RangeVoting.org is an advocacy site for range voting (also known as score voting), full of (sometimes bewilderingly academic) information arguing in favor of range voting and other ratings-based election methods.
RangeVoting.com is a brand-new-today site created by FairVote, an instant runoff (and other voting reforms) advocacy group, specifically for the purpose of mocking range voting and other voting methods that FairVote disagrees with.
I can't rouse anything other than disappointment over FairVote's blatant attempt to hijack the name "range voting" in such a search-engine-friendly way. It's sad how an organization, one that so often cries about its opponents being unfairly combative, would take such a low road to try and make its arguments seem more legitimate.
On the other hand, I guess it means range voting advocacy is making an impression. I often go to the trouble of checking the comments on pro-IRV editorials, and in the past I would always be the first to present any counter arguments or range voting advocacy. But recently, and more and more often, I find someone, or even two or three someones, have already beaten me too it. This new ".com" shows that FairVote feels that ignoring us is no longer an option.
Couldn't FairVote afford better than just a blogspot? Not saying it isn't them, but that's uncharacteristically unprofessional.
ReplyDeleteIn a first-past-the-post plurality system, it's hard for there to be more than one dominant alternative to the status quo and there's incentive for negative campaigning.
ReplyDeleteIs it true that if voter preferences are unimodal among candidates that the diffs between rank choice based rules and other alternatives to FPTP are reduced considerably?
dlw
I do hate these types of pieces that selectively pick out certain characteristics of elections as "definitive" and dump on (almost)all of the others....
ReplyDeleteBut it is true that Approval Voting works best when lots of voters get how it should work, but that's neither here nor there.
In my opinion, the powers that be don't like hedging over too many candidates and incumbents want to keep their incumbency advantage and so my support for IRV3/AV3 is probably "strategic". I've read enough of the electoral reform lit to be a relativist when it comes to pushing for alternatives to FPTP...
dlw
Hey, Dale,
ReplyDeleteThere's nothing nefarious about having an easy URL to give out people who have questions about a voting system that you and others always raise as a proposal. No one owns the rights to the name of a voting method.
Few serious reformers actually trying to get things passed see approval voting and range voting as serious. And the fact that neither method is used for ANY governmental election at ANY level ANYWHERE in the world (with absolutely no indication of that changing) is evidence that we just may be right.
So no, we're not "laughing" at you. The piece is designed to explain why you're wasting your time. If you read one part of it, read the part about Bucklin Voting and its failures. Hard for me to believe you can read that and imagine approval voting really working.
f you don't like IRV and want third parties to be able to run hard and have a shot at winning, then take on winner-take-all elections with proportional representation (as we do along with IRV). Now that actually would make some sense.
As to DLW, being highly vulnerable to strategic voting isn't just "another property." That's the whole point of our analysis. If a system is highly vulnerable to strategic voting, your reform is dead in the water.
Rob,
ReplyDeleteThe point of my analysis is that it's hard to have more than one alternative election rule (at a time) in a system dominated by the use of FPTP.
I agree that few serious reformers actually trying to get things passed are pushing for Approval Voting. I think that this is more due to pragmatic reasons than the intrinsic superiority of IRV, which is fine by me. At the end of the day, what matters is that we diversify our electoral system as soon as possible. But I do think that for bigger elections that you could trim the time it takes to count the vote considerably and answer the valid critiques of Approval Voting advocates by using a limited form of Approval Voting in a first stage of IRV.
dlw
DLW -- The analysis in facts makes your point about potential utility for approval voting. For example, it could be part of a system with repeated voting in getting a kind of "thermomter" of where things voters stand -- and __potentially __could be a way to winnow the field.
ReplyDeleteBut that's quite different from proposing approval to replace plurality voting, runoffs or IRV for "one-shot", one-round elections.
One recent telling example is discussion in San Francisco's mayoral race about no one having figured out how to game IRV. Well... that's great. Otherwise, the system wouldn't work. You can be absolutely sure that this would not be said if approval or range were in place.
To be clear on "intrinsic superiority", that phrase gets into what we call theoretical flaws. Being directly and consistently vulnerable to tactical voting is different because.. it means it just won't work. That's the point that its advocates won't accept, seemingly.
And a final word for Dale, et al: I truly don't understand why you don't do more to advance proportional voting. It seems like something you would want to do. If you ever get around to that, get in touch.
Rob, I just wish us electoral reformers could get along... I've had these frustrating email exchanges with Stephen Brams...
ReplyDeleteI understand that strategic voting is possible with all election rules.
I also understand that strategy matters a lot when one can't tell from navel gazing what one's cardinal utilities over all of the candidates are and that what works well for public votes by Olympics judges on a specific concrete athletic feat doesn't per se work for private votes by regular voter based on the melee of issues/matters of character (with a slew of misinformation thrown in for good measure) involved with evaluating political candidates.
We agree on what matters most.
Bayesian Regret analysis is too utopic to be the definitive basis for the evaluation of electoral rules and practice should trump theory in this regard.
As for tactical voting, I think that if a handicapped version of Approval Voting that made folks need to approve of three of the seven candidates, it would remove this consideration from the equation. And theoretically, almost all folks could learn to vote well with approval voting, but that's neither here nor there in the immediate future, which is everything for the future of the US's democracy.
dlw
Dale, do you think Approval Voting advocates like yourself might strategically support Fair Vote's advocacy for (3 or 4 seat) Proportional Representation in US Congressional Elections, even if they push for STV or cumulative voting?
ReplyDeleteI think the specifics of what types of options we give voters is trumped in importance by the need to use a mix of single seat and multi seat elections.
dlw
Sorry for taking so long to get back to these (busy month, but I'm a home owner now!)
ReplyDeleteTo focus on the positive:
Yes. I, and many other approval/range advocates would support proportional representation, even if it were STV.
Any sort of proportional method is probably better than any sort of single-winner method, and the differences between proportional methods are probably less significant than the differences between single-winner methods.
(Although, we do have some ratings-based proportional methods; re-weighted range (which could easily be converted re-weighted approval) for instance.)
Getting into the negative:
I disagree, very strongly, that approval is more tactically vulnerable than other voting methods, for a host of reasons (a future blog post there...) but the major one being that "bullet voting" doesn't happen at anywhere near the rate that your intuition suggest, and so it is IRV that more-strongly degenerates to plurality results (and so, two-party domination), not approval. Which is why *I* don't understand why *you* keep pushing IRV.
Dale,
ReplyDeleteThese might be a "go read the FAQ" style questions, but I was wondering, what affect (if any?) is approval and/or range voting projected to have on marginalized groups, such as low income voters? Would it be projected to increase their voting power, reduce it, keep it the same?
Or a better way of putting it, I'm curious how would approval/range voting help (or hinder, or not affect at all) the ability for groups to organize more effectively to press their issues into the political mainstream diaglogue.
Hopefully my wording wasn't too nebulous?
I'd hesitate to make any specific claims about marginalized groups.
ReplyDeleteWhat approval/range should allow, is for a candidate with views which are popular, but unsupported by either of the two dominate parties, to more-easily gain support, since they would not immediately be branded as a spoiler.
To the extent that marginalized groups are in their position because of active hostility towards them, approval/range would perhaps not be much help. But if support for them is popular but ignored, then they may see some benefit.
Dale, as you know I am an electoral relativist. I think most single seat alternatives to FPTP are worthy of support if they are pursued by reformers. Top two primary is an exception, but I think it could be gamed at the primary and general election levels to push for better electoral reforms.
ReplyDeleteI think that when one considers the political economy of electoral reform that the such strengths of Approval Voting could become liabilities in a two party dominated system. I only push for electoral reform changes that do not challenge the fact we have two major parties.
dlw
"I only push for electoral reform changes that do not challenge the fact we have two major parties."
DeleteYou might as well have said "I don't support electoral reform".
Hey Dale, is it true that if preferences are unimodal that the diffs between rank choice and approval/score-voting are reduced significantly?
ReplyDeleteAs I learned from the book Majority Judgment, this preference assumption can be empirically verified and so presumably it could be verified that IRV is a worthy electoral reform for even folks who prefer Approval Voting to support. Also, I'd add that if preferences are unimodal then non-monotonicity should be a non-issue since folks are very unlikely to experience such a radical change in political preferences.
What say you?
dlw
My apologies, I meant this blog-post when I linked to the political economy above...
ReplyDeletedlw
Dale, would it be okay if reposted
ReplyDeleteTo focus on the positive:
Yes. I, and many other approval/range advocates would support proportional representation, even if it were STV.
Any sort of proportional method is probably better than any sort of single-winner method, and the differences between proportional methods are probably less significant than the differences between single-winner methods.
(Although, we do have some ratings-based proportional methods; re-weighted range (which could easily be converted re-weighted approval) for instance.)
Getting into the negative:
I disagree, very strongly, that approval is more tactically vulnerable than other voting methods, for a host of reasons (a future blog post there...) but the major one being that "bullet voting" doesn't happen at anywhere near the rate that your intuition suggest, and so it is IRV that more-strongly degenerates to plurality results (and so, two-party domination), not approval. Which is why *I* don't understand why *you* keep pushing IRV.
at my blog.
I promise to repost it in its entirety, instead of cherry-picking...
dlw
@DLW
ReplyDeleteUni-modal preferences don't eliminate the differences between ranked and rated methods. I don't recall off-hand which came out with a greater divergence, but they were all in the same ball park.
But uni-modal preferences certainly do NOT eliminate non-monotonicity in IRV elections.
Example: 3 candidates, left-to-right, A,B,C. There are only 4 votes consistent w/ uni-modal preferences. 21 voters, distribute thus: 7:A>B>C, 3:B>A>C, 3:B>C>A, 8:C>B>A. C wins. Now, increase C's support by raising C from bottom to top on 2 of the A>B>C ballots. C loses.
--
I'm not sure what you're getting at with the political reform article. Which strengths of approval do you see as being liabilities? That it challenges the existing party structure, and therefore will be opposed by it? I suppose that's a legitimate concern; but I have (perhaps false?) hope that the truth will win out of obvious cynicism.
--
And yeah, I guess quoting that comment of mine would be fine; be sure to provide the direct link:
http://leastevil.blogspot.com/2011/07/then-they-laugh-at-you.html?showComment=1314746847726#c1476761526538675620
DSH: I don't recall off-hand which came out with a greater divergence, but they were all in the same ball park.
ReplyDeletedlw: I'd be interesting in checking such out, especially if the number of candidates is reasonable(like 4), given that there's a costly campaign and endogenous entry. There may be like 7 candidates on the ballot often, but many of those aren't serious and so it'd be more useful to omit them from the model.
DSH:But uni-modal preferences certainly do NOT eliminate non-monotonicity in IRV elections.
Example: 3 candidates, left-to-right, A,B,C. There are only 4 votes consistent w/ uni-modal preferences. 21 voters, distribute thus: 7:A>B>C, 3:B>A>C, 3:B>C>A, 8:C>B>A. C wins. Now, increase C's support by raising C from bottom to top on 2 of the A>B>C ballots. C loses.
dlw:I said non-monotonicity was a non-issue since unimodal preference means that for someone in the fourth case to re-rank A from third to first place is unlikely to happen in real life. It'd be like a supporter of Michele Bachmann switching to like Ralph Nader best or vice versa.
DSH:I'm not sure what you're getting at with the political reform article. Which strengths of approval do you see as being liabilities? That it challenges the existing party structure, and therefore will be opposed by it? I suppose that's a legitimate concern; but I have (perhaps false?) hope that the truth will win out of obvious cynicism.
dlw: The bottom line is that while quite often in the past electoral reforms were driven by the desire of those in power to stay in power, this has been diminished some and yet that doesn't mean we can avoid the need for political jujitsu or picking our battles such that those in power are wiser to accommodate us than to fight us.
Thanks for letting me quote you and the good honest feedback.
dlw
"I said non-monotonicity was a non-issue since unimodal preference means that for someone in the fourth case to re-rank A from third to first place is unlikely to happen in..."
ReplyDeleteI'm gonna stop you right there.
Non-monotonicity isn't about what a voter MIGHT CHANGE. There is no changing. It is about an incongruity in what the outcome of the election IS.
Drop, for a moment, the "all honest, uni-modal" requirement. You have two election results; either of them could be valid and accurately express the will of the people. In one, a candidate with low-ish support wins, in the other, that candidate, with higher support, loses. That's nonsense.
One of those two elections must be WRONG. I don't know which, but I know it has to be (at least) one of them.
Non-monotonicity creates doubt and confusion over how to express your vote in an effective way. That's a recipe for voter repeal (and indeed, that Burlington's last IRV election was non-monotonic certainly contributed to the success of the repeal effort there.)
As I understand from email exchanges with Stephen Brams, "the work of Robert Norman, a mathematician at Dartmouth, indicates that the probability of nonmonotonicity in a 3-candidate, close election is not low (about 20 percent)."
ReplyDeleteBut that's actually a close election between three candidates and such is a lot less likely than a close election between the top two candidates. And even if it were such, 80% of the time one's rankings would have their intended effect. In our world of Heisenberg uncertainty, why should we expect our votes to always have their intended effects?
As for, "You have two election results; either of them could be valid and accurately express the will of the people. In one, a candidate with low-ish support wins, in the other, that candidate, with higher support, loses. "
dlw: This seems to relate more to BR-regret stuff than non-monotonicity to me. Could you clarify the connection for me?
And do you really believe that all who pushed against IRV in Vermont did so out of the right intentions? As you know from campaigns against the adoption/use of Approval Voting, it's easy to isolate aspects of a voting rule and blow their significance out of proportion in order to protect or bring back the status quo use of First-Past-the-Post.
As such, I stand by my previous remarks: with a unimodal model, folks would never change their preferences such that non-monotonicity would come into play. And one can empirically test for the usefulness of the unimodal model. In the study cited in "Majority Judgment", they found that among the 4 serious candidates there was statistical evidence for unimodality tending to hold.
And so if FairVote were to push for Proportional Representation and the use of a modified form of IRV that uses a limited form of Approval Voting to whittle down the number of candidates to three then wouldn't that be something an intelligent person like yourself could get behind?
dlw
The Pew study shows a good cross-section of the electorate. Part of the problem with it is that it its focused on "value statements", and this hides the fact that if you poll different positions on individual issues, you can get the majority people agreeing with 'moderate' positions.
ReplyDeleteAbortion is a classic case. Approximately half of voters will identify as pro-life, and the other half as pro-choice; but all polls show that large majority in both groups believe that abortion should be illegal in the third trimester; and a majority of Americans also support parental notification laws. But you'll find similar results with issues like gay marriage; where people will opt for civil unions; for marijuana; where people will opt for decriminalization over legalization; and on fiscal issues, where people support combinations of spending reductions, spending increases, tax cuts, and tax increases; depending on the case.
Still, in a political environment, it often comes down to to the "value statements", since these voting blocs have to form coalitions. Currently those coalitions are what we have now, Republicans/Democrats.
The "Libertarian lite" that TiradeFaction mentioned would be a proposal for a new coalition; the people in that group think they have the best chance of capturing the center and so, that in mind, portray themselves as "centrist". I think paradoxically they represent the center least, though.
Besides the public at large being more moderate on social issues, they also tend to be protectionist and against free trade agreements, and also tend to support things like campaign finance restrictions, which libertarians believe is an infringement on free speech.
Perot appealed to those type of views 90s, but they're still pretty common today, just obscured by the fact that people argue in absolutes all the time, because they're arguing values rather than policy.
The problem with a "moderate" third party isn't that most Americans would disagree with it, but the way politics actually works in the real world -- coalition forming -- makes it difficult.
It would be interesting to see what the formal opposition to a "Libertarian lite" party would look like.
@redfish
ReplyDeleteI think you posted this in the wrong post.
"It would be interesting to see what the formal opposition to a "Libertarian lite" party would look like."
Absent of the Democrats and Republicans? I'd imagine a center left party. Keep in mind when I said "Libertarian lite", it was a broad brush that doesn't include inevitable nuance. "Center rightists" I suppose would be more accurate. That faction of moderates seem the most eager to organize (and the ones getting funding from big capital), so I suspect their next round will have Perot's protectionism absent. I could be wrong though.
It's worth keeping in mind the only effective differences between the Democrats and Republicans nowadays in terms of policy are the "value" stances you bring up, or culture war wedge issues as DLW likes to call them. Both this factor, and the fact the national dialogue on the economy has shifted increasingly right ward for the past 20-30 years are two factors I often see "centrists" overlook or are unaware of.
@TiradeFaction:
ReplyDeleteI assumed when you said "libertarian lite" you meant the type of "fiscally conservative, socially liberal" voter, that doesn't take his positions to such a principled extreme that he's a libertarian. I know a lot of people with this orientation and they usually pitch their positions as "centrist". The Tea Party voter is somewhat different, they're pretty culturally conservative; though they're trying to appeal with the same principle by ignoring social issues.
My view on that, like I said, was just that it doesn't really represent the actual policy views of the center at all (which are pretty moderate); it just represents some vague idea of their "values". The majority of people's policy views are left out in any case.
And you're right, politics is divided by cultural wedge issues. Which doesn't have anything to do with social issues, like people often assume. Social issues absent, the fiscal debate has come about cultural issues; framed as between people who hate business versus people who hate unions, people who like the European model versus people who like small-town America, and so on.
The opposition to the centrist-posing Tea Party conservatives are the also centrist-posing "No Labels" groups, which rallied with Jon Stewart and Stephen Colbert. None of these are really new coalitions though; they're just new ways conservatives and liberals are marketing themselves.
"I assumed when you said "libertarian lite" you meant the type of "fiscally conservative, socially liberal" voter, that doesn't take his positions to such a principled extreme that he's a libertarian."
ReplyDeleteYes, I did mean that actually. While they're a significant part of people who identify as "centrist" (which on economic views actually puts them to the right of a lot of popular economic positions), they're only a part of that bunch. When it comes to social views, there's the ones that are socially progressive (not exactly the same as Libertarian), vaguely socially libertarian, and some authoritarians as well (some support the Patriot act and so forth). I was just trying to point out I'm not being detailed in my generalizations. (As generalizations rarely are)
Another problem though I think that faces "centrists", is they often over inflate how much consensus "centrists" have. Some honestly believe they'll be able to rally 60% of the electorate under one united banner. That's wishful thinking, to say the least.
"In our world of Heisenberg uncertainty, why should we expect our votes to always have their intended effects?"
ReplyDeleteBecause we can and because it's better that way.
Don't be trying to shoehorn Heisenberg someplace he doesn't belong (especially in a world where Arrow's work exist!)
"As for, "You have two election results; either of them could be valid and accurately express the will of the people...."
This seems to relate more to BR-regret stuff than non-monotonicity to me. Could you clarify the connection for me?
BR is inapplicable here; we're not assigning utilities, and so we can't measure regret.
Non-monotonicity is simply the realization that if certain voters had liked the winning candidate MORE, then they would not have won. (Or, that if certain voters had liked a losing candidate LESS, then they would have won.)
No one is changing their votes. No one is choosing to vote dishonestly. We are simply realizing that, if the electorate had liked a candidate BETTER, that would have been WORSE for the candidate. That's a perverse, non-intuitive, counter-productive aspect of a voting system.
Trying to call it back to your uni-modal model and saying "but they won't change their votes that way" doesn't matter; that's not the problem. (Also, you're trying to turn the map into the territory.)
You asked if an assumption of uni-modal preferences eliminated the problem of non-monotonicity in IRV. I answered you, correctly, that it does not.
I think you should go play with this vizualizer: http://zesty.ca/voting/voteline/
Enable Hare, enable either Condorcet or approval (they give identical results under this model's assumptions), enable a third candidate. Put one candidate around 30% from the left, put a second about 35% from the right, then slide the third candidate back and forth between them. You will see what non-monotonicity looks like.
"do you really believe that all who pushed against IRV in Vermont did so out of the right intentions?"
No. Some of them were just pissed off. But the reason they were pissed off (even if they couldn't put it into these words) was because of a spoiler and non-monotonicity.
"Because we can and because it's better that way."
ReplyDeleteOr you think you can and believe such would be (a lot) better...
"Don't be trying to shoehorn Heisenberg someplace he doesn't belong (especially in a world where Arrow's work exist!)"
dlw: I think it does belong. Just as once you relax the assumption of cardinal utility, there's an indeterminancy over which candidates someone gives their approval votes to, there's plausibly a(n unlikely) situation where there's a 20% chance that some people's votes could have the opposite of their intended effect.
"Non-monotonicity is simply the realization that if certain voters had liked the winning candidate MORE, then they would not have won. (Or, that if certain voters had liked a losing candidate LESS, then they would have won.)"
Yes, but it's not just any voters who have to change their preferences/votes. It's folks who, upon the assumption of unimodality, really didn't like said candidate.
"No one is changing their votes. No one is choosing to vote dishonestly. We are simply realizing that, if the electorate had liked a candidate BETTER, that would have been WORSE for the candidate. That's a perverse, non-intuitive, counter-productive aspect of a voting system."
dlw: You are over-generalizing. If x set of voters who really didn't like candidate y were to change their mind then y might not win. This tends to happen when there's a close 3-way between candidates. In which case, the differences among the candidates would be less variable and it'd still be a step up from FPTP.
"Trying to call it back to your uni-modal model and saying "but they won't change their votes that way" doesn't matter; that's not the problem. (Also, you're trying to turn the map into the territory.)"
dlw: It does matter. Only plausible possible scenarios are worthy of consideration.
"You asked if an assumption of uni-modal preferences eliminated the problem of non-monotonicity in IRV. I answered you, correctly, that it does not."
dlw: If you reread my question, I didn't say it eliminated the problem. I said something akin to whether it "neutralized" the problem.
"dsh:No. Some of them were just pissed off. But the reason they were pissed off (even if they couldn't put it into these words) was because of a spoiler and non-monotonicity."
dlw: Cuz there wouldn't have been a spoiler problem if they had been using FPTP? That's what they're going to be using now. I mean, Follow the money! In the next election, enuf Moderate Republicans would have ranked number one the Democratic party's candidate to get the "right" outcome. It was just a matter of learning by voting... Yet, those who were financially up$et by the IRV outcome seized the moment to bring back FPTP.
Maybe I'm(You're) wrong. Maybe I'm (You're) right. The point is we'd need more evidence to say either way... My contention is that this case study does not refute my basic contention that IRV is a significant improvement over FPTP that is worthy of support even if one prefers Approval Voting.
dlw
On Burlington, IRV-proponents said "You'll be able to vote your conscience without worrying about spoilers!"
ReplyDeleteMeanwhile, IRV-opponents said "IRV still has spoilers, and adds a host of new problems, like non-monontonicity. IRV's promises will fail, and, once bitten, further improvements to the voting system will be impossible for years to come; most likely, they'll go back to crappy plurality for a generation."
One of those groups was right, and the other was wrong.
So no, I don't think you're going to convince me that IRV is a significant improvement over FPTP worthy of my support, certainly not based on Burlington.
On Burlington, IRV-proponents said "You'll be able to vote your conscience without worrying about spoilers!"
ReplyDeletedlw: Fair Vote has a comparative advantage in marketing electoral reform. This entails often stating things that tend to be true as if they were always true. And for most folks who dissent from the two major parties at times, this is very much true when IRV is used.
DHS:Meanwhile, IRV-opponents said "IRV still has spoilers, and adds a host of new problems, like non-monontonicity."
dlw: If there are a "host" of problems with IRV, non-monotonicity does not deserve to be listed at the top of the pack for the practical reason I mentioned above.
DHS: IRV's promises will fail, and, once bitten, further improvements to the voting system will be impossible for years to come; most likely, they'll go back to crappy plurality for a generation."
DLW: IRV's promises did not fail! It wasn't permitted to continue to work in Burlington because of an opportunistic campaign against it.
It is true that IRV doesn't end effective two party rule, but it does make it so that those two parties are more apt to be centered around the true center. What was going on in Burlington was a shift towards two different major parties effectively being in power at the local level. Since the Republicans couldn't get elected anymore by FPTP, the odds are its moderates would join the Democratic party and the progressive democrats would join the Progressive party, in which case the Republican party would be the effective third party who would benefit from the use of IRV since they could help elect the Democratic party's candidates.
In which case the better candidate would come to win [more often], while the center would have tilted to the left economically due to how the Progressive Party would be the second major party. Now, with the use of FPTP, the Democratic party might be in power, but it'll tilt to the right not the left.
DSH:One of those groups was right, and the other was wrong.
dlw: What if both were using "political slogans" that selectively focused on aspects of reality that served their conflicting purposes (with the latter oft having more so other purposes than keeping the god-awful IRV rule from ruining elections.)
DSH:So no, I don't think you're going to convince me that IRV is a significant improvement over FPTP worthy of my support, certainly not based on Burlington.
dlw: I'm sorry to hear that... Burlington is just one case and it got stopped after only two IRV elections! There are lots of other cases where IRV has been used and it been retained. So why are you focusing so much on this small sample of one(or so)? It seems like a very non-engineer-like thing to do to me.
dlw
Dale,
ReplyDeleteI did look at the link you gave me that looked at the effects of the different election rules under different assumptions. It was cool. I showed it and emailed it to my father who is a prof of Math and Statistics and he liked it too.
As for the specific example you asked me to look at (candidates A, B, C at 30%, 65% and somewhere in between). It lacked realism given how the two major party's tend to be close to the de facto center (for robu$t reasons). For I discount the non-monotonicity problem due to realism.
Given the de facto position of IRV as the best understood and marketed alternative to FPTP in the USA for single-seated elections, the burden of proof is on those who claim that non-monotonicity ought to be a deal-closer to prove its plausibility.
As I see it, a 20% chance of occurrence in a close 3-way 3-candidate election would not upset voter incentives to vote sincerely enough. I fear that sincere opponents of IRV, not unlike yourself, were being played by pragmatic proponents of FPTP during the anti-IRV referendum in Vermont last year.
dlw
I'm simply surprised that you don't think it's a problem if 1 in 5 elections returns an obviously incoherent result. That seems quite significant to me.
ReplyDeleteAnd the problem isn't just present in close elections, by the way; it's present in 7-15% of all elections with 3 candidates.
And if you pre-condition the results to only look at elections where the plurality result and the IRV result differ (i.e., the elections where IRV "matters") it shoots up somewhere into the 10-35% range.
And non-monotonicity isn't the only perverse outcome possible. This page goes over some others, and their probabilities. The headline-grabbing number is, in elections where IRV matters, over 70% of elections return a paradoxical result of some kind.
"I'm simply surprised that you don't think it's a problem if 1 in 5 elections returns an obviously incoherent result. That seems quite significant to me."
ReplyDeletedlw: It's not incoherent. It's an outcome whereby if some folks were to experience an unlikely radical change in voting preferences it could have unintended consequences, contrary to their intent. But 80% (or 100-p%) of the time, their actions would have their intended effects.
"And if you pre-condition the results to only look at elections where the plurality result and the IRV result differ (i.e., the elections where IRV "matters") it shoots up somewhere into the 10-35% range."
dlw:The moral of the story being that who gets eliminated first matters, given how only that candidate's votes would get transferred. In the Vermont case, it's still far more likely that Republicans would come to vote Democratic than Progressive and so a change of hearts would have their intended effect.
DSH:And non-monotonicity isn't the only perverse outcome possible. This page goes over some others, and their probabilities. The headline-grabbing number is, in elections where IRV matters, over 70% of elections return a paradoxical result of some kind.
dlw: I'm not sure which page you are referring to, I'm sure that if IRV3/AV3 were used instead that this would be trimmed. And, my intuition is that a good deal of these paradoxes are "sour grapes" phenomena, whereby some folks would've gotten an outcome they'd prefer if they'd only ranked candidates somewhat differently.
I don't see these as deal-closers against supporting the replacement of FPTP with IRV3 or IRV3/AV3. I don't see also why the cases whereby it got repealed justify its retirement among electoral reformers. Just because a new medical product has some quasi-adverse side-effects some times and its use is consequently rescinded a (small) fraction of the time, does not mean it should be prohibited by the EPA. Its use, relative to other treatments, needs to be weighed carefully on the basis of a lot of stuff. I concur with Rob Richie and other practicing "hands on" electoral reform activist that it(IRV3 or IRV3/AV3)'s worth using.
I'd argue from pragmatism that incumbency advantage matters a lot more from a politics of electoral reform perspective, since likely incumbents are the ones who literally adopt or propose a referendum for an alternative electoral reform. I believe AV trims the incumbency advantage a lot. I would not be surprised if this accounts for how when it gets adopted, it has often been replaced not too long afterwards. I prefer pushing for the use of AV in contexts where retaining some of the incumbency advantage is not important. An example would be in <a href="http://anewkindofparty.blogspot.com/2011/09/how-new-zealand-may-change-its.html>the second part of a referendum to replace the existing electoral system with an alternative as will be taking place in NZ. </a>
dlw
"since likely incumbents are the ones who literally adopt or propose a referendum for an alternative electoral reform. "
ReplyDeleteUnless it's a jurisdiction that allows citizens themselves to put a proposal via referendum on the ballot, like California. The dynamics change a bit then.
My (butted in) 0.02.
For the record, I support IRV.
I'd thereby argue that it's quite likely that the way we do referenda in the future may be with Approval Voting...
ReplyDeletedlw
"I'd thereby argue that it's quite likely that the way we do referenda in the future may be with Approval Voting..."
ReplyDeleteMaybe, but I've not seen any reason to suggest this, but it'd be interesting, and similar to what Benjamin Barber years ago suggested in his book "Strong Democracy".
I was more referring to the fact that incumbency advantage doesn't matter *as much* as in jurisdictions that allow forth referendums that do not need legislative approval. While they're very rare, they exist here and there, and most relevantly, in the United States.
I stand corrected.
ReplyDeleteI think pragmatism is more important than the attempts by electoral theorists to determine "the" right election rule that folks should rally around.
Democracy is an ongoing experiment.
Electoral reform is a key ingredient, but so is a detente among advocates for IRV and advocates for AV.
dlw
Okay... here's a "I've only had one cup of coffee" rant.
ReplyDeleteDale -- you say you're going to prove that approval voting is less prone to tactical voting than instant runoff voting.
Ridiculous You simply will fail. As a start, put your chalkboard aside and talk to real people involved in politics. ANY decent political consultant would tell you how they would go about trying to win an election with IRV vs. an election with approval voting.
Rather that just call our approval voting analysis (www.approvalvoting.blogspot.com) names, please confront its real-life examples. Please dispute Steve Pond's analysis of the Oscar voting for best picture. Please explain the failure of the Bucklin voting system and its contemporary analysts who concluded that the key problem with Bucklin voting was that indicating support for a lesser choice counted directly against a first choice.
On the point about you folks putting away your theoretical chalkboards and talking to people, you would learn a lot if you talked to real people in places like Burlington and Portland.
For example, you say that Burlington was some big failure because in 2009, the Republican who nearly won the election (it was close enough in the final round for him to call fora recount) was a "spoiler." Yet... he almost won! His backers didn't see him as a spoiler because they wanted him to win -- Burlington is a city where most of the backers of the three major parties sees its side as clearly better than the other two.
So there was no gaming of the vote in Burlington during the election. And if the election had been re-run, the Republican Wright's backers simply would have redoubled their efforts to get a few more votes, not voted tactically for a lesser of 2 evils.
Contrast that with approval voting. If the third-place Montroll had won because of approval votes from backers of the Republican, those Republicans would have been really frustrated if they found out that by voting for just their favorite Wright, he would have won. If the race were re-run, they would have cast a bullet vote-- but, more accurately, they would have figured out the math of approval voting before the election and never voted for Montroll in the first place.
Ayyone who says that Montroll would have won in Burlington in 2009 with approval voting, range voting or Bucklin voting just does not know Burlington politics. Much more likely is that the Condorcet loser Wright would have won.
I can tell you this in part because in places implementing IRV, we repeatedly need to emphasize that there's no advantage to casting bullet votes -- that indicating a lesser choice will not hurt your first choice. Even so, we still see some voters doing it because they think it's a point system (like Borda) or a Bucklin-type system. If it in fact WAS one of those system, there would be NO answer to their strategic plan -- and you would see massive loss of confidence in the system.
As to Burlington's "failure." You say that plurality voting would have been better, but with plurality, the Condorcet loser Wright would have been elected because he led in first choices. People voting for the 4th place independent (who won more than 10%) would have been told not to "waste" their vote on him. You say those outcomes are better?
Meanwhile, IRV has elected the Condorcet winner in EVERY other election where the full rankings have been provided, including a large number of very close elections in the Bay Area and several races where the IRV winner was second or third in first choices.
And you say plurality voting is better? And IRV is awful? Do you have any sense of why I and other IRV backers might you see you as a whacked-out idealogue who really just wants to defend the status quo for all you talk of reform?
And after that rant, let me congratulate Dale on at least saying he'd theoretically work for proportional voting. That's one theory that makes sense to me.
Your defense of IRV hasn't aged well, Robbie.
Deletehttps://www.rangevoting.org/RichieRV
https://www.rangevoting.org/RichieOnApproval
This comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDelete"the fact that neither method is used for ANY governmental election at ANY level ANYWHERE in the world"
ReplyDeleteWhat about the UN Secretary General?
http://rangevoting.org/UNsecyGen.html
FairVote has always been sleazy and dishonest. Always. They've never seriously responded to criticisms and have always promoted IRV despite mathematical data showing it's as flawed as plurality voting is.
ReplyDeleteMy suspicion is that FairVote is being funded and promoted by some combination of the major two political parties in an attempt to:
a) Pollute the discourse surrounding legitimate election reform attempts. Their website and discourse act in a directly adversarial tone and is positioned directly against RangeVoting.org because they know that RangeVoting.org is promoting a competent election method, so they funnel their efforts into a slick website and marketing campaign. RangeVoting.org is a humble website and certainly not flashy in design, but it does not need to be; it is backed by insurmountable amounts of data, and that's what matters. Or at least, it's what should matter. Unfortunately, much of what goes on in politics is a cheap popularity contest. In confusing an average audience about what election systems are worthwhile, they present the image that it's a burdensomely complex issue and that there are many competing election systems to choose from, thereby making it difficult to generate a public push to promote one election system.
B) Prevent a competent election system from coming into power in the worse case scenario that there is mass support case for election reform. If somehow the general public raise public consciousness enough that there's a widespread demand for election reform in America such that the major parties in power are forced to enact it, the 2 parties in power can maintain their power by selecting Instand Runoff Voting, which is known to lead to and maintain 2 party domination. This will act as a feel-good measure, satisfying the general public's demand for reform (those in the public who do not realize IRV is as flawed as plurality is) without actually changing the balance of power in a way that threatens their current status.