Election Day Registration:
Election Day registration is already proven to be a success in Maine, Wisconsin, Minnesota, New Hampshire, and several other states. It guarantees eligible citizens who can provide identification and proof of residence need never be turned away from voting. Election day registration has been used for over 30 years in hundreds of elections by millions of voters.A recent study shows up to 225,000 new voters will be able to turn out and vote in Massachusetts if Election Day registration is passed. Not since 18-year-olds were granted the right to vote has a single measure stood to increase the number of new voters.
It's estimated that new voters will include:
- 139,000 people who make under 40,000 per year
- 105,000 people under 35 years old
- 27,000 people who moved in the past six months
National Popular Vote:
Do you think the President of the United States should be elected by popular vote? Shouldn't everyone's vote count equally, rather than voters in a handful of so-called swing states deciding the outcome?The National Popular Vote plan creates an agreement among states that guarantees the presidency to the candidate who receives the most popular votes in all 50 states and the District of Columbia. The Massachusetts bill would only take effect once similar bills are enacted by enough states to combine a majority of the electoral votes (270 or more).
So far Maryland, New Jersey, Illinois, and Hawaii have all passed National Popular Vote bills. We need your help to get this bill passed in Massachusetts. Four times throughout our history, most recently in 2000, the presidential candidate who received the most votes was not elected. America deserves better, and fortunately the Founding Fathers left us with a mechanism for states to adopt a better system.
This is a critical time since the votes on both bills is expected on Thursday. Your State Senator needs to know that you care about strengthening our democracy and making access to voting easier for the working public. Please call your State Senator right now.
CLICK HERE TO FIND YOUR SENATOR
State Senate Switchboard
(617) 722-2000
So what are you waiting for?
Thanks, Lynne. I hope everyone calls their Senator to support both of these important bills.
<
p>I borrowed the title of this comment from Katrina vanden Heuvel’s recent essay in The Nation in which she lays out an ambitious agenda for electoral reform that includes both EDR and NPV.
<
p>We can spend our time fighting for better health care, fairer tax policies, improved environmental safeguards … pick your issue. But the best way to ensure better policy is to fix the broken electoral processes that stand in the way of fundamental reform. A more democratic process will give rise to better outcomes.
<
p>Why?
<
p>I am still figuring out what I think about the NPV bill, but I see your argument pretty often, and I wonder why we should accept it. There are good arguments for the idea that the majority should rule–the best argument, probably, is a fairness argument, namely that in fairness, everyone’s preferences should count equally. But this seems to me to be wholly independent of the quality of the policy that results.
<
p>If by “democratic process” you mean an educated and informed citizenry that takes an interest in public policy, that debates the issues in the press and other forums, and that votes accordingly, then I agree with you. But if by “democratic process” you mean just “majority rules,” I don’t.
<
p>TedF
To the “swing states get all the attention” problem. It never fails. Look at the amounts spent in just the swing states. It’s disgusting. If you’re a state that is “safe” than the election isn’t run for you. Even if you’re a populated state.
<
p>There is no perfect solution. However, the current system is insane, and 2000 (Florida) and 2004 (Ohio) bear that out.
But I think this just goes to show that the merits of the NPV relate to issues like “swing states” versus “safe states” rather on the notion that we’ll get better policy as a result of the proposed change.
<
p>TedF
that I find persuasive, is that having more votes in play–rather than just the undecideds in the battleground states–will change the political discourse for the better.
<
p>Certainly, it will nationalize the election.
<
p>I know you could argue it both ways, but consider–a much larger and more diverse pool of voters, less stampedable by targeted media campaigns. That’s an improvement.
<
p>I think it will change the way candidates campaign for the better.
Our foreign policy with respect to Cuba is effectively held hostage by the Cuban exile community in and around Miami, because Florida is (usually) a swing state. This magnifies the effect of said community (as well as everybody else in the state) to absurd levels.
<
p>If it were a truly national election, then alienating that community with a sane and reasonable policy (instead of our current kneejerk “communism is bad (except in China)” policy) might be feasible. You’d only stand to lose thousands of votes out of 100+ million (which could be as much as 0.1% of the outcome, instead of 27 out of 538 (5% of the outcome).
<
p>This is but one example of how we would get better policy out of this procedural change.
in some form. I’m a Democrat, sure enough, but competition can only make us stronger (after all, if we don’t deserve the support, we should have to get to work for it). Instant runoff voting sounds weird at first, but it solves one of the age-old problems of a one-man-one-vote system: that of taking the lesser of two evils (not weevils).
<
p>If you are going to the voting booth, you are often faced with a choice – vote the party that mostly or sort of holds your values even if it’s failed you or has moved from them or does not match perfectly, because if you vote for the smaller third party which more closely follows your values, that party that totally is the opposite of your values could win. So, many people hold their nose when voting, and don’t vote their true values in fear of throwing the election to the other party. This happens on both sides (for example, Greens or Libertarians) and so third parties are never really viable ever, unless there’s some sort of dramatic catastrophic shift.
<
p>Anyway, more here. Other countries do use it, as do some localities in the US.
<
p>Unfortunately, our US system, unlike parliamentarian systems, is flawed in the aforementioned way. However, it’s fixable, albeit in a dramatic change-the-way-we-vote sort of way. I would really get behind a movement that pushed some form of runoff voting.
Thanks, Lynne. As you might know, ONE Lowell and AALDEF are seriously considering a push for Proportional Representation on the Lowell City Council, probably via “Choice Voting” (aka the “Single Transferable Vote”) like Cambridge. That method is essentially a generalization of IRV to elect multiple people at once, and it is very well-regarded by activists and academics alike. I don’t know whether that effort would be combined with the effort to elect the Mayor of Lowell via IRV. In any case, it’s a local movement worth getting behind.
just to defeat it. Thanks for the heads up.
Proportional voting has been extremely successful in Cambridge.
EaBo, not sure whether it’s proportional representation, IRV, or both that you’re against. If it’s IRV, it might interest you that the “Barr effect” (the degree to which Bob Barr will hurt John McCain) is measured by Zogby to be greater than the Nader effect against Obama this year. If John McCain loses due to third-party presence in the race, I would believe that to be a poor outcome. With IRV, Barr supporters could rank Barr first, McCain second. Then, when Barr is eliminated due to low first-choice support, people who chose McCain second would have their vote instead count towards McCain.
would get their national 3%+ of votes nationally to remain a viable party, and could grow their party from there much easier, get better ballot access, etc.
<
p>Same with Greens, or any other third party.
PR is not a generalization but a subset of preferential balloting (the old-fashioned name for IRV). And a quirky subset.
<
p>I like it but a buddy of mine–who served two terms on the Cambridge City Council–became a skeptic. “It’s like the Knesset,” he once told me, referring to the Council’s ideological diversity, an intended consequence of PR’s empowerment of minorities.
<
p>The generalized case would be if Cambridge counted the ballots nine times to discover nine candidates, each one elected by a majority of votes cast (instead of 1/9th + 1).
Proportional Representation is an idea, not a specific system. There are many voting systems that achieve different forms of PR. The Knesset and many western democracies use party-list systems of PR. Cambridge and several countries use “Choice Voting” (technically known as the “Single Transferable Vote”). Choice Voting is my preferred means of PR, and it is also the preferred means of the leading electoral reform organizations in the world.
<
p>When Choice Voting is applied to elect a single candidate, it becomes Instant Runoff Voting. So Choice Voting is indeed a generalization of IRV to elect multiple candidates.
<
p>It is true that some locations that use IRV and/or Choice Voting refer to it as “preferential voting.” That is an unfortunate term, because there are a variety of systems in which voters rank their choices in preferential order, some of which are quite different from IRV and Choice.
here’s a great page on it…
<
p>http://rangevoting.org/PropRep…
It’s better than first-past-the-post, but still produces some bizarre results. If we’re going to reform our ballots, I’d rather see us look towards range voting (Wikipedia).
<
p>I’ve been meaning to do a BMG write-up on range voting for a while now, but haven’t found the time. Gotta do that soon…
Range has several very bad properties, particularly an extreme susceptibility to strategic voting. This is what caused voting scholar Professor Nicholas Tideman, in his most recent book on voting theory, to rank Range among the most “unsupportable” of voting systems.
<
p>Range even fails to satisfy the most basic criterion of any democratic voting system: the Majority Criterion. That is the simple criterion that a voter who is the first choice of a majority of voters should win. Even our current plurality system (except for the Electoral College) satisfies that, and of course so does IRV.
<
p>One of the reasons for Range’s susceptibility to strategic voting is that, unlike IRV, every vote in Range to a candidate other than one’s first choice, hurts the chances of that first choice. This criterion that Range fails is called “later-no-harm.” Thus, if Range were in place, expect any candidate with a chance of winning to ask their supporters to give them the highest score and everyone else the lowest. If this became the defacto manner of voting, then it would be no better than plurality.
<
p>Due to its failure of later-no-harm, Range has big problems with vote splitting. Imagine a three-way race between Obama, Clinton, and McCain. How should an Obama supporter vote on a Range ballot? Should the voter give Obama a 10 and give Clinton and McCain a zero, in order to maximize the chance that Obama is the winner. Or should the the voter may give both Obama and Clinton a 10 and McCain a zero, to maximize the chance a Democrat is elected? If the voter applies the wrong strategy, s/he winds up at a disadvantage. This predicament was coined the “Burr dilemma” by Professor Jack Nagel, and it is the key reason he abandoned his prior support for Approval/Range in favor of IRV.
<
p>Range also still suffers from a spoiler effect. Consider someone in the 2000 election who prefers Nader first, Gore second, and Bush third. If Nader is not in the race, the voter gives Gore a 10 and Bush 0. When Nader enters the race, the voter gives Nader a 10, Gore 5, and Bush 0. The entrance of Nader into the race hurt Gore’s chances of beating Bush — thus, the spoiler effect is still alive and well with Range Voting.
<
p>Furthermore, Range isn’t currently used to elect any public official anywhere in the world. Before it is advocated for public office, I don’t think it unreasonable to ask that it be vetted in small organizations before the public are asked to be guinea pigs. The limited track record of Approval Voting (the simplest case of Range Voting where voters score each candidate 0 or 1) has not been encouraging.
In 1991 there was a now-famous race for Governor of Louisiana. LA, as you are probably aware, uses a run-off system for all state-wide offices, and no primary. All candidates are on the first ballot, and if nobody gets more than 50%, the top two vote-getters face off in a secondary election.
<
p>In 1991 the three major candidates were David Duke (yes, that David Duke), Buddy Roemer and Edwin Edwards. Here’s what the first round looked like:
CandidateVote
Edwards523,096
Duke491,342
Roemer410,690
<
p>Prior polling found that, in a Roemer-Duke runoff, Roemer would win. It also established that, in a Roemer-Edwards runoff, Roemer would win. Yet, in the three-way, Roemer came in last place, got eliminated, and the final election was between Edwards and Duke. Edwards won.
<
p>Furthermore, in this election, Duke was far right, Edwards far left, and Roemer in the moderate middle. Supporters of Duke and Edwards hated and reviled the other group.
<
p>So, suppose that 41,000 Duke supporters (that’s fewer than 1 in 10) “betrayed” their candidate and voted strategically for Roemer. The tally would have been thus:
CandidateVote
Edwards523,096
Roemer451,690
Duke450,342
<
p>Duke would have been eliminated in the first round and Roemer would have won. Duke votes actually led to the election of the hated enemy Edwards.
<
p>This is the sort of real-life example of strategic voting and voting paradoxes that IRV will lead to — if IRV is put in place, pre-election polls will ask questions that would lead to this data being uncovered, and people will use it, I promise.
<
p>As for the rest of your issues — as I said, I’ve been meaning to address range voting in a diary entry, and I do still intend to do so. You’ve given me food for thought and an excellent place to start, and I appreciate that greatly.
Yes, I’ve seen this example given many times. What surprises me most about this example is that there are hundreds of IRV elections conducted every year, many within the United State, and yet the supposedly best example of IRV’s flaws is this non-IRV election. You would think, if these flaws were so prevalent in practice, that opponents would point to at least one of any of the thousands of IRV elections conducted over the years.
<
p>What would the IRV result been in this election? It’s hard to say. There were eight other candidates running who would have been eliminated before any of the top-three, and depending on their distribution of votes, Roemer could have won. For example, had the voters for the 4th place candidate Holloway ranked Roemer second, as I would expect, that would have been enough to push Roemer into the final tally. Also, because it was a two-round event, Roemer supporters could have unwisely sat out the first round, thinking he would be a shoe-in for the inevitable runoff.
<
p>So we don’t really know what would have happened with IRV. What we do know for sure is that this was not an IRV election.
<
p>If, as you claim, that IRV “will lead to” strategic voting, then why isn’t it even a blip on the radar screen anywhere IRV is used? There are pre-election polls in Australia, for example, but this isn’t any kind of issue there. It hasn’t come up in Ireland, San Francisco, or Burlington, VT, or anywhere else IRV is already in place. These supposed flaws are just vanishingly rare in practice.
Greg’s post is deceptive, and careless with facts.
<
p>The Center for Range Voting was co-founded by a Princeton math Ph.D. named Warren D. Smith, who has performed arguably the world’s most extensive computer simulations of elections, to show that Range Voting is quite robust to the problem of strategic voting. In fact, Range Voting performs about as well with 100% strategic voters as IRV performs with 100% honest voters. This is explained in incredible detail over the span of numerous pages at RangeVoting.org, such as:
<
p>http://rangevoting.org/StratHo…
<
p>Greg’s mention of the work of Nicholas Tideman is especially dishonest, because he’s well aware of flaws in Tideman’s analyis, not to mention the fact that Tideman also called IRV unsupportable. A full review by Warren Smith is here:
<
p>http://rangevoting.org/Tideman…
<
p>Tideman’s biggest fundamental flaw was that he treated all vulnerabilities to strategy as being equally significant, when in fact they differ radically in severity. By contrast, Smith’s Monte Carlo computer simulations measure the net impact of various strategies in terms of how much they decrease the average voter’s satisfaction with election results. Smith has pointed out these flaws to Tideman.
<
p>Continuing to the point about the later-no-harm criterion, Greg’s argument becomes even more specious and ironic. Imagine a Ralph Nader supporter has marked down a perfect “10” score for Nader, and now deliberates on how to score the Democrat and Republican. Does he, as Greg insists, give the Democrat a zero, out of fear that a higher score might cause the Democrat to defeat Nader? In most cases, no. The voter’s bigger fear is that by NOT helping the Democrat, he’ll help the Republican. That explains why, according to NES polling data, around 90% of Nader’s supporters betrayed him and voted for someone else (mainly Gore) in 2000. If Greg’s point was correct, they would have listened to their favorite candidate, Ralph Nader, and voted for him. But they didn’t, because most people don’t want to “throw away” their vote. So with Range Voting, they would have the same incentive to help other candidates besides just their favorite. Greg’s argument is wrong, and out of touch with reality.
<
p>In fact, far from being a benefit, passing the later-no-harm criterion means that a voting method necessarily ignores all information about strength of preference, and looks only at order of preference. This makes a voting method susceptible to some serious flaws that have been articulated in two famous mathematical proofs about election theory: Arrow’s theorem, and the Gibbard-Satterthwaite theorem.
<
p>http://rangevoting.org/ArrowTh…
http://rangevoting.org/GibbSat…
<
p>So it is a good thing that Range Voting “fails” later-no-harm. This is central to the ability of Range Voting to resist strategic voting, by utilizing what economists call “revealed preference”. Revealed preference is basically just a way to reward someone for being more truthful. For instance, say you tell me that you prefer oranges to apples. Maybe you’re lying. So I offer you an orange or an apple, to see which one you’ll actually prefer if given the choice. We could do the same experiment with a Nader-betrayer, by making him a “dictator” whose ballot trumps all others. Maybe he voted for Gore for strategic reasons, but if we give him this dictator opportunity, he’ll have no reason to do that. He’ll want to be truthful about who his real favorite is.
<
p>How does that apply to Range Voting? Well for one thing, Range Voting never gives you a reason not to give the highest score to your favorite candidate, and the lowest score to your least favorite. So if Nader’s your man, you’ll be able to give him a “10” with no hesitation. But with IRV, a group of voters who sincerely prefer X>Y>Z can get their least favorite candidate, unless they “betray” their sincere favorite, and vote Y>X>Z — which causes Y to win, and rewards them for their favorite-betrayal.
<
p>Going deeper, say you and I are using Range Voting, and those 3 candidates all have a realistic chance of winning, and we have the same order of preference, but different intensity of preference.
<
p>I’d score them: X=10, Y=7, Z=0
You’d score them: X=10, Y=2, Z=0
<
p>To IRV, or any other ranked method, our ballots would look the same. With Range Voting, they would be different, to express the differences we really feel. But what if we vote strategically? Then our ballots will be
<
p>Me: X=10, Y=10, Z=0
You: X=10, Y=0, Z=0
<
p>See? Even with strategic voting, Range Voting “forces” us to reveal something important: the fact that I think Y is about as good as X, and you think Y is about as bad as Z. Greg speaks as though the “Burr dilemma” is a great problem for Range Voting, but casting a reasonably good strategic vote with Range Voting is not too difficult, as explained here: http://rangevoting.org/RVstrat…
<
p>And for those who do think it’s difficult, that’s all the more incentive for them to vote honestly! While it might come as a surprise, honesty is actually not a bad strategy with Range Voting: http://rangevoting.org/RVstrat…
<
p>Greg’s argument that a voter who fails to apply a good strategy is at a disadvantage is especially misguided. To demonstrate, I’ll use some real social utility efficiency metrics which compare average voter satisfaction based on whether the electorate uses IRV or Range Voting. I’ll include a realistic, but “made-up”, difference in outcome which represents the effect of a particular voter’s decision to vote expressively vs. strategically.
<
p>Range – expressive voter : 85.84%
Range – strategic voter : 85.85%
IRV – expressive voter : 57.77%
IRV – strategic voter : 57.78%
(data from http://rangevoting.org/vsi.html)
<
p>So is an expressive Range Voting user at a disadvantage? Depends on what you’re comparing to! Compared to a strategic Range Voting user, yes he is at a slight disadvantage – slight because the odds his vote will swing the election are next to zero. But compared to an IRV user, he’s at a huge advantage. In fact, with Range Voting, he could not even vote, and still be better off than with IRV. And let’s not forget that strategic voting is also a factor with IRV, a point which IRV enthusiasts ardently ignore or even deny.
<
p>=> http://rangevoting.org/TarrIrv…
<
p>As for spoilers, it is true that the spoiler effect is “alive and well” with IRV and Range Voting. However it is much more severe with IRV, and Range Voting eliminates the spoiler effect insofar as votes are cast. That is, given a set of Range Voting ballots, it is impossible to change the winner by eliminating any candidates from consideration. If Bush is the winner, and we erase Nader’s name from the ballots, Bush must still be the winner. IRV (in fact, all rank-order methods) can fail even this most basic test. Imagine that the ballots have been cast, and Bush is the winner, but now we take Nader off those ballots — but leave all rankings for all other candidates untouched. Now it is possible for the winner to change. This kind of insane behavior has been observed in figure skating competitions, such as when Michelle Kwan’s performance did not earn her a medal, but caused two higher ranked competitors to swap positions after they had already competed. This is the kind of behavior that Range Voting elimina
tes.
<
p>Range Voting has been extensively tested in computer simulations, based on real election data and real polling experiments. In many ways, these computer experiments are superior to anything that can be done in real life, since we have no way to read voters’ minds
=> http://rangevoting.org/WhyNoHu…
<
p>Range Voting has been vastly more studied in the present day than IRV had been at the time it was first implemented at the government level. And there is no serious question that it is a vastly improvement over the plurality system that we’ve used for two centuries.
<
p>These kinds of delicate points are missed on the hand-waving contingent who support IRV in the face of all contrary evidence. Greg’s numerous errors have been explained to him in great detail in the past, but in a style that has become emblematic of the IRV movement, he has ignored that and re-posted the same fallacious and misleading arguments for another audience.
<
p>Clay Shentrup
San Francisco, CA
clay@electopia.org
206.801.0484
<
p>Yes, “arguably” indeed. The fact remains that Smith, although he has a PhD in Math, is a wannabe political scientist who has never been published in the field of social choice theory. As Smith himself admits, his submissions on Range to peer-reviewed journals have all been rejected.
<
p>
<
p>I am aware of what you and Warren call “flaws”, and Tideman is aware of your criticisms as well, but we both disagree with Warren that they are flaws at all.
<
p>
<
p>As you know, Tideman ranks voting methods by degree of supportability. He ranks Range last as completely unsupportable. He ranks IRV second best, only surpassed by a complicatd Condorcet method. I agree the Condorcet method is very good, though I prefer IRV, in part due to the complication of the Condorcet method and a few other factors. I don’t disagree with how Tideman scores different methods with respect to each criterion, though I do weigh the criteria a bit differently than he does.
<
p>
<
p>A blatant error. In my example, the voters gives Gore a 5, not a zero. Look before you leap. If they give Gore anything less than a 10 — the score they would gave given him had Nader not been in the race — then they are hurting Gore’s chances. That’s a spoiler effect.
<
p>
<
p>You are mistaken again. It’s fairly trivial to construct a voting method that satisfies both later-no-harm and considers strength of preference. I do prefer methods that only consider order of preference, but those two properties (later-no-harm and strength of preference) are not fundamentally incompatible.
<
p>
<
p>I don’t agree that the flaws are “serious”, but they do exist. Criteria are tradeoffs. That’s life — there is no perfect voting system. I happen to think IRV, of all single-winner systems, makes the best tradeoffs and Range makes close to the worst. You happen to disagree, but we’ll see who wins.
<
p>
<
p>Performing the calculation you link to requires knowing with accuracy the probability that each candidate will be elected. For starters, most races throughout the country (those at the local and state level) have no public polling to speak of, so knowing the probability a candidate has to be elected is a task that can be performed accurately by only political insiders. Moreover, even when polling is conducted, you would require the voter to have surveyed the polls to make an accurate guestimate. That would be an effective disenfranchisement for low-information (mostly low-income) voters throughout the country.
<
p>
<
p>Yes, just vote honestly — in the meantime, the elite are able to vote strategically and, thereby, have effectively a stronger say in the election than you do. That’s not a fair system. It’s a system that puts honest voters, low-information voters, and voters that are less mathematically inclined at a distinct disadvantage.
<
p>
<
p>If the spoiler effect is “alive and well” in IRV, where is it? Point to a real IRV election where if a candidate not been in the race, you honestly believe another candidate would have won.
<
p>
<
p>Your chalkboard ramblings are no match for real, empirical data from real elections. With a proper user study, you could measure voters satisfaction with the system, how well they understood it, and whether they would want to use it again.
<
p>
<
p>Where’s the “contrary evidence” from real elections? In which election did IRV choose the wrong winner? Point us to an IRV election where candidate A won but you believe candidate B was a better reflection of the voters preferences. Seeing none over thousands and thousands of real IRV elections, I’m forced to conclude the flaws — to the extent that they’re real flaws — are vanishlingly rare in practice.
<
p>
Greg,
<
p>Despite your assessment of Warren Smith as being a “wannabe” political scientist, the fact remains that his Monte Carlo simulations are the most extensive yet performed. If you are aware of others, particularly having contrary results, please discuss them. Similarly, if you can find any valid reasons for Smith’s academic submissions to have been rejected, please present them, rather than simply implying they must have been legitimate.
<
p>Regarding the analysis by Nicholas Tideman, I repeat that you are entitled to your own opinions, but not your own facts. If you look at Smith’s review of Tideman’s work, under the heading Tideman’s “Strategy Resistance” Measure is Flawed, you will see a laundry list of proven flaws. For instance, Tideman addresses strategic voting post-election, when in fact voters must base their strategies only on what they know prior to an election. Another especially noteworthy example is that Tideman mistakenly treats all strategy effects as having equal weight. Here’s a simple counter-example that disproves him. If the result is “the candidate unanimously agreed to be worst is elected,” as in the DH3 pathology, that is a major effect that causes a huge decrease in average voter satisfaction. Whereas, “the second-best candidate is elected” is a much less severe effect. This is similar to the effectiveness of the fare inspectors in our public transportation system here in San Francisco. Say that they decrease the frequency of their random car inspections, but increase the punishment for fare evasion to include prison time. That could actually have a much stronger deterrent effect, even though Tideman’s blind-to-severity perspective would say otherwise.
<
p>As a quick sanity check, we can observe that Tideman’s figures show plurality voting to be superior to approval voting – a ridiculous notion in light of the fact that approval voting was designed to be strategy-resistant.
<
p>The biggest fundamental flaw across Tideman’s analysis is the arbitrary nature of it. He picks and chooses what he wants to call a flaw, and how significant that flaw is. Smith’s work instead puts things in objective “economic” terms, using Bayesian regret. These metrics tell us how well off a voter can expect to be as a result of the voting method his society adopts. There are no arbitrary decisions about which criteria are important, or how important. Smith’s simulations test not only the criteria that someone like Tideman arbitrarily considers, but all criteria — even those that have not been invented. And voting methods are tested not simply on a pass/fail basis, but on a scale of quality that also considers frequency. Compared to this sane and grounded method, Tideman’s analysis is subjective and misleading. When you correct Tideman’s flaws, his own data supports range voting.
<
p>Bottom line: range voting performs about as well with 100% strategic voters as IRV with 100% honest voters.
<
p>
<
p>Tideman, e.g (on p.240), brands plain IRV as “unsupportable” (provided it is “feasible” to construct a pairwise-table, which it is) on the grounds that it is dominated by other methods in terms of its (precise and imprecise) properties.
– Smith
<
p>
<
p>That’s precisely the problem with subjective weights placed on subjective criteria. Smith’s simulations fix that problem, giving us an objective measure of how well we’ll be served by the different voting methods. Range voting, approval voting, Borda, Condorcet, and many other methods are substantially better than Instant Runoff, which sits near the bottom with plurality.
<
p>
<
p>I’ll refer you to your previous quote:
if Range were in place, expect any candidate with a chance of winning to ask their supporters to give them the highest score and everyone else the lowest
<
p>I did indeed miss the part where you said “with any chance of winning”, but your claim is still wrong. Say we have a range voting election, and Hillary Clinton runs as an independent. If she’s neck-and-neck with Obama and McCain, then a voter who prefers Obama just slightly over Clinton, but prefers them both to McCain, will not want to give Clinton “the lowest” score just because Obama says so. That would be strategically unwise, since helping Clinton beat Obama is less harmful to that voter than helping Clinton lose to McCain.
<
p>
<
p>Which is a much larger problem for IRV than for range voting.
<
p>
<
p>Well, no. To satisfy later-no-harm, we may only consider a ballot’s support for any given candidate once all more-preferred candidates have been eliminated from consideration. Thus there can be nothing against which to compare the candidate currently being considered. For example, say that we use a variant of range voting in which we only consider a voter’s highest-scored non-eliminated candidate. Say I have a ballot where A=10, B=3, C=0. But my B=3 does not actually say anything about strength of preference, since it will only be considered once A has been eliminated, and will not be compared against my score for C. So what you call “fairly trivial” I do not see as being possible, unless you can show me an example.
<
p>
<
p>As Arrow (and others before him) famously proved, the correct social utility function cannot have this restriction. So you admit to hold a belief whose validity has already been disproved via formal math. Telling.
<
p>
<
p>I don’t just “disagree”, I have massive amounts of evidence. Range voting..
<
p>- yields much better results than IRV
– is able to be performed on all standard voting machines, unlike IRV
– is precinct-countable, unlike IRV
– makes it always safe to support one’s favorite candidate, unlike IRV
– reduces the amount of spoiled ballots, whereas IRV increases them
<
p>Contrast this with the IRV camp, who essentially hasn’t got a single valid argument to their credit, as you amply demonstrate.
<
p>
<
p>Assuming candidates and news media don’t pay those insiders to give voters the inside scoop on that poll data, then you are simply arguing that range voting makes strategy harder. Great! The more expressive voters, the better for the voters overall.
<
p>
<
p>Available information says that’s completely implausible. Say we have (hypothetical, but reality-based) social utility expectations such as:
<
p>IRV – low income voter – 52.05%
IRV – medium/high income voter – 54.67%
<
br />RV – low income voter – 73.63%
RV – medium/high income voter – 79.34%
<
p>Here I’m being generous to you with my hypothetical figures, so that there’s a bigger disparity for range voting. The low income voters aren’t “disenfranchised” by range voting. They’re massively better off with it. Your mistake is to compare them to other voters within the same voting system, rather than to themselves in another voting system. IRV proponents constantly make this mistake.
<
p>
<
p>You’re continuing the same mistake that I just discussed. Voting isn’t about how much “say” (power) you are given. Power is only a means to an end, which is utility, aka “satisfaction”. If, just for the sake of argument, range voting gives the rich a stronger “say” than IRV, but still makes the poor better off than with IRV, then it’s inaccurate to say that range voting “disadvantages” or “disenfranchises” the poor. A rational poor person would prefer range voting over IRV, even if your alleged disparity were accurate. The simple explanation for this is that range voting gives less say to the effects of randomness and ignorance than IRV. But that’s an esoteric explanation that is basically lost on the unsophisticated realm of IRV advocacy.
<
p>Going further in that vein, you ignore at least one substantial manner in which range voting counteracts elitism. That is by decreasing the importance of money in elections (making big corporate donors less significant). That is, range voting does not require that a candidate be “electable” in order to make it safe to vote for him. If your favorite candidate is a no-hoper like Mike Gravel or Bob Barr, you’re free to vote for him without fear. Whereas IRV can cause a faction of voters who prefer X>Y>Z to be better off voting Y>X>Z if they don’t think X is electable. IRV maintains two-party domination, and makes it important not just to be a good candidate, but to appear electable. As we saw throughout the recent primary races, having a big war chest is crucial to presenting this appearance of electability. So IRV does little to lessen the money factor. Whereas range voting dramatically diminishes it.
<
p>
<
p>An honest, low-information, mathematically disinclined voter is better off with range voting than with IRV. You can’t dispute that.
<
p>
<
p>This is another very confused question. The IRV spoiler scenario is to be expected frequently, according to statistics. Therefore it is incumbent upon you, if you are to argue that reality is somehow biased against such an occurrence, to present us with full-ballot-data analysis of a representative sample of IRV elections that demonstrate the lack of the IRV spoiler scenario. Of course you won’t do that, because as you so conveniently ignore, there is extremely limited access to full ballot data from IRV elections, in an easy-to-analyze format.
<
p>The other huge problem that you ignore is the same one responsible for the relatively small number of plurality elections which exhibit the spoiler phenomenon. It’s called strategic voting. A lot more third party and independent runs have the potential for a spoiler effect than actually cause one. That is because most voters do not like “wasting” their votes, and so they vote strategically for their favorite between the Dem and GOP nominee. If people are doing this in IRV elections, then the actual frequency of spoiler scenarios is being hidden. The nice thing about range voting is that such strategic behavior cannot be expected to maintain two-party duopoly, or stop alternative parties from getting truly representative election numbers.
<
p>
<
p>That is simply false. There is no “real, empirical data” about voter utilities from real elections. We cannot read people’s minds to compare how happy different voting methods left them over the course of zillions elections – both because we do not have brain scanners and because elections are relatively infrequent. Computer simulation is the solution to that. It also allows you to run zillions of simulated elections, and to read the exact preferences from the voters’ digital minds.
<
p>Of course that would be irrelevant if practical considerations, such as complexity, were not addressed. But I have personally conducted a range voting experiment in Beaumont, Texas. There was no confusion that I observed. People scored the candidates on a 0-10 scale the same way millions of people rate products on Amazon.com. It’s not rocket science. In fact, experiments show that people find range voting simpler, in that they less frequently spoil their ballots with range voting. Whereas with IRV, spoilage becomes about 7 times more common. Range voting is also simpler to count.
<
p>It is nice to know that range voting is not only vastly better than IRV in terms of performance, but also simpler and easier, making it the more practical choice.
<
p>
<
p>We’ve discussed numerous real elections that demonstrate the susceptibility of IRV to a host of problems. And we’ve demonstrated results from computer simulations, for example, that give us far more explicit information than any real elections possibly could.
<
p>
<
p>In which elections has IRV chosen the right winner? You need to get your burden of proof straight, cause you’ve currently got it precisely backward.
<
p>
<
p>Let’s put the burden of proof in the right place. Point me to a representative sample of the history of IRV elections, showing zero examples of this.
<
p>
<
p>Considering that you have no idea about the extent of strategic voting in those elections, nor full ballot data for virtually all of those elections, nor any knowledge of how much the candidate pool was pre-constrained by the inherent properties of IRV, you have absolutely no way to make such an assessment with any certainty.
<
p>The fact remains that such problems are extremely common in the statistical model. For them to actually be rare in practice would require reality to somehow be biased non-randomly – in just the right way to counteract those problems. Do you even have an inkling of a theory that would explain that?
<
p>I don’t think you do, or else you would have discussed it by now.
I forgot to respond to one of Greg’s most problematic arguments:
<
p>
<
p>This is where I embrace the saying, “You’re entitled to your own opinions, but not your own facts.” While Greg presents the Majority Criterion as a principle whose near-universal appeal puts it virtually outside the realm of question, mathematicians such as Charles Dodgson (aka Lewis Carroll) long ago disproved this assumption via a logical device called “reductio ad absurdum” (reducing to an absurdity). Put simply, if a proposition can be shown to contradict itself, then it cannot be a correct proposition. Consider the following simplified example of electoral preferences for 3 candidates, A, B, and C:
<
p>% of voters – their ordered preferences
35% A > B > C
33% C > A > B
32% B > C > A
<
p>Now let us ask Greg (or anyone else who insists on the sanctity of the Majority Criterion) to tell us which of these 3 candidates is “best”, based on these preferences. No matter who he picks, he’s contradicting the very Majority Criterion he espouses.
<
p>For instance, say Greg decides that C is the best (C actually would be the winner under IRV). If B and C were the only ones to enter the race, then a huge 67% majority of voters would pick B over C. According to the Majority Criterion, that Greg insists is inviolable, B would therefore have to be better than C (for society as a whole). But that contradicts the original premise that C is the best. And this happens no matter who you decide is best – proving that the Majority Criterion is not valid.
<
p>Greg and other IRV proponents, such as FairVote’s Rob Richie, have been very clever with these sorts of arguments, exploiting the intuitive notions people often hold before having more deeply researched them and thought them through. By dumbing down the discourse on this crucial issue, IRV evangelists like Greg have stifled the progress of electoral reform. It’s sad and irresponsible, and interested parties (such as minority political parties) must work hard to stifle their misinformation campaign.
In your example, no candidate has a majority of first-choice support, so the majority criterion does not require any candidate be elected. Of course there are many examples, like yours above, in which no candidate receives a majority of first-choice support. In those cases, we should fall back on other criteria like Condorcet loser. But when there is such a candidate who receives a majority of first-choice support, I and many others believe that candidate should be elected — that democracy demands it. If you disagree with that, fine, but I think the vast majority of people agree with me.
<
p>On the contrary, I listed no less than three cases where a candidate has a majority of first-place support. In my example, A runs against B, and has a first-place majority. Likewise, B runs against C, and C runs against A.
<
p>But no matter which candidate you name as the “best” for society, in one of those match-ups, the Majority Criterion causes us to elect the wrong candidate.
<
p>
<
p>And I just showed you a mathematical proof that this is wrong.
<
p>
<
p>I don’t disagree, I disprove. That’s why I said you’re entitled to your own opinions, but not your own facts.
To make this even more clear, I ask you to simply tell me which of those 3 candidates is the best for society, based on those ordered preferences. Once you have done that, I will then prove that you disagree with the majority criterion.
And while I’m talking about the simple proof against the Majority Criterion, here’s an additional construction to make the point. Imagine the following two elections.
<
p>Year one: A vs. B : A wins
You prefer A and are in the majority.
<
p>Year two: X vs. Y : Y wins
You prefer X and are in the minority.
<
p>Now say that the amount by which you prefer X over Y is much larger than the amount by which you prefer A over B. If you are economically rational, you would want to relinquish your majority in the former election and have minority rule in both elections. That would actually make you better off. Range voting vastly increases a voter’s expected welfare, and one small cost of that benefit is that he gives up majority rule.
<
p>But range voting actually makes this more democratic. Here’s how.
<
p>Imagine that we use range voting, and all voters start by casting a “10” for their favorite candidate – because they know that this will guarantee satisfaction of the Majority Criterion (and they think that’s very important). Candidate X has 51% at this point, while Y has 48%, and Z has 1%.
<
p>Now suppose that as they deliberate further on it, many of X’s supporters decide that they would like to support Z, their second choice, just in case they do not actually have a majority, and Y (the least favorite for most X supporters) does. Let’s say that many Y supporters do the same thing, resulting in a win for Z, in spite of the fact that X actually had a majority.
<
p>The question is, was it wrong to violate the Majority Criterion here? Well, not according to democratic consensus, since a huge number of voters had to decide that utility was more important than the Majority Criterion in order for this to have happened.
<
p>This is what economists call “revealed preference”. Sure many voters will tell you, if you ask them casually, that they insist on the Majority Criterion. But if you ask them to put their money where their mouth is, you’ll get a different story.
<
p>So, in practice, either range voting satisfies the Majority Criterion, or there’s a democratic consensus that utility trumps the Majority Criterion, QED.
<
p>No it doesn’t.
<
p>Contrary to these popular myths, IRV does not make it safe for you to vote in the most expressive way for your sincere favorite.
<
p>Here’s an explanation.
<
p>Some more pages to consider:
<
p>http://rangevoting.org/CFERlet…
http://rangevoting.org/IrvPath…
http://rangevoting.org/IrvExtr…
http://rangevoting.org/AusAbov…
<
p>Don’t believe the IRV propaganda. Look at the scientific evidence for yourself. I’m not just saying this as a score voting proponent. Condorcet, Approval, and Borda are 3 other methods that are also better than IRV.
trying to circumvent the US Constitution in the first place. The writers of the Constitution knew the people would want various changes from time to time. So they created mechanisms to make changes but with safeguards to prevent impulsive reactions. These safeguards delay any kind of action/changes which ensures the change has real merit and withstands the change process. This attempt to change things is borderline unconstitutional and certainly unethical… IMO. If the backers of this movement really believed it was worthy, they would do it the right way!
Hmmm heres an idea lets try an experiment,Lets actually read what the constitution actually says regarding presidential elections.shall we?
<
p>Article II
Section 1
Each state shall appoint, in such manner as the Legislature thereof may direct, a number of electors, equal to the whole number of Senators and Representatives to which the State may be entitled in the Congress: but no Senator or Representative, or person holding an office of trust or profit under the United States, shall be appointed an elector.
<
p>The National Popular vote law still uses the the state electors but the State directs the electors to the winner of the national popular vote(in a manner as the legislature may direct!).
<
p>I sorry but I have yet to see a rational argument against the NPV.