagree among themselves), or truncate all 
of them.

Thus it'll only be a problem if there's a top group that is somewhat 
well known (or strategized to appear so), but enough candidates rank 
neither of them or rank them all equal. That's less realistic than just 
"ranks a subset".

To see this more easily: consider the case where every voter only ranks 
a single candidate. Obviously, a Condorcet cycle is impossible in this 
case. Now consider the case where every voter submits a full ranking. 
Again, obviously, a Condorcet cycle can exist in this case.

> Or there's the garbage-in-garbage-out problem. 
> IRV3 assumes that lower rankings are less likely to be meaningful than 
> higher rankings and thereby limits the info coming in and prioritizes 
> the use of the top rankings.  

So does Condorcet, in its own way. If there are ten candidates, a top 
rank counts against nine candidates. A second rank counts against eight, 
and so on. What Condorcet doesn't do is exclude from consideration lower 
ranks simply because a particular candidate is ranked first, and place 
the effective threshold as to which preferences it'll consider in which 
rounds in a nonlinear manner.

If you have slight GIGO, then the departure from full information isn't 
enough to be bothered about. If you have moderate, Brian Olson's graphs 
show IRV is affected more than Condorcet is. If you have heavy garbage, 
no algorithm short of a mind-reading one could work.

(If there's persistent strategy, use an IRV-Condorcet hybrid. The 
resulting method will be more resilient than either alone.)

>> KM: As for "muddying the waters", I said that to the extent it does so, 
>> it cuts against IRV. First, I introduced the Yee diagram, where IRV has 
>> much more complex win regions than Condorcet. The thinner and more 
>> convoluted the "border" between win regions, the greater the chance that 
>> an election result will fall in the wrong "country". 
> 
> dlw: ie, there's greater chance of center-squeeze in a close three way 
> with IRV than one that uses more info.  

Not just that, but that there are more cases where IRV would give one 
result as is, and another result if a single voter voted differently. 
The uncertainty given by noise could easily be greater than "a single 
voter voting differently".

That holds even in IRV3. The win regions for IRV limited to three 
candidates are more complex than those for Condorcet, for instance.

>> KM: Second, I pointed at IRV's amplification dynamic, where near-ties in 
>> one round could lead the method on a completely different path in a 
>> later round. 
> 
> dlw: This is mitigated by IRV3/AV3.

Yes, granted on that aspect.

(While you're altering IRV by limiting the number of candidates and 
prefixing it by an Approval round, why not add a bottom-two runoff to 
fix a lot more? You're only *that* far from something that would have 
got the right result in Burlington without having to mess with all the 
readjusting-center tatonnement. Or are you right at the edge of what 
FairVote will accept as IRV, so you can't afford to change it further?)

>> KM:So I was saying "Alright, you think that rational choice is too 
>> simplistic on account of fuzziness. Well, here's what happens if you 
>> take noise into account, and it's not favorable to IRV".
> 
> dlw: I apologize for my fuzzy thinking on this matter.  I think I'm 
> mostly prejudiced against static models of voter preferences and their 
> purported implications.  This is me being middle-brow.  In the end, the 
> diffs among the Xs of election rules tend to be relatively small and so 
> it's damn hard to elevate the Ps.  If there is a high P then it's better 
> to go forward with something close to it than to try to get a better 
> rule and elevate its P.

As I've said before, all elections are static at the moment of that 
election. Dynamics can alter this if the parties or voters learn to stay 
away from where the method behaves unreasonably. But as I also said, 
above, is it worth it? Should you burden parties and voters just because 
of FairVote?

So your middle-brow-ness doesn't make the picture shown by the Yee 
diagrams any less correct. Regardless of your brow position, what 
matters is whether the voters will often stumble upon the weirdness 
(shown in those diagrams) when they vote honestly, whether they'll 
compensate by twisting their votes, and whether the outcome given by 
those twisted votes is any good.

>> KM:Are they? I don't think the opponents of electoral reform know about 
>> Condorcet, much less Majority Judgement, Range, Approval, or the likes. 
>> The greater you think the order of magnitude in p_IRV >> p_Condorcet, 
>> the less of a point you'd think there would be for the opponents to even 
>> care about Condorcet.
> 
> dlw: They don't need to know much about such to understand that args 
> that X_Condorcet_etc >> X_IRV tend to lower p_IRV.  

If p_IRV >> p_Condorcet, Condorcet would be off their radar. Why bother 
splitting a few EM list members' opinions when they can get more bang 
for their buck by simply going directly to the objections of 
nonmonotonicity and failure to elect the CW?

>> dlw: I agree that the evidence suggests the D candidate was the CW and 
>> that this made it somewhat easier to overturn IRV with a deceptive 
>> campaign against IRV.

Okay. I'll strike the "deceptive" because I don't find the Burlington 
arguments that unreasonable (except perhaps some of the no-show ones), 
but that's a quibble. We can agree that IRV's failure to pick the CW 
made it weaker whether or not the campaign was deceptive.

>> KM:If FairVote continues on its marketing of IRV and we do nothing, yes, 
>> IRV is more likely to be adopted than Condorcet, at least in the short term.
> 
> dlw: I disagree about doing nothing.  There's always enforcing a truce 
> on IRVish for single-winner and  support for American forms of PR!  That 
> is hardly nothing.  

I should have clarified. If non-FPTP-non-IRV people do nothing, either 
IRV will win or FPTP will retain its position. Yes. If non-IRV people do 
something, they/we may overtake IRV. It'll take time, but this thing 
should be done right. We may not get another shot in a really long time, 
should "reform" be linked to IRV and then found lacking.

> KM: However, I think that would be unfortunate in two ways. First, I 
> don't think IRV improves Plurality enough that it'll matter. It'll keep 
> major parties safe as long as minor parties are minor enough, but not 
> beyond that point. Therefore, if you do end up with IRV, you keep your 
> uncontested two-party rule.
> 
> dlw: Unless, of course there's such serious unhappiness with the two 
> current major parties in the US that we'd get two different major 
> parties with (somewhat) better rules a lot faster?   And then there'd be 
> scope for further experimentation later down the road...

There's the rub. Will IRV give you two different major parties "a lot 
faster"? You think so. You keep saying so. I keep saying not.

> KM: Second, you may not even keep IRV. If IRV gets it wrong often 
> enough, or reproduces Plurality's results often enough that it doesn't 
> seem to be worth it, then the option of reverting it can seem quite 
> tempting. If FairVote claims that it's a runoff-without-runoff or that 
> you can vote as you wish without fearing spoilers, and that turns out to 
> be false, then IRV may not last; and if IRV is considered equal to 
> ranked balloting, then the immediate reform chance is lost. You might 
> have to go a far more circuitous route involving augmenting Plurality 
> with MMP - and that wouldn't help executive positions like Governor or 
> President.
> 
> dlw: Well, it's not likely

Why?

 > and if we stand by IRV as significantly
> better than FPTP for single-winner elections rather than puff up the 
> import of non-monotonicity, it's not going to happen as much.  

The flaws will still be there. IRV's failure to handle the center when 
it gets crowded is both a real (X) and a marketing (p) problem because 
people are going to be disappointed that they have to strategize even 
though the method was supposed to free them from it.

And as long as these problems exist, those who oppose the method can 
stumble across them - or voters can go "hey, this doesn't work as 
advertised". Better to defend something with fewer weak spots than to 
gather around to shield those weak spots from those that eventually will 
see them anyway.

(If a voter says that IRV is bad because it gets the winner wrong or he 
correctly feels that it did, he's not going to be very happy if you say 
"oh, but then you should just strategize or tell your party to not be so 
close to the center".)

> dlw: I believe my point is that for "more local" elections, IRV or other 
> single-winner elections won't be that much useful due to de facto 
> segregation.  We need multi-winner elections to make these elections 
> more competitive.  

So work for PR. No need to tie it to single-winner reform (unless you 
really have to have FairVote support).

>> KM:The lack of a heir apparent might not be that bad. Committees like 
>> the Rhode Island one might pick the best among near-equals according to 
>> what they deem important. This kind of approach has worked in New 
>> Zealand and led to election reform referenda in parts of Canada (though 
>> the voters there decided not to go for it in two cases, and had a 
>> majority but not a supermajority in the third).
> 
> dlw: If overall the Xs aren't that different then it's easy to attack 
> whatever gets selected on the basis of some criterion.  I myself blogged 
> about NZ. 
> <http://anewkindofparty.blogspot.com/2011/09/how-new-zealand-may-change-its.html>  
> They recently were proffered 4 options in what I felt was a manipulative 
> manner.  I hope they didn't force upon them the use of a plurality vote 
> among the alternatives. 

Yup. Some parts of NZ society probably didn't like losing their power. 
It does seem that MMP will stay in service, though. See, for instance, 
https://en.wikipedia.org/wiki/New_Zealand_voting_system_referendum,_2011 
which shows that a majority of advance votes are in favor of MMP, and a 
majority was in favor of MMP in all the 2011 polls.

My point here is that the 1992/1993 referenda showed how to have the 
people decide which voting system they'd prefer, even within the 
constraints of FPTP. In 1992, the first referendum question was whether 
to change at all; then the second question was, "if there's going to be 
a change, which do you want?". Thus there was no vote splitting between 
FPTP and non-FPTP. The 1993 referendum then followed up with a question 
of FPTP vs MMP (where MMP was what was picked in 92). MMP passed.

> In the US, we have a plurality voting system, 
> which implies you gotta have strong agreement on which alternative to 
> FPTP is going to get pushed.  

No. You can do it like New Zealand did in 1992 and 93.

>> KM:If you want electoral reform to happen from the bottom up, you don't 
>> need the national government to set up the committee. A state can do so, 
>> or a more local area like a city.
> 
> dlw: But those committees may not have the right motives and their 
> decisions might not carry much weight.  

They would carry weight within their jurisdictions.

> dlw: three way competitive races are not common.  They are hard to 
> sustain.  Thus, it is unlikely that IRV is going not to elect the CW 
> regularly, thereby making it vulnerable to anti-IRV campaigns.  I don't 
> want to risk letting a thousand flowers bloom makes it damn hard to 
> rally enough people around an alternative to FPTP.  The evidence 
> supports that IRV moves the political center so as to strengthen 
> democracy and weaken negative campaigning tendencies in FPTP elections.  

Three-way races are not common *today*. This is the same kind of 
thinking that led to the IRV patch in the first place: "if we use 
something that will fix the immediate problem, we've done it". It did 
fix the immediate problem of very minor parties that would otherwise 
split the vote (it probably would have elected Gore in 2000), but it 
doesn't fix the next problem when people become emboldened and vote for 
third parties.