I think having two parties equidistant from the center is a stable equilibrium, for some distances, so it can arise without the two colluding. For a toy model, say that your chance of winning the primary is proportional to your distance from the party's center and your chance of winning the general is proportional to your distance from the electorate's center, as a ratio to your distance and your opponent's distance. So, with 0 being extremist and 1 being perfectly moderate, a position of 0.5+d means your chance of winning an election is (1-d) * d/(d + opponent's d). Which is stable with d=1/3 on both sides, I think.
And this isn't restricted to just the party elites and politicians themselves. Many voters will vote for whoever they like the most in primaries, but many will question their general electability (see Clinton vs Sanders). So I think this fits and makes a lot of sense.
It's also important that the voters aren't actually evenly distributed, they're something more like bimodal. Or arguably trimodal on a two-dimensional model, along the lines of DW-NOMINATE, where there's an economic dimension, and a second dimension that roughly maps to being socially progressive and egalitarian, or protective of traditional hierarchies.
Historically, Southern Dems were lefty on the economic dimension (they liked the New Deal programs, and stuff like the Tennessee Valley Authority), but right-wing on the social dimension. There are still a lot of voters like that, and they like Trump, who offers right-wing red meat on culture war stuff, while also saying he won't cut Social Security and Medicare.
Modern Democrats are mostly lefty on both the social and economic dimensions.
There's also a tranche of hardcore traditionalist voters (like the Christian Dominionist types) who are right-wing on both. Trump won in 2016 by holding together the hard-right faction, and the econ-left social-right faction. Biden beat him in 2020 by peeling off some of that econ-left social-right faction, and perhaps somewhat by goosing turnout from the left-left faction, some of whom stayed home had been mad that Hillary wasn't Bernie and hence stayed home (or even defected to Trump, because they thought burning down the system sounded good).
Unfortunately, there are only the tiniest sliver of voters who are socially left, but economically right -- neoliberals, libertarians, etc. In favor of lower taxes, and thinking about social insurance from the perspective of avoiding large distortions (so, like, we should get rid of a lot of disparate programs that create weird income cliffs, and just do a UBI and a negative income tax). This is why the distribution is trimodal instead of tetramodal. Sigh.
There is, BTW, a nifty tool for modeling the electorate and candidates' positions relative to the electorate, that was created by Ka-Ping Yee (who is probably known to some of the folks who hung around Berkeley Rationalist circles in the aughts).
One example: gay marriage. Obama could not even run on supporting gay marriage because it was believed it would make him unelectable. 8 years ago, you probably couldn't win most elections, even in fairly red areas, opposing gay marriage. As the tides seem to be shifting more right-wing lately, I see a definite undercurrent of opposition rising.
Also: Abortion. Overturning Roe is a liability in many elections for Republicans so you see most of them distancing themselves from it or proposing more moderate positions, in contrast with their prior stated views.
> Overturning Roe is a liability in many elections for Republicans
I'm not sure about that. Democrats love to make that claim, based on a small sample size of outcomes in low-turnout off-year elections that can be easily swayed by deep-pocketed interest groups mobilizing single-issue voters. But because of the low turnout, off-year elections very commonly produce results that are quite different from the preferences of the electorate in general.
Pro-life policies have historically been a winning position for the last 50 years or so, and every time some state-level policy restricting abortions that was duly passed by the people's representatives, after they got elected promising to restrict abortions, got overturned by judicial review on "this violates Roe" grounds, — in suits brought by activists of the party nominally claiming to be in favor of democracy! — it brought more and more people around to the realization that the root-cause problem that needed to be fixed was Roe itself.
A couple low-turnout elections are not going to hard-reverse half a century of ever-increasing momentum. If Democrats believe otherwise, they're in for a rude awakening.
There is just no way democrats should have had any sort of midterm in 2022 that wasn't a catastrophe and even in deep red state pro-choice ballot measures are winning. It's crystal clear that many people who were "pro-life" only was long as Roe was enact.
What's crystal clear is that those "low-turnout elections"(midterms) determine control of Congress half the time. But then, who cares about control of Congress, right?
The marketing of the "return to Roe" faction has been very strong. When they polled this issue in the past, most voters landed at some compromise short of what Roe mandated. A 20 week ban like the one at issue in Dobbs doesn't seem that threatening, there aren't a ton of abortions after 20 weeks and nearly all of those would qualify under common exceptions, so nobody rational should be imagining "oh gee what if I need an abortion and can't get one", bc those circumstances are very unlikely. I don't think people are actually afraid they won't be able to have 3rd trimester abortions, I think instead that the Roe revivalists have managed to convince people that the prohibitions are broader than they actually are, and as a result gotten them to vote for maximalist initiatives that guarantee later access to abortion than anybody ever said they wanted.
Even though the "this is a private health decision between a woman and her doctor" framing has been around quite awhile, I'm noticing a lot of mostly non-political women picking it up and echoing it who wouldn't have said that 10 years ago.
Appealing to single women voters will be an interesting test of the median voter theorem, as the GOP obviously can't meet single women entirely on the social safety net, other safetyism like gun control, or abortion policy (not without just becoming the Tories and losing any reason to exist) -- but the GOP can sure ramp up "strangers are killing your children and dogs" and whatever anti-crime backlash is coming in the nest few years, if that's all that's left that's what we should see all the focus being.
>I think instead that the Roe revivalists have managed to convince people that the prohibitions are broader than they actually are
I think the crux of it is that, while there's only a minority of the electorate that thinks an abortion of a healthy baby at the moment of birth should be permissible, if people think there's any real chance that an abortion restriction could apply to themselves or someone they care about -- including in situations that aren't actually an abortion, like causing doctors to make bad decisions in cases of a miscarriage -- they'd rather just have no restrictions on abortion at all.
Those cases don't even have to be true, because who has time to investigate? The people in the fuzzy middle just have to think there's a real chance they could be personally affected to decide whatever vague pro-life convictions they have aren't worth the risk. Which is a pretty easy sell for the pro-choice side, low burden of proof. In this way I think the trend towards continued atomization and cultural disintegration will favor that side politically.
ETA: my prior is that this is almost certain to be true - in the US specifically - because if you actually let doctors carry out abortions whenever they might be necessary to save the mother's life, pro-choice doctors will abuse that to carry out unnecessary abortions, and the pro-life factions writing the laws really really don't want that to happen. (As the saying goes, "the ability to make a decision is inseparable from the ability to make the wrong decision".)
Yeah, like the ones who empower the worst of men among us by pushing, in the name of "autonomy," for policies that make it easy for bad men to indulge their worst impulses and pressure women into making the consequences disappear.
Yes, some women, like some men, live and vote by moral principles, and some live and vote by pure selfishness. In a remotely humane and decent society we'd all agree to only give respect to the first group. What truly terrifies me is that people like you seem to actually want more respect for the second!
I really hope you're a troll. If not, how does it feel to be a literal sociopath?
The kind of "person" who could be confronted with an act of violence against a child and not only make no effort to justify it but actually respond with a cheerful "have a great day" is a person for whom I'm not going to begin to list what they dessrve to have happen to them. Just use your imagination.
One day, in a more humane time, people will look at monsters like you the way they look at the ones who cheered on slavery, and shake their heads in inexpressable disgust.
Hopefully, this will happen in your lifetime and you'll face some real justice.
People who are killed in the womb don't care about having been killed in the womb, due to being dead. And their next of kin seem to care less than in other cases of manslaughter. As do their non-related acquaintances.
At the same time we don't treat the killing of other creatures of equivalent or greater intelligence as crimes. Instead we treat it as bacon. This makes it hard for me to take secular moralist arguments against abortion seriously from anyone other than vegetarians. Are you a pro-life vegetarian?
I don't think opinions have changed, just the status quo has changed.
A few years ago, the median voter's preferences on abortion were somewhat to the right of the overall national status quo. Now the median voter's preferences on abortion are somewhat to the left of the overall (measured somehow) national status quo.
But it's one of those rare cases where the voters didn't change their minds, the status quo changed, and without any direct intervention from politicians.
I think the issue is that the median person has inconsistent opinions about Abortion. They think having an abortion is bad but are very opposed to actual stories of people being denied care.
Nobody's actually proposing a moderation of state-level abortion bans except Trump, who won't say specifically what he actually wants. (He only said Florida's ban was too harsh.) And now the GOP has decided to do this, being incapable of waiting three weeks until after the election:
You make an interesting point. Yes, some years ago, no mainstream politician dared campaign with a pro-LGBT platform. However, a lot of society has shifted leftwards, and now LGBT issues are fine. You say that tides are shifting towards the right - but this is just a rather slight correction of the leftward shift over the past years. Yet people on the left somehow think it is some sort of earthquake.
I'll point to trans issues as an example. Totally off the table 10 or 20 years ago. Became widely accepted. Now the right wants some limits, such as prohibiting trans (MTF) participation in women's sports. Small correction to a large leftward shift.
...It's not going to last. Once the right is in power, they will absolutely crack down on the presence of trans people. There's no reason for them to tolerate such mockeries of the sexes.
One reason would be that the median voter wants them to tolerate the presence of trans people (if that's indeed the case). "Tolerating the presence" is a pretty low bar.
A majority of Americans still think changing gender is morally wrong. https://news.gallup.com/poll/1651/gay-lesbian-rights.aspx Even the readership of left-leaning papers like the NYT barely try to hide their disdain. Sure, they claim the problem is "social contagion" and "compromising women's rights," but it's pretty clear that it goes deeper than that.
The only force that is forcing everyone to pretend to accept the presence of transgenderism is the left. The allure of Trump is that he has shown people that you don't need to pretend to be tolerant of loathsome things just for the sake of "civility". The left will be dealt with, and with it will go transgenderism.
> Last week, the Republican governor of Tennessee signed into law a bill prohibiting gender-affirming care for minors as well as one imposing new limits on drag performances, which have become a target for Republicans. Mississippi also enacted a ban on treatment for transgender youth while Republican state lawmakers in Kentucky advanced a similar measure, following a charged debate over a separate proposal allowing teachers to refuse to use students’ preferred pronouns.
Until recently, most legislation banning transgender healthcare was aimed at minors, but Republicans are increasingly pushing proposals that would limit treatment for adults...
...But the speech that LGBTQ advocates found the most chilling came from Michael Knowles, a rightwing political commentator for the Daily Wire, who declared that “for the good of society … transgenderism must be eradicated from public life entirely”. A range of voices, including public officials, experts and observers of rightwing rhetoric, condemned the remarks as inflammatory and dangerous, with some calling them “genocidal”. (Knowles insisted on Twitter that he was not referring to trans people, but “transgenderism” which he has described as a “false” ideology.)
Yet the intense focus on transgender rights at CPAC this year – nearly every speaker raised it – suggests it is likely to be an animating issue in the coming presidential election.
I will point out that during a part of the Trump presidency, the right was in power in presidency and both chambers of Congress. If they rounded up all the trans people then, I must have missed that.
I would argue what really broke the (almost) 1:1 mapping between gender roles and biological sexes was effective contraceptives. Before they were a thing, it was the fate of perhaps 45% of the post-pubescent population to get pregnant and then (because of various reasons) end up raising their children. Gendered behavioral archetypes (the pushy suitor, the chaste maiden, the tastefully-hot woman, the caring mother, the providing father) come from this time.
Thankfully, with contraceptives, the mapping between gender roles and sexes broke down. Humanity hacked its reward function. Today, most sex in first world societies is non-reproductive, an evolutionary imperative we follow even though its original purpose has been lost. What remains of genders is likewise an vestigial echo of our evolutionary past. In accordance with that past, most will prefer to have sex with people presenting as one gender (perhaps in a specific life stage), or having one set of secondary or primary sex organs, or perhaps even having the corresponding genotype.
In 1700, being a woman (e.g. presenting as a chaste, yet hot maiden so that you would attract a high-quality husband) was a very different role from being a man (e.g. learning a trade or going to war to earn enough of a fortune to be an acceptable husband). Today, little of that difference remains. Our sex chromosomes no longer determine our lives to a comparable degree.
The way I see it, gendered titles and pronouns are almost as vestigial as noble name prefixes. Where in 1500, the difference between Mr von Grauen and Mr Grauen might be the difference between dying in a duel or getting beaten up when giving offense, and the difference between Miss Grauen and Mr Grauen was the difference between wooing (or raping) someone and slaughtering them on some battlefield, the days of nobility are just as gone as the days of strong gender roles, and good riddance to either.
I am a straight, cis-by-default guy. I don't really care what sex bits most of the people around me have in their pants, be it dicks, pussies, nothing or tentacles, pubes or shaved. If I foresee the possibility that I will be interacting with said sex bits of a person, I will make tasteful inquiries, but I actually tend to fuck just a small fraction of the people I encounter. If a few people feel really strong about belonging to some medieval class or gender role, let them acquire their von or Mr or Ms or whatever. Like gay marriage, it does not hurt me in the least and seems to make them happy, so I am a-ok with it.
While I think a long-term right win would bring issues, I'm skeptical that they'll crack down in a truly significant manner without some major disruption (Republicans win the next several elections, Trump somehow manages to perform a coup). I don't expect a major disruption like that.
I expect they'll most likely crack down on even more controversial topics, like puberty blockers, which earns them applause from their base which has far less resistance from our existing laws about people doing what they want with their body.
I heard much of the same during 2016 about a variety of topics and it simply didn't happen.
There does seem to be a shifting tide on trans issues where the NYT is now publishing things that horrify activists. But gay marriage seems cemented in place.
The median voter theorem is a statement about how rational political parties would act based on an assumed one dimensional political spectrum.
First of all, it's a model, and the onus is on a model to describe real life, not the other way around.
Second of all, "evil" (as well as "good") is a mechanism for explaining why some political parties did not play rationally - for example, why the house stayed democratic for many decades.
I believe many politicians were not able to move along the ideological spectrum to optimally play the median voter theorem, because their individual values prevented them from taking optimal positions.
However, it is worth noting that while the house did stay democratic for several decades, the Republicans eventually won in the long term, and have achieved many of their goals since taking the house under Gingrich. Barry Goldwater may have lost his election decisively in '64, but his ideas won in the long term. So perhaps we shouldn't be too critical of Republican strategy during that period - they won eventually.
If we say for the sake of argument that the axis of politics goes from good to evil, this does not contradict the median voter theorem. It would just result in both parties being equidistant from the median moral voter, such that half the country was better than him and half worse.
Also, remember the orthogonality thesis, evil people need not (in the game theoretic sense) be irrational
Then you need to also bring in "good", and might as well call it "values" as you do in the comment above: that certain politicians or whole parties will not shift as much as they "should" from the point of view of optimal electability because they're *constrained* by those values, beyond the obvious "we can't give up on position X even tho it's a minority position, because it will lose is the most faithful and generous supporters".
Elections aren't simple games, they're iterated games. One reason for not playing an optimal median voter election, is that many individuals do not simply have inherent preferences somewhere in the left right spectrum, but are influenced by what positions political parties have taken in past elections.
Many voters are "leftists", "rightists", or "centrists", *relative to the positions of the two political parties*. This is "Overton window shifting". So, it may be worth not playing to the median voter in today's election, because that same voter may change his politics to be "more centrist" in future elections in a way which pushes policy in a particular direction.
>I believe many politicians were not able to move along the ideological spectrum to optimally play the median voter theorem, because their individual values prevented them from taking optimal positions.
This can explain why an individual politician didn't move, but wouldn't that lead to them eventually losing to a candidate whose values/positions put them in a different place along the ideological spectrum?
The candidate generation process is not necessarily efficient - people who run for office aren't uniformly sampled from the electorate. Most people aren't able to campaign for public office - those who do are therefore different from the norm, and these differences are not uncorrelated to political opinions
Even evil people want to achieve their goals; losing elections doesn't help with that. Even if you're right that some politicians are too principled to moderate their positions for tactical reasons, I don't see why you'd describe that as evil.
Shouldn't "evilness" increase the sway this model holds? And wouldn't "evil" people be more likely to outcompete "good" people on all levels of politics, after controlling for intelligence?
It just means you have to apply Conflict Theory more than you are currently doing. For example, you seemed surprised when Nancy Pelosi intervened to get Gavin Newsom to veto the AI bill. You shouldn’t have been.
Please. One of the biggest problems facing the West today is the inability to discuss and understand each other's positions. Your opponents are not wrong, they are *evil*. This is not a productive way to conduct politics, and is something we need to work against.
This sounds like something you would need to explore if you wanted to know whether it was true, no? And so I don't think that works as a rebuttal of what MM said.
To what ends? Sadism? You think there is a significant political faction of abstract sadists who get off on looking at a spreadsheet where they can see how many lives they have made worse?
Part of what's relevant to designating something as "evil" is whether it's beyond the range of potentially convinceable by argument.
If Hitler is going to do the Holocaust, I don't think I can have a rational debate with him and convince him not to. Hitler's just going to do the Holocaust because he's Hitler, i.e., evil
But that would be true of any strongly held conviction whether good or evil. I don't know whether you're pro or anti abortion, but what are the odds of someone changing your mind about it with debate? And we don't know how hard it would be to persuade Hitler with debate either; I don't think he exactly had a lot of friends pitching, "hey Adolf, maybe Jews are actually great". You may be assuming that bad people and your ideological enemies are more rigid in their beliefs than your allies. People always think their own side is rational and the other side won't listen to reason.
People can be rendered unpersuadable by delusion, mental illness, or manipulation by another agent. At the same time, a genuinely evil person often will be persuadable by appealing to self interest (e.g. a sadistic sociopath who gets finite utils from causing suffering can still be bargained with as long as they also like money). Unpersuadability is neither necessary nor sufficient for designation as evil of someone who does bad things.
I don’t think people who disagree with me are evil, but I think the politicians they vote for are. I’m just here in the comments section trying to get them to realise it
Are you saying that members of one or both parties wake up in the morning thinking 'how can we decrease the global human utility sum and turn our world into the torment nexus?'
If you think that, then you have a very shallow model of your political opponents. Sure, there are sadistic psychopaths who enjoy spreading suffering where they go. But these are relatively rare, and they do not cooperate well enough to dominate. (No, not even the NSDAP was majority psychopaths.)
Take abortion, a typical hot topic where people on either side portrait the other side as evil.
If a fetus has the same moral value as a teenage human, then forcing women to carry their fetus to term (until we get artificial wombs, at least) seems perfectly acceptable. We don't allow parents to murder their teenagers either, even if murdering them was really convenient because they make a great an organ donor for you.
If a fetus is just some tissue without intrinsic moral value, then forcing a woman to carry it to term and thereby creating a kid which has a worse starting hand genetically and socially than average is monstrous towards both the woman and the kid.
So the other side is evil only if your own assumptions are true. Granted, you will likely find the odd psychopath on either side, like a pro-lifer who does not care about unborn children at all, but really wants to punish women who had sex, or a pro-choicer who does not care about women, but just rejoices at any life ending, but these are minorities.
>Donald Trump specifically is pretty clearly a sadistic psychopath
It's not really clear at all. Boorish, boastful, philandering, greedy, lying, even foolish I could see arguments for. But what cause do we have to believe he is a psychopath, or a sadist? He certainly seems to experience the normal range of human emotion, and what I've heard of his sexual exploits did not seem to prominently feature infliction of pain.
There are forms of sadism other than "inflicting physical pain in an obviously sexual context." The man's catchphrase is "you're fired," contextually meaning "with a momentary whim, in two syllables, I have destroyed what took you years to build, and most of your hope for prosperity in years yet to come."
Words have meaning, and if firing people on a game show makes you a sadist then the word has little meaning left. Trump is certainly no saint who humbly loves his enemies, but I don't see him as particularly taking pleasure in the pain and suffering of others.
> 'how can we decrease the global human utility sum'
Isn't classical evil more like 'how can I increase my own utility & happiness by any means including decreasing that of everyone else?" Whether it's loud or quiet, hidden as a usually unconscious motive or 100% embraced, I do think there's plenty of *that* out there. (And not just in politicians.)
I fully agree with the rest of your comment though.
I think that most people put a higher weight on their own utility than on the utility of a random person. In a way, this is okay -- a person who is just likely to buy any stranger some food as he is to buy his own dinner will starve in short order, which will seriously hamper his efforts to increase the utility of mankind.
I would define evilness as the quotient of utility weights between oneself and a stranger being above a certain threshold. For example, the evilness of someone who gets paid 10M$ to kill an innocent is much lower than the evilness of someone who kills innocents just to pass the time, like that gunner in Apocalypse Now ("How can you kill women and children" - "It's easy, you just don't lead them so much").
--
Unrelated, I think that 'my opponent is evil' acts as a curiosity stopper, just like murderism [1]. Sometimes this is appropriate: there are likely few insights to be gained from digging deeply into the prioritization the Nazis had for their genocides (unless they are still at large and you are wondering when your turn will be). On the other hand, while Jan 6 was imho clearly an evil act of Trump, there are myriad other ways in which he could have acted. "Evildoers do evil, and how they pick their acts is forever beyond the understanding of decent folks" seems to be a bad frame to understand politics.
That type of evil wouldn't stand in the way of the theorem presented in the article though. Maximizing electability as an avenue towards personal power and safety still makes sense.
Obviously, "Many voters and politicians make their decisions simply because they are evil" is not self-explanatory, so you'll hsve to expand on that concept.
Personally, I would argue diametrically that noone thinks that they are evil, although from an outside perspective, their argument may look very thin. So I have problems to understand your position.
But if you want to convince others that politicians really do act out of a pure will to just hurt people, I'd have to see some examples. Please provide cases where you think this is the underlying reason.
If primary voters myopically pick their favorite primary candidate without considering general electability, then you'll always get a median-democrat and median-republican nomination regardless of what positions individual candidates adopt.
Ok, but those two would be equidistant from the actual median only when the electorate is very close to 50/50. That seems to have been true for a while, but I’m not sure I can explain why.
If the two parties' candidates are typically equidistant from the median voter, that also explains why a similar number of voters belong to each party and vote in its primaries, which in turn explains why their candidates are typically equidistant from the median voter.
In theory this is an unstable equilibrium though (under our assumptions that primary voters mostly don't vote tactically, and people vote in the primaries of the party they prefer). If a bigger fraction of the electorate belongs to one of parties, that should make its primary voters more moderate on average, which should result in its candidates being closer to the median voter, which should make even more people switching to that party. Though this is mitigated if, say, center-right voters keep voting in the Republican primaries even if they vote Democratic in the general election because they perceive Democrats as more moderate, or vice versa.
Yeah, the problem is the apparent stability of the 50/50 split. I always fear that the reason for it is that there *are* no median voters -- it's a bimodal distribution of groups that see no validity in the other group's positions, and never the twain shall meet.
I'd like to blame the primary system for this. Smoke-filled rooms have a bad rep, but at least the party leaders have a terminal goal of winning the election, which we gather primary voters often do not. But we have not had the knife-edge 50/50 split for as long as we've had the primary system, so if primaries are the problem the effect has been gradual and cumulative.
This assumes that the primary voters represent the general election voters more or less 1:1, which is generally not the case. Primaries, and especially caucuses, attract people who are politically engaged, and they tend to activist-y and have stronger, more choate opinions on politics.
We can look at 2024 for some evidence against that idea. Trump currently leads in AZ polls by nearly 2 points. AZ Republican primary voters selected Kari Lake (~409k votes) for the 2024 Senate seat over Mark Lamb (~293k) and Elizabeth Reye (~38k). That works out to Lake winning 55.3%, all numbers per NBC News.
Per 538, Gallegos, the Democratic Senate nominee, is leading Lake by nearly 7 points. With races having tightened across the country, Trump has nearly 10 full points on Lake even after she won her primary comfortably.
The AZ primary is an excellent example of primary voters not considering general electability. Lake's weakness in the general was very well known in the few weeks leading up to the primary. GOP strategists trying to win the Senate were screaming about it.
The process for parties staying equidistant from the median voter doesn't have to be fully organized top-down. It can be partially an emergent phenomenon, in which candidates who do so the best are successfully elected, and candidates who don't, aren't. This applies to candidates for all positions - state, local, and federal. Some of these positions aren't term limited, so successful politicians can stick around (even when they can't stick around in a given position, they can leverage their success to move to different political offices). The collective of elected candidates from a given political party will therefore approximate median voter preferences. This collective then creates norms that other members of the party can default to. These norms will shift over time as the collective of elected officials shifts per shifting voter preferences.
This model could create an inertia that keeps electoral results relatively close, even in instances when a given candidate doesn't do as good a job of matching the collective party line. Since per this model, party norms develop based on the collective tendencies of successfully elected members of that party, the public can develop party loyalty and vote for the party candidate even if they veer somewhat from the norm. Of course, party loyalty isn't absolute, so we'd expect such candidates to lose some support. They'd then be less likely to get elected and join the collective of elected officials from their party, which then shifts the makeup of the collective slightly, to maintain equilibrium.
Party loyalty in the partisan era could probably explain the relatively even electoral results even if one doesn't adopt the aforementioned model.
1. If three reliably blue states are admitted to the union you still only need 54 senators (or 53+VP) to control the Senate.
2. Tony Blair became the person that said "We Labour politicians whould veer right" after 17 years of conservative dominance and it worked. He even successfully removed achieving socialism from the party goals.
Bill Clinton did the same in 1992. 90s Democrats were extremely conservative by most metrics we would use to measure that before or after his presidency.
He also veered left on a bunch of social issues, like the whole Christian agenda.
The only thing he's veered right on is immigration, and even that is mostly just a return to the classical consensus of "hey why don't we actually enforce the law?"
He’s used a lot of rhetoric that scans as right-wing, particularly about immigration - but some of his related rhetoric on international trade would have scanned as left wing a few decades ago. He’s just systematically anti-foreigner in the way he talks, which isn’t really right or left.
I guess that depends where you put nationalism on the spectrum. The far-left often present themselves as anti-nationalist, yet in historic examples they were willing to drop that stance and opportunistically veer nationalist. And nationalism usually becomes more and more pronounced the further right you go in modern terms, yet the original right, i.e. royalty, nobility and clergy with temporal power, were very much not nationalist and only occasionally played towards nationalist sentiments for gain or to boost general morale.
Republicans prior to Trump didn’t say, “I’m running exclusively to represent the interests of the rich.” It’s not a winning electoral tactic because even if you get the vote of every single voter in the top 1%, that won’t get you to 50%, which is what you need to win.
Once in office, however, Republicans tend to favor the rich. Trump’s major legislative achievement was a tax cut which primarily benefits the wealthy. During the campaign, promised to eliminate the carried interest loophole (which lets hedge fund managers play lower taxes), but he didn’t try to add a provision to the tax bill to do that.
Trump was more willing to impose tariffs than previous Republican presidents, but because of the marginal utility of money decreases with wealth, higher prices due to tariffs affect the wealthy less than they do people lower down on the income scale. So I don’t think Trump’s tariffs mean that the Republican Party is no longer the party of the rich.
RE tariffs - typically, at least part of the rationale behind them is to protect domestic manufacturing (existing or emerging), and more manufacturing jobs should be beneficial to the working class.
So won't count it as policy benefiting the rich...
A minor quibble; electoral college advantage isn't static, but relates to the composition of the two parties' electoral coalitions (it tilted towards the dems as recently as 2012), so the proper electoral-advantage-adjusted measure isn't national popular vote minus a static adjustment, it's margin of victory in the marginal state.
Yes this is correct. And it will show an even closer margin in the last 20 years than 2% I believe. At least 2016 and 2020 tipping point state differences are less than 1%.
If I calculated correctly the average in the years 2000 to 2020 was 3%, mostly because of Obama’s decisive victories with tipping state margins of 9.0% and 5.4%.
Yeah I was kinda surprised Scott applied that lazy 2% hack rather than just showing the margin in the tipping point state, especially since the table already includes which state that is.
Yup, the electoral college favoring Republicans has much more to do with 'Winner Takes All' and much less that small states get extra electoral votes that is usually the go to explanation. The effect of that changes as the parties coalitions change (big driver of it getting more favorable for Republicans lately is the increasing education polarization). Small states are surprisingly relatively evenly split (RI,VT,HI, DE etc. are all tiny and D dominated for example). Nate Silver has talked about this a couple of times.
Right, the big problem with the electoral college is the winner takes all nature of it. If states allocated electoral votes roughly proportional to the vote shares, the electoral college would be more or less fine.
Do you by chance have a link to one or more of Silver's discussions of this? It's not an idea I've run across before, and I'm interested to learn more.
Another interesting social choice result (though less applicable to the real world) is the "McKelvey–Schofield chaos theorem". It says that if one uses a sequence of pairwise election to choose a winner, then someone can arbitrarily control who wins just by adding more choices and changing the order that the pairwise elections are held in: https://en.wikipedia.org/wiki/McKelvey%E2%80%93Schofield_chaos_theorem
"Maybe after decades of humiliation and failure you could convince someone, but I don’t see the history of parties suffering decades of humiliation and failure before finally agreeing to turn around."
You could look at the UK if you want to see that: Labour lost the 1979, 1983, 1987 and 1992 elections before turning around, choosing Tony Blair as leader and winning 1997 by a landslide; similarly, the Conservatives lost that 1997 election, then 2001 and 2005 before choosing moderate David Cameron and winning in 2010. The process since then has been complicated by Brexit, but Labour were out of power 2010-2024 and chose moderate Keir Starmer and won.
Brexit works in that it traded how the Tories picked up the parts that constitute a median voter. Johnson was also more 'left' on some issues (legal immigration, regional development, green investment).
Indeed: UK elections that have been close (since the consolidation of the Labour/Conservative two-party system in 1945): 1950, 1951, 1964, 1974A, 1974B, 1992, 2010, 2015, 2017. Not close: 1945, 1955, 1959, 1966, 1970, 1979, 1983, 1987, 1997, 2001, 2005, 2019, 2024. I don't see a pattern in them becoming more or less close in there.
If you assess this based on just the two party system it will make it very difficult to analyse- combined two party vote share varies massively. The recent election wasn't in any meaningful way similar to 1997. Labour's vote share barely budged compared to 2019 (and was down in absolute terms), and was massively down vs 2017. But they won a landslide mostly due to the exodus of RW/working class voters to Reform (who had nearly entirely voted Tory in 2019).
The high point of the two party system (1945-1970) saw mostly closer elections than the period after, with 1974 as a transition point. The Lib Dems acted as an amorphous centrist party that could peal votes off either group depending on the issue of the day. With the 2024 election seeing 30% of the vote going on new-ish left and right wing challengers on either side of the three older parties, suddenly we're in a slightly different world. Last time this sort of happened (2015) the Brexit referendum acted as a new position to triangulate around, and the old system re-emerged. This doesn't seem to be an option now.
Because you have this pressure valve of third/fourth/fifth parties it makes it hard to assess the relative merit of centrism of candidates. The Conservatives shifted back and forth on social and economic centrism (moving toward the median voter position in 2019 of centrist economics and right wing social values) but saw continuous vote share increases until 2024.
In the UK, whenever the Conservative Party sees itself losing votes to Labour or LibDems they always talk about "creating clear blue water" - ie going right to appease their core voters (who would never vote for anybody else anyway (I'm talking pre-Farage here)). Dunno the logic, if there is any: maybe they get more volunteer canvassers that way. Maybe it's just a stress response.
That looks like a standard reaction of people with very strong positions: If they do something extreme and it does not work, clearly it does not mean they were wrong, it means they weren't extreme enough.
I see it all the time, especially in politics. Noone is ever wrong, instead they either didn't explain their correct position to the voter enough, or they foolishly let moderates dilute their pure position, thus sabotaging it.
This part is clearly not true. A lot of the smaller states are in the northeast, which is traditionally much more progressive. The large states out west tend to be much more rural and conservative, with the obvious exception of California, and I guess the Seattle/Portland area. By the list of smallest states, we have:
Rhode Island, Delaware, Connecticut, New Jersey, New Hampshire, Vermont, Massachusetts, Hawaii, Maryland, West Virginia... ok, by state 10 we get conservatives. But there isn't some grand strategy where small states vote Republican because the electoral college favors smaller states. The popular vote gap is an artifact of the hordes of progressives in California, if anything.
Pretty sure that "small states" here is meant to refer to small population, not land area. That said, there's still a mix of red and blue in the least populous States:
Wyoming (red), Vermont (blue), Alaska (red-ish?), North Dakota (red), South Dakota (red), Delaware (blue), Rhode Island (blue).
If you're smart enough to look at the ordering of states by area, you're smart enough to understand that, especially with talk of elections, people use "small states" to mean "lesser populated states."
You're measuring size by area; you should be measuring by population if you want to understand the electoral college influence. But it's certainly true that there are loads of low-population ("small") states that lean left: Vermont, Delaware, Rhode Island, Maine, New Hampshire, Hawaii, for instance. Though there are similar numbers of small states that lean right: Wyoming, Alaska, North Dakota, South Dakota, Montana, West Virginia and Idaho, for instance (that's with a cut off of states with just one or two house seats). Equally, if you look at really big states, while California is blue, the next two (Texas and Florida) are red.
Your last sentence is right, though: the real reason for the electoral college advantage is that the Democrats win big states by big margins - California, New York and Illinois above all, while the biggest state that the Republicans win by more than 10% is Tennessee. That means that a significantly larger fraction of the Democrats' vote is "wasted" by piling up a big margin.
The obsession with small states also results in people misdiagnosing the Democrats' problem in the Senate, where the real advantage is in the middle: if you take the eighteen states from 5th largest to 22nd largest (Pennsylvania to Minnesota), they split 13-5 to Biden in 2020. Take the next eighteen (23rd to 40th, South Carolina to Hawaii) and they split 13-5 to Trump in 2020. That excludes the four large states (California, Texas, Florida and New York) which split 2-2, and also the ten smallest states, which split 6-4 to the blue side. (Note: in an election that Biden won by a 4.5% margin, he only won the states 26-24; if the presidency were determined by the same rules as the Senate, he would have won by the narrowest possible margin, rather than by a relatively comfortable EV margin)
The reality is less about size and more about population density. In the US, the less densely an area is populated, the more Republican it tends to be. This is almost fractally true. Rural states are more republican than more urbanized states, but their cities will be more democratic than the states overall, and the denser the neighborhood the more democratic it votes.
1. This appears to assume that voters are rational and rationally vote for candidates based on their stated positions. I would suggest that voting is largely a cult of personality and thought-terminating cliches.
2. There is a theory in poltiical science that any first past the post winner takes all political system will eventually evolve into a two-party system.
For my part, if I wanted to implode either major US political party, the surest way to do so is to give them unfettered power. Team R and Team D need one another, otherwise there will be no excuses as to why we haven't reached Nirvana yet, and there won't be enough spoils to keep all the interest groups making up Team D or Team R happy, not to mention, some of those demands may be incompossible. For instance certain libertarians and religious conservatives may vote Team R, but their preferred policies are very different and there is no easy way to keep both happy, especially as there is no long a Team D to enforce unity.
> I would suggest that voting is largely a cult of personality and thought-terminating cliches.
I think that it's "largely" this, in that at least 50% of voters run this way.
But you just need a big enough minority of people in the middle who can say "y'know, that shit is just too crazy for me, I'm switching my vote" for the MVT to work.
(I love the word incompossible now. Thanks for teaching it to me.)
Britain has been functionally a two *dominant* party system for the majority of it's modern history, with the exception of the interwar period, and the 1974-2015 period. It is true that any third party has been generally punished by the electoral system. But similarly to India, what sustained the Lib Dems was geographic differences where in certain areas of the country the party could sustain itself as either the number 2 or number 1 party (Wales, the South West).
What is happening currently in the UK is a breakdown along the lines of Scott's example where the two main parties become too centrist and people stay home (which he discounts as possible). But this is mostly what happened, except they didn't stay home, they voted to split the vote by going for third or fourth parties who *did* meet them on issues. The last time this happened in any large number (2015) it caused a major political realignment after the Brexit referendum, with voting coalitions shifting closer to a US style Blue tribe-Red tribe split. This collapsed in 2024 as White working class voters abandoned the Tories that (in their view) betrayed them, and young, often Muslim, voters didn't turn out for Labour in their 2017 numbers due to Israel/Palestine.
I haven't looked in a while, but I thought Britain used two systems, one of which was national votes and one of which was local votes? The local votes seem to be very two party, while the national votes allowed smaller parties to get some representation.
Also, a few extremely geographically located parties that could win local elections on a very specific platform.
I don’t know what you are talking about. I’m talking about elections to the Westminster parliament which are via first past the post voting in single representative constituencies. There is no proportional representation. Third (and fourth) parties which win nontrivial numbers of parliamentary seats exist, and sometimes end up kingmakers in a hung parliament.
Yes there are two dominant parties, but if lesser parties get nontrivial parliamentary representation and sometimes end up kingmakers then I think this is meaningfully a multiparty system.
The analogy would be a US system where the libertarians and greens each had a dozen Congress critters and a couple of senators, and occasionally won a couple of states in the electoral college. This I would consider a multiparty system albeit with two dominant parties.
Also geography is a lousy excuse since US geography is more varied than Britains. If lib Dems can carve out a niche in the Southwest and SNP in Scotland, then why can’t libertarians carve out a niche in the mountain west and greens on the pacific coast?
For the "why": isn't the primary election system a major reason? Thanks to it, every somewhat mainstream ideology can hope to have some voice in at least one of the two major parties, so there's less motivation to form and vote for minor parties. If libertarianism or green politics are popular in some regions, they can just elect libertarian Republicans or green Democrats.
Perhaps the FPTP presidential elections also make a difference? They produce a national two-party system for the presidential election, and once voters align themselves with one of the main two parties for president, they mostly end up identifying with that party and don't bother to vote for a minor party in the other races either?
Yes, ‘FPTP plus presidential system plus nationalized politics produces a two party system’ is consistent with the data (or at least I don’t know of any counter examples). My point was just that FPTP alone is empirically insufficient, since there are lots of counter examples of countries that use FPTP but have multiparty systems.
This at a minimum is a necessary rider. Whether FPTP voting and a presidential system is sufficient, idk. As best as I understand, Brazil and Indonesia both have presidential systems and multiparty democracies, but neither is strictly FPTP, and arguably neither is mature. But then if we demand FPTP and a presidential system and a mature democracy, then maybe the US is the only example, so we are reasoning from n=1.
I can't speak to Britain necessarily, but the issue with India is that *at the state level* it's largely (not entirely, but largely) a two party system. The reason it's a multiparty system at the federal level is because the two-party contests are different in each state, and more generally because Indian states are really *extremely different from each other*, to a degree that dwarfs anything in the United States.
Indian states, for the most part, don't share a common language, in many cases they differ either ethnoracially or religiously, they don't share much of a common history until the British forcibly conglomerated them, and they are as far appart in terms of economic development as Russia and Burundi.
If you consider, say, the social democrats / liberal democrats in the Congress Party, the Tamil/Dravidian nationalists, the Communists, and the caste-based parties, all of these groups are fundamentally opposed to the Hindu nationalist narrative of history, and the latter three of them at least could broadly be considered on the "Left" (Congress was certainly a center-left party from the 1950s through the 1980s, but it's hard to consider them anything but centrist today). But, these four groups are never going to consolidate into a single party, because not only does each of them have a sharply different ideological basis and narrative of history, but each of them also has a different geographic stronghold. It would be totally counterproductive for each of them to give up their distinct existence at a party, when each of them has a reasonable shot of being dominant *at the state level*.
I don't think this is true? e.g. Uttar Pradesh (the largest state by population) appears to have three major parties (as measured by `have held the office of chief minister in the 21st century'), plus the Congress as a significant fourth player. Maharashtra (the second most populous state) also appears to have three major parties (by the same metric). The first two entries on the list both yielded multiparty systems so I suspect the thesis that Indian elections are two party affairs on the state level is false.
Maybe just for due diligence I will check state #10, seeing as how I have ten fingers. According to Wikipedia that would be Andhra Pradesh. Again, three different parties have held the chief ministership of Andhra Pradesh in the 21st century. OK, I declare the thesis that `Indian elections are two party affairs on the state level' to be false.
More likely, the basic thesis of `FPTP produces two party systems' is just wrong for parliamentary systems. Maybe it's right for presidential systems, but it's hard to say, because I think the set of large mature presidential democracies with FPTP voting has n=1.
Why is voting for personality not a valid preference as a rational voter, i.e. a voter that votes for what they want.
Even if some voters voted by flipping a coin that's still a valid preference - and rational. An irrational voter would be one that didn't have transitive preferences. I think the onus is on you to prove preferences are not transitive across the left-right spectrum. A voter who wants a far left candidate, will usually settle for a less left candidate given the choice. We see this in my own electoral system (Australia) with instant runoff and single transferrable vote.
Edit: To be clear personality is also transitive. If you're voting based on who you like the most, it is rational to pick the candidate you like the most, or if they aren't available, the next most likable.
You have a different defintion of rational that what most people in here do. A rational agent has a meaning and if you make a top level comment using another definition I think that's unhelpful for discussion.
I agree you can't create a mathematical argument about a fair coin being transitive but the point is about rational choice and preferences.
If you vote in *this upcoming election* by flipping a coin it means you have an identical preference for each candidate. It's just an extra thing I added to what I was saying. It's not the core of my argument. Yes, if someone lived their entire life as a random number generator they wouldn't satisfy rational choice but that's not very interesting in the context of this discussion.
The point is against this:
> I would suggest that voting is largely a cult of personality and thought-terminating cliches.
As evidence that voters are not rational. Which I think is wrong. Expressing a preference for personality is a rational preference by the definition.
Most notable FPTP lack the extreme bipartisan division in the US. You've got Libdems in UK, NDP in Canada, a host of ethnic/regional parties in various Third World countries (and, of course, BQ and SNP in Canada/UK).
What really causes America's very strident two-party system is probably a combo of FPTP and open primaries, which basically makes the two parties into placeholders for competition of group and coalitions that would form different parties in other systems.
I think it's also geography. Namely, the US has a lot of it - so much that there's room for multiple very disparate subcultures within, compounded with having had open immigration for so much of its history. (Immigration restriction is, I predict, bringing the US closer together, though there are also forces pushing society in the other direction.)
The UK mostly just has an island roughly the size of California. Canada has plenty of land, but most of it isn't densely populated (90% of Canadians reside within 100 miles of the US border). But even so, there's a lot of difference between British Columbia and Quebec, and sure enough, that shows up in their politics.
In the UK, Keir Starmer pretty much attempted the strategy of saying one thing to win the party leadership, and then switching. Yes, the party membership noticed and we’re Not Happy. Well, approx 50% of them were Not Happy, as you’ll expect from the median voter theorem.
Missing from Scott’s analysis of the effect of primaries… if the switch after the primary is big enough that more than 50% of your party’s supporters switch to a new third party, you lose. Rarely happens, possibly because they’d lose if they tried it.
That's basically the point covered by the Second... paragraph. Explicit party defection or not, alienating potential voters who are closer to you than your opponent works out as a turnout (dis)advantage.
I think it's implicit that this analysis applies pretty much only to countries that fall into Duverger's law and have a bipartisan system. Once you have multiple viable parties you're probably talking about multiple median voters and many different axis rather than just left-right.
Last UK election, the Conservative Party didn’t just have to worry about losing votes to Labour from their left, but also Reform from their right. (And Labour was against Jeremy Corbyn etc. to their left, and the Conservatives to their right).
Primaries are mostly American and mostly in the past 50 years. So there's even less reason for the Republicans to have stayed minority party -- they could have re-aligned themselves in a smoke-filled room without worrying about red-meat voters.
I don't see much credence in claims that the Median Voter Theorem isn't valid. What we in the U.S. see as extreme is only extreme as measured against our fairly narrow Overton Window compared to other countries. Extreme candidates pretty much never make it through the primary process. Let's look at the history: who was the most extreme major party candidate in recent (say last 50 years or so) history? And I'm talking about policy positions. I can't stand Trump personally, but none of his policy proposals are particularly extreme. His tariff proposals are bad economics, but hardly extreme given the U.S. history of tariffs. Truthfully, I can't think of a single candidate that has met even an honorable mention as an extremist. Reagan might be the most, but he governed as a right centrist and looking back on his first campaign from hindsight it actually appears pretty mild in retrospect.
The Median Voter Theorem is pretty much alive and well - and let's hope it stays that way.
Australia is notorious for being far closer to MVT results than the USA is, and we also have the conditions for the MVT to fully bind (compulsory instant-runoff voting, so the base *cannot* defect without explicitly falsifying its own preferences). Our politics largely actually run on scandals, specifically because the two big parties are so close to each other on policy.
My first thought about Scott's post was "What about places with compulsory voting?" If there's no incentive to focus on turning out your base, you have to go after the centrists instead. (Of course, you could have compulsory voting without first-past-the-post, and have a multiparty system instead, which complicates things.)
MVT doesn't fully bind for compulsory plurality voting, because extremists on your side can vote third-party to waste their vote. It's the combination of IRV with compulsory voting that makes "get out the vote" useless.
Another interesting thing about Australian politics is that independents can frequenly pick up seats by being even more centrist than either of the two parties. In the last election the loosely affiliated "teal independents" managed to pick up a whole lot of safe Liberal seats by giving off the general vibe of "nicer than the Liberals".
Loudness is often mistaken for extremism. People think that US politics is extreme because both parties spend a lot of time shouting very loudly about how the other one is so eeeevil and defeating them is vital to the future of the Earth and this is the most important election ever and how could anyone possibly think about voting for someone so eeeevil.
But it's very easy to be extremely loud without actually having any particularly extreme policy positions, and this is the situation of the US today.
If I could draw one of Scott's median voter theorem plots, it would have a red dot and a blue dot close to the middle, but each dot would be loudly yelling at its respective side "OMG THOSE OTHER GUYS ARE TERRIBLE". This is how you persuade the people way over on the left or right of the plot to turn out for you -- you fire them up with rhetoric about how evil the other guys are so they don't notice you're barely distinguishable from them.
There are also often people who confuse being consistent for being extreme. Thus a relatively more moderate politician in a parliament who just consistently votes party line (and against the other party's line) would look more "extreme" than an actual extremist who ends up taking "odd" positions on some mainstream issues due to the particular nature of their extremism. (ie. if you somehow had a genocidal-level right-wing racist in who votes pro-choice because he believes this leads to less minorities or so on)
>But the GOP has about a two percentage point Electoral College advantage; that is, the popular vote total most likely to produce an electoral tie is one where the Democrats lead by 2. Adjust for this advantage (we’ll talk later about whether we should), and it looks like this:
You misinterpreted the Centerforpolitics.org piece on this you linked. The GOP has enjoyed a ~2% advantage *only in the most recent elections* and this cannot be projected backwards past 2016. The Democrats had the electoral college advantage between 2000 and 2016.
I’d argue that this is actually the true margin of victory. If I calculated correctly the average is 3% for the years 2000 to 2020, mostly because of Obama’s decisive victories with tipping state margins of 9.0% and 5.4%.
I think a challenge with the precise application of the median voter theorem is that voters don't vote solely (or even necessarily mainly) on policy. Parties also serve as identity groups, and a person might identity as a Republican or Democrat for many non-policy reasons. I think much of the increasing closeness of elections is due to people's politics becoming a stronger part of their identity, so they have become less likely to consider voting for the other party. This could have many causes, from the changing media landscape to the declining centrality of religious and local identities in American life.
But even with polarization, something keeps elections balanced and prevents one party from just securing a permanent lead by locking in the right supporters. I would guess that it is mainly thermostatic public opinion. When bad things happen, people tend to blame the party in power (without giving them as much credit for good things), and this nudges them toward identifying with the other party. So the longer a party is in power, the more supporters the other party picks up. At equilibrium, this means both parties have about the same number of supporters. Of course, the median voter still plays a role, but it's a dwindling number of swing voters who are truly up for grabs, guaranteeing close elections.
Unrelatedly, in the chart showing electoral college closeness, it'd probably make more sense to show the vote margin in the tipping point state (which mathematically represents the median electoral college voter, at least post-facto), rather than adjusting the national margin by some factor, because the electoral college advantage shifts over time and also doesn't have a single unambiguous definition.
> This could have many causes, from the changing media landscape to the declining centrality of religious and local identities in American life.
My hope is that it's downstream from the lack of construction, manufacturing, and other "atoms" jobs. People without material challenges anchoring their daily life get increasingly caught up in immaterial ("extremely online") zero-sum conflicts.
Energy is scarce enough that it must be used efficiently, meaning the few things which are still built here are obsessively optimized, "no user-serviceable parts inside." Those who didn't make the cut for initiation into one of those specialist priesthoods are sloshing around in marginal positions, make-work or not quite worth the hassle to automate, strategically irrelevant, nowhere to go but down.
Abundant cheap energy from solar and wind will change that, open up fields where personal initiative is a meaningful advantage against overengineered perfection. Give 'em something to gnaw on besides each other.
> If some people dislike Trump because he committed 10,000 felonies and an attempted coup, do the Democrats enter that into their calculations and veer slightly further left? If Biden is demented, do the Republicans enter that into their calculations and veer slightly further right?
Historically, both of these seem to have happened, with Democrats shifting left over Trump's term (to the point that Biden, who was still further left than any recent democratic candidate, was seen as the moderate option). And Trump may have decided to pick Vance as his VP candidate over options seen as more moderate in response to Biden's failed debate performance (this was before Biden dropped out). More big picture, the Biden admin was already fairly unpopular a year ago - I can imagine a world where Biden is doing well in the polls and Republican primary voters decide to go with a more electable candidate.
My mental model for the median voter theorem is that it's a pressuring factor towards equilibrium. There's a lot of other influencing factors (and just random noise in who your candidates end up being), so people don't end up at the exact median, but the further out they are the more strongly they get pushed towards it.
This doesn't seem super complicated to me: People vote on vibes.
Biden was losing badly with a policy-driven, median-voter-tracking strategy in part because he wasn't campaigning well, but in part because nobody cares about that stuff or believes him (even though his campaign was mostly "truthful" in the sense that what they were saying was literally true).
Harris was able to make big gains by switching the conversation to "Republicans are weird." They're still tracking the median electoral voter with their policy positions but her public appearances are 100% about projecting a kind of solidarity with suburban moderates in her attitude, not her policy. They've taken a more laid back approach to challenging even Trump's most ridiculous claims, because the democrats screaming "He's destroying democracy!" seem equally hysterical to someone who doesn't live and breathe this stuff (note: I am a "He's destroying democracy!" liberal).
In addition, people in the social media age have absolutely *atrocious* understanding of what the "other" party actually believes. They don't even know the other party's platform. Often they don't know their own party's platform. So even if Republicans moderated considerably, my social group (probably including me) would still see them as extremists, because that's how the news and memes I'm regularly exposed to paint them.
The median voter theorem holds fairly well on simple, high-profile issues. But mostly people don't care about the issues at all - Republicans want someone brash, powerful, and spiteful; Democrats want someone "tolerant (TM)," technocratic, and non-threatening. And moderates want someone who doesn't challenge their sense of what is and is not normal in any way.
It's worth noting that this isn't necessarily wrong or irrational - even if Trump adopted every policy I support today, I likely wouldn't vote for him just because he's him. I remember 2020 when he made extremely spurious claims about election integrity and pressured local officials to distort the results of the election. That level of interference with an election outweighs all other potential issue for me, and I won't vote for someone who did it whatever their positions.
> In addition, people in the social media age have absolutely *atrocious* understanding of what the "other" party actually believes. They don't even know the other party's platform. Often they don't know their own party's platform.
I’d add that it has become effectively impossible to know either party’s platform, and this is particularly true for Democrats. Their platform weighs in at 92 standard pages, the Republican platform at 16. Both are inexcusably poor documents. Platforms have been increasing in size and decreasing in comprehensibility since at least the civil war.
If you want to know what each party really stands for it’s effectively impossible and it’s been made intentionally so, particularly in regards to the Democrats from the 1990s onward when they became notably less populist and more technocratic.
Even this can be a fracture point, because I agree with the premise but disagree with the cause. I see the issue as the nation's problems becoming increasingly complicated and the right pushing simplistic solutions that will obviously fail. In that framing, Democrats aren't necessarily obfuscating, they're responding to an obfuscating situation.
Though have to admit that I'm much more likely to see the flaws in things I understand than in things I don't, and that politicians often retreat into complexity to avoid criticism, so probably the answer is not that simple.
You make a good point. There’s undoubtably some truth to it. Though I think the Democrat’s shift from unions and labor and blue collar to higher education, technical expertise, and white collar has made the party so philosophically confusing that they need 92 pages to try to convince the voter that in the end it all makes sense.
Nevertheless I think I largely agree with you. The biggest divide between the average Trump and Harris voter remains education level, and the Republican Party’s effective strategy to appeal to less educated voters is clearly present in the tone, vocabulary, formatting and relative brevity of its current platform.
Most Americans, Democrat and Republican alike, can masquerade as a member of the other party well enough to pass an ideological Turing test judged by members of the other party. So maybe people's understanding of what the other party believes, isn't that atrocious after all.
This is interesting! I'd seen the opposite reported - at least that people who spend a lot of time on social media have poor understanding of the opposing party. But now I can find no source for that, and the reason it stuck in my head is that it's counter-intuitive. Maybe it came to me in a dream :/
There's lots of reasons for this but the easy, partially wrong answer is that the parties are following the median voter theorem, just for affect instead of policy.
I think the theorem has an implied "complete/instant propagation of information, or otherwise static information landscape" assumption. Removing that assumption, it reads more like "parties will attempt to steer towards what they perceive as the median voter, in order to gain/retain power". Wordier and squishier, but makes sense.
Yeah if you assume a lag in both learning information about the electorate and making changes towards policies you would expect to see something similar to what we see (e.g. having a polling miss in 2016/2020 lead Democrats to work less hard at appealing to the center because the center they were appealing to was wrong)
My pet theory is some of the weirdness of the real world vis-a-vis Median Voter Theorem relates to the fact that its less products being bought and sold (like Hotelling's Law for ice cream) but rather subscriptions being sold and bought (like AT&T vs Verizon, or Netflix vs HBO, but for 4-year terms and party identification).
Not sure how specifically it would do so but that must muck up the generic 'product discovery' process.
* First, don't discount the problem of measurement. Prior to the daily tracking polls, campaigns had to operate with much less data; they could reasonably be mistaken about whether voters perceived them as centrist.
* Second, I think you're on to something with your comment about reducing things to a single straight line. We know that high-dimensional spaces are awfully counterintuitive. Candidates have more options than moving towards and away from the center: they can "rotate" around the space to try to capture voters.
The curse of dimensionality is made worse because the election is a dynamic game, and issues that are important today (e.g. a candidate's alleged dementia or penchant for felonies) weren't issues a few presidential elections ago. Campaigns try to make issues more or less salient at least as much as they try to (re)position themselves along the hyperspectrum.
Median voter theory doesn't require a single dimension; no matter how many axes you plot on, there will still be an averaged 'center' of public preference, and the logic applies that a Condorcet voting method should converge on it.
That being the key factor not mentioned here - median voter theory only applies to Condorcet voting methods (such as ranked choice voting), it does not apply to first past the post voting, which is what we use in the US.
But anyway, US politics gets reduced to a line colloquially because it's a two-party big-tent system (which is an inevitable result of first-past-the-post voting). No mater how many dimensions you plot politics in, the two parties will each have a discrete point location in that space, and you can always draw a straight line that passes through them.
Whatever this line happens to be at the moment, becomes what that generation conceptualizes as 'left vs right'.
> No mater how many dimensions you plot politics in, the two parties will each have a discrete point location in that space, and you can always draw a straight line that passes through them.
I think this is building in too much intuition from Cartesian geometry. It seems very unlikely to me that the shortest line that passes between the parties in such a space is the same straight line that you'd get from linear interpolation.
Let's think non-parametrically. For each of the two candidates in the race, order all of the voters by winnability, starting with those requiring the smallest change to win over or alienate.
The median voter theorem would assert that *these lists are the same*. If Alice and Bob are the candidates, the set of marginal voters that Alice and Bob might win or lose with small positioning changes are the same set.
I submit that in multiple dimensions, these lists differ. By making small changes, Alice can win over Bob-voters that would otherwise be safe for him, and vice versa. In wibbly multidimensional space, this is how the candidates rotate around each other.
Take the current US presidential race as an example. One issue where the candidates differ is in their stance on whether presidents should profit from business ventures that attract foreign money. Can Harris really win over any marginal voters by saying she'll accept *a little bit* of Saudi investment in her nascent t-shirt business?
> If so, how come when Biden was replaced with the less-demented Kamala and the Democrats’ betting odds went way up, Trump didn’t change any of his positions AFAICT?
It seems to be widely accepted that Trump spouts meaningless syllables and one cannot take anything he says at face value. His supporters regularly make claims to this effect in response to people asking their opinion on Trump's latest utterance; including here in these comments sections.
It is unclear to me how people who claim that position actually determine what Trump's policies are, but this process cannot be related to the statements he makes if no statements he makes are to be taken at face value. IDK, perhaps they read tea leaves or something. In any case, if nothing he says or does can change what people think of him, why change what he says or does?
It has an electoral system (SNTV) which combines both "choosing a candidate within the party" and "choosing a party over the other party" into a single vote, and it has a bunch of formalised internal factions, which means that any voter wanting to vote for (or against) one faction has to vote for the party in order to pick between the factions.
This lets it target multiple different positions on the electoral spectrum at once.
Japan hasn't used SNTV for their more powerful lower house since 1993. They use the majoritarian parallel voting there, and a mix of SNTV & party-list PR for their senate. I believe Taiwan and South Korea also experimented with & ultimately discarded SNTV.
In the Japanese case before the 90s, they found that it lead to corruption- politicians within the same party didn't have much to compete on ideologically, so they competed on pork barrel spending and dubious payoffs to local power brokers
I hear that part of it's that because the LDP's always in power, and it's got only weak ideological goals, if an opposition party comes up with a popular policy idea the LDP just implements it immediately and takes the credit.
I think there was a bit where they bribed the owner of 2ch to covertly shape discussions in their favour, too.
A Japanese-language general-purpose BBS. 4chan is somewhat descended from it as an English-language imitator (technically, it's an imitator of a splinter of 2ch that added image support).
2ch is/was huge, with millions of posts per day at its peak (and almost all of that from Japan), so it's significantly more relevant than any single English-language BBS ever was (though still not on par with the social media titans).
If, for presumably cultural reasons, the ruling party is more capable of rapidly altering positions (or equivalently, the voters shift relatively slowly), they can stay close enough to the median that no one cares enough to switch. (This assumes some level of hysteresis, which I think is a reasonable model.)
"As per the Median Voter Theorem, most people prefer the (centrist) Democrat to the (more extreme) Republican. But the far left doesn’t care enough to vote, so the Republicans carry the day."
You've just explained 2010. It's more complex than that, everything is, but those are the blurry contours of 2010.
What’s going on right now is just Trump. He’s extremely unpopular but it’s not because of his policies. It’s mostly the things he says, the way he acts and J6. The policies he actually enacted in office were moderate. But the left pushed further to the left during and after his presidency, which is why Trump is still competitive. The median voter hates Trump but does not like the Democrats either, hence the stalemate.
You seem to have gotten the causality backwards. The only reason that Trump was seen as remotely viable in 2016 was because the Dems had already been going leftward full-speed for years, making the notion of "owning the libs" very attractive in some quarters.
Exactly. A lot of people think that "Trumpism" is all about Trump, but Trump is a businessman first and foremost. He saw a market full of unmet demand, decided to meet it, and built a successful brand.
In other words, it's the precise opposite. Trump is about the "Trumpists."
It’s both. Trump won the primaries because of immigration. The Republican Party had been moving towards amnesty for years beforehand, which was unpopular for Republican voters. Trump promised to stop this and gained supporters because of it.
But because Trump was so unpopular with Democrats, they decided afterwards to pivot towards an extreme and incoherent position where they opposed any enforcement of immigration laws but claimed they didn’t support open borders.
So Trump was filling an unmet demand among Republicans and then the left reacted by opposing that.
I don’t think Hillary lost because she was seen as an extremist on policy. She lost because she was seen as corrupt and held many Americans in contempt.
Obama also pretended not to be extreme at first, but didn't do anything to curb extremists on his side, and eventually started fanning the flames himself, particularly on race. I don't think anybody expected Hillary to be any different.
I'm always up for a good "the Democrats started it first" but Trump's big breakout hit was the fact that neither party wanted to speak against immigration and he was able to capture that market. Also the mainstream Republicans refused to drop out splitting the mainstream vote.
More generally, much of Trump's support is a reaction against the perception by Republican primary voters that conventional politicians of both parties had been colluding to enact unpopular policies favored by "elites", particularly with regard to immigration and free trade. There was a similar reaction among Democratic primary voters, but a combination of different primary systems and different candidate environments lead to Trump winning and Sanders losing their respective nominations. One key difference, as you allude to, was the other candidates: the Republican field was wide open, while the pro-establishment Democrats were united behind Clinton. Trump's celebrity status probably made a difference as well. And there was a huge impact from the way the parties allocated delegates: Democrats allocated proportionately subject to a 15% vote threshold, plus a large number of current and former officeholders able to act as ex officio unpledged delegates, so Sanders probably would have needed a solid majority of primary votes (or at least a very strong plurality) to win even in a three or four way race. And Republicans mostly used "winner take all" or "winner take most" allocations so Trump could run up large delegate tallies with 25-40% of the vote in any given state.
Tangentially, Sanders and Trump used to have quite a bit more in common policy-wise before moving towards the centers of their respective parties in order to win the primaries. Sanders has always been protectionist on trade issues, and was fairly strongly nativist (on economics/class-warfare grounds) up until early 2016. And Trump had been strongly pro-choice and had advocated for a wealth tax when he'd previously flirted with running for President in 2000.
Can you tell me more about Trump flirting with a wealth tax? That sounds so weird. I can't google anything because google is flooded with how Trump wants to not tax the wealthy.
This would seem to fit the argument though, Obama won by pretty large margins (biggest since Reagan iirc). By the MVT that would suggest there was room for activists to push the Dems left (which I think we saw over Obama's terms) and perhaps overshoot.
That is at most half the reason Trump was viable in 2016. The other half, at least, was Hillary Rodham Clinton being so monumentally anti-charismatic that to many voters it didn't matter what her policies were if she was the one mouthing them.
In 2016 Trump claimed that we would protect medicaire and social security which opened a lane for people who care a lot about those things but dislike other democrat policies to consider voting for a republican.
Wait, shouldn't median voter theorem work *less well* in a highly polorized society?
When the polarization is low, we actually have this "median voter personality" - someone who could realistically vote for either democrats or republicans. And, therefore, appealing to them is a great idea.
But when the polarization is high such median voters cease to exist. Moderate republicans will keep voting republican, no matter how far right the republican candidate goes, because the thought of voting democrat is just unconceivable due to the polarization. And likewise with modern democrats.
It seems logical that in highly polarized societies the radicals will have more and more power making the "centrist Democrat’s nightmare scenario" more of a reality. Do we have a more up to date version of this analysis that showed that such situation almost never happens? I'd like to check the dynamics and therefore put my hypothesis to the test.
The relevant change is how much "you can reduce everything to a single straight line." I agree "polarization" is the wrong word for it: "dimensional reduction" perhaps?
No, the issue is that Scott is describing a greater reliance by voters on party labels, and saying this should make them more willing to vote for a guy with the wrong label who's closer to their personal position. This makes no sense; greater reliance on labels makes people less likely to vote for someone with the wrong label.
Moving from a situation where the median voter theorem has no meaning because there's more than one political issue, to a situation where the median voter theorem has no meaning because there's only one political issue and it isn't continuous, is not actually an improvement in the explanatory power of the median voter theorem.
No matter how many dimensions your politics has, a two party system will always consist of two points somewhere in that space, and you can always draw a line between them. That line them becomes 'left vs right' for the current moment, regardless of its contours.
The reason we're stuck with a two party system in the US is that we use first past the post voting, which means spoiler effects (and some other technical stuff) drives out any third parties. It's also why median voter theorem doesn't work well here - median voter theory only applies to Condorcet voting methods (like ranked choice or Approval voting), and teh US doesn't use a Condorcet method.
Revisit The Myth of the Rational Voter by Bryan Caplan. Nearly 20 years later it’s still an excellent guide to critically assess the idea of a rational voter. To the extent politicians are rational, they know the voters are biased and become rational in exploiting it.
> Instead, Democrats are often pretty far left, and Republicans pretty far right.
That might be narcissism of small difference? From the outside looking in, there's a remarkable consensus on many topics in American politics. You guys just don't talk about the things you agree on, because it's boring?
Eg a while ago Germany privatised their snail mail and removed any monopoly rights. In contrast, you'd be hard pressed to find any mainstream politician in the US who would contemplate such a move. Even suggesting that is far outside the overton window of polite conversation in the US.
Similar for the German railways (Deutsche Bahn): in that case, they were about to privatise is but then the Great Recession hit, and they never got around to it again. But it's organised as a private company that just happens to be 100% owned by the government. And there's plenty of other companies competing with them for passenger rail traffic.
In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties.
Those are examples where Germany looks a bit more 'right' than the US. There's plenty of examples where Germany looks more 'left'. Eg, if you want to open a computer repair business in Germany, you need to do 3 years of government approved training, then work for a few years, and then do another two years of government approved training. That's nuts, and wouldn't fly in the US. (Though occupational licensing is becoming worse and worse in the land of the free, too.) See Meisterzwang, if you want to read more. Germany also taxes software engineers differently depending on what kind of software they write, but the difference is only something judges made up, nothing any real software engineer understands or supports. (See 'Freiberufler'.)
For a different topic: German has regulations that all pork that's sold must be safe to be eaten raw. Because Germans like to eat raw minced pork on breadrolls. Trying to introduce such regulation in the US would mark you as quite the weirdo. That wouldn't really fall into a standard left-right spectrum.
"In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties."
Would it, though? Maybe in the Acela Corridor, but in the rest of the country "Defund Amtrak" wouldn't be even remotely controversial for a Republican.
"In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties."
On the other hand, suggesting the rail infrastructure should be nationalised - as is the case in pretty much every European country - is well outside the Overton Window, and you don't get rail traffic competition without that.
" And there's plenty of other companies competing with them for passenger rail traffic."
Well, it's more accurate in most cases to say that there are other companies competing with them for government contracts to operate local rail services - which is the case in several US cities as well.
" But it's organised as a private company that just happens to be 100% owned by the government."
Is the postal monopoly something a strong consensus of Americans strongly support, rather than something few people care one way or the other strongly enough to make waves about it? Beyond the few hot button issues in politics, there are thousands upon thousands of accumulated laws and regulations that stick around mostly because few people care strongly about them one way or the other, as voters will only pay attention to a limited number of issues at a time. Often also there are small interest groups that care strongly about preserving them, while few others care to oppose them. In America it's further complicated by a bicameral legislature, presidential veto and Senate filibuster making it hard to change laws; if there is no prospect of an overwhelming support to change a law, it's not even worth starting to push the issue. Is the USPS monopoly something the vast majority of Americans strongly support, rather than one of these thousands of regulations few people pay attention to?
> Is the postal monopoly something a strong consensus of Americans strongly support, rather than something few people care one way or the other strongly enough to make waves about it?
There isn't really a difference between those two ideas. People won't vote for anything they don't care about.
Every so often someone tries to legalize self-service gas stations in Oregon, and it gets defeated by a scare campaign, despite the fact that believing anything the scare campaign said would require you to be far too stupid to be able to feed yourself. It's not like this is an untested policy.
It doesn't matter for the argument as long as you can still veer towards the center and get more popular approval. Harris could become less favorable to abortion, though still more than Trump. The question is why doesn't she. For this, it doesn't matter the size of the spectrum.
There does seem to be one unspoken assumption here. This states that (hypothetically at least) there is a one-dimensional line of positions from right to left with voters having preferences somewhere on this line, but does not state (and seems to simply take for granted) that the voters are evenly distributed along that line.
This is obviously not the case. Intuitively, you'd naively expect there to be a regular distribution with the bulk of voters clustered around the center. Of course, this would only make the naive Median Voter Theorem's predictions that much stronger, which is clearly wrong.
A better explanation would be that there are *multiple* regular distributions. Research in Moral Foundations Theory supports this idea: there are five fundamental, inborn "axes" used by human beings to evaluate questions of morality, and there's a strong tendency for liberal people to only reason with two of them at the expense of the other three, while more conservative people use all five more or less equally. So with two fundamentally different ways of looking at the world and evaluating the issues, you would expect to see two regular distributions of voters along the left-right line, with the largest groups clustered near the local maximum of each distribution, and the two parties staking out positions pretty close to those two local maxima.
>This states that (hypothetically at least) there is a one-dimensional line of positions from right to left with voters having preferences somewhere on this line
No matter how many dimensions your political model has, a two party system has just 2 points within that space as viable options on the ballot, and a straight line can always be drawn between those 2 points. Coalitions are then forced to form themselves relative to that line, and that line becomes what we mean when we say 'left vs right' for the span of that election.
Every individual voter has one point on that line which is closest to their actual position in higher dimensional space. That is their point of preference on that line.
>but does not state (and seems to simply take for granted) that the voters are evenly distributed along that line.
Not at all. The median voter theorem says elections that use Condorcet voting methods (which we do not) will converge on the weighted average of public preference, ie the spot that minimizes total regret across all voters (as measured by the distance between their personal point on the line and the point which we call the center).
Basically there's an optimal frontier at which moving left to satisfy people on the left would create more dissatisfaction than that on the right, and moving right to satisfy people on the right would create more dissatisfaction than that on the left. That equilibrium point where you can't minimize total regret any more by moving along the line is the center of public opinion, and it exists for any possible distribution of preferences across the line.
This all assumes that there isn't a "knowledge problem", and that both parties know what the exact center is so they can make their perfect plans. More importantly, it assumes that the center doesn't change.
But as you know, coalitions change, the electorate changes, and parties are constantly making electoral guesses about which combination of factors they can compromise on to build a winning platform. This can swing dramatically from one election to the next in ways that might never have been predictable ahead of time.
For example, in 2004 when Republicans were running on finishing the job in Iraq and Afghanistan, you'd never have predicted that they'd be running on "no new wars" two decades later, or that Democrats would be running on building up a stronger military and support from Dick Cheney.
In 2022, Republicans found out that abortion was a devastating electoral issue for them. Trump reacted accordingly, but soon found that he couldn't escape the issue because his donors demanded he push it in the opposite direction his polls suggest is a winning strategy. Yet, just 2 years ago there was a serious national attempt by Republicans to push a national ban through ... before the issue beat them up. Sure, there are still some Republicans whose constituencies favor a ban, but the party saw big losses from it in 2022 and recalibrated accordingly. The point is that they didn't recalibrate until after the losses, suggesting there was a knowledge problem prior to the election.
If this hypothesis is true, we should expect election-winning issues from midterm elections (especially in swing states) to be adopted in presidential platforms two years later.
This was more of an issue before the digital age, which answers Scott's question of 'why were the margins so much bigger back then, but way closer now?'
These days information gets distributed and integrated pretty quick; even if normal people can't name specific policies and pieces of legislation, changes in their party's 'vibes' get transmitted to them through elite culture production in a million little channels pretty rapidly and effectively.
Also, it's not too important for voters to have highly accurate knowledge on an individual level. Basic experimental power calculation - we average across tens or hundreds of millions of voters. As long as there is *some* signal by which actual changes in the parties affect some people's understanding of those parties, it doesn't matter if the average voter is extremely noisy and unreliable, we will extract teh signal when averaging over millions of data points (votes).
But the underlying problem remains: the sampling mechanism is infrequent (once every 2-4 years) and therefore all reasoning is based on a very small n. The glut of new methods doesn't change this. You still have to figure out what's relevant to the metric you care about, or how to adjust your sampling method to more accurately reflect the 'true' outcome, just like pollsters do. The fundamental question isn't "who's posting what to social media?" That only matters to the extent it tracks with the more important questions, "Who is going to show up on election day?" and "How are they planning to vote?" The answers to the questions we care about are still just as infrequent as they've ever been.
Whether it's polls, cell phone tracking, app usage, social media, combined aggregation, etc., you still have to grapple with the same question: how well does this new metric predict the outcome of the election?
All these new methods suffer from the same knowledge problem that polls suffer from: you can't calibrate statistical results from a single data point. So they do past-casting from old data points ... just like polling organizations have been doing for decades.
But which ones? How much do mid-terms matter? What about off-cycle elections, recalls, run-offs, etc.? Will it be relevant to incorporate old data, and how far back does an election result become irrelevant to the current cycle? Were important issues from last cycle fads, such that we should expect the electorate to revert to their old habits, or are they enduring trends that will continue moving forward? Are there new fads we're not picking up on, or are the new fads something that reflect in the news/social media but won't impact polls?
The answer to each of these questions is not "Yes" or "No", but somewhere between. The amount a pollster should recalibrate on any of these questions is always a judgement call, whose validity can only be confirmed in retrospect. Except how much can you 'confirm' a statistical model from a single data point? These problems are not 'solved' with new prediction methods.
"But we have so many more sources of information! Surely that improves our accuracy." Aggregate a dozen data sources calibrated from the same one or two hard data points and it's garbage-in/garbage-out.
Off-topic: This problem bears a surprising similarity to the Fermi Paradox and the Drake Equation. How much should you calibrate your estimate of how much each factor matters (e.g. presence of a gas giant, or a moon, or a magnetic radiation shield, or even presence of liquid water) in your estimate of intelligent life developing? We only have one data point, so it's impossible to calibrate your results, but always possible to generate new variables (e.g. Great Filter).
Mathematically, the median voter theory doesn't require a single axis, like left vs right, to drive the analysis. You can calculate a multidimensional distance that takes into account many issues or even how intensely voters care about them.
All models are wrong, but some are useful. A one-dimensional model is useful to some extent, but it's also interesting to have a look at higher-dimensional models and see if anything new pops out.
A two dimensional model (say, "social issues" vs "economic issues") gives you interesting new behaviours that you don't get in one dimension. If the views of voters on these issues are not perfectly correlated then you'll have voters in weird corners and you can expect to see the parties circle each other a bit as they try to sweep up these weird-corner voters without losing touch with their base. This explains why the direction of the Republican-Democrat axis in the US feels like it's currently rotating, something you don't get in a one-dimensional model.
Going beyond two dimensions is a more realistic model but probably doesn't give you enough new insights to justify the extra complexity.
It's more that in a two-party system, you can always draw a line between the position of the two parties, no matter how many dimensions those points are plotted in.
It's more relevant in systems with more than two choices, that the theory still works with an arbitrary number of dimensions and choices.
You and John are talking about different sets of dimension, I think.
John is talking about dimensions of preference, like economic vs social.
The single axis you are referring to is about the space of candidates/choices; it shows what happens when you cannot draw a single axis through all of the parties.
You can indeed have an arbitrary number of dimensions of preference/politics/issues/etc. As long as you only have two candidates, they will each be at some point in that higher-dimensional space, and you can draw a single axis through them. The median voter theory will then work on that singular axis.
Re: control of the House, for most of the last couple decades Republicans have had an inherent House advantage. This reflects 1) more and better gerrymandering. Ballot initiatives and judges blocked Dem gerrymanders in CA and NY, nothing stopped the Republicans in TX and FL. 2) The inherent propensity of democrats to live in hyper concentrated pockets. There are urban House districts that vote 90 - 10 Dem, but even extremely republicans areas (TX panhandle, NW Georgia) don't reach that level. Those are basically wasted votes in a first past the post electoral system.
Interestingly 2022 broke this trend somewhat, with a narrow Republican popular win leading to a narrow Republican House majority. I think this reflected greater Republican support in black and particularly Latino ghettos (where they still lost anyway), some rural areas becoming more lopsidedly Republican, and suburbs moving Democratic.
What is important about collusion is what is at stake. If you're running the Conservative Party or worse, you are Israel's Yair Lapid, you need to worry about never losing elections, because there's little constitution and little courts, Labor or Likud can just win and make all the stuff you care about illegal.
But because of three distinct powers, a reasonably strong constitution, and the filibuster, if one party loses, the other party won't just undo everything they have done. What you're optimizing for is getting a trifecta, maybe a 60+ trifecta like Obama had. In this sense, "Democrats can have school funding" and "Republicans can have military funding", the way you described.
But you can't run Likud, Labor Party, Yesh Atid, or the Conservative Party this way!
In my country, Brazil, we have a really big constitution and even more judiciary oversight. If Israel is in one extreme for how much power one given election has, U.K is still bad, and the U.S. is in the side where it isn't too bad, Brazil is in the other extreme. The Supreme Court could just run the country without any oversight because as the Constitution is so extense (the 2nd most extense in the world!), there's little you can do once elected. And indeed, Congress is an amorph group of centrist people with little ideologic beliefs. (See the wikipedia article: https://en.wikipedia.org/wiki/Centr%C3%A3o )
And indeed, all presidential candidates that run since the down of the 6th republic and 2018 were very firm in saying they weren't right-wing. Indeed, Bolsonaro big contribution to Brazil politics is the discovery that you can win elections by being right-wing.
Somehow now politics resemble more U.S. politics. The first-round is expected to be a primary, and if you can't pander to the median voter of your coalition, you're toast. Why? I don't know. Maybe politics has become an sport where people really like winning but for little policy subtantiation? Or people don't realize there's little they can achieve once elected. Idk.
How are Brazil's Supreme Court justices appointed and how much latitude do they have to reinterpret or emphasize parts of the constitution over others?
So on the theory that parties try to decide just how crazy they can be:
Parties are coalitions of many interest groups. Those interest groups are sometimes pointing in sort of opposite directions, sometimes they are vaguely pointing the same direction, but they're all competing for the same amount of oxygen.
So each individual interest group would accept an increased 0.5% chance of their party losing the election (c'mon, will it really be that close?) if it meant locking their party into their interest group's positions. Repeated many times this leads to some real stupid things, stupid things you can see in the other party even if you can't see it in your own.
(This isn't even bringing in the Iron Law of Institutions.)
I’m a fan of this, but it doesn’t explain how it stays so close to 50/50. Surely one party has interest groups which are more willing to make that bargain, right?
So one of the big things missing from this article is that the median voter theory only applies to Condorcet voting methods, which the US does not have. That's the main reason the US doesn't align well with the theory.
Australia uses Instant Run-Off voting, which is also not Condorcet. In practice, the first-past-the-post voting method used in the US, and the IRV method used in Australia, are nearly identical in terms of outcomes in most real-world cases.
There are Condorcet methods that use preferential ballots, an any system that uses them will create outcomes in line with the median voter theorem. But in practice I don't think any countries currently use these methods at scale, because the elites who decide the voting method are rarely advantaged by outcomes that match the center of overall public opinion.
thank you 🙏🏻. Voted in our state election today. Numbered all candidates as I preferred from 1 to 7. For a valid vote, we have to number ALL … can’t leave a blank against the absolutely not wanted at all candidates/parties. Which nutter to put last is the challenge!
The collusion theory where both parties trust each other to run on unpopular things is on the right track in that it acknowledges that parties care about things other than winning. Usually both parties are really OK with losing, especially since fundraising goes up when a party is not in power.
But they can’t lose horribly or repeatedly because they want to be taken seriously; they both want special interests to court them and special interests don’t want to bribe a side that doesn’t have a chance of winning.
Scott noted that the Republicans don’t seem like they are putting much effort into calibrating to balance out the Democrats. Well, you actually only need one mover to play the game and find balance, and Democrats have been the ones doing that recently.
Replacing Biden with Harris was the result of these incentives playing out pretty perfectly: someone just likable enough to run a tight race but not so likable as to abandon the party’s interests and beat Trump easily.
To be clear, I don’t think Democrat politicians are literally thinking through this exact logic in their heads per se; rather, at scale this is how the incentives play out in the system.
I think that the reasons the Democrat chose Harris rather than "a better candidate" was based on Harris being the obvious Schelling point candidate just as Biden was the obvious Schelling point candidate in 2020.
Much like "professionals think logistics" politicians think turn out to vote. There is no "median voter". Professionals look at the spectrum as "your mother on one side and your ex-spouse on the other". Your mother will always vote for you, your ex will never vote for you. There are hard limits in each district on the persuadable voter. There is also a hard limit on how much fraud will occur in each district (yes it occurs). Each median voter is an amalgam of multiple single issue concerns. There is no "median voter" in reality, just as no one is "average" in all dimensions. The overriding concerns will vary depending on the immediate concerns at that election. Your candidate just needs to be less of a turn off compared to the opposition in a 2 way race. As explained by a national consultant " A company thinks a 45% market share is great. In a 2 way race it is called loser". Where US politics went bad is when the courts ruled that majority election, ie 50% +1 needed to win was discriminatory. 50%+1 meant the more mainstream candidate won the primary and then the general. Mere highest vote total leads to winning candidates outside the mainstream. (Ranked choice is worse. remember no method of voting satisfies all criteria). If we want more "mainstream" median politicians we need to have party primaries where the winner needs 50%+1 to win, run offs of the top 2. The election winner needs 50%+1, run offs as needed. Fraud can be kept as low as possible with same day in person voting, Government issued IDs, and paper ballots in transparent boxes, ie just like Europe.
Wave elections occur when one sides voters can not bring themselves to vote for "their sides candidate" and just stays home. No one actually shifts parties other than very few. A 10% shift in a groups preferences are historic and rare.
"When the courts ruled that majority election, ie 50% +1 needed to win was discriminatory."
What is this a reference to? I don't think that the Courts handed down the rule that pluralities count as wins, but I could be misinterpreting something.
>There is also a hard limit on how much fraud will occur in each district (yes it occurs).
I’m not sure I understand. Are you claiming voter fraud in the U.S. plays a significant role in election outcomes, or that due to the “hard limit” it doesn’t? If you think it does, can you link to the data that suggests that?
> Fraud can be kept as low as possible with same day in person voting, Government issued IDs, and paper ballots in transparent boxes, ie just like Europe.
The assertion you make here about European countries requiring same-day in-person voting is false. Many European countries do allow mail-in voting (Germany, UK, Sweden, Spain to name just a few).
I don't believe you're a professional on this issue. Scott discussed turnout vs swing voters, and you haven't actually corrected anything he said (instead you acted like you hadn't read it).
I think the easy explanation for the 55-81 Democratic control of Congress is a lot of southern Democrats were elected Blue but voted Red. If you look at the voting record of those congressional years it will look a lot more 50/50 than party affiliation alone would predict. (I suspect the same is true of the other years that don't look balanced either)
No, they didn't. Per DW-NOMINATE the primary axis of politics was economic and regional/social issues were the secondary axis of politics. Your link still shows the parties primarily in separate clusters (even if less cleanly so) in the past, and doesn't mark out southerners at all.
No, my link doesn't show separate clusters - it shows lots of blue dots on the right side of the scatterplot. As someone who was alive and very much politically active during that time period I can confidently state that many southern Democrats were essentially Republicans in all but name. Many even went so far as to switch parties.
Is it possible that elections were farther from 50/50 in the past not due to information scarcity? Today's politicians don't need to guess how popular they are, they know what voters say about themselves, they can just open Twitter and get more information about the state of the election than anyone could have dreamed of 40 years ago.
It seems intuitive to me that if both candidates had perfect information about the electorate, they would find some kind of strategic equilibrium that made elections tend towards a toss up, but I can't really formalize that intuition.
Yes, this is very likely correct. Information abundance has made both parties more 'efficient' in terms of ability to measure public opinion and respond to it quickly.
When both parties are doing this at maximum efficiency, the natural equilibrium is 50% (or 50/50 in the electoral college). The median voter theorem shows why - they should both want to be very close together at the center of public opinion.
And absent some culture war issues where the public is divided 50/50 and therefore it is safe for the parties to signal their allegiance by disagreeing with each other loudly, this basically is what has happened. Despite culture war over whether immigrants are scary or inspirational, both parties are currently touting their tough-on-the-border credentials. Despite culture war over whether cops are bastards or heroes, both parties are running tough-on-crime fund-the-police-more platforms. Despite culture war over whether China should be hated and feared or if hating and fearing China is racist, both parties are advocating trade tariffs and strategic industry subsidies to fight China. Etc.
I don't think Guam would be a lock for 2 democratic senators. The current non voting representative is a Republican, as was the previous-to-current governor.
The median voter usually isn’t someone whose opinions cluster in the center of two ideologically sorted poles. They are someone whose views don’t sort ideologically. They might think we should deport all immigrants and nationalize healthcare; or that we should end the Fed and subsidize organic farms. They are just as likely to think a centrist candidate doesn’t reflect their values as the other candidates.
What might explain the 1955-1981 Democratic congressional dominance is lack of polarization combined with a huge incumbency advantage during that period. One party happened to be in power when this incumbency advantage suddenly took hold. Why might such an advantage suddenly take hold in 1955? Well, it corresponds with the first year the majority of Americans owned TV sets. If congressmen were advertising on TV back then -- and I don't know that they were -- it may have created a positive feedback loop of winning -> gaining name and face recognition -> raising more money to advertise on TV. This inertia worked for Democrats because they had first mover advantage until Reagan came along and changed the game with more TV charisma than anyone had had yet. After Reagan, party polarization gets real.
What about state level politics? How does the Median Voter Theorem work in, say, Wyoming, where both the state house and senate have been Republican since 1967? Or Maryland where they have been Democratic since 1920? Surely the Maryland Republican party has an incentive to move to the left, and the Wyoming Democrats to the right, but this doesn't seem to have happened despite decades of effectively one-party rule.
The GOP and DNC have a lot of control over who runs in state races, you can't put your name down as a Republican on the ballot if the GOP doesn't affirm you as a member of their party.
They're not going to help a Republican senator get elected if he's going to vote against them in the Senate all the time, even if that's what would be needed to win his district.
And without an R or D next to your name and with zero funding and resources from the national party, you can't win anything larger than a small-town mayorship.
So local politicians are beholden to the national parties, and can't diverge too far from them.
One obvious answer is that primary voters consider both ideological purity and ‘electability’ and that median voter theorem comes in through the second channel in a thermostatic wisdom of crowds kind of way. So if you run Dukakis and get annihilated then next primary you weight electability much more heavily (and conversely if you run Romney and still lose then next primary you say ‘fuck it, let’s roll the dice.’
All of this assumes voters sum all their interests together, build an average and vote based on that. This is not how it feels lately. Here in Germany, the latest elections were mostly about the stance on migration (even local elections where the elected political bodies have no influence on this matter), less so on the cost of climate change politics and whether or not the new right is electable. For the US, it seems a lot is still on abortion.
And everywhere, the split seems to be around wokeness
> Here in Germany, the latest elections were mostly about the stance on migration (even local elections where the elected political bodies have no influence on this matter)
That doesn't make sense. It's like saying that women couldn't influence politics before getting the vote, or that merchants couldn't influence politics... and still can't.
Having or lacking statutory authority over something just doesn't have much to do with how much influence you have over that thing.
Brexit was determined by a non-binding referendum. How non-binding was it?
What I am saying is that city, county and state elections were about things like migration issues. But that is a federal issue, a county or even state parliament cannot change the asylum or migrant benefits laws
I don't see how that responds to my comment at all. County officials definitely can change the asylum or migrant benefits laws. The fact that there is no legal channel for them to do so doesn't mean that they can't; it doesn't even mean that they aren't the primary means by which such changes happen.
And... that's what my original comment says. What were you responding to?
Again, I'm going to have to ask you to respond to something in my comment. Why do you believe that "These are federal (and partially EU) laws that cannot be changed locally" is a relevant way to reply to "Having or lacking statutory authority over something just doesn't have much to do with how much influence you have over that thing"?
Do you think that I've overlooked part of your point? If so, what is it?
Do you think that you've overlooked any part of my point?
Why do you assume the opposite? Why would a county assembly have any influence over the federal immigration policies? They deal with building permits, infrastructure maintenance and whether (big topic here) the freshwater supply can sustain the expansion of the Tesla factory. All they can do about immigration is to push back on the state and federal government that they cannot afford to house the next batch of immigrants. The federal government can safely ignore that and just reply "but you have to, it's the law"
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”. Doesn’t feel like the kind of thing that would get lots of votes. Maybe after decades of humiliation and failure you could convince someone
Look at the last two competitive primaries. In 2020, most of the Democratic candidates were running left, but the winner was the guy who said “actually we should veer right”.
And in 2016, most of the Republicans were running right, but the winner was the guy who said “actually we should protect social security and Medicare, maybe even increase them, and cancel all wars”. He didn’t *spin* it as veering left, but on the traditional political substance Trump is definitely to the left of most Republicans in the past few decades. He just has an angry culture war persona that makes him *feel* right wing, while campaigning on undoing the core Republican platforms of cutting social security and Medicare and expanding wars.
> In 2020, most of the Democratic candidates were running left, but the winner was the guy who said “actually we should veer right”.
...and then didn't. And did anyone *really* believe he would?
I think people are starting to wake up to the common pattern that Democrats these days always seem to run as reasonable moderates, but then when elected they invariably govern as extreme-left as they can get away with.
I'd say that a lot of the "seem to run as reasonable moderates" is entirely due to the heavy media spin that the "mainstream" democrats inherit by right. Because the main media outlets are *openly and blatantly* in the tank for the Democratic party establishment. It doesn't matter what they say or do, the main media outlets are going to paint them as saints who are moderate as apple pie and will paint the Republican as the second coming of Hitler-Satan.
And the right-leaning alt-media will paint the Democrat as being the second coming of Stalin-Mao, so that part is totally bipartisan. The right-leaning media is a bit more fractious in not *always* lock-step supporting the Republican--GOPe candidates get tarred as Democrat-lite these days.
The main things where he is left of Obama are ones he inherited from Trump, like his harsh anti-free-trade policies, and his big cash giveaways at the beginning of his term. There’s also a lot of mid-level appointees that he got from the Warren team who implemented lots of small left things, that I agree are left of anything from Obama or Trump. But the main Biden policy bills are Manchin-approved centrism. He didn’t do anything as big as Obamacare or same sex marriage.
> But the main Biden policy bills are Manchin-approved centrism.
The ones that *passed* are. Now look at the ones that he wanted but that didn't get past Manchin and Sinema.
> He didn’t do anything as big as Obamacare or same sex marriage.
- Open borders, and all the gigantic messes downstream from it.
- Title IX abuse that goes far beyond anything Obama did.
- Criminal prosecution of political opponents, a tactic straight out of the USSR and Communist China. (For all that Democratic scaremongering claimed President Trump wanted to do this to Hillary Clinton, he never did it. But Biden did!)
- The repeated, entirely lawless attempts to nullify student loans
I haven’t seen any evidence of open borders, much as I would like them.
And there is no universe in which he is prosecuting political opponents. Who are you even thinking of? Bob Menendez? Donald Trump? Rod Blagojevich? Hunter Biden? These are criminals, who should be prosecuted, and any competent government would prosecute them. You would have to be Berlusconi’s Italy to *not* prosecute people like this.
In other words, if you plot voter preferences in one dimension, the median voter is a point, but if you plot them in higher dimensions, wouldn't you get a higher-dimensional interface with lots of points? That is, aren't there actually tons of very distinct policy platforms that would win half the voters?
If you only have two candidates to choose between, then no matter how many dimensions your preferences are plotted on, you can draw a single line between the position of those two parties in the higher-dimensional space, and every voter will have some point on that line closest to their actual position in the higher-dimensional space. Median voter theorem will then apply to that line and people's proxy location on it.
If you have more than two candidates to choose between, then yes, it is possible to construct arrangements where you cannot reduce the higher dimensions down to a single line, and public opinion can have multiple balancing ('center') points which each point at a different party.
This happens more often in thought experiments than in reality, though. Political parties don't actually choose random positions in the higher-dimensional space, those dimensions are typically correlated to each other in a lot of ways that constrains the space of internally-coherent parties. And those parties tend to converge towards center points in order to get more votes. You need fairly particular configurations to get multiple center points which each favor a different party, it doesn't happen by chance too often in practice.
But two parties cannot occupy the exact same position of median voter, it would make them indistinguishable. So there is a force that pushes them apart, at least on "branding" issues. If it pushes them equally, we get the observed result, no?
I know my fellow residents on the left and right wouldn't agree but you can see a real 'median voter theorem' effect in some sense in Canada where the parties are very very powerful and top-down - where we have more than 2 parties (usually ~2.5).
Liberals have done well since the 1900s as the 'centre' party (not centrist) and they often get booted out to crash and burn extremely painfully once people get tired of economic mismanagement.
The Tories come in focusing on the 'basics' but being the sum of various anti-Liberal factions across the country results in them imploding in the effort to appeal to median voters.
The UK seems to be pretty similar in this regard too, and it seems its a result of the stronger parties.
The US has not only weak parties but various 'backbenchers' (in that they are not part of Cabinet or top ranks of the deparment) like Rand Paul or Cortez who appear to be more popular and visible than their own partie - simply for being expressive virtue signallers in Congress.
"Parties and candidates seem to do a suspiciously good job staying equidistant from the median voter, far beyond the pathetic amount of effort they explicitly put in. I still can’t tell if it’s all coincidence, or whether there are deeper currents at work."
Baseball managers, before the analytics explosion, naturally adjusted to the run scoring environment. They used one run strategies more in low scoring seasons and less in high scoring seasons. I think very few of them thought of it in those terms, they were simply rewarded with wins if they adjusted and losses if they didn't.
"If some people dislike Trump because he committed 10,000 felonies and an attempted coup, do the Democrats enter that into their calculations and veer slightly further left?"
FWIW, this is in basically Matthew Yglesias's position on what happened — Trump is a below-average quality candidate, because he's an offensive nutcase, so this gave Democrats latitude to shift left on policy (relative to Obama, let's say) while maintaining about a 50% chance of winning. (Separate question is whether this was a good idea.)
Yes, and I think the primaries essentially ensure that democratic leadership always *wants* to push as far to the left as they can, because they were selected to represent the median democrat.
Vance's pick as VC also seems to have been done under the expectation that Trump would face a weak democratic candidate. Why didn't he switch later? I'm assuming because Trump has this whole personal loyalty thing going on.
That, and he *is* facing a weak Democratic candidate. As the burst of Democratic euphoria from the abrupt switch dies down, that's becoming more and more obvious, and Trump continues to gain in the polls a little bit more every day. The latest RCP polling has him (narrowly) ahead in every one of this year's swing states.
> Why didn't he switch later? I'm assuming because Trump has this whole personal loyalty thing going on
There's no real mechanism by which he could switch, after the convention. Vance is the Republican Party's nominee for Vice President, and being the nominee for President doesn't give you the right to change that.
Some of the questions around “Why don’t the parties simply X?”, where X is some action that would definitely get the more votes, boil down to the fact that what appears to be a spectrum is actually composed of specific issues, many of which are extremely emotionally charged and polarizing. Some of these issues are so deeply psychologically motivating that some voters actually think more in terms “punishing” the other side for having such a terribly immoral and evil stance in on their specific issue.
Other commenters have brought up the way in which attitudes on gay marriage shifted substantially in the past decades as an example of the parties successfully maneuvering around the central axis. But this was possible because gay marriage was not one of these hypercharged emotional issues for very many people. It was never the sort of issue that would make a preponderance of republicans say, “I could never, in a million years, vote for someone who endorsed this policy. I would rather vote for higher taxes than vote for gay marriage.” Very few republicans felt that way about gay marriage, to the consternation of many conspicuously religious conservative pundits, who felt confused and betrayed by their own party’s abandonment of this issue.
The thing that confuses me about the present moment is that both parties seem to commit what appear to me to be frequent own-goals by willingly adopting policy positions that are widely loathed by huge numbers of Americans. In the frame of this article, they’re pandering to the most extreme elements of the base in a way that actively infuriates and enrages pretty much everyone who isn’t an extremist. And both sides do this! Why doesn’t one party simply choose to be the one that doesn’t uphold *any* policy positions that the median voter thinks are insane/stupid/evil?
Because as Scott mentioned, motivating your base, your donors and your volunteers/staff are three disparate activities. The most successful compromise is apparently to make a platform with mostly normal positions, then mix in a few insane ones and hope it doesn't spoil the voters' appetite.
You can see this working by contrasting with mayoral platforms. When elections are small enough that a candidate can furnish all the funds and labor themselves, platforms tend to be all sane positions or all insane positions, not this inorganic mix.
> If so, how come when Biden was replaced with the less-demented Kamala and the Democrats’ betting odds went way up, Trump didn’t change any of his positions AFAICT?
I guess there wasn't a huge change between June and August, but Trump this time around has made moves towards the middle. He is going to specific interest groups and promising policies. He went to the Libertarian convention and promised to free Ross Ulbricht, a super niche issue. He basically told the GOP party platform they were banned from mentioning abortion.
That said, Trump IMO is a result of the primary system. He's not a very popular general election candidate, has high net negative favorability, but he has a strong base in the primary. I think the Median Voter theorem is true, but within the constraints of the current political system, which are numerous. In a parliamentary system, I suspect Republicans would have replaced him.
By the time Biden dropped out it was too late to change anything, but I think that if Biden had dropped out a month earlier then Trump wouldn't have picked Vance as a running mate.
Vance was the pick of a Trump who is confident in his poll lead and can afford to pick someone for alignment rather than electability. If Biden was out and Kamala was in the middle of her media-puffed bounce then Trump would have "tacked to the centre" by picking for electability.
I think the simplest explanation for parties not having identical platforms in the middle of the policy space is that a single dimension of policy is distortingly idealized. There are many different issues that actually mattered to different voters and many voters have constellations of preferences that are not perfectly aligned with one another. Consequently, coming up with a platform to appeal to a maximum number of voters is not a simple problem. You end up with some things that the parties agree on and a lot of things that they disagree on. All in order to maximize their chances of appealing to a majority of voters the right combination of policies can in theory appeal to substantially more than 50% if the other party does, a bad job of calibrating their own policies.
The two American parties have wildly different rhetoric, but isn't it a common complaint from their bases that they don't govern all that differently when they're in power? That could be because of inability (split government, filibuster, bureaucratic inertia, other checks and balances), but also because they cater to the median voter. Though I don't see why the median voter theorem would apply more to policy than to rhetoric; the opposite would make more sense.
Total post-hoc speculation: perhaps voters intuitively know that parties won't shift policy all that much one way or the other if they get elected, so the radical rhetoric doesn't put them off. At the same time, while they govern similarly, the exaggerated rhetoric helps voters know how the parties are positioned relative to each other (as well as helping candidates win primaries). If, say, Republicans govern in a centrist way as do Democrats, then if Republicans didn't have distinctly right-wing rhetoric either, right-wing voters wouldn't know they have to vote Republican to signal to politicians that they want more right-wing governance.
--------
One more reason for major parties to not be perfectly centrist is to keep minor parties at bay. If the major parties were indistinguishable, people wouldn't worry about wasting their votes on minor parties, which could allow them to grow until they can actually challenge the two major parties; or at least until they can spoil the major party closer to them.
If, say, the major parties were both centrist, but Republicans were still recognized by the public as more right-wing, and a more radical right-wing minor party started gaining awareness, then some right-wing voters would turn to that party, while left-wingers would stay with the Democrats, so Republicans would lose (like how Reform spoiled the too centrist Conservatives in the recent UK election). This possibility incentivizes Republicans to be somewhat more radical.
--------
Maximum collusion between parties wouldn't push them to be maximally extreme in different directions. If they both want unpopular policies, they can also collude to both support both unpopular policies (in the example, both school and military funding), so voters have no choice. They may not collude to be perfectly equivalent to keep third parties at bay and to win primaries, but there's no incentive to be maximally different.
Indeed, if they were maximally extreme in one direction and the other, that would presumably also include maximally undoing the other party's policy, e.g. one party would increase military funding, but slash school funding well below where a centrist compromise or perhaps even the median voter would put it, and vice versa, so in expectation there would be no benefit to the parties over colluding to stay moderate (though not indistinguishable).
--------
It's not exactly right that candidates in the primaries have to cater to the median of their party, since presumably some primary voters are tactical enough to care about their candidate's electability in the general election, even though many primary voters don't do that. So in the primaries candidates' target should be somewhere between their party's median and the general electorate's median.
To begin with, the 'wildly different rhetoric' is ussually restricted to issues where the country is already divided close to 50-50, and ussually issues that are not very consequential. They tend to actually be close together on most important issues, and just not talk about those very much.
Can we safely assume that the overwhelming majority of voters vote on candidate positions, rather than on something else?
We know people vote based on other things as well - candidate personality, media coverage, the demographic of the voter, voter inertia, in some cases the demographic of the candidate. Is it fair to assume that all of this is insignificant in the picture?
The median voter theorem applies to voter preferences, not voter policy preferences. Anything that makes voters prefer one option over another is a dimension on which the theorem operates, including everything you listed.
Trump did say something like, "If it was popular vote it would have been a different campaign"
I think Canada has some good things going on with campaign finance. The Liberals banned corporate and large donations and instituted a per vote party subsisdy
subsidy
The Conservatives banned union donation. I really like that median voters can have an effect on party finances without directly donating, which is a harrowing experience of all caps paper mailouts that are hard to unsubscribe from, that appear to be designed to activate very intense supporters
I agree that the evenness of the splits is surprising but I don’t find the median voter theorem plausible. How do trump’s felonies or biden’s senility fit into that? When you factor in trends like, declining trust in all kinds of institutions, and massive increases in debt, I think you’re watching some long term play itself out, whereby a polity splits in two as it dies, and something in that process seems to equilibrate the split along roughly even numbers.
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”. Doesn’t feel like the kind of thing that would get lots of votes.
This strikes me as a serious failure of imagination. Trump is way to e.g. Romney's "left" on entitlements, gay rights, marijuana, and trade policy. He's successfully captured the support of more minority voters and averted the "permanent Democratic majority" scenario that Republicans were panicking about during the Obama years. Also, as this example demonstrates, politics is not one-dimensional—Trump brought his party's platform closer to the median voter by moving "to the right" on immigration. Primaries have functioned pretty effectively at producing electable candidates recently; I don't see any reason to expect that to suddenly change.
Get out of one dimensional thinking. Think of a two dimensional political map where equality is the left-right axis and amount of liberty is the vertical axis. (see: https://holisticpolitics.org/NewParty/ ) The party positions are rotating counterclockwise. Trump has pushed the Republican Party kicking and screaming into the long neglected upper left quadrant. Lower-right Republicans (Liz Cheney) are now voting Democrat.
But no matter where the parties are in higher-dimensional space, you an always draw a straight line between them.
That straight line becomes the 'left vs right' axis for that election.
And yes, it might be very different from the 'left vs right' line from 20 or 10 or 2 years ago! But we are so polarized and stereotyped that we will call it the 'left vs right' axis no matter where it actually goes between, and form coalitions in line with it.
All part of the madness of a two-party system, which is the inevitable side effect of first-past-the-post voting.
"But no matter where the parties are in higher-dimensional space, you an always draw a straight line between them."
You've said that at least half a dozen times. It is past tiresome, and even more tiresome if it incites people to try and rebut you in half a dozen different subthreads. Maybe focus your attention on one or two, and otherwise let it go that there are people who are Wrong On The Internet.
Why did we see Dems win in Alaska and Arizona, Republican majority states, over the last four years? Because the Republican choices were too far right for moderate Republicans and Independents. And the Dem choices were not far-left. Seems pretty straight forward.
The "reasonable assumptions" of the median voter theorem assume a reasonable voting method. Literally anything better than the mess the US is currently using would qualify. Plurality voting and its variants suffer from the *spoiler effect*. The spoiler effect leads to *center squeeze* and polarization.
Also, you will have a hard time finding a political scientist who is educated in this regard. The topic of voting methods is a kind of niche one even within science. Most experts are from various fields: economics, statistics, math, self-educated activists, etc. Only a few are political scientists.
Thanks for these awesome links. I suspected this whenever I start to talk to political science people and they get this "I hope he stops talking" look when I mention Arrow's theorem.
This is especially striking in the Senate. Of the 50 states, 31 are to the right of the nation as a whole (see here <https://en.wikipedia.org/wiki/Cook_Partisan_Voting_Index#By_state>), so naively you'd expect 62/100 senators to be Republicans on average. Instead it's closer to 50. That's because Democrats have done a really impressive job of getting high-quality moderate candidates in swing states (and sometimes red states). Democrats have the vast majority of swing-state Senate seats (11 out of 14) and a few more senators in red states than Republicans do in blue states (though that could change this year).
Also -- quick fact check: in the six presidential election cycles you mention in Part II (2000 through 2020), the average electoral college advantage for the GOP has been 0.6 points, not 2 points. It was:
All this chicanery and maneuvering regarding trying to strategically appeal to different hypothetical voters seems to serve as a pretty good argument against electoral democracy in general as a desirable system.
Median Voter Theorem is like the Efficient Market Hypothesis: It's about equilibrium situations. Elections are far more noisy and slow feedback loops than the market, so it can take a long time to reach equilibrium: Republicans losing the house a lot is a PART of that feedback process that gets us to where we are now. We've only been having elections for a couple hundred years, and we've also not been in a stable political, economic, or social equilibrium for most of that time.
You also keep talking about parties being coordinated centrally but that's not necessary. Marginal new members of each party (whether from independence or from the other party) can gradually shift the positions of a party over time without any central plan.
Political scientist here... :-). Mostly just want to say how much I enjoyed your thoughtful take on this important concept within our discipline. But, I did feel compelled to point out one considerable error. The electoral college had a strong R bias in 2016 and 2020 but that is most definitely not a feature over all elections post 2000. In fact, in 2008 and 2012 there was a clear D bias. So, there's no way you should simply be subtracting 2 percentage points from the D margin every year since 2000. Actual figures nicely laid out here: https://centerforpolitics.org/crystalball/a-brief-history-of-electoral-college-bias/
The Median Voter Theorem fails IRL because the initial assumption of a one-dimensional political spectrum is false. It's not even a decent approximation. It's not close to true.
Politics is not even 2-dimensional, as depicted in the common square diagrams beloved of libertarians. You probably need 5 or 6 dimensions to get even a decent model.
One way to see this is that it is common for two candidates with very similar positions, perhaps even the same positions, to perform very differently based other factors. Those other factors are simply manifestations of dimensions other than the left-right dimension.
The missing ingredient of the Median Voter Theorem is "political capital." Regardless of the electorate's preferences, professional politicians are obsessed by the concept. To summarize quickly, it essentially means "popularity points" and leaders love to have them but their parties hate to accumulate too many because they want their core constituencies happy by spending their political capital on them. That's why parties often spend political capital, always trying to just keep enough to win the next election, by 0.1% if possible. That's the strategist's real dream.
This was most evident to me during Bush II's second term. Having won clearly in the 2004 presidential election, he started off with a lot of political capital, which his aides at the time urged him to spend in core rightist concerns (at the time, mostly law and order, school vouchers, stuff like that) but Bush essentially tried to maintain that political capital longer than many thought wise, and it was eventually consumed by negative newsflow, things like Iraq and Katrina -- as his aides feared.
I don't think "we should spend our political capital, because that is the point of having it" is the same idea as "we should spend our political capital, because it deteriorates over time".
What if our switch to voting the "lesser of two evils" and fine tuned national campaigns is the culprit? In the past we would, I think, vote for the candidate we liked, with less regard for party affiliation. Then national campaigns got better at scaring us about our opponents, but no better at making us like their candidate. So over time we find ourselves agreeing with the negative campaigning and feel pressure to not vote in the worst candidate. Using wedge issues (immigration, abortion, police/crime, whatever works) to make the opponent's look as bad as possible. Positions that are too unpopular disappear, leaving just the wedgiest of wedge issues, pushed as hard as possible. Over time this creates the culture war, because these topics have been optimized to find the hardest edge to push against an opponent.
It seems that negative campaigns are far more successful than positive, as much as I hate that. Jeb Bush was probably a very competent administrator and would likely have been a good president. But Trump calls him "low energy" and suddenly he's toast.
>But we may still find this surprising. The parties - especially the Republican Party - don’t feel like masterminds executing a complicated dance where they determine the exact amount of extremism the voters can tolerate.
They may not be doing math, but politicians (and parties) are doing this all the time. Why do you think the RNC scrubbed abortion completely from their platform this year? Did they all have a change of heart? No, they figure the median voter is currently more worried about abortion restrictions right now than they are concerned about fetuses, so goodbye pro-life, we can’t afford you this year.
I think you're leaving out at least two important factors:
1) Polls exist, giving parties information continuous information about where the center is, and parties must publicly signal positions to voters which also signals positions to the other party. As a result, no collusion (as commonly considered) is required for triangulation to occur. Also note that the US holds *very* long elections by global standards so there's a lot of time for the triangulation to occur.
2) polls are quite noisy, so it's difficult to identify an exact median position on any one issue. This uncertainty is amplified by the multidimensional nature of political positioning: combining positions on (eg) immigration, taxes, and abortion) there's theoretically *several* median cumulative positions available that a candidate might aim for based on their priorities.
Example: most Poli sci people I read thought (based on lots of polling!) That Democrats would need to tack to center on abortion to take advantage of the post-Dobbs situation, but... That's basically been wrong.
Adam Mastroiani of Experimental History just published a study today on how Dems and Reps pass the "idealogical Turing test"; basically they can't tell when a statement was written by a real Dem/Rep or someone who is pretending.
His take away was that each side actually understands the other side's point of view, at least to a degree. But I think it more likely supports your half-baked idea that the political ideologies are emergent and incoherent, and that we Americans are all trained on the same political data...so we mostly regurgitate the same nonsense that everyone already knows.
That's not the Turing test. The ideological Turing test is whether you can write a statement explaining the views of the other side that is accepted as accurate by the other side. My recollection is that the studies I have seen that tested this found Republicans more able to pass this test than Democrats
I was just doing a quick summary of someone else's post. If you must correct someone, I'd suggest reading the post I'm referring to and correcting the author. Thanks!
Single-axis left-to-right as a model of political positions is an oversimplification, and enough people deviate from "choose the candidate who can win whose closer on this axis" to throw a bit of a wrench in the predictions that model would make. Notably, swing voters sometimes have a mix of left-wing and right-wing policy preferences (which makes "closest" hard to calculate"), make decisions on the basis of something that the parties don't particularly select for (which changes their preference from somewhat arbitrarily), have a bias towards change versus status quo (which changes their preference depending on who's currently in charge), or have other structural preferences that outweigh the usual left-to-right concerns (at least sometimes; e.g. Bernie/Trump voters, the Cheneys apparently).
I wish you had bothered to examine whether the median voter theorem is even true at all. The available data indicates that it is not. Most swing voters are economically left and socially center-right. And then you have a small sliver of people who are economically right and socially left. But the truest swing voters do not have a consistent ideology. This is why rhetoric and marketing are far more important than party positions.
Voters do not fall on a single left to right spectrum. There are multiple spectra on different issues. Also multiple interest group conflicts that do not fall on the left-right spectrum. Parties are complex coalitions. The median voter theorem is correct but too simplistic to account for more than a part of .the behavior of national political parties.
There's an interesting dynamic, too, in American politics (which is maybe basically what Scott referred to, just from another POV): very frequently, some party tries to inch closer to the median voter, then loses, and all their base say, "See?! We told you so! America will only support us once we go full-bore right / left!"
The point is that I think unsuccessful attempts to deploy the median voter strategy can lead the median *primary* voter to a more extreme place.
>In 1964, LBJ beat Goldwater 61-39. In 1972, Nixon beat McGovern 61-37. Even as recently as 1984, Reagan beat Mondale 59-41.
>But we already know that partisanship was weaker in those days. The Median Voter Theorem only works if you can reduce everything to a single straight line. If voters don’t care about the right-left spectrum, they might judge based on criteria like “Reagan is more charismatic”, and then if everyone agrees that he’s charismatic you can get 59-41 or 90-10 or whatever numbers you want.
Alternate possibility: Maybe polling has gotten more accurate over time, and the parties are more precisely following optimal tactics than they were before 2000.
Each party is trying to win while spending the minimum amount of money in the campaign. So if a party is ahead in the poles, they start spending less, and vice versa.
This can by itself explain the near 50-50 outcomes, though reality is probably more complex.
Why does it make sense for them to do that? If a party wins while having lots of money left over, surely that will signal to the donors to donate less next time? If I were a campaign manager I'd try to win while spending all of the money I have.
I'm noticing a few factors that can lead to heterogenous candidates.
One, I saw Majromax mention: the measurement can be off. If I'm a candidate, the MVT says I want to adopt the median position; but that doesn't mean I know what it is. That's why I run polls. Polls are imperfect, though, and maybe my polls tell me something different from what my opponent's polls tell him. The catch here is that the R and D polls would need to consistently tell each candidate that the median position is R or D respectively, when in reality, one would expect them to update toward each other and be nearly identical by now, and they aren't.
Two, in the US, it's not just a median position, but a median position altered by an electoral college effect. If the median position is against tariffs on apricots but nearly all of those people live in four states, then the non-median position might get more electoral votes anyway. And this is also a reason each party might get different poll results; they can't afford to poll everything everywhere. Though again, one would expect them to have learned which issues are most important and probably agree on that by now, and again, they don't. So all this explains is why the true median might not be the one sought by candidates - in other words, why party platforms, even if you average their positions, will be consistently off from whatever a pure population poll would tell you - but I suspect this is a minor effect even so.
Three, the median position isn't just one issue, but many (as many comments have noted). A candidate can be median on five positions and still lose because they weren't the most important positions. Yet again, though, one would expect each party's pollers to know this by now and be equilibrated.
Four, each position waxes and wanes in importance relative to the others, and this can change in a matter of weeks. Interestingly, general positions matter, on the economy, civil rights, defense, education, and so on, and sure enough, most candidates *do* take the same position on the basics, which tells me that they *do* conform to the MVT! They only differ on more topical issues such as a war somewhere or bioresearch reform or ethanol subsidies or which metal should back up the dollar or whatever, and these will froth around so chaotically that the topicality-adjusted median position (TAMP) will drift, and each party *will* turn up different results depending on when and how they poll.
It might even explain why each party will appear to be pro-war one decade and anti-war the next; it depends on why each war started, and which party is out of power, and therefore able to flex fastest toward the TAMP. It also applies to other issues; if a party's position is sort of virgin and untested, it's also pliable, and able to adapt to new polls. So imagine the Median Voter as having a general position that both parties adopt, with a bunch of variations on it that the parties can only pivot to when they're not too committed.
Five, is the sort of flip side of One: each party can exert pressure on the median position. This works to the extent that voters often don't even have a position on some issue, but will go along with whatever they think the median position is, so here's this handy party machine to tell them, if it dares. Wellp, it dares. But only so far. No Democrat is going to say, know what, most of the US is pro-gun after all, so let's repeal all the gun laws except for felons! Each party is still anchored to its previous commitments (you could say the median position is strongly against waffling), and can't change them without a plausible justification. But that still leaves a lot of room to wiggle, for anyone who isn't a purist extremist.
And this helps each party. If the TAMP is against changing major positions and slightly tolerant of changing minor ones, then it's easier for a party to convince voters that the median position is the one that party has long favored, than for that party to move itself to the TAMP, which drifts from week to week anyway. And since it's easier to stick with the same messaging than to change it every week, it explains why parties have consistent, heterogenous depictions of the median.
In this frame, one can view the MVT as producing a median window, and each party trying to persuade voters that that median is narrower and closer to its side of that window.
If most voters are fairly content with the status quo then I've no doubt the Median Theorem is pretty much true.
But, as the Tory party in the UK recently found out the hard way (although, incredible as it sounds, most of the fools still haven't grasped this and learned their lesson!), centrism is a fair weather creed.
Just as beach goers enjoying the sun will scatter for cover in all directions when storm clouds gather and the rain starts, if the economic climate darkens or causes of discontent proliferate then votes will polarise and for all practical purposes one median splits into two, on left and right, like a cell dividing!
1) Discussions of rational intent for parties fall down when they turn to Trump specifically. Trump is not a rational actor.
2) California Democrats' distaste for the national Republican party (or, mutatis mutandis, other such distastes) matters less in a gubernatorial race than in one for Congress, because a governor does not become part of a legislative caucus. That's a reason Arnold could get elected governor at a time when no Republican could any longer be elected senator from California.
3) Any consideration of why the Republicans couldn't win Congress in 1955-81 has to take into consideration the fact that, for most of that period, Congress was in fact controlled by "the conservative coalition," between most Republicans and conservative Southern Democrats. (This coalition was actually founded around 1938.) The erosion of the tradition that Southern conservatives had to be Democrats reached a tipping point around 1980, and that's one reason the Republicans started to win the body, because now they had Southern conservatives on their side of the aisle, and not just in informal coalition with them.
Political science is no more a science than economics.
A much simpler explanation: Both US political parties are primarily focused on being the 2 halves of the rulers of the United States.
It doesn't matter who wins for most federal apparatchiks so long as both parties defer to said apparatchiks.
This is also noticeable in the way both parties are against populists whether right populist like Trump or left populist like RFK Jr and Tulsi Gabbard.
The oligopoly of political parties dynamic describes Europe very nicely: for all the nice talk about diversity of representation - the machinations against AfD and Le Pen make it clear that the mainstream parties simply just don't want anyone else horning in on their action.
This also explains why both legacy Democrat and legacy Republican parties have paid exactly zero attention to what Americans overall want. These parties are not there to serve the people, they are there to ensure they can oligopolize the levers of power.
But all good things come to an end - the right and left populists have now merged under Trump, RFK Jr and Tulsi Gabbard. The dinosaur neocon warmonger Republicans are in the process of migrating to the Democrat party, which in turn is becoming ever more transparently the tool of the Federal Bureaucracy including literally dozens of ex-military, ex-FBI, ex-CIA types in office or running for office under the Democrat shingle: https://covertactionmagazine.com/2024/10/21/class-of-2018-cia-pentagon-democrats-continue-to-advance-hawkish-policies-in-congress/
The median voter theorem assumes that in a one-dimensional two-party setting, both parties will glue themselves to the median voter (known as Hotellings law).
However, in this situation the median voter has no incentive to join either party. Since s/he will get their preferences served without having to lift a finger.
Those who have an incentive to lift a finger (spend time and other limited resources on political activity) are those whose preferences are different from the median voter. Let’s call them firebrands. Both parties have them, and the further away they are from the median, the more firebrand they are.
Now introduce the concept of the median party member. The party members are those choosing the candidates that face the voters. (Simplifying, but that is what models are for.)
The median position within the party depends of the relationship between the firebrands in the party (who want to drag the party toward their own non-median preferences) and the pragmatists, who want power above anything else. (“A politician without power is an abomination” ). The latter will fight the firebrands to appeal to the median voter.
However, since the pragmatists have to compromise with the firebrands within each party (and sometimes the firebrands sweep the floor- in particular in gerrymandered districts, where there is no risk of losing to the opposition party anyway), the political profiles of the candidates in both parties will diverge from the median voter. In particular in non-marginal districts & constituencies & states.
And from there on you can do your own combinations.
There will be a long-run tendency back toward the median voter in two-party systems, but there will be a lot of hiccups along the way. As the political history of countries like Great Britain, New Zealand (before NZ changed to a version of proportional representation in 1993) and the US show in abundance. Edit:
One type of hiccup occurs when one party is conquered by a firebrand faction (defeating the pragmatists), which then pulls the party toward an extreme. In theory (following Hotelling's law) this should prompt the other party to move in, to occupy the median ground. However, sometimes this instead fuels/provokes the firebrands in the other party to move their party further in the other direction (defeating their pragmatists, who want to move center-wise).
An example was the radicalization of Labour in Great Britain after Thatcher and her neoliberal faction conquered the Conservative Party in 1979. It took until 1994 before the Labour pragmatists were able to defeat the Labour firebrands, and elect the moderate Tony Blair as their leader. Resulting in victory a few years later.
…An even more dramatic type of hiccup in two-party systems occurs when a faction of firebrands conquers one of the parties, and then move it across the aisle to the extreme of the opposition party! This happened in New Zealand between 1984 and 1988, when a neoliberal faction conquered the NZ Labour party from within and moved it to the right of the National (conservative) party. This resulted in a “I am more neoliberal than thou” competition between the parties in the following years, until exasperated voters in a referendum voted to change the whole system to a version of proportional representation. Which resulted in a shift away from extreme neoliberalism and back toward the median.
The US has arguably been through a light version of something similar. The Democatic party under Clinton moved toward a neoliberal position (Clinton’s welfare reform of 1996 being a high water mark in this regard), and from then on until 2016 the two parties competed about who was most consistently neo-liberal. When Trump, who saw an opening for a political entrepreneur, conquered the Republican Party from within, and shifted the party profile toward immigration control, protectionism and disengagement from non-vital wars – very far from neoliberalism. And understandably more popular with the US working class, where the median voter is located.
...Which in its turn has resulted in the Democratic Party now scrambling back toward the median to regain their working class constituency, who now suddenly have another option. This process back toward the median was evident under Bidens’ “build back better” renewed industrial policies. And regardless of who wins November 5th , this “return to the US median voter” is likely to continue.
This theorem is an interesting angle to look at the elections and democracy in general. But I think some points are missing here.
1. I think/hope democracy is about power to the people. Insofar it makes sense for a politician to adapt their opinions on certain topics to better cover the opinions of the majority. But on the other hand, politics and elections especially should be a competition of ideas, shouldn't they? A credible politician is the one who can convince the voters of the superiority of their (own, not the voter's) stance in the political spectrum, or not? Isn't it a bit of a misery that election seems to be the voting for the best weather vane, at least according to the median voter theorem? I mean doesn't it look like deception when power-hungry politicians change their position just to get elected? Anyone here who wants to be ruled/represented by such a person?
To say the least, the median voter theorem assumes a static public opinion. To what extent is this actually the case?
2. I also think that media is playing an important role here. The center/mainstream of accepted opinions/major consensus narrative has become narrower and shifted more and more to the right in the last thirty years (at least in Germany, how about the US?). I think everywhere as a reaction (Covid has been a catalyst) people retreated to their respective filter bubbles. As an example, in Germany, the main topic in the media is (illegal) immigration. This is however, in my opinion, pretty detached from the problems this country actually has. Therefore, it is a question, in which way the actual policies are still influencing the voters' opinions/votings because of the media's influence and how much it is just the making of clever spin doctor's. I remember the "Dancing Queen" Theresa May, it only extended her time as prime minister for one year.
People are flooded with impressions on a daily basis, they are weather vanes themselves. On the other hand, I am to young to compare with the "good old times".
3. I also want to make the point that it pays off to care about people. In Graz, Austria's the second largest city, the mayor is a communist. And this in a country where the neo-fascists have gained the majority in the last nationwide parliamentary election. I guess the local communist party doesn't have particularly communist positions rather social-democratic ones, I would say. Still they had to overcome a pretty high threshold, but their persistent actions and support for common people in health care and affordable housing made them credible and electable. But I'm afraid this is limited to be effective in local/regional politics.
Are there common sense voters in this age of heightened partisanship and what seems to be strong tribal affiliation? I for one do not fit neatly in to the Dem or Repub Parties, but will not vote for Harris and if i vote will vote for Trump. I noted somewhere that Martin Gurri who sat out the last two elections said he will vote and it will be for Trump.
Isn’t it difficult to veer left or right credibly? If the Republicans started supporting all the Democrat positions on hot issues, which party would the voters that care about those issues vote for? And the same in reverse.
The Republicans have lately tried to paint themselves as pro-union defenders of the working class. It has sort of worked, but at the cost of basically purging the Republicans elite. And the Democrats have actually been assisting, by clearly supporting policies and displaying attitudes that favor the elite.
Yes, not having any baggage of a history as a Republican politician helped him in the general election. But we were talking about how one can win the primary while tacking to the center.
Sometimes “moderating” makes them a maverick, sometimes it makes them a RINO.
I thought we were talking about the general. Why would tacking to center even be a temptation in the primary?
Was Trump tacking to center, or rotating the axes of the left/right spectrum? He tried to get support from traditionally Democratic voters, but not by adopting Democrat policy ideas. Trade and immigration policy was barely a spectrum, more of a consensus, before Trump trampled it.
Okay, but was he tacking left or realigning the axis, redefining what the left-right divide means (or maybe taking off the mask that hid what it means)? The median voter theorem is not about redefining the terms of debate. It is about trying to occupy the center of a stable and well defined spectrum of voter opinion. Trump was more disruptive than that. He did not try to make his ideas sound like a more sensible version of what the Democrats have been saying.
How about multiparty systems and non-first-past-the-post voting systems?
In the aftermath of the 2001 elections here in Italy I would often hear the opinion that the mainstream left was only defeated because the extremist Communist Refoundation party received 5% of the popular vote.
That sounds like a clear-cut case of what Scott says doesn't happen (extreme voters punishing the mainstream parties for being moderate).
This happened during a period in which Italy was experimenting with a mixed first-past-the-post and non-first-past-the-post system. It's possible that people were still voting out of habit under the assumption that smaller parties had a chance, as if there were no first-past-the-post.
"Maybe after decades of humiliation and failure you could convince someone, but I don’t see the history of parties suffering decades of humiliation and failure before finally agreeing to turn around."
I don't know enough but weren't the Republicans on the left until about 1920? And didn't they veer right in the 1970s to get votes from Southerners discontented with civil rights reforms?
No, per DW-NOMINATE the Federalists are the original right-wing party before the Republicans (the primary axis of politics is economic), and indeed John Adams is recognizably conservative (if not as far right as Hamilton) while Jefferson was a comparative radical who was enthusiastic about the French revolution and thought we needed periodic revolutions rather than being bound by the dead hand of a Constitution written by previous generations. Nor were the Republicans left of the Democrats prior to 1920, McKinley was definitely to the right of Bryan. Woodrow Wilson, who gave us the income tax and tried to bring us into the League of Nations, was on the progressive left (part of why he's ranked so highly by historians despite being a segregationist).
Mapping current labels onto historic parties leads to poor analysis. The Federalists were a pro business, pro big government party. In 1820 Monroe won every state basically unopposed. In 1824 there was a new split in the Democratic Republican party due to the contested election.
In 1900 Republicans were to the left of Democrats but in 1912 everyone was a variant of progressive (by early 1900s definition). Because Teddy Roosevelt ran against Taft, his followers drifted from the Republican party and eventually became aligned with the followers of FDR in the same way that many of the Anti-Trump republicans are becoming Democrats.
The Federalists were electorally viable when the US was a timocracy where only property-owners could vote. Democrats dominated after property restrictions fell away. The government at that time relied on tariffs for revenue, so prior the 16th amendment it couldn't tax the rich to buy off voters the way big government could later.
> In 1900 Republicans were to the left of Democrats
The federalist were not viable because the major issues that defined the difference between the Federalists and Democratic Republicans ceased to matter (the end of the French Revolution). The Democrats split from the Democratic Republicans (the Whigs also split from the Democratic Republicans). The Democrats dominated until the civil war because there was not clear ideological groupings.
In 1900 Republicans were the party of big government, while the Democrats favored ending the gold standard and were opposed to imperialism (the American occupation of the Philipines). Republicans were still the party of civil rights. Ideologically the Republicans of this time were probably most similar to Late Obama and Hillary's campaign (pro business pro government) while the Democrats were closer to Trumponomics.
William Jennings Bryan was the Democratic candidate in 1900. He had advocated a graduated income tax, just as the Socialist Labor Party and Populist Party (which also nominated him) had done. The Populists also advocated that the government create deposit-taking banks so people wouldn't have to use private ones, and that the telegraph system be nationalized, and that the government employ laborers for public works during times of economic depression.
> The common-sense answer is probably right, but one thing still bothers me: how did the Republicans keep losing both houses of Congress every year from 1955 to 1981? Wouldn’t they have shifted left during this time? I don’t know, this was before our current partisan era, and there were lots of weird deals around civil rights going on.
I think the way to understand this is that US political parties were not meaningfully coherent entities. They were brands, but lacked sufficient overt hierarchical control or more decentralized perceived-unity-of-interest-within (by voters, influence, politicians) to model as a single interested actors making strategic choices. Today, the two parties still mostly lack overt hierarchical control (although Donald Trump may be changing this for Republicans), but perceived-unity-of-interest-within is strong, so they do approximate strategic actors.
Partisan outcomes before 1980 were more side effects of an "all politics is local" (or factional) world. Shifting coalitions within (and across) the parties were the strategic entities. Whether the result was a "win" for Democrats or Republicans was a by-product of all that, not orthogonal by any means, but not strongly correlated either with the players that mattered more in those times.
Some good political science that describes this shift is Lee Drutman, "Breaking the Two-Party Doom Loop".
The parties were stronger in the sense of being able to choose their national candidates. But that's all they could do. All those delegates at all those conventions were pursuing local power, primarily over local concerns, often in ways that copartisans in a different geographies would find abhorrent. Dixiecrats and North Eastern liberals shared little but a party. Goldwaterites and Rockefeller Republicans the same.
What do you mean to suggest? The graph shows the two parties were lots more similar to one another than now, when you average through them. (It'd be fun to see a breakdown of factions for Republicans as much as Democrats).
Party differences were less than now, cross-partisan factions more frequent. Party membership was much less predictive of behavior and alignments than now, even though yes, national parties could select their candidates more strategically.
I think what this misses is the median voter does not have median *opinions*. The median opinion on abortion is that it should be legal, probably up through some threshold time like viability or sometime in the 2nd trimester, and legal after for the health of the mother. And the median opinion on climate change is that it's real and the US should try to greenify energy production but also make gasoline as cheap as possible.
The median *voter* thinks abortion should be illegal always and we should outlaw coal tomorrow. Or that abortion should be legal in all cases no questions asked but also wind should be illegal because it kills birds. And while sure there's some theoretical multi-dimensional hypercube median voter you can reduce to a number line, it's not clear that adopting the median opinion on every issue is the way to get the median voter (even setting aside the areas you discuss).
I think one of the big thing that has changed, Ezra Klein's Why We're Polarized is great on this, is that previously, political parties and people were less aligned across issues than they are now. We know that abortion being legal is the left-coded opinion and that cutting taxes on the rich is the right-coded opinion, but there's nothing necessarily connected between those two issues. You could imagine a country with a far-left economically, liberation theology Catholic voting bloc with strong pro-life views but also strong income inequality & progressive taxation policies. It's just not the society we happen to live in.
Polls used to show that voters opinions on diverse range of issues were less ideological and less correlated than they are now. One fascinating factoid I remember is in the 90s, Democrats and Republicans (I don't remember if controlling for race or not) had indistinguishable views on the OJ Simpson trial. Compared to now (or recent history), and there are huge polling differences between the parties on whether American Sniper or Selma should win Best Picture. A person's opinion on abortion used to tell you very little about their views on immigration; now it tells you a lot.
Abortion was never broadly legal in Nicaragua, as far as I know, including during the Sandinista era in the 1980s, but yes they toughened up the laws in 2006 or so (banning it without exception, including in cases where the mother's life was at risk), and yes the Sandinistas did support that.
Abortion is also illegal in communist-ruled Laos, although as a general rule, worldwide, "Left" political parties have been more aligned with feminist values, and thus more likely to favor abortion rights at least to some degree.
I think immigration is actually a better example: questions of nationality and ethnicity really do seem to me to be completely orthogonal to the economic left vs. right, whereas you can at least draw a reasonable case why someone who cares about socialism should also care about women's rights (starting with the historical note that 19th century socialists often were critical of traditional gender roles and expectations).
> I said above that the real world doesn’t follow the Median Voter Theorem. But is that true? Certainly both parties don’t have indistinguishable and perfectly centrist platforms. But here are the last twenty years of election margins of victory
This is not well stated. As the image itself states, it shows the share of the national popular vote going to each party.
But the margin of victory isn't the difference in number of votes you get. That would only be true if you could win by getting more votes. The margin of victory is the number of votes that would need to change in order to change the winner.
This drew a lot of commentary in 2020, when Biden got 5 million more votes than Trump did at the same time that his margin of victory was 0.08 million votes. (See e.g. https://www.washingtonpost.com/graphics/2020/elections/vote-margin-of-victory/ ; take a look at that URL.) Differences in popular vote totals are just an epiphenomenon. Winning by 80,000 votes out of 158,000,000 votes cast is an absurdly thin margin.
> The common-sense answer is that it would take a while, there would be a lot of negotiation, but five or ten or thirty years later the Republicans would get their act together, shift left, and start winning elections again.
You know, one of my favorite bits of video game dialogue comes from a platformer in which the protagonist leaves a conversation by pleading "pressing business far, far to the right of here".
Left and right, with their total lack of any objective difference, are good labels for political parties.
This example is a case where the assumption of one-dimensionality makes examining the problem pointless. You need a different model. There is zero chance, in the scenario you describe, that the Republicans respond by "shifting to the left". Rather, policies will move around between the parties, and "left" and "right" will then be defined by whatever state the party platforms end up in. Juicing the political power of Guam, Puerto Rico, and Washington DC should end up pulling both parties to the left (as defined today) on economic redistribution at the same time it pulls them both to the right (...) on social conservatism.
Polarization might have fucked with this. I have a relative who was as 10% off median as you can get; 60/40 split D/R. Real traditional god bothering farmer type but also very concerned with environmental issues.
In 2000, he was very excited to vote for Mccain, but history took the stupid path and we got dubbya. Next comes 2008, and after 8 years of dumb, venal administration here comes Mccain again! And his running mate is who it was.
At this point he hard committed to only ever voting D until team R de-Palinized. He made a solemn vow: If St, Francis (R) was running for dog catcher, he was voting for Caligula (D).
I think the 2008 election was an inflection point where the parties stopped sharing a common reality, and any median voters that still exist only exist because they are low information. Unless there is some sort of correction, it will probably get worse over time instead of better.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
This isn't true. California did elect a Republican Governor, twice, while being a solid blue state. It's fairly common for solidly red or solidly blue states to elect governors of the party opposite their overall lean. This happens less frequently with Senators, but still happens a nontrivial amount (with Democrats generally being more successful at getting elected Senators in red states than the other way around, related to your observation about how the Senate should be stacked against Democrats, but doesn't actually have a Republican lean in practice).
Schwarzenegger is the only recent example in California only because one state doesn't go through Governors that often. The current Governor of Vermont is a Republican, and the Governors of Kansas and Kentucky are Democrats. Go back to some random year in recent history, and you'll probably find more examples like this that were in office at the time.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
In previous decades, southern Democrats could run under a national label while adjusting their campaigns to match the local tendencies. Why can't Californian Republicans adjust in the same way?
Perhaps it's that state-level Republicans want to "graduate" to the national party.
But I think it's mostly tribalism. Republicans in Cali would rather maintain their alignment with the national Republican tribe then take power by severing that connection. Nobody wants to be Joe Manchin or Larry Hogan.
The conclusion is that it wouldn't work, as Californians would still vote for the official Democrat even if the California Republican party took to nominating Democrats.
I think it's because the large cities outweigh the rest of the state, which tends (citation needed) to vote Republican. To capture the urban vote, the Californian Republicans would need to move so close to the Democrats' positions that they'd probably lose the rural/non-urban vote, and wouldn't be very Republican by national standards.
In which case, why vote for ersatz Democrats when you can vote for real Democrats?
I suspect the answer to the civil rights era question about Republicans shifting to gain congressional control is that because there was less partisanship, they cared about *conservatives* controlling congress, not *Republican conservatives.*
I believe DW-NOMINATE still showed that the primary axis of political disagreement at that time was the one that roughly mapped onto party labels rather than regionalism. So they most likely cared more about having Republicans in charge even if they were northern moderate Republicans rather than having a southern Democrat.
Part of the tricky thing about applying this theorem, I think, is that there is a qualitative difference between the median voter and the strictly partisan voter. The median voter is not a guy who has an equal balance of right/left positions or is equidistant on the issues from each extreme. The median voter is a guy who maybe has an elementary, abstract understanding of like one or two issues (usually acting as signifiers of moral character in a particular candidate rather than a schema for policy) and is voting based mostly on how much money he has in his pocket after a 2-4 year run. Strictly partisan voters in America tend to be more engaged.
I think that the main reason that the median voter theorem does not apply is that nothing is ever one-dimensional in politics.
Consider people voting on how spicy a pasta sauce should be. The MVT applies, you have exactly one axis, "concentration of capsaicin" and both the spicy and the bland party compete for the median spiciness voter, so he gets his wish.
Even for a sauce, this would not be realistic. People will want different amounts of garlic, mushrooms, animal products and all kind of stuff, so you quickly go to a high-dimensional space where the MVT does not apply.
I think politics is better modeled as building a coalition of various special interests. In two-party systems like the US, the coalition building happens before the election, while in proportional representation systems the special interests are kinda represented by parties, which can negotiate coalition agreements after the votes are cast.
The dynamic which pushes the coalition size towards 50% for either party would be that if you have a secure majority, strive within the coalition becomes the main political battlefield. After all, the other members of the coalition do not exactly share your interests, and if you managed to kick some of them out, you could enact more of your policies. If you are polling at 75%, that will increase the expected amount of political impact your special interest group has. If you coalition is polling at 45%, you actually need their votes to win, so it is in your best interests to make concessions to them.
I think that you can sometimes see similar effects in interstate anarchy international relations as simulated by Paradox games. It would be in the security interest of every polity to conquer everyone else, so they can feed on their riches and don't risk getting conquered themselves. However, they generally lack the military power to accomplish that. So they form (and sometimes, break) all forms of political alliances. While I am not an expert here, I think that a typical outcome is that you get blocks of roughly equal size.
For example, if you have an island with three polities with powers of 50, 30 and 20, the natural blocks would be 50 vs a coalition of 30 and 20. 50 would have little to gain from entering an alliance with either, its security interests are met adequately, and if its enemies were to war among themselves, why would it stop them? Likewise, 20 will recognize that an alliance with 50 is foolish: after 50+20 defeat 30, 20 will find itself in a very unfavorable situation. (Of course, election politics are only vaguely similar. In particular, while a military defeat can critically weaken a country for decades and thus shift the power balance, defeating a special interest group at the voting booth will generally not shift the power balance. Sometimes, the opposite is true: while Roe was the law, being pro-choice just meant defending the status quo of 1973, which did not seem to hinge on any election outcome. With Dobbs, moderate pro-choicers are more likely to align their votes along that issue because they want to change things back to the way they were.)
Or consider two local polities (A and B) in a long-running conflict with two large blocks (0 and 1) expanding their areas of interest to cover them. If A gets support from block 0, then B would be foolish to try to also get support from 0. After all, if the block has done its homework, it likely has sound strategic reasons to prefer an alliance with A, and to convince them that B would make the better partner is bound to fail. The better strategy would be to form an alliance with block 1, which has a direct interest in seeing the efforts of 0+A fail. In election politics, this is analogous to an issue becoming partisan. For whatever reason, the pro-lifers (A) become part of the Republicans (0). Then the logical step for the pro-choicers (B) is to become part of the Democrats (1).
Of course, real life in both international relations and domestic politics is obviously more complex than all that, with shared history, cultural, religious or ideological ties gating which alliances are feasible and which are not.
Looking it up, the origins seem to be obscure. Wikipedia says it was popularised in Rome in the 1950s-60s when spicy food was popular. Another site claims it was invented in Sicily in the 1920s.
Since most "this recipe was invented by X in year Y" lore is invented or muddled, I'm going with "this was a new recipe invented in the 50s to cash in on a fad".
Depending on how long we want something to be in existence before it becomes 'traditional', then Arrabbiata may or may not be 'traditional'. Is a Hawaiian pizza traditional Italian cooking? 😁
Regarding your last paragraphs, it's possible for one party to get stuck in a bad equilibrium where they can't win running on their current platform, so everyone who's pragmatic/a hack leaves for the other party, leaving only people who are too ideologically tied to the current platform to be willing to change it to make the party more competitive. This persists until the party collapses and is replaced by another, a coherent group of outsiders who like a weaker version of the party's ideas do a hostile takeover and make the platform more pragmatic, or there is some major change in the political environment which suddenly makes the previously hopeless platform viable again.
The recent convergence is probably driven by improvement in data quality. In reality, everyone's making very approximate guesses at where the median voter is and where potential party nominees are relative to that median voter. In the last 30 years, compute has become vastly cheaper, allowing ever-more-powerful analysis of fine-grained data to determine exactly where that median voter is and how to target optimally. A lot of the lopsided elections of the 20th century simply stemmed from bad estimation of where median voter is.
There's also a dynamic of investing in the future: trying to move the median voter to you, rather than the reverse. A party will implicitly or explicitly lose an election taking a principled stand in order to grow the movement and win in future elections. Goldwater in 1964 can be understood this way.
Your diagram shows 50% Republicans and 50% Democrats leading to a candidate at the 25th percentile along the partisan axis against a candidate at the 75th percentile. Unfortunately, things are worse than this. A whopping 50% of American adults aren't members of either party. If we assume these people are mostly in the center and closed partisan primaries we get candidates at the 12.5th percentile and the 87.5th percentile. There's a whole political advocacy organisation focused on allowing unaffiliated voters to choose a primary and vote because having 25th and 75th percentile candidates would be SO MUCH BETTER than where we are now.
I don't think much more than 50% of American adults even vote in most elections. Unless you mean 50% of registered voters aren't members of either party?
> Democrats and Republicans don’t have platforms exactly identical to each other and to the exact most centrist American.
Sure they do!
They both think that slavery should be illegal. They both think that capitalism is better than communism. They both think we should be a democracy rather than a monarchy or theocracy. They both think that women should be allowed to own property and that marital rape is a crime. They both believe in closed borders. Etc.
The two parties *feel* radically far apart to someone stuck living under them, but this is just the standard emotional reaction to inter-factional conflict (see the old Emo Philips joke).
Sure, they disagree on... check notes... which sports team trans kids should play on, I guess? And a few other narrow wedge issues that are specifically chosen to be things that are close to 50/50 splits in the population, bimodal distributions where the 'centrist' position is itself unpopular or incoherent.
But out of the space of all possible platforms, or even just all police positions that actual political factions have taken across human all human cultures and history? The venn diagram of the two party's policy platforms is indistinguishable from a circle, at any level of zoom larger than an electron microscope.
(another way of saying this is - the theorem implies we should have a narrow Overton Window, and we in fact do. Whether the parties are close together or far apart depends on what scale you are using, and is not well correlated to how close or far they *feel* to a partisan.)
The voters in the two parties disagree on the distribution of scarce resources in society. One party want most of it to go disproportionally to "traditional Americans" and the other party wants it to go disproportionally to marginalized communities.
Yes, that's true. However, I think most political parties attempt to adopt platforms that appeal to the most voters. Hence, the platforms of the Republicans and that of the Democrats broadly reflect the priorities of their base.
> Instead, Democrats are often pretty far left, and Republicans pretty far right. What’s going on? I think at least three things.
None of these three points are wrong/objectionable in their own right, but I think you are missing a central point here, which the WIkipedia entry mentions at the end of the first paragraph: the theory assumes the use of a Condorcet voting method.
First past the post is not a Condorcet method.
From the intro to Duncan Black's original paper on the theory:
>To develop our theory, we must make some further assumptions. Our major assumption will be that each member of the committee ranks the motions in a definite order of preference, whatever that order may be.
The theory is premised on ranked-choice voting. It does not straightforwardly apply to first past the post voting, as we use in the US.
This is why, as the wiki says, modern political scientists often talk about the median voter PROPERTY rather than the median voter THEOREM. The median voter property is possessed by any voting system which selects the candidate closest to the center of public preference. Many voting systems achieve this property, but the system we use in the US does not.
The general idea is that this is one of many properties it is desirous for a voting system to have, and it is used as a point to argue for or against different voting methods in political science.
I disagree with the framing of the parties' actions as product of some rational thought process informed by the median voter theorem. It's much simpler than that: politicians' most important skill is to accurately read the room, and they are doing it constantly. It's just a constant feedback process.
Take Roe V Wade being overturned: to argue whether it's a R+0.1%, D+0.1% or a D+2.0% is necessary to do the calculus, but that's not what went down in the GOP's elites' minds. Instead, it was more like "polls say it's unpopular, media coverage is terrible, even Fox isn't defending it much... unequivocal support only by our crazies. Well, that went worse than expected. Ok, drop "fetus is a human life" and go with the "states' rights" angle"
Incidents are individually consistent with a theory do not prove it. Seeking them out is just an example of confirmation bias. Instead, scientifically we seek out examples that can DISPROVE a theory. This idea of falsification goes back to Popper.
Which was all hunky-dory when the decisions were favouring the popular causes of the left. Once the court stopped being an arm of the Democratic Party, suddenly it was a national disgrace that it didn't represent the demographic makeup of the nation and there should be term limits for judges etc.
I see it as an artefact of neoliberalism: both parties essentially agree on all the important factors (i.e. being neoliberal). This explanation works even better in other countries, none of which have the divided two-stage (primary) election system. This is also why the median voter theory breaks down prior to the neoliberal era.
Edit: I reckon my argument doesn't need to define neoliberalism to make sense, but since that's what people seem to want:
Neoliberalism is return of liberal ideas, previously in vogue from about 1860 to 1930, when socialism became the predominant ideology of those seeking to improve society with *new* ideas (conservatism always exists and seeks to improve society with *old* ideas, and isn't automatically wrong to think so).
Liberalism is based on the idea of the maximisation of the right of the individual, chiefly through two mechanisms: the removal of laws the restrain individuals from taking actions that do not obviously harm others (in the short term); and the strengthening of individual or corporate property rights as opposed to government appropriation or some common good.
In the US, before the coming of Maga populism, conservatism and liberalism were largely the same thing, because the US was established more or less on liberal principles, so that is the political tradition here.
Yes, the US has always stuck a lot closer to its founding liberal principles than most countries. But FDR's New Deal era did hew much closer to socialist principles while it lasted. The US wasn't immune to international trends.
Trump probably did move left when Biden withdrew, but did so by not announcing any policies as extreme as they would have been. It’s a bad look to shift course too visibly. But my sense is that people’s platforms evolve over time, as they reveal their policies a bit at a time and see the electorate’s reaction. And those changes would be shaped by how popular a party is. A politician would be more scared to follow an unpopular principle when they’re doing badly in polls. But I think this looks like doing more vs fewer unexpected things, not like doing unexpected things in different directions
Katherine Gehl and Michael Porter seem to support something along the lines of the collusion hypothesis, describing America’s two party dynamic as a duopoly benefiting the existing players in the multi-billion dollar politics industry. See https://gehlporter.com/
So I guess the question is how does the Democratic party gain from a challenge by Trump? Does he help them mobilize their core, thus preserving the Blue status of Blue states?
If that's the answer, will they ride this horse all the way over the cliff?
Currently I'm seeing something bruited about online that an ex-chief of staff (or something) from the Trump administration is saying that Trump expressed admiration for Hitler, and at this point I'm going "You know what? Sure, whatever"and I'm not even reading the stories because I can't muster up the energy to be outraged.
I had an exchange with a person on here about "What will Harris do once elected?" and their main (and only) point was "She won't be Trump. Vote for her because Not Trump. Whatever she does or doesn't do if elected, will be What Trump Wouldn't Do".
There's been eight years of constant whipping up hysteria about him, and I've always thought that ignoring him after 2020 would have been the best response: "Donald who? Oh, that has-been? He's not relevant to anything now". But no, apparently the best thing the Democratic Party can think of is "Vote for us, we're Not Trump, because did you know? He's the Devil!"
Right over that cliff all the way down to the bottom and then keep on going down till you hit China.
Classical economics is built on the 'rational economic agent' that moves always in the direction (in choice space) which maximises its utility. While this model has some predictive power, Kahnemann and many others showed that people hardly ever act like this. And hence economists cannot predict recessions etc. because their models don't reflect reality.
Social choice theory has the same deficits. Arrow's Impossibility theorem is fairly trivial mathematically (and might show up in an undergrad course on discrete topology or similar, but is otherwise never mentioned). It gets talked about a lot and somehow entered the popular consciousness, but it has hypotheses that are never satisfied, and so doesn't tell you much about what's going to happen in the world. The median voter theorem is a response to Arrow: under still unrealistic assumptions, you can satisfy the median voter. But only if the median voter is a mathematical abstraction (insert the old joke about the mathematician predicting the winner of a horse race by assuming the horses are identical spheres).
The US system is particularly archaic, complex and unrepresentative. (I notice that most of the responses to the post are pointing out details of the US electoral system that they feel were glossed over in the post.) There's an understandable desire to reduce it to a mathematical formalism which can be analysed. But this type of analysis throws all of the details that matter away - I don't see how it could lead to meaningful insights. The type of analysis that Nate Silver did in 2016 and 2020 was, to my mind, a more sensible way of understanding and predicting outcomes for the election. The voting reform movements in the US, particularly at the time of the last redistricting used some interesting mathematics to show that some districts were gerrymandered. But that's slightly different than this discussion.
I guess it's important to work out what the goal is of the political scientists. I think it's less to build models that describe or predict real world electoral systems (particularly that of the US). It's more like mathematical research: here are axiom and here are conclusions. The applications of that to the real world are problematic... but where the media interest is.
This is interesting as I have thought for years that when one party moves to a more extreme position, the default response seems to be for the opposition to move in the opposite direction, when it would make so much more sense to move either to the centre or to cross the centre into enemy territory a little. That way you keep all your own voters and pick up voters from the other side who aren't happy about the new extremist policies. Let's take an example just for the sake of argument. Let's say the Democrats promise completely open borders. Would the Republicans be better off promising closed borders and mass deportations, or better off promising limited immigration for well-qualified and vetted applicants? You can think up lots of similar examples.
1) Both parties in the US are indeed quite similar from an outside perspective. From a French point of view, they offer both slightly different flavour of ultra-libertarian economic policy for example. There is extremism and polarisation in the US but there is also a broad agreement around an extremely narrow overton window on many issues.
2) As is often the case, your viewpoint is very US centric - not your fault, but you would greatly benefit from a deep dive into the political system and history about half a dozen other democracies. And indeed the median voter theorem seems pretty applicable to countries like France (where the two main parties were so convergent that Macron basically replaced both of them with his own), Germany or the UK. Don't mistake the specific of the US system for deep laws of political science.
2) There is obviously a feedback loop between what the voters want and what the parties propose. Parties move toward the median voter but voters also gravitate toward the issues that are emphasized by their preferred party.
Blue states and red states create large constituencies that are against their party moderating because those specific constituencies have little to gain from it. What I mean is... if you're a California Democrat, why would you moderate? If you're a Kansas Republican, why would you moderate? Your party dominates the state you live in. Yes, during federal elections, this lack of moderation can hurt. But it tends to cancel out to some degree, with both parties suffering some from it. Republican governors of red states often have no real reason to moderate, and likewise with Democrat governors of blue states. And these are influential figures even at the national level (Newsom and DeSantis are good examples of this). The Republican politicians and voting blocs that determine the overall direction of the federal party are more likely to arise out of red states, and even when they don't (like Trump), the primary process forces these blue state Republicans to shift some to the right. Similar situation for the Democrats.
If you look through the history of American politics, you'll see many similar periods to what we have today here, particularly on a North vs. South basis.
Now, the median voter theorem is undermined in the opposite direction as well. Democrats in red states can feel a strong need to signal that they're good progressives even though they live in a relatively conservative part of the country. Republicans in blue states can be similar here as well.
So many charts and number lines that put Democrats on the right and Republicans on the left. 😔
Makes me wonder if subconsciously the people creating these charts are going “well of course democrats are positive and republicans are negative! Of course democrats are right!”
Something this simple isn't going to work very well in describing a recursive system. You have to factor in that voters are watching the polls, talking to other voters, and strategically calculating how best to organize their votes. You see this in the primary period discussion we always have over "electability" vs. doctrinal purity. If some set of voters perceive the other side as "going too far" (this describes both sides today) then there will be a tendency to match that with their own veer toward more extremism. If, however, they perceive that the other side is playing it safe for strategic reasons (presumably so they can veer more extreme after the election, those scoundrels) , then there will be more pressure within their own party to do that too. Note that these are popular *perceptions*, not realities, and so are susceptible to the full range of propaganda and misinformation that bubbles up every election cycle.
I am old enough to remember the era when everyone was complaining that there weren't real differences between the two parties. No matter who won, they used to say, you ended up with more or less the same policies. That ended with Reagan and Gingrich.
We are still playing out what happened in the 1990's.
“Isn’t this weird? In a world with so many numbers - 55-to-45, 80-to-20, 99-to-1 - all of the past three elections have been approximately 49-51. So maybe the Median Voter Theorem does work.”
I don’t see why this result suggests that the median voter theory works.
If both parties are trying to win the election, and they can respond to each others actions, we could expect that the margins of victory would be small, and would get smaller over time as parties get better at strategizing, responding to information, etc.
The top baseball two teams going into the world series tend to be roughly as good as each other, because a competition for who’s best at something is likely to give those results.
So sure, we see close results. This seems true in a world where the median voter theorem works, and in one where it doesn’t. If, say, the best strategy to win elections was to make the most delicious cookie, we would expect both parties to get better and better at cookie-baking until they got pretty close to a 50-50 match. If the solution was to run the tallest candidate, we’d expect them to find the tallest and second-tallest candidate.
If the solution was, say, a really complicated mix of getting votes out, and persuading people, and catering to median votere, and also fringe voters, in a careful balance - it does not seem hard to imagine that both parties might work equally hard at that very complicated strategy, and we’d see very close elections, no?
Close elections seem compatible with *any* strategy, not just median-voter.
So though I understand everything you're saying here. None of it makes sense to me in the current political climate. There is no middle position anymore. It's all extremes, and to try and live in some middle position, is to get attacked from both sides. And trying to have some middle position is mostly not allowed.
Sometimes it seems like we're being played by the politicians. It's the midterms and we need to raise more money, so you member from the other party start some abortion bill, and I'll start a gun bill, and then you can fund raise trying to stop my gun bill, and I'll do the same with yours. We both win... more money. The perfect play for this is when it's all split 50/50. Then 'everything' is always close to going the other way. 'everything' is in quotes because it doesn't include everything, just those things that most divide us. We never try and solve the important things... the budget/spending, big business, pharma, defense, too much money in politics!
I mean Donald Trump was the guy who said "hey we should veer left". He dropped the most conservative and least popular republican policy: cutting social security and medicare in order to balance the budget. It worked pretty well!
TLDR: As the US demographically splinters, voting behaviour decouples, big electoral alliances form that for game theoretic reasons each encompass around 50% of the electorate and the presidential nominee is one factor in this alliance formation.
--------------
Let me explain in more detail! It's not the parties that choose a candidate that appeals to roughly 50% of the population.
The candidates are downstream (and interdependent) with the culture war dynamic that splits the American public roughly in half.
Earlier in the 20th century the largest bulk of the electorate were white, protestant middle class, whose voting behaviour was very correlated. All candidates had to at least somewhat appeal to them if they wanted to be politically viable. Often, one candidate's message just resonated somewhat more with that demographic and this candidate landed a huge landslide.
As American society grows more racially diverse, marriage rates go down (decorrelating men and women), people move to cities and form subcultures, the electorate fractures into different alliances.
And these alliances have a habit to form about 50% each. Why? Because if you are a group in an alliance (e.g. black women, LGBT, Mormons), your ideal alliance has 51% of the power, so to win the spoils but share them amongst the smallest group. And in a two-party democracy, political power often just means voters.
When one group becomes too influential, the members at the lower ranks can feel the heat. You can see this in the progressive alliance who started questioning the merit of the "white, male cis gays" or the "straight black men are the white people of the POCs". Why are black men moving towards Trump?
So the culture war just preselects the electoral alliance. The pres. candidate of each party is a part of that culture war. Their selection cements but also shapes the alliance that the party represents.
So why are there so many undecideds? Because a lot of people don't neatly fall into one category and they're not actively part of the culture war. They just get served two unappealing candidates through the dynamics of the culture war and terminally online people who then wonder how there can possibly be any undecideds.
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”.
I think that Trump did this in 2016. Not on the left/right spectrum but obviously politics is more complicated than a one dimensional spectrum. But Trump was significantly less culturally/religiously conservative and significantly more economically populist than previous Republican candidates.
Seems relevant that polling has improved a lot (in particular finding more representative samples and correcting for ways in which a sample is not representative) and many previous landslide elections were surprises (often to everyone involved I think). This suggests that the weak form of the median voter theorem where every election is pretty close, because the candidates are about equidistant from the median is holding more true now because the candidates now have a better idea where the median actually is. In the past there was a lot more uncertainty on this, some people guessed where the median was wrong (due to say bad polling) and got destroyed in the election.
> Parties and candidates seem to do a suspiciously good job staying equidistant from the median voter, far beyond the pathetic amount of effort they explicitly put in
I suspect the real answer here is something akin to Adam Smith's "invisible hand" - which is to say, democratic elections are a free market (of sorts), and free markets are very good at producing results that appear coordinated and well thought-out, but are actually just the result of all of the actors individually acting in their own best interests.
RE: research on the topic, I'm not sure *precisely* what you're looking for, so it could be that you're already doing this, but one reason that you might have had trouble could be because most literature on the topic, especially when modeling party competition, cites Downs instead of Hotelling or the descriptive model names ("median voter", "spatial"), so if looking for "median voter" specifically you'd probably only get less than half of the available research. You'd get things like Jones et al. (2022), of course, but the search would probably not find papers like Dodd (2015), which as far as I can tell do not have the words "median voter" together. Grofman's work is a bit older (2004 to 2006) but it also covers a lot of relevant scenarios (and just due to length, a 300 page book is going to cover much more than a 30 page journal article). Though, I believe Grofman does mention "median voter", I'm not sure how well that would show up in search results.
Jones, Matthew I.; Sirianni, Antonio D.; Fu, Feng (2022-02-02). "Polarization, abstention, and the median voter theorem". Humanities and Social Sciences Communications. 9 (1): 1–12. https://doi.org/10.1057/s41599-022-01056-0. ISSN 2662-9992.
Dodd, Lawrence C. (April 2015). "Congress in a Downsian World: Polarization Cycles and Regime Change". The Journal of Politics. 77 (2): 311–323. https://doi.org/10.1086/680041. ISSN 0022-3816.
Owen, Guillermo; Grofman, Bernard (2006-05-11). "Two-stage electoral competition in two-party contests: persistent divergence of party positions". Social Choice and Welfare. 26 (3): 547–569. https://doi.org/10.1007/s00355-006-0087-1. ISSN 0176-1714.
Adams, James F.; Merrill III, Samuel; Grofman, Bernard (2005-03-21). A Unified Theory of Party Competition: A Cross-National Analysis Integrating Spatial and Behavioral Factors (1 ed.). Cambridge University Press. https://doi.org/10.1017/cbo9780511614453. ISBN 978-0-521-83644-9.
That weird period between 1955 and 1981 is unlikely to ever repeat itself again. The Democrats had been the party of the South and the Republicans the party of the North. The parties really didn't represent progressives and conservatives, respectively - rather each had their own progressive and conservative wings.
That meant that Republicans couldn't win in the South, no matter how conservative. Additionally, conservative Democrats in the South wouldn't get kicked out of the party, no matter how conservative.
Political scientists at the time decried this state of affairs - without some degree of partisanship, you can't have the median voter theorem. Or, put differently, if I can't tell whether the Democrat is on the left or the right, how am I going to vote? Thus political scientists in the 1950s started calling for *more* political polarization. And we are living in the world they imagined!
Several people have brought up that the distribution is actually bimodal and I think the argument is not getting enough traction / is shot down too readily.
If the population falls into two disjunct clusters of roughly 50% it is absolutely possible to be the closest candidate (of two) to roughly 50% of the population without being in the middle.
Now this only works with what we might call perfect polarisation but I find that to be congruent with every observation we make. Viewed from the outside the US seem to have managed to perfectly polarise itself.
In fact intuitively it would seem easier to polarise the population than to be at the exact center of opinion. Getting marginally closer to the center when your opponent follows the same strategy appears to have quickly diminishing returns. Polarising the marginal voter you way when your opponent has the same strategy by contrast is a positive feedback loop.
I can see the counterargument that this would allow one candidate to move to the other side of the center and getting votes from both clusters. But that is completely unrealistic. Real people, when sufficiently polarised, would rather not vote or give their vote to an outsider candidate than to vote "technically closest candidate with chances of winning but wrong side of the center", no?
To loop around, I don't really understand why the bimodality argument is so quickly shot down in the comments. Wasn't this the blog that had the "scissor statement" post? Am I missing something?
I've always thought this phenomenon of "both parties approaching the center" was true, but I've never written it up, given it a name, or read too deeply into the existing literature about it. But I think it is true due to something most of us rationalists have trouble understanding: Politicians and activists are just going on feelings, or to use a more recent term, "vibes." They don't consciously move toward the center. They just start to get a vague, collective feeling that, if they don't, they'll lose. Think of religious Republicans and gay marriage. They probably still hate the idea of it, but they've just sort of vaguely given up on fighting that battle because they know they won't win.
> [ES: I am not a political scientist. I’ve tried to find good political science on these topics and failed; maybe you can point me in the right direction]
I would suggest starting with the Wikipedia article you linked, which states in the opening that the assumptions used aren't very reasonable—namely, the median voter theorem is a theorem about how majority-choice (Condorcet) rules act, and is limited to this class of fancy voting rules. There are generalizations that extend this to include strategic approval voting or honest score voting, but see the line that says:
> It cannot be applied to systems like ranked choice voting (RCV) or first-past-the-post at all, even in two-party systems.
In particular, both RCV/IRV and FPP have a polarized equilibrium, because the center is "too crowded" to hang out around there.
Neither party normally has a unified platform for congressional candidates.
The organizations which we call parties in the USA are long term coalitions of factions which would be separate parties in a parliamentary system.
The northern branch of the Democratic Party threw the southern branch under the bus when the Great Migration of blacks from the south to the north threatened their decades of single party rule. They found other ways besides segregation to neuter the newcomers.
Upward mobility among their constituencies threatens their coalition. They have doubled down on the grievance politics in an attempt to retain minorities while not alienating their biggest single constituency; government workers.
One thing I feel is missing in a lot of these discussions is that there are different goals for politicians and while public choice/rational voter are useful models they are not complete. Many politicians have sincere ideological commitments (e.g. https://www.vanityfair.com/news/2017/03/paul-ryan-medicaid, I can't find a link but the Democrats who voted for the ACA knowing they would lose re-election). A party wants to be the largest coalition that can accomplish the goals of it's members and the smaller the coalition the less conflict there is between different groups (e.g. there is no policy that the Democrats can take on Israel/Palestine that will not cause them to lose votes from part of their coalition and both parts of the coalition would prefer for the Democrats to kick out the other part).
Also because people can't remember anything before 2016 in 2008 and 2012 the electoral college favored the Democrats.
The bubble and bias theory on why parties don't pursue the median voter efficiently.
many in the party apparatus are too cloistered and blind to honesty understand the median voter.
additionally, many of the party faithful are even more blind, making it much harder to pursue Vs the party supporters.
source: I've witnessed discussions in the opposition in the following countries: UK 2016 to present. US left. Israel anti Netanyahu block 2015 - present.
it's quite obvious that the opposition in all those cases were ridiculously oblivious to the median voter views and preferences, and multiple times gave up on practically open goals due to such blindness.
I don't have time to read all of the comments so someone may have already mentioned this, but I believe the problem is in the voting system (first past the post). There are types of voting systems deployed in the real world that have resulted in all candidates being essentially centrist mirrors of each other and far less adversarial.
I feel like an analysis of this topic without digging into why the mechanism of voting seems to have such a big impact will result in potentially incorrect conclusions.
Two-party systems sometimes result in hiccups that delay the return toward the median voter. I gave examples in a comment above of such delays from Britain 1979-1997, New Zealand 1984-1993 and the US 1993-2020 (but still ongoing).
Multi-party systems can show more stability across time, due to the stabilizing influence of “center parties”, who control the crossover vote in parliaments. A center party is a party that can make an allegiance both to the left and the right, without alienating its core of loyal voters. Center parties can be small but none the less control the political game, since they control the median vote in the national parliament.
But proportional representation (which leads to multi-party systems) is no guarantee for median-voter politics. That is only the case if the center parties are moderate parties - as they are, for example, in Scandinavia. If the center parties instead are religious extremist single-issue parties, multi-party systems may instead give a boost to religious extremism – as seen in Israeli multi-party politics.
Another possible source of collusion: it's not the parties per se that collude, but the political campaign industry. Most of the campaign messaging I see is along the lines of "this election is really really close, we desperately need your money". Some of that money presumably ends up in the pockets of campaign employees. And because it's an easily measurable signal, those employees are probably in large part *evaluated* on how much money they bring in.
Basically, the campaigns are trying to raise as much money as possible, and only secondarily to win the election.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
But it does happen. Schwarzenegger was governor of California. Christie was governor of New Jersey. Giuliani was mayor of NYC.
I reached some of the same hypotheses, but also including irrationality of candidates/parties, and donations/activism declining with a veer towards the center, even if individual voting behavior does not.
I also had a different but similar visual choice for the graphs. :)
The House is an unusual case, because of gerrymandering, and there are some rather extreme gerrymanders in some states. In principle, a gerrymander can give you a majority of the representatives with only 34% of the popular vote. It also helps that Democratic voters tend to be more geographically concentrated in urban areas, making gerrymandering easier. An individual politician in a gerrymandered district, especially a "safe seat" held by the minority party, is going to win the general election anyway, so they only have to appeal to the median primary voter, making the national party seem more extreme. Additionally, oarties also have ways (some underhanded) to discourage primary challengers against incumbents. (For example, according to what I've read, if your company or organization works for a primary campaign against an incumbent Democrat, the rest of the Democratic party will blacklist you and will never hire you or your organization to work for a campaign ever again.)
The way I model it is that candidates are typically pulled between the base and the center. When they start to look like they're losing, the base gives them permission to move towards the center. But the second they look like they're winning, they start facing pressure to slide back towards the preferences of their base (which in practice means many of their friends and family). Thet aren't actively seeking equilibrium, but the natural result is both candidates will sort of bob around 50% all the time
I know nothing about political science, it is the first time I read about the median voter theorem, but still I'm arrogant enough to strongly dislike it.
Firstly, it is one-dimensional. Multi-dimensional spaces are surprising, you cannot rely on intuition tuned for one-dimensional space to reason about them, even 3-dimensional intuition can fail you badly with n-dimensional spaces. But the issue is deeper, I don't see easy way to generalize the theorem to N-dimensional space because dimensions will not be equal by their strength. People will prioritize them somehow, they will stop considering some of them because some other are more important for them.
Secondly, I generally do not believe in "average" or "median" people, because attempts to rely on them before failed spectacularly. Mathematically speaking issues relying on "average person" stem from multidimensionality, but even without mathematics people are so vastly different, that "average" or "median" person just a non sense. But I suspect with the median voter there will be more complex problem, because they are not form a normal distribution in n-dimensional space. I think there are a lot of clusters, which tend to remain clusters. The drift from one cluster to the other is possible, but mostly people stick to their cluster. I don't know will it break the median voter theorem by itself, but read the next paragraph.
And thirdly, this theorem ignores completely that there is no fixed median voter. What is important for people depends a lot on what they hear, on what important for other people around them, it depends on external factors which are not fixed for a lot of people. Yes, some deeply partisan voter would live in their informational bubble and would not hear anything that can shift their priorities. But some people would read news, and each piece of news will shift their priorities, maybe just temporarily, only for five minutes, but it will. Candidates can _change_ median voter by shifting their priorities of what is important and what is not. Now if people are grouped in clusters and clusters tend to drift on a "political spectrum" (or rather in the political multidimensional space) as a whole, then what?
All in all, my uneducated opinion that the median voter theorem looks more like "it is the only mathematical thing we can tackle" not like "it is the model that grasps all the important facets of the reality". Social sciences use such models sometimes, and one of the signs that this is probably the case is the impossibility to apply this theorem to the reality. You cannot know the multidimensional space of preferences of voters, you cannot measure politicians and place them in the multidimensional space, so the question is: how did scientists set up the experiments to validate the theorem? In my experience it pays to be vary every time when you see a mathematical quantitative model that cannot be tested on real data.
I wouldn't want to underestimate the role of simple miscalculation. For example, it sure looks like Harris thought she was winning and was just afraid of messing it up somehow. Turns out this was very wrong. There's going to be a lot of that in every election. Everyone has their theory of where the median voter is, but we don't really know until we see who showed up to vote. Often those theories are wrong.
(1) catholics won 12/20 of the US elections, lost 2/20 and then 5/20 are ambiguous/split (i think 2020 is actually ambiguous/split. fox says 49D/50R as opposed to others that says 52D/49R.)
(2) assuming the 5/20 are wins and assuming 2000 & 2016 count as wins considering the popular vote vs electoral vote thing, catholics have won nearly almost all of the US elections post ww2 namely 18/20 as predicted by the median voter theorem - the remaining 2/20 are 1968 & 1952. this means catholics have a 14 win streak starting 1972
(3) catholics are 20% of the US but bipartisan. protestants as a whole are 40% and always republican. atheists are 30% and always democrat.
non-orthodox jews, jews as a whole, atheists, muslims, non-abrahamic religions are always democrat.
haredi orthodox jews, mormons, evangelicals, christians as a whole are always republican.
i think mainline protestants, modern orthodox jews and maybe orthodox christians are bipartisan. idk.
I think having two parties equidistant from the center is a stable equilibrium, for some distances, so it can arise without the two colluding. For a toy model, say that your chance of winning the primary is proportional to your distance from the party's center and your chance of winning the general is proportional to your distance from the electorate's center, as a ratio to your distance and your opponent's distance. So, with 0 being extremist and 1 being perfectly moderate, a position of 0.5+d means your chance of winning an election is (1-d) * d/(d + opponent's d). Which is stable with d=1/3 on both sides, I think.
And this isn't restricted to just the party elites and politicians themselves. Many voters will vote for whoever they like the most in primaries, but many will question their general electability (see Clinton vs Sanders). So I think this fits and makes a lot of sense.
It's also important that the voters aren't actually evenly distributed, they're something more like bimodal. Or arguably trimodal on a two-dimensional model, along the lines of DW-NOMINATE, where there's an economic dimension, and a second dimension that roughly maps to being socially progressive and egalitarian, or protective of traditional hierarchies.
Historically, Southern Dems were lefty on the economic dimension (they liked the New Deal programs, and stuff like the Tennessee Valley Authority), but right-wing on the social dimension. There are still a lot of voters like that, and they like Trump, who offers right-wing red meat on culture war stuff, while also saying he won't cut Social Security and Medicare.
Modern Democrats are mostly lefty on both the social and economic dimensions.
There's also a tranche of hardcore traditionalist voters (like the Christian Dominionist types) who are right-wing on both. Trump won in 2016 by holding together the hard-right faction, and the econ-left social-right faction. Biden beat him in 2020 by peeling off some of that econ-left social-right faction, and perhaps somewhat by goosing turnout from the left-left faction, some of whom stayed home had been mad that Hillary wasn't Bernie and hence stayed home (or even defected to Trump, because they thought burning down the system sounded good).
Unfortunately, there are only the tiniest sliver of voters who are socially left, but economically right -- neoliberals, libertarians, etc. In favor of lower taxes, and thinking about social insurance from the perspective of avoiding large distortions (so, like, we should get rid of a lot of disparate programs that create weird income cliffs, and just do a UBI and a negative income tax). This is why the distribution is trimodal instead of tetramodal. Sigh.
There is, BTW, a nifty tool for modeling the electorate and candidates' positions relative to the electorate, that was created by Ka-Ping Yee (who is probably known to some of the folks who hung around Berkeley Rationalist circles in the aughts).
http://zesty.ca/voting/voteline/
http://zesty.ca/voting/sim/
One example: gay marriage. Obama could not even run on supporting gay marriage because it was believed it would make him unelectable. 8 years ago, you probably couldn't win most elections, even in fairly red areas, opposing gay marriage. As the tides seem to be shifting more right-wing lately, I see a definite undercurrent of opposition rising.
Also: Abortion. Overturning Roe is a liability in many elections for Republicans so you see most of them distancing themselves from it or proposing more moderate positions, in contrast with their prior stated views.
> Overturning Roe is a liability in many elections for Republicans
I'm not sure about that. Democrats love to make that claim, based on a small sample size of outcomes in low-turnout off-year elections that can be easily swayed by deep-pocketed interest groups mobilizing single-issue voters. But because of the low turnout, off-year elections very commonly produce results that are quite different from the preferences of the electorate in general.
Pro-life policies have historically been a winning position for the last 50 years or so, and every time some state-level policy restricting abortions that was duly passed by the people's representatives, after they got elected promising to restrict abortions, got overturned by judicial review on "this violates Roe" grounds, — in suits brought by activists of the party nominally claiming to be in favor of democracy! — it brought more and more people around to the realization that the root-cause problem that needed to be fixed was Roe itself.
A couple low-turnout elections are not going to hard-reverse half a century of ever-increasing momentum. If Democrats believe otherwise, they're in for a rude awakening.
The median voter wants neither no abortion nor abortion on demand.
Abortions for some, miniature American flags for others!
There is just no way democrats should have had any sort of midterm in 2022 that wasn't a catastrophe and even in deep red state pro-choice ballot measures are winning. It's crystal clear that many people who were "pro-life" only was long as Roe was enact.
There's nothing "crystal-clear" at all about the result of low-turnout elections.
I think the results of the 2002 midterm, in which Republicans gained seats the year after 9/11, are pretty clear.
What's crystal clear is that those "low-turnout elections"(midterms) determine control of Congress half the time. But then, who cares about control of Congress, right?
The marketing of the "return to Roe" faction has been very strong. When they polled this issue in the past, most voters landed at some compromise short of what Roe mandated. A 20 week ban like the one at issue in Dobbs doesn't seem that threatening, there aren't a ton of abortions after 20 weeks and nearly all of those would qualify under common exceptions, so nobody rational should be imagining "oh gee what if I need an abortion and can't get one", bc those circumstances are very unlikely. I don't think people are actually afraid they won't be able to have 3rd trimester abortions, I think instead that the Roe revivalists have managed to convince people that the prohibitions are broader than they actually are, and as a result gotten them to vote for maximalist initiatives that guarantee later access to abortion than anybody ever said they wanted.
Even though the "this is a private health decision between a woman and her doctor" framing has been around quite awhile, I'm noticing a lot of mostly non-political women picking it up and echoing it who wouldn't have said that 10 years ago.
Appealing to single women voters will be an interesting test of the median voter theorem, as the GOP obviously can't meet single women entirely on the social safety net, other safetyism like gun control, or abortion policy (not without just becoming the Tories and losing any reason to exist) -- but the GOP can sure ramp up "strangers are killing your children and dogs" and whatever anti-crime backlash is coming in the nest few years, if that's all that's left that's what we should see all the focus being.
>I think instead that the Roe revivalists have managed to convince people that the prohibitions are broader than they actually are
I think the crux of it is that, while there's only a minority of the electorate that thinks an abortion of a healthy baby at the moment of birth should be permissible, if people think there's any real chance that an abortion restriction could apply to themselves or someone they care about -- including in situations that aren't actually an abortion, like causing doctors to make bad decisions in cases of a miscarriage -- they'd rather just have no restrictions on abortion at all.
Those cases don't even have to be true, because who has time to investigate? The people in the fuzzy middle just have to think there's a real chance they could be personally affected to decide whatever vague pro-life convictions they have aren't worth the risk. Which is a pretty easy sell for the pro-choice side, low burden of proof. In this way I think the trend towards continued atomization and cultural disintegration will favor that side politically.
My understanding is that this is in fact true? For example, https://www.bmj.com/content/386/bmj.q2073
ETA: my prior is that this is almost certain to be true - in the US specifically - because if you actually let doctors carry out abortions whenever they might be necessary to save the mother's life, pro-choice doctors will abuse that to carry out unnecessary abortions, and the pro-life factions writing the laws really really don't want that to happen. (As the saying goes, "the ability to make a decision is inseparable from the ability to make the wrong decision".)
Abortion referenda are on the ballot this year, will you update if pro-life loses just like it has lost in the off-year elections?
Yet another dude downplaying the importance of reproductive health/control in the lives of American women.
Umm... you *do* know that the pro-life movement has always been driven predominantly by women, right?
You do know that women aren't a hive mind, so some subset of them can work against the interests of other subsets, right?
Yeah, like the ones who empower the worst of men among us by pushing, in the name of "autonomy," for policies that make it easy for bad men to indulge their worst impulses and pressure women into making the consequences disappear.
Yes, some women, like some men, live and vote by moral principles, and some live and vote by pure selfishness. In a remotely humane and decent society we'd all agree to only give respect to the first group. What truly terrifies me is that people like you seem to actually want more respect for the second!
"Ethically screwed up" doesn't begin to cover it.
Yet another person who's been born downplaying the importance of not being killed in the womb.
And euphemistically describing as "reproductive rights" acts of brutal violence that will never happen to you.
Thanks for the mansplaining! Have a great day :)
I really hope you're a troll. If not, how does it feel to be a literal sociopath?
The kind of "person" who could be confronted with an act of violence against a child and not only make no effort to justify it but actually respond with a cheerful "have a great day" is a person for whom I'm not going to begin to list what they dessrve to have happen to them. Just use your imagination.
One day, in a more humane time, people will look at monsters like you the way they look at the ones who cheered on slavery, and shake their heads in inexpressable disgust.
Hopefully, this will happen in your lifetime and you'll face some real justice.
People who are killed in the womb don't care about having been killed in the womb, due to being dead. And their next of kin seem to care less than in other cases of manslaughter. As do their non-related acquaintances.
At the same time we don't treat the killing of other creatures of equivalent or greater intelligence as crimes. Instead we treat it as bacon. This makes it hard for me to take secular moralist arguments against abortion seriously from anyone other than vegetarians. Are you a pro-life vegetarian?
National opinions have shifted since Roe was repealed. Sort of a thermostatic effect, combined with declining religiosity.
I don't think opinions have changed, just the status quo has changed.
A few years ago, the median voter's preferences on abortion were somewhat to the right of the overall national status quo. Now the median voter's preferences on abortion are somewhat to the left of the overall (measured somehow) national status quo.
But it's one of those rare cases where the voters didn't change their minds, the status quo changed, and without any direct intervention from politicians.
I think the issue is that the median person has inconsistent opinions about Abortion. They think having an abortion is bad but are very opposed to actual stories of people being denied care.
Nobody's actually proposing a moderation of state-level abortion bans except Trump, who won't say specifically what he actually wants. (He only said Florida's ban was too harsh.) And now the GOP has decided to do this, being incapable of waiting three weeks until after the election:
https://www.forbes.com/sites/alisondurkee/2024/10/17/gop-states-are-still-trying-to-restrict-abortion-pill-mifepristone-in-court-heres-how/
They're really eager to make the GOP into a regional party for left-behind areas.
"left-behind areas" -- awesome play on words ;)
You make an interesting point. Yes, some years ago, no mainstream politician dared campaign with a pro-LGBT platform. However, a lot of society has shifted leftwards, and now LGBT issues are fine. You say that tides are shifting towards the right - but this is just a rather slight correction of the leftward shift over the past years. Yet people on the left somehow think it is some sort of earthquake.
I'll point to trans issues as an example. Totally off the table 10 or 20 years ago. Became widely accepted. Now the right wants some limits, such as prohibiting trans (MTF) participation in women's sports. Small correction to a large leftward shift.
...It's not going to last. Once the right is in power, they will absolutely crack down on the presence of trans people. There's no reason for them to tolerate such mockeries of the sexes.
One reason would be that the median voter wants them to tolerate the presence of trans people (if that's indeed the case). "Tolerating the presence" is a pretty low bar.
A majority of Americans still think changing gender is morally wrong. https://news.gallup.com/poll/1651/gay-lesbian-rights.aspx Even the readership of left-leaning papers like the NYT barely try to hide their disdain. Sure, they claim the problem is "social contagion" and "compromising women's rights," but it's pretty clear that it goes deeper than that.
The only force that is forcing everyone to pretend to accept the presence of transgenderism is the left. The allure of Trump is that he has shown people that you don't need to pretend to be tolerant of loathsome things just for the sake of "civility". The left will be dealt with, and with it will go transgenderism.
From your lips to God's ear.
I hear you saying that; I don't hear anyone influential on the right saying that.
https://www.theguardian.com/us-news/2023/mar/07/cpac-anti-trans-rhetoric
> Last week, the Republican governor of Tennessee signed into law a bill prohibiting gender-affirming care for minors as well as one imposing new limits on drag performances, which have become a target for Republicans. Mississippi also enacted a ban on treatment for transgender youth while Republican state lawmakers in Kentucky advanced a similar measure, following a charged debate over a separate proposal allowing teachers to refuse to use students’ preferred pronouns.
Until recently, most legislation banning transgender healthcare was aimed at minors, but Republicans are increasingly pushing proposals that would limit treatment for adults...
...But the speech that LGBTQ advocates found the most chilling came from Michael Knowles, a rightwing political commentator for the Daily Wire, who declared that “for the good of society … transgenderism must be eradicated from public life entirely”. A range of voices, including public officials, experts and observers of rightwing rhetoric, condemned the remarks as inflammatory and dangerous, with some calling them “genocidal”. (Knowles insisted on Twitter that he was not referring to trans people, but “transgenderism” which he has described as a “false” ideology.)
Yet the intense focus on transgender rights at CPAC this year – nearly every speaker raised it – suggests it is likely to be an animating issue in the coming presidential election.
"crack down on the presence of trans people"
How so specifically?
I will point out that during a part of the Trump presidency, the right was in power in presidency and both chambers of Congress. If they rounded up all the trans people then, I must have missed that.
I would argue what really broke the (almost) 1:1 mapping between gender roles and biological sexes was effective contraceptives. Before they were a thing, it was the fate of perhaps 45% of the post-pubescent population to get pregnant and then (because of various reasons) end up raising their children. Gendered behavioral archetypes (the pushy suitor, the chaste maiden, the tastefully-hot woman, the caring mother, the providing father) come from this time.
Thankfully, with contraceptives, the mapping between gender roles and sexes broke down. Humanity hacked its reward function. Today, most sex in first world societies is non-reproductive, an evolutionary imperative we follow even though its original purpose has been lost. What remains of genders is likewise an vestigial echo of our evolutionary past. In accordance with that past, most will prefer to have sex with people presenting as one gender (perhaps in a specific life stage), or having one set of secondary or primary sex organs, or perhaps even having the corresponding genotype.
In 1700, being a woman (e.g. presenting as a chaste, yet hot maiden so that you would attract a high-quality husband) was a very different role from being a man (e.g. learning a trade or going to war to earn enough of a fortune to be an acceptable husband). Today, little of that difference remains. Our sex chromosomes no longer determine our lives to a comparable degree.
The way I see it, gendered titles and pronouns are almost as vestigial as noble name prefixes. Where in 1500, the difference between Mr von Grauen and Mr Grauen might be the difference between dying in a duel or getting beaten up when giving offense, and the difference between Miss Grauen and Mr Grauen was the difference between wooing (or raping) someone and slaughtering them on some battlefield, the days of nobility are just as gone as the days of strong gender roles, and good riddance to either.
I am a straight, cis-by-default guy. I don't really care what sex bits most of the people around me have in their pants, be it dicks, pussies, nothing or tentacles, pubes or shaved. If I foresee the possibility that I will be interacting with said sex bits of a person, I will make tasteful inquiries, but I actually tend to fuck just a small fraction of the people I encounter. If a few people feel really strong about belonging to some medieval class or gender role, let them acquire their von or Mr or Ms or whatever. Like gay marriage, it does not hurt me in the least and seems to make them happy, so I am a-ok with it.
Personality differences are LARGER between the sexes in Western Educated Industrialized Rich Democratic societies. https://www.overcomingbias.com/p/klyde-the-barbarianhtml
While I think a long-term right win would bring issues, I'm skeptical that they'll crack down in a truly significant manner without some major disruption (Republicans win the next several elections, Trump somehow manages to perform a coup). I don't expect a major disruption like that.
I expect they'll most likely crack down on even more controversial topics, like puberty blockers, which earns them applause from their base which has far less resistance from our existing laws about people doing what they want with their body.
I heard much of the same during 2016 about a variety of topics and it simply didn't happen.
There does seem to be a shifting tide on trans issues where the NYT is now publishing things that horrify activists. But gay marriage seems cemented in place.
I think you're being entirely too charitable and bipartisan. Many voters and politicians make their decisions simply because they are evil.
How does this contradict or affect the Median Voter Theorem?
The median voter theorem is a statement about how rational political parties would act based on an assumed one dimensional political spectrum.
First of all, it's a model, and the onus is on a model to describe real life, not the other way around.
Second of all, "evil" (as well as "good") is a mechanism for explaining why some political parties did not play rationally - for example, why the house stayed democratic for many decades.
I believe many politicians were not able to move along the ideological spectrum to optimally play the median voter theorem, because their individual values prevented them from taking optimal positions.
However, it is worth noting that while the house did stay democratic for several decades, the Republicans eventually won in the long term, and have achieved many of their goals since taking the house under Gingrich. Barry Goldwater may have lost his election decisively in '64, but his ideas won in the long term. So perhaps we shouldn't be too critical of Republican strategy during that period - they won eventually.
Part of why I want to bring in "evil" here is that part of politics is moral, and I don't believe you can completely reduce it to simple math.
If we say for the sake of argument that the axis of politics goes from good to evil, this does not contradict the median voter theorem. It would just result in both parties being equidistant from the median moral voter, such that half the country was better than him and half worse.
Also, remember the orthogonality thesis, evil people need not (in the game theoretic sense) be irrational
The game-theoretic sense is clearly not the one they're using, "irrational" here is being (ab)used to mean "not maximizing electoral success".
Then you need to also bring in "good", and might as well call it "values" as you do in the comment above: that certain politicians or whole parties will not shift as much as they "should" from the point of view of optimal electability because they're *constrained* by those values, beyond the obvious "we can't give up on position X even tho it's a minority position, because it will lose is the most faithful and generous supporters".
Elections aren't simple games, they're iterated games. One reason for not playing an optimal median voter election, is that many individuals do not simply have inherent preferences somewhere in the left right spectrum, but are influenced by what positions political parties have taken in past elections.
Many voters are "leftists", "rightists", or "centrists", *relative to the positions of the two political parties*. This is "Overton window shifting". So, it may be worth not playing to the median voter in today's election, because that same voter may change his politics to be "more centrist" in future elections in a way which pushes policy in a particular direction.
>I believe many politicians were not able to move along the ideological spectrum to optimally play the median voter theorem, because their individual values prevented them from taking optimal positions.
This can explain why an individual politician didn't move, but wouldn't that lead to them eventually losing to a candidate whose values/positions put them in a different place along the ideological spectrum?
The candidate generation process is not necessarily efficient - people who run for office aren't uniformly sampled from the electorate. Most people aren't able to campaign for public office - those who do are therefore different from the norm, and these differences are not uncorrelated to political opinions
Even evil people want to achieve their goals; losing elections doesn't help with that. Even if you're right that some politicians are too principled to moderate their positions for tactical reasons, I don't see why you'd describe that as evil.
How are we defining evil here?
What policy favored by Goldwater do we have now?
Shouldn't "evilness" increase the sway this model holds? And wouldn't "evil" people be more likely to outcompete "good" people on all levels of politics, after controlling for intelligence?
It just means you have to apply Conflict Theory more than you are currently doing. For example, you seemed surprised when Nancy Pelosi intervened to get Gavin Newsom to veto the AI bill. You shouldn’t have been.
Please. One of the biggest problems facing the West today is the inability to discuss and understand each other's positions. Your opponents are not wrong, they are *evil*. This is not a productive way to conduct politics, and is something we need to work against.
How can you so confidently rule out "evil" as a motivation?
At the very least, "evil" is like "a wizard did it". It means you're not interested in any further explanation or exploration.
That may in fact be true; you may already have pierced their arguments and decided there is no merit to them. But most people haven't done that.
Evil people aren't even trying to make arguments, they're trying to use power to hurt people.
That's not something that's about debate, that's a naked use of power. They either have the power to hurt, or they don't
This sounds like something you would need to explore if you wanted to know whether it was true, no? And so I don't think that works as a rebuttal of what MM said.
To what ends? Sadism? You think there is a significant political faction of abstract sadists who get off on looking at a spreadsheet where they can see how many lives they have made worse?
Because "evil" isn't anyone's motivation, even very evil people. Even in cartoons, villians usually have a better motivation than "I want to be evil".
I think evil people usually fall into four categories:
1. Doing evil because they think it's good (classic Mao/Hitler/whatever)
2. Doing evil because they think it's justifiable because reasons (waah I had a broken home so it's fine if I steal these shoes), and
3. Doing evil because they're emotional and can't help themselves (classic domestic violence), or
4. Doing evil because it benefits them and they don't care (organised crime etc)
Nobody does evil because "hooray evil".
Part of what's relevant to designating something as "evil" is whether it's beyond the range of potentially convinceable by argument.
If Hitler is going to do the Holocaust, I don't think I can have a rational debate with him and convince him not to. Hitler's just going to do the Holocaust because he's Hitler, i.e., evil
But that would be true of any strongly held conviction whether good or evil. I don't know whether you're pro or anti abortion, but what are the odds of someone changing your mind about it with debate? And we don't know how hard it would be to persuade Hitler with debate either; I don't think he exactly had a lot of friends pitching, "hey Adolf, maybe Jews are actually great". You may be assuming that bad people and your ideological enemies are more rigid in their beliefs than your allies. People always think their own side is rational and the other side won't listen to reason.
And I can't convince you that there might be a justfiable reason to commit a genocide, so does that make you evil as well?
People can be rendered unpersuadable by delusion, mental illness, or manipulation by another agent. At the same time, a genuinely evil person often will be persuadable by appealing to self interest (e.g. a sadistic sociopath who gets finite utils from causing suffering can still be bargained with as long as they also like money). Unpersuadability is neither necessary nor sufficient for designation as evil of someone who does bad things.
As a time traveller, I think I could show to him that his Malthusian belief about Germans starving without more lebensraum was simply incorrect.
Over the last decade, my belief in "evil" as a motivation has steadily risen based on what I've observed in this country.
I'll bite. How are you defining "evil" (either with or without the scare quotes) here?
Enjoyment of the suffering of others
I don’t think people who disagree with me are evil, but I think the politicians they vote for are. I’m just here in the comments section trying to get them to realise it
Agree
https://slatestarcodex.com/2018/01/24/conflict-vs-mistake/
Are you saying that members of one or both parties wake up in the morning thinking 'how can we decrease the global human utility sum and turn our world into the torment nexus?'
If you think that, then you have a very shallow model of your political opponents. Sure, there are sadistic psychopaths who enjoy spreading suffering where they go. But these are relatively rare, and they do not cooperate well enough to dominate. (No, not even the NSDAP was majority psychopaths.)
Take abortion, a typical hot topic where people on either side portrait the other side as evil.
If a fetus has the same moral value as a teenage human, then forcing women to carry their fetus to term (until we get artificial wombs, at least) seems perfectly acceptable. We don't allow parents to murder their teenagers either, even if murdering them was really convenient because they make a great an organ donor for you.
If a fetus is just some tissue without intrinsic moral value, then forcing a woman to carry it to term and thereby creating a kid which has a worse starting hand genetically and socially than average is monstrous towards both the woman and the kid.
So the other side is evil only if your own assumptions are true. Granted, you will likely find the odd psychopath on either side, like a pro-lifer who does not care about unborn children at all, but really wants to punish women who had sex, or a pro-choicer who does not care about women, but just rejoices at any life ending, but these are minorities.
Counterpoint: Donald Trump specifically is pretty clearly a sadistic psychopath
The Republicans are so personalist at this point it's not really meaningful to talk about a republican party distinct from Trump
>Donald Trump specifically is pretty clearly a sadistic psychopath
It's not really clear at all. Boorish, boastful, philandering, greedy, lying, even foolish I could see arguments for. But what cause do we have to believe he is a psychopath, or a sadist? He certainly seems to experience the normal range of human emotion, and what I've heard of his sexual exploits did not seem to prominently feature infliction of pain.
I do not think it is productive to have further discussions if you cannot agree on the above comment. Have a nice day.
If that is your response, then I would have to agree: it would not be productive.
There are forms of sadism other than "inflicting physical pain in an obviously sexual context." The man's catchphrase is "you're fired," contextually meaning "with a momentary whim, in two syllables, I have destroyed what took you years to build, and most of your hope for prosperity in years yet to come."
Words have meaning, and if firing people on a game show makes you a sadist then the word has little meaning left. Trump is certainly no saint who humbly loves his enemies, but I don't see him as particularly taking pleasure in the pain and suffering of others.
> 'how can we decrease the global human utility sum'
Isn't classical evil more like 'how can I increase my own utility & happiness by any means including decreasing that of everyone else?" Whether it's loud or quiet, hidden as a usually unconscious motive or 100% embraced, I do think there's plenty of *that* out there. (And not just in politicians.)
I fully agree with the rest of your comment though.
I think that most people put a higher weight on their own utility than on the utility of a random person. In a way, this is okay -- a person who is just likely to buy any stranger some food as he is to buy his own dinner will starve in short order, which will seriously hamper his efforts to increase the utility of mankind.
I would define evilness as the quotient of utility weights between oneself and a stranger being above a certain threshold. For example, the evilness of someone who gets paid 10M$ to kill an innocent is much lower than the evilness of someone who kills innocents just to pass the time, like that gunner in Apocalypse Now ("How can you kill women and children" - "It's easy, you just don't lead them so much").
--
Unrelated, I think that 'my opponent is evil' acts as a curiosity stopper, just like murderism [1]. Sometimes this is appropriate: there are likely few insights to be gained from digging deeply into the prioritization the Nazis had for their genocides (unless they are still at large and you are wondering when your turn will be). On the other hand, while Jan 6 was imho clearly an evil act of Trump, there are myriad other ways in which he could have acted. "Evildoers do evil, and how they pick their acts is forever beyond the understanding of decent folks" seems to be a bad frame to understand politics.
[1] https://slatestarcodex.com/2017/06/21/against-murderism/
That type of evil wouldn't stand in the way of the theorem presented in the article though. Maximizing electability as an avenue towards personal power and safety still makes sense.
Obviously, "Many voters and politicians make their decisions simply because they are evil" is not self-explanatory, so you'll hsve to expand on that concept.
Personally, I would argue diametrically that noone thinks that they are evil, although from an outside perspective, their argument may look very thin. So I have problems to understand your position.
But if you want to convince others that politicians really do act out of a pure will to just hurt people, I'd have to see some examples. Please provide cases where you think this is the underlying reason.
If primary voters myopically pick their favorite primary candidate without considering general electability, then you'll always get a median-democrat and median-republican nomination regardless of what positions individual candidates adopt.
Ok, but those two would be equidistant from the actual median only when the electorate is very close to 50/50. That seems to have been true for a while, but I’m not sure I can explain why.
If the two parties' candidates are typically equidistant from the median voter, that also explains why a similar number of voters belong to each party and vote in its primaries, which in turn explains why their candidates are typically equidistant from the median voter.
In theory this is an unstable equilibrium though (under our assumptions that primary voters mostly don't vote tactically, and people vote in the primaries of the party they prefer). If a bigger fraction of the electorate belongs to one of parties, that should make its primary voters more moderate on average, which should result in its candidates being closer to the median voter, which should make even more people switching to that party. Though this is mitigated if, say, center-right voters keep voting in the Republican primaries even if they vote Democratic in the general election because they perceive Democrats as more moderate, or vice versa.
Yeah, the problem is the apparent stability of the 50/50 split. I always fear that the reason for it is that there *are* no median voters -- it's a bimodal distribution of groups that see no validity in the other group's positions, and never the twain shall meet.
I'd like to blame the primary system for this. Smoke-filled rooms have a bad rep, but at least the party leaders have a terminal goal of winning the election, which we gather primary voters often do not. But we have not had the knife-edge 50/50 split for as long as we've had the primary system, so if primaries are the problem the effect has been gradual and cumulative.
This assumes that the primary voters represent the general election voters more or less 1:1, which is generally not the case. Primaries, and especially caucuses, attract people who are politically engaged, and they tend to activist-y and have stronger, more choate opinions on politics.
Michael Podhorzer has a great write-up on his Substack wrt 2022: https://www.weekendreading.net/p/red-wave-blue-undertow
We can look at 2024 for some evidence against that idea. Trump currently leads in AZ polls by nearly 2 points. AZ Republican primary voters selected Kari Lake (~409k votes) for the 2024 Senate seat over Mark Lamb (~293k) and Elizabeth Reye (~38k). That works out to Lake winning 55.3%, all numbers per NBC News.
Per 538, Gallegos, the Democratic Senate nominee, is leading Lake by nearly 7 points. With races having tightened across the country, Trump has nearly 10 full points on Lake even after she won her primary comfortably.
The AZ primary is an excellent example of primary voters not considering general electability. Lake's weakness in the general was very well known in the few weeks leading up to the primary. GOP strategists trying to win the Senate were screaming about it.
The process for parties staying equidistant from the median voter doesn't have to be fully organized top-down. It can be partially an emergent phenomenon, in which candidates who do so the best are successfully elected, and candidates who don't, aren't. This applies to candidates for all positions - state, local, and federal. Some of these positions aren't term limited, so successful politicians can stick around (even when they can't stick around in a given position, they can leverage their success to move to different political offices). The collective of elected candidates from a given political party will therefore approximate median voter preferences. This collective then creates norms that other members of the party can default to. These norms will shift over time as the collective of elected officials shifts per shifting voter preferences.
This model could create an inertia that keeps electoral results relatively close, even in instances when a given candidate doesn't do as good a job of matching the collective party line. Since per this model, party norms develop based on the collective tendencies of successfully elected members of that party, the public can develop party loyalty and vote for the party candidate even if they veer somewhat from the norm. Of course, party loyalty isn't absolute, so we'd expect such candidates to lose some support. They'd then be less likely to get elected and join the collective of elected officials from their party, which then shifts the makeup of the collective slightly, to maintain equilibrium.
Party loyalty in the partisan era could probably explain the relatively even electoral results even if one doesn't adopt the aforementioned model.
1. If three reliably blue states are admitted to the union you still only need 54 senators (or 53+VP) to control the Senate.
2. Tony Blair became the person that said "We Labour politicians whould veer right" after 17 years of conservative dominance and it worked. He even successfully removed achieving socialism from the party goals.
Bill Clinton did the same in 1992. 90s Democrats were extremely conservative by most metrics we would use to measure that before or after his presidency.
Trump veered to the left, economically. Republicans used to be the party of the rich. They are not anymore.
He also veered left on a bunch of social issues, like the whole Christian agenda.
The only thing he's veered right on is immigration, and even that is mostly just a return to the classical consensus of "hey why don't we actually enforce the law?"
He’s used a lot of rhetoric that scans as right-wing, particularly about immigration - but some of his related rhetoric on international trade would have scanned as left wing a few decades ago. He’s just systematically anti-foreigner in the way he talks, which isn’t really right or left.
I guess that depends where you put nationalism on the spectrum. The far-left often present themselves as anti-nationalist, yet in historic examples they were willing to drop that stance and opportunistically veer nationalist. And nationalism usually becomes more and more pronounced the further right you go in modern terms, yet the original right, i.e. royalty, nobility and clergy with temporal power, were very much not nationalist and only occasionally played towards nationalist sentiments for gain or to boost general morale.
Republicans prior to Trump didn’t say, “I’m running exclusively to represent the interests of the rich.” It’s not a winning electoral tactic because even if you get the vote of every single voter in the top 1%, that won’t get you to 50%, which is what you need to win.
Once in office, however, Republicans tend to favor the rich. Trump’s major legislative achievement was a tax cut which primarily benefits the wealthy. During the campaign, promised to eliminate the carried interest loophole (which lets hedge fund managers play lower taxes), but he didn’t try to add a provision to the tax bill to do that.
Trump was more willing to impose tariffs than previous Republican presidents, but because of the marginal utility of money decreases with wealth, higher prices due to tariffs affect the wealthy less than they do people lower down on the income scale. So I don’t think Trump’s tariffs mean that the Republican Party is no longer the party of the rich.
"republicans aren't the party of the rich" is based on rich people shifting to net-support for the democrats, see eg https://www.cambridge.org/core/journals/perspectives-on-politics/article/polarization-of-the-rich-the-new-democratic-allegiance-of-affluent-americans-and-the-politics-of-redistribution/E18D7DAE3A1EF35BA5BC54DE799F291B. It's obviously sensitive to exactly how you define rich.
RE tariffs - typically, at least part of the rationale behind them is to protect domestic manufacturing (existing or emerging), and more manufacturing jobs should be beneficial to the working class.
So won't count it as policy benefiting the rich...
Who funds the tariffs? Rich CEOs? Foreign countries?
Domestic end consumers, mostly.
A minor quibble; electoral college advantage isn't static, but relates to the composition of the two parties' electoral coalitions (it tilted towards the dems as recently as 2012), so the proper electoral-advantage-adjusted measure isn't national popular vote minus a static adjustment, it's margin of victory in the marginal state.
Yes this is correct. And it will show an even closer margin in the last 20 years than 2% I believe. At least 2016 and 2020 tipping point state differences are less than 1%.
Yes, I made the same point elsewhere in the discussion, I think we all agree that the margin in the tipping point state is the appropriate view. To save others the effort, here’s a link to the table on Wikipedia: https://en.wikipedia.org/wiki/Tipping-point_state#List_of_tipping-point_states_by_election
If I calculated correctly the average in the years 2000 to 2020 was 3%, mostly because of Obama’s decisive victories with tipping state margins of 9.0% and 5.4%.
Yeah I was kinda surprised Scott applied that lazy 2% hack rather than just showing the margin in the tipping point state, especially since the table already includes which state that is.
Yup, the electoral college favoring Republicans has much more to do with 'Winner Takes All' and much less that small states get extra electoral votes that is usually the go to explanation. The effect of that changes as the parties coalitions change (big driver of it getting more favorable for Republicans lately is the increasing education polarization). Small states are surprisingly relatively evenly split (RI,VT,HI, DE etc. are all tiny and D dominated for example). Nate Silver has talked about this a couple of times.
Right, the big problem with the electoral college is the winner takes all nature of it. If states allocated electoral votes roughly proportional to the vote shares, the electoral college would be more or less fine.
The ME/NE system might be a good nearer-term change target, if for no other reason than that it has already been tested by real-world implementation.
I agree!
The big worry about that method is just that it supercharges the value of congressional gerrymandering.
It could, or it might reduce it (if the partisan-optimal gerrymander for Congress differs from that for the Electoral College).
Even if it did, it'd still be better than either the current system or a national popular vote.
Why would it be better than national popular vote?
Do you by chance have a link to one or more of Silver's discussions of this? It's not an idea I've run across before, and I'm interested to learn more.
https://www.natesilver.net/p/mad-about-the-electoral-college-blame and https://www.natesilver.net/p/its-2004-all-over-again cover some.
Thanks!
The problem that primaries in elections drive candidates away from the middle is called "Center Squeeze" in social choice theory: https://en.wikipedia.org/wiki/Center_squeeze
Another interesting social choice result (though less applicable to the real world) is the "McKelvey–Schofield chaos theorem". It says that if one uses a sequence of pairwise election to choose a winner, then someone can arbitrarily control who wins just by adding more choices and changing the order that the pairwise elections are held in: https://en.wikipedia.org/wiki/McKelvey%E2%80%93Schofield_chaos_theorem
"Maybe after decades of humiliation and failure you could convince someone, but I don’t see the history of parties suffering decades of humiliation and failure before finally agreeing to turn around."
You could look at the UK if you want to see that: Labour lost the 1979, 1983, 1987 and 1992 elections before turning around, choosing Tony Blair as leader and winning 1997 by a landslide; similarly, the Conservatives lost that 1997 election, then 2001 and 2005 before choosing moderate David Cameron and winning in 2010. The process since then has been complicated by Brexit, but Labour were out of power 2010-2024 and chose moderate Keir Starmer and won.
Brexit works in that it traded how the Tories picked up the parts that constitute a median voter. Johnson was also more 'left' on some issues (legal immigration, regional development, green investment).
Many recent UK elections have not been particularly close at all; the trend of all recent elections being very close in basically a US phenomonon.
Indeed: UK elections that have been close (since the consolidation of the Labour/Conservative two-party system in 1945): 1950, 1951, 1964, 1974A, 1974B, 1992, 2010, 2015, 2017. Not close: 1945, 1955, 1959, 1966, 1970, 1979, 1983, 1987, 1997, 2001, 2005, 2019, 2024. I don't see a pattern in them becoming more or less close in there.
If you assess this based on just the two party system it will make it very difficult to analyse- combined two party vote share varies massively. The recent election wasn't in any meaningful way similar to 1997. Labour's vote share barely budged compared to 2019 (and was down in absolute terms), and was massively down vs 2017. But they won a landslide mostly due to the exodus of RW/working class voters to Reform (who had nearly entirely voted Tory in 2019).
The high point of the two party system (1945-1970) saw mostly closer elections than the period after, with 1974 as a transition point. The Lib Dems acted as an amorphous centrist party that could peal votes off either group depending on the issue of the day. With the 2024 election seeing 30% of the vote going on new-ish left and right wing challengers on either side of the three older parties, suddenly we're in a slightly different world. Last time this sort of happened (2015) the Brexit referendum acted as a new position to triangulate around, and the old system re-emerged. This doesn't seem to be an option now.
Because you have this pressure valve of third/fourth/fifth parties it makes it hard to assess the relative merit of centrism of candidates. The Conservatives shifted back and forth on social and economic centrism (moving toward the median voter position in 2019 of centrist economics and right wing social values) but saw continuous vote share increases until 2024.
In the UK, whenever the Conservative Party sees itself losing votes to Labour or LibDems they always talk about "creating clear blue water" - ie going right to appease their core voters (who would never vote for anybody else anyway (I'm talking pre-Farage here)). Dunno the logic, if there is any: maybe they get more volunteer canvassers that way. Maybe it's just a stress response.
That looks like a standard reaction of people with very strong positions: If they do something extreme and it does not work, clearly it does not mean they were wrong, it means they weren't extreme enough.
I see it all the time, especially in politics. Noone is ever wrong, instead they either didn't explain their correct position to the voter enough, or they foolishly let moderates dilute their pure position, thus sabotaging it.
> If we assume small states tend conservative...
This part is clearly not true. A lot of the smaller states are in the northeast, which is traditionally much more progressive. The large states out west tend to be much more rural and conservative, with the obvious exception of California, and I guess the Seattle/Portland area. By the list of smallest states, we have:
Rhode Island, Delaware, Connecticut, New Jersey, New Hampshire, Vermont, Massachusetts, Hawaii, Maryland, West Virginia... ok, by state 10 we get conservatives. But there isn't some grand strategy where small states vote Republican because the electoral college favors smaller states. The popular vote gap is an artifact of the hordes of progressives in California, if anything.
Pretty sure that "small states" here is meant to refer to small population, not land area. That said, there's still a mix of red and blue in the least populous States:
Wyoming (red), Vermont (blue), Alaska (red-ish?), North Dakota (red), South Dakota (red), Delaware (blue), Rhode Island (blue).
> Pretty sure that "small states" here is meant to refer to small population, not land area.
Maybe that's what Scott meant, it would make more sense.
If you're smart enough to look at the ordering of states by area, you're smart enough to understand that, especially with talk of elections, people use "small states" to mean "lesser populated states."
You would hope so.
Also HI (blue), NH (blue-ish?), Maine (blue-ish?), Idaho (red) for your list (I think that hits all the states with 3 or 4 electors).
You're measuring size by area; you should be measuring by population if you want to understand the electoral college influence. But it's certainly true that there are loads of low-population ("small") states that lean left: Vermont, Delaware, Rhode Island, Maine, New Hampshire, Hawaii, for instance. Though there are similar numbers of small states that lean right: Wyoming, Alaska, North Dakota, South Dakota, Montana, West Virginia and Idaho, for instance (that's with a cut off of states with just one or two house seats). Equally, if you look at really big states, while California is blue, the next two (Texas and Florida) are red.
Your last sentence is right, though: the real reason for the electoral college advantage is that the Democrats win big states by big margins - California, New York and Illinois above all, while the biggest state that the Republicans win by more than 10% is Tennessee. That means that a significantly larger fraction of the Democrats' vote is "wasted" by piling up a big margin.
The obsession with small states also results in people misdiagnosing the Democrats' problem in the Senate, where the real advantage is in the middle: if you take the eighteen states from 5th largest to 22nd largest (Pennsylvania to Minnesota), they split 13-5 to Biden in 2020. Take the next eighteen (23rd to 40th, South Carolina to Hawaii) and they split 13-5 to Trump in 2020. That excludes the four large states (California, Texas, Florida and New York) which split 2-2, and also the ten smallest states, which split 6-4 to the blue side. (Note: in an election that Biden won by a 4.5% margin, he only won the states 26-24; if the presidency were determined by the same rules as the Senate, he would have won by the narrowest possible margin, rather than by a relatively comfortable EV margin)
The reality is less about size and more about population density. In the US, the less densely an area is populated, the more Republican it tends to be. This is almost fractally true. Rural states are more republican than more urbanized states, but their cities will be more democratic than the states overall, and the denser the neighborhood the more democratic it votes.
https://www.niskanencenter.org/wp-content/uploads/2019/09/Wilkinson-Density-Divide-Final.pdf
https://engaging-data.com/election-population-density/
1. This appears to assume that voters are rational and rationally vote for candidates based on their stated positions. I would suggest that voting is largely a cult of personality and thought-terminating cliches.
2. There is a theory in poltiical science that any first past the post winner takes all political system will eventually evolve into a two-party system.
For my part, if I wanted to implode either major US political party, the surest way to do so is to give them unfettered power. Team R and Team D need one another, otherwise there will be no excuses as to why we haven't reached Nirvana yet, and there won't be enough spoils to keep all the interest groups making up Team D or Team R happy, not to mention, some of those demands may be incompossible. For instance certain libertarians and religious conservatives may vote Team R, but their preferred policies are very different and there is no easy way to keep both happy, especially as there is no long a Team D to enforce unity.
> I would suggest that voting is largely a cult of personality and thought-terminating cliches.
I think that it's "largely" this, in that at least 50% of voters run this way.
But you just need a big enough minority of people in the middle who can say "y'know, that shit is just too crazy for me, I'm switching my vote" for the MVT to work.
(I love the word incompossible now. Thanks for teaching it to me.)
How does (2) explain Britain (very old FPTP system and not quite two party) or India (also FPTP and extremely multiparty)
Britain has been functionally a two *dominant* party system for the majority of it's modern history, with the exception of the interwar period, and the 1974-2015 period. It is true that any third party has been generally punished by the electoral system. But similarly to India, what sustained the Lib Dems was geographic differences where in certain areas of the country the party could sustain itself as either the number 2 or number 1 party (Wales, the South West).
What is happening currently in the UK is a breakdown along the lines of Scott's example where the two main parties become too centrist and people stay home (which he discounts as possible). But this is mostly what happened, except they didn't stay home, they voted to split the vote by going for third or fourth parties who *did* meet them on issues. The last time this happened in any large number (2015) it caused a major political realignment after the Brexit referendum, with voting coalitions shifting closer to a US style Blue tribe-Red tribe split. This collapsed in 2024 as White working class voters abandoned the Tories that (in their view) betrayed them, and young, often Muslim, voters didn't turn out for Labour in their 2017 numbers due to Israel/Palestine.
If ‘modern history’ means post WW1 (say), then your ‘exceptional’ periods make up two thirds of the total. Seems like the exception IS the rule.
IIRC, it was Labour and Troies, there were the Liberals and Tories, then before that, the Whigs and the Tories.
Lib Dems, and snp exist and win nontrivial numbers of parliamentary seats. Hung parliaments sometimes happen.
Yes, and Ralph Nader is alleged to have played spoiler in 2000. For that matter, Bill Clinton would be a footnote in history, but for H. Russ Perot.
I haven't looked in a while, but I thought Britain used two systems, one of which was national votes and one of which was local votes? The local votes seem to be very two party, while the national votes allowed smaller parties to get some representation.
Also, a few extremely geographically located parties that could win local elections on a very specific platform.
I don’t know what you are talking about. I’m talking about elections to the Westminster parliament which are via first past the post voting in single representative constituencies. There is no proportional representation. Third (and fourth) parties which win nontrivial numbers of parliamentary seats exist, and sometimes end up kingmakers in a hung parliament.
Yes there are two dominant parties, but if lesser parties get nontrivial parliamentary representation and sometimes end up kingmakers then I think this is meaningfully a multiparty system.
The analogy would be a US system where the libertarians and greens each had a dozen Congress critters and a couple of senators, and occasionally won a couple of states in the electoral college. This I would consider a multiparty system albeit with two dominant parties.
Also geography is a lousy excuse since US geography is more varied than Britains. If lib Dems can carve out a niche in the Southwest and SNP in Scotland, then why can’t libertarians carve out a niche in the mountain west and greens on the pacific coast?
For the "why": isn't the primary election system a major reason? Thanks to it, every somewhat mainstream ideology can hope to have some voice in at least one of the two major parties, so there's less motivation to form and vote for minor parties. If libertarianism or green politics are popular in some regions, they can just elect libertarian Republicans or green Democrats.
Perhaps the FPTP presidential elections also make a difference? They produce a national two-party system for the presidential election, and once voters align themselves with one of the main two parties for president, they mostly end up identifying with that party and don't bother to vote for a minor party in the other races either?
Yes, ‘FPTP plus presidential system plus nationalized politics produces a two party system’ is consistent with the data (or at least I don’t know of any counter examples). My point was just that FPTP alone is empirically insufficient, since there are lots of counter examples of countries that use FPTP but have multiparty systems.
There are some ways (regional accents for example) in which England is actually more geographically divided than the US.
Parliamentary vs Presidential systems.
This at a minimum is a necessary rider. Whether FPTP voting and a presidential system is sufficient, idk. As best as I understand, Brazil and Indonesia both have presidential systems and multiparty democracies, but neither is strictly FPTP, and arguably neither is mature. But then if we demand FPTP and a presidential system and a mature democracy, then maybe the US is the only example, so we are reasoning from n=1.
I can't speak to Britain necessarily, but the issue with India is that *at the state level* it's largely (not entirely, but largely) a two party system. The reason it's a multiparty system at the federal level is because the two-party contests are different in each state, and more generally because Indian states are really *extremely different from each other*, to a degree that dwarfs anything in the United States.
Indian states, for the most part, don't share a common language, in many cases they differ either ethnoracially or religiously, they don't share much of a common history until the British forcibly conglomerated them, and they are as far appart in terms of economic development as Russia and Burundi.
If you consider, say, the social democrats / liberal democrats in the Congress Party, the Tamil/Dravidian nationalists, the Communists, and the caste-based parties, all of these groups are fundamentally opposed to the Hindu nationalist narrative of history, and the latter three of them at least could broadly be considered on the "Left" (Congress was certainly a center-left party from the 1950s through the 1980s, but it's hard to consider them anything but centrist today). But, these four groups are never going to consolidate into a single party, because not only does each of them have a sharply different ideological basis and narrative of history, but each of them also has a different geographic stronghold. It would be totally counterproductive for each of them to give up their distinct existence at a party, when each of them has a reasonable shot of being dominant *at the state level*.
I don't think this is true? e.g. Uttar Pradesh (the largest state by population) appears to have three major parties (as measured by `have held the office of chief minister in the 21st century'), plus the Congress as a significant fourth player. Maharashtra (the second most populous state) also appears to have three major parties (by the same metric). The first two entries on the list both yielded multiparty systems so I suspect the thesis that Indian elections are two party affairs on the state level is false.
Maybe just for due diligence I will check state #10, seeing as how I have ten fingers. According to Wikipedia that would be Andhra Pradesh. Again, three different parties have held the chief ministership of Andhra Pradesh in the 21st century. OK, I declare the thesis that `Indian elections are two party affairs on the state level' to be false.
More likely, the basic thesis of `FPTP produces two party systems' is just wrong for parliamentary systems. Maybe it's right for presidential systems, but it's hard to say, because I think the set of large mature presidential democracies with FPTP voting has n=1.
Why is voting for personality not a valid preference as a rational voter, i.e. a voter that votes for what they want.
Even if some voters voted by flipping a coin that's still a valid preference - and rational. An irrational voter would be one that didn't have transitive preferences. I think the onus is on you to prove preferences are not transitive across the left-right spectrum. A voter who wants a far left candidate, will usually settle for a less left candidate given the choice. We see this in my own electoral system (Australia) with instant runoff and single transferrable vote.
Edit: To be clear personality is also transitive. If you're voting based on who you like the most, it is rational to pick the candidate you like the most, or if they aren't available, the next most likable.
For the same reason most of us learn not to hire likeable incomps, especially if we never personally will meet the hire.
You have a different defintion of rational that what most people in here do. A rational agent has a meaning and if you make a top level comment using another definition I think that's unhelpful for discussion.
By your logic, anything I like is "rational" because it appeals to what I like.
It's not very hard to satisfy rationality by the definition. I agree with you that you have rational preferences. Nearly all humans do.
Rational has at least two meanings.
> Even if some voters voted by flipping a coin that's still a valid preference - and rational.
> An irrational voter would be one that didn't have transitive preferences.
> Edit: To be clear personality is also transitive.
This is an odd collection of statements. Coin flip decisions aren't transitive.
I agree you can't create a mathematical argument about a fair coin being transitive but the point is about rational choice and preferences.
If you vote in *this upcoming election* by flipping a coin it means you have an identical preference for each candidate. It's just an extra thing I added to what I was saying. It's not the core of my argument. Yes, if someone lived their entire life as a random number generator they wouldn't satisfy rational choice but that's not very interesting in the context of this discussion.
The point is against this:
> I would suggest that voting is largely a cult of personality and thought-terminating cliches.
As evidence that voters are not rational. Which I think is wrong. Expressing a preference for personality is a rational preference by the definition.
Most notable FPTP lack the extreme bipartisan division in the US. You've got Libdems in UK, NDP in Canada, a host of ethnic/regional parties in various Third World countries (and, of course, BQ and SNP in Canada/UK).
What really causes America's very strident two-party system is probably a combo of FPTP and open primaries, which basically makes the two parties into placeholders for competition of group and coalitions that would form different parties in other systems.
I think it's also geography. Namely, the US has a lot of it - so much that there's room for multiple very disparate subcultures within, compounded with having had open immigration for so much of its history. (Immigration restriction is, I predict, bringing the US closer together, though there are also forces pushing society in the other direction.)
The UK mostly just has an island roughly the size of California. Canada has plenty of land, but most of it isn't densely populated (90% of Canadians reside within 100 miles of the US border). But even so, there's a lot of difference between British Columbia and Quebec, and sure enough, that shows up in their politics.
In the UK, Keir Starmer pretty much attempted the strategy of saying one thing to win the party leadership, and then switching. Yes, the party membership noticed and we’re Not Happy. Well, approx 50% of them were Not Happy, as you’ll expect from the median voter theorem.
Missing from Scott’s analysis of the effect of primaries… if the switch after the primary is big enough that more than 50% of your party’s supporters switch to a new third party, you lose. Rarely happens, possibly because they’d lose if they tried it.
That's basically the point covered by the Second... paragraph. Explicit party defection or not, alienating potential voters who are closer to you than your opponent works out as a turnout (dis)advantage.
I think it's implicit that this analysis applies pretty much only to countries that fall into Duverger's law and have a bipartisan system. Once you have multiple viable parties you're probably talking about multiple median voters and many different axis rather than just left-right.
Last UK election, the Conservative Party didn’t just have to worry about losing votes to Labour from their left, but also Reform from their right. (And Labour was against Jeremy Corbyn etc. to their left, and the Conservatives to their right).
Primaries are mostly American and mostly in the past 50 years. So there's even less reason for the Republicans to have stayed minority party -- they could have re-aligned themselves in a smoke-filled room without worrying about red-meat voters.
I don't see much credence in claims that the Median Voter Theorem isn't valid. What we in the U.S. see as extreme is only extreme as measured against our fairly narrow Overton Window compared to other countries. Extreme candidates pretty much never make it through the primary process. Let's look at the history: who was the most extreme major party candidate in recent (say last 50 years or so) history? And I'm talking about policy positions. I can't stand Trump personally, but none of his policy proposals are particularly extreme. His tariff proposals are bad economics, but hardly extreme given the U.S. history of tariffs. Truthfully, I can't think of a single candidate that has met even an honorable mention as an extremist. Reagan might be the most, but he governed as a right centrist and looking back on his first campaign from hindsight it actually appears pretty mild in retrospect.
The Median Voter Theorem is pretty much alive and well - and let's hope it stays that way.
> What we in the U.S. see as extreme is only extreme as measured against our fairly narrow Overton Window compared to other countries.
Yep, compare what I wrote at https://www.astralcodexten.com/p/secrets-of-the-median-voter-theorem/comment/73763567
However, I'm not saying that the US has a particularly narrow Overton window. Just that it's not the same as other countries.
Australia is notorious for being far closer to MVT results than the USA is, and we also have the conditions for the MVT to fully bind (compulsory instant-runoff voting, so the base *cannot* defect without explicitly falsifying its own preferences). Our politics largely actually run on scandals, specifically because the two big parties are so close to each other on policy.
My first thought about Scott's post was "What about places with compulsory voting?" If there's no incentive to focus on turning out your base, you have to go after the centrists instead. (Of course, you could have compulsory voting without first-past-the-post, and have a multiparty system instead, which complicates things.)
MVT doesn't fully bind for compulsory plurality voting, because extremists on your side can vote third-party to waste their vote. It's the combination of IRV with compulsory voting that makes "get out the vote" useless.
Another interesting thing about Australian politics is that independents can frequenly pick up seats by being even more centrist than either of the two parties. In the last election the loosely affiliated "teal independents" managed to pick up a whole lot of safe Liberal seats by giving off the general vibe of "nicer than the Liberals".
I remember some Australian complaining constantly about the teals being NIMBYs or something and I didn't know what he was talking about.
Loudness is often mistaken for extremism. People think that US politics is extreme because both parties spend a lot of time shouting very loudly about how the other one is so eeeevil and defeating them is vital to the future of the Earth and this is the most important election ever and how could anyone possibly think about voting for someone so eeeevil.
But it's very easy to be extremely loud without actually having any particularly extreme policy positions, and this is the situation of the US today.
If I could draw one of Scott's median voter theorem plots, it would have a red dot and a blue dot close to the middle, but each dot would be loudly yelling at its respective side "OMG THOSE OTHER GUYS ARE TERRIBLE". This is how you persuade the people way over on the left or right of the plot to turn out for you -- you fire them up with rhetoric about how evil the other guys are so they don't notice you're barely distinguishable from them.
It was noticed. https://www.youtube.com/watch?v=Ll3iyvbsRDM
Matthew Yglesias helpfully refers to that as "unhinged moderation": https://www.slowboring.com/p/unhinged-moderation
There are also often people who confuse being consistent for being extreme. Thus a relatively more moderate politician in a parliament who just consistently votes party line (and against the other party's line) would look more "extreme" than an actual extremist who ends up taking "odd" positions on some mainstream issues due to the particular nature of their extremism. (ie. if you somehow had a genocidal-level right-wing racist in who votes pro-choice because he believes this leads to less minorities or so on)
>But the GOP has about a two percentage point Electoral College advantage; that is, the popular vote total most likely to produce an electoral tie is one where the Democrats lead by 2. Adjust for this advantage (we’ll talk later about whether we should), and it looks like this:
You misinterpreted the Centerforpolitics.org piece on this you linked. The GOP has enjoyed a ~2% advantage *only in the most recent elections* and this cannot be projected backwards past 2016. The Democrats had the electoral college advantage between 2000 and 2016.
I agree with this which is why I only discussed the past three elections in this context.
Then you should probably only apply that modifier to the election table numbers for those elections.
I think what Scott really wants is to see how close the elections were and and how optimally the parties aligned themselves to the voters. I think a better approach than applying a modifier is to look straight at the margins in the respective tipping point states: https://en.wikipedia.org/wiki/Tipping-point_state#List_of_tipping-point_states_by_election
I’d argue that this is actually the true margin of victory. If I calculated correctly the average is 3% for the years 2000 to 2020, mostly because of Obama’s decisive victories with tipping state margins of 9.0% and 5.4%.
I think a challenge with the precise application of the median voter theorem is that voters don't vote solely (or even necessarily mainly) on policy. Parties also serve as identity groups, and a person might identity as a Republican or Democrat for many non-policy reasons. I think much of the increasing closeness of elections is due to people's politics becoming a stronger part of their identity, so they have become less likely to consider voting for the other party. This could have many causes, from the changing media landscape to the declining centrality of religious and local identities in American life.
But even with polarization, something keeps elections balanced and prevents one party from just securing a permanent lead by locking in the right supporters. I would guess that it is mainly thermostatic public opinion. When bad things happen, people tend to blame the party in power (without giving them as much credit for good things), and this nudges them toward identifying with the other party. So the longer a party is in power, the more supporters the other party picks up. At equilibrium, this means both parties have about the same number of supporters. Of course, the median voter still plays a role, but it's a dwindling number of swing voters who are truly up for grabs, guaranteeing close elections.
Unrelatedly, in the chart showing electoral college closeness, it'd probably make more sense to show the vote margin in the tipping point state (which mathematically represents the median electoral college voter, at least post-facto), rather than adjusting the national margin by some factor, because the electoral college advantage shifts over time and also doesn't have a single unambiguous definition.
> This could have many causes, from the changing media landscape to the declining centrality of religious and local identities in American life.
My hope is that it's downstream from the lack of construction, manufacturing, and other "atoms" jobs. People without material challenges anchoring their daily life get increasingly caught up in immaterial ("extremely online") zero-sum conflicts.
Energy is scarce enough that it must be used efficiently, meaning the few things which are still built here are obsessively optimized, "no user-serviceable parts inside." Those who didn't make the cut for initiation into one of those specialist priesthoods are sloshing around in marginal positions, make-work or not quite worth the hassle to automate, strategically irrelevant, nowhere to go but down.
Abundant cheap energy from solar and wind will change that, open up fields where personal initiative is a meaningful advantage against overengineered perfection. Give 'em something to gnaw on besides each other.
> If some people dislike Trump because he committed 10,000 felonies and an attempted coup, do the Democrats enter that into their calculations and veer slightly further left? If Biden is demented, do the Republicans enter that into their calculations and veer slightly further right?
Historically, both of these seem to have happened, with Democrats shifting left over Trump's term (to the point that Biden, who was still further left than any recent democratic candidate, was seen as the moderate option). And Trump may have decided to pick Vance as his VP candidate over options seen as more moderate in response to Biden's failed debate performance (this was before Biden dropped out). More big picture, the Biden admin was already fairly unpopular a year ago - I can imagine a world where Biden is doing well in the polls and Republican primary voters decide to go with a more electable candidate.
My mental model for the median voter theorem is that it's a pressuring factor towards equilibrium. There's a lot of other influencing factors (and just random noise in who your candidates end up being), so people don't end up at the exact median, but the further out they are the more strongly they get pushed towards it.
Now let's add a data point: what year did electronic voting machines become common?
This doesn't seem super complicated to me: People vote on vibes.
Biden was losing badly with a policy-driven, median-voter-tracking strategy in part because he wasn't campaigning well, but in part because nobody cares about that stuff or believes him (even though his campaign was mostly "truthful" in the sense that what they were saying was literally true).
Harris was able to make big gains by switching the conversation to "Republicans are weird." They're still tracking the median electoral voter with their policy positions but her public appearances are 100% about projecting a kind of solidarity with suburban moderates in her attitude, not her policy. They've taken a more laid back approach to challenging even Trump's most ridiculous claims, because the democrats screaming "He's destroying democracy!" seem equally hysterical to someone who doesn't live and breathe this stuff (note: I am a "He's destroying democracy!" liberal).
In addition, people in the social media age have absolutely *atrocious* understanding of what the "other" party actually believes. They don't even know the other party's platform. Often they don't know their own party's platform. So even if Republicans moderated considerably, my social group (probably including me) would still see them as extremists, because that's how the news and memes I'm regularly exposed to paint them.
The median voter theorem holds fairly well on simple, high-profile issues. But mostly people don't care about the issues at all - Republicans want someone brash, powerful, and spiteful; Democrats want someone "tolerant (TM)," technocratic, and non-threatening. And moderates want someone who doesn't challenge their sense of what is and is not normal in any way.
It's worth noting that this isn't necessarily wrong or irrational - even if Trump adopted every policy I support today, I likely wouldn't vote for him just because he's him. I remember 2020 when he made extremely spurious claims about election integrity and pressured local officials to distort the results of the election. That level of interference with an election outweighs all other potential issue for me, and I won't vote for someone who did it whatever their positions.
> In addition, people in the social media age have absolutely *atrocious* understanding of what the "other" party actually believes. They don't even know the other party's platform. Often they don't know their own party's platform.
I’d add that it has become effectively impossible to know either party’s platform, and this is particularly true for Democrats. Their platform weighs in at 92 standard pages, the Republican platform at 16. Both are inexcusably poor documents. Platforms have been increasing in size and decreasing in comprehensibility since at least the civil war.
If you want to know what each party really stands for it’s effectively impossible and it’s been made intentionally so, particularly in regards to the Democrats from the 1990s onward when they became notably less populist and more technocratic.
Even this can be a fracture point, because I agree with the premise but disagree with the cause. I see the issue as the nation's problems becoming increasingly complicated and the right pushing simplistic solutions that will obviously fail. In that framing, Democrats aren't necessarily obfuscating, they're responding to an obfuscating situation.
Though have to admit that I'm much more likely to see the flaws in things I understand than in things I don't, and that politicians often retreat into complexity to avoid criticism, so probably the answer is not that simple.
You make a good point. There’s undoubtably some truth to it. Though I think the Democrat’s shift from unions and labor and blue collar to higher education, technical expertise, and white collar has made the party so philosophically confusing that they need 92 pages to try to convince the voter that in the end it all makes sense.
Nevertheless I think I largely agree with you. The biggest divide between the average Trump and Harris voter remains education level, and the Republican Party’s effective strategy to appeal to less educated voters is clearly present in the tone, vocabulary, formatting and relative brevity of its current platform.
Coincidentally seen on the intertubes just this morning: https://www.experimental-history.com/p/ideological-turing-test
Most Americans, Democrat and Republican alike, can masquerade as a member of the other party well enough to pass an ideological Turing test judged by members of the other party. So maybe people's understanding of what the other party believes, isn't that atrocious after all.
This is interesting! I'd seen the opposite reported - at least that people who spend a lot of time on social media have poor understanding of the opposing party. But now I can find no source for that, and the reason it stuck in my head is that it's counter-intuitive. Maybe it came to me in a dream :/
But then why are elections close? Did these preferences magically descend from heaven split about 50-50?
There's lots of reasons for this but the easy, partially wrong answer is that the parties are following the median voter theorem, just for affect instead of policy.
I think the theorem has an implied "complete/instant propagation of information, or otherwise static information landscape" assumption. Removing that assumption, it reads more like "parties will attempt to steer towards what they perceive as the median voter, in order to gain/retain power". Wordier and squishier, but makes sense.
Yeah if you assume a lag in both learning information about the electorate and making changes towards policies you would expect to see something similar to what we see (e.g. having a polling miss in 2016/2020 lead Democrats to work less hard at appealing to the center because the center they were appealing to was wrong)
My pet theory is some of the weirdness of the real world vis-a-vis Median Voter Theorem relates to the fact that its less products being bought and sold (like Hotelling's Law for ice cream) but rather subscriptions being sold and bought (like AT&T vs Verizon, or Netflix vs HBO, but for 4-year terms and party identification).
Not sure how specifically it would do so but that must muck up the generic 'product discovery' process.
But people aren't buying products. They're voting, and everyone gets the same result no matter who they voted for (of if they voted at all).
I have two thoughts on this subject:
* First, don't discount the problem of measurement. Prior to the daily tracking polls, campaigns had to operate with much less data; they could reasonably be mistaken about whether voters perceived them as centrist.
* Second, I think you're on to something with your comment about reducing things to a single straight line. We know that high-dimensional spaces are awfully counterintuitive. Candidates have more options than moving towards and away from the center: they can "rotate" around the space to try to capture voters.
The curse of dimensionality is made worse because the election is a dynamic game, and issues that are important today (e.g. a candidate's alleged dementia or penchant for felonies) weren't issues a few presidential elections ago. Campaigns try to make issues more or less salient at least as much as they try to (re)position themselves along the hyperspectrum.
Median voter theory doesn't require a single dimension; no matter how many axes you plot on, there will still be an averaged 'center' of public preference, and the logic applies that a Condorcet voting method should converge on it.
That being the key factor not mentioned here - median voter theory only applies to Condorcet voting methods (such as ranked choice voting), it does not apply to first past the post voting, which is what we use in the US.
But anyway, US politics gets reduced to a line colloquially because it's a two-party big-tent system (which is an inevitable result of first-past-the-post voting). No mater how many dimensions you plot politics in, the two parties will each have a discrete point location in that space, and you can always draw a straight line that passes through them.
Whatever this line happens to be at the moment, becomes what that generation conceptualizes as 'left vs right'.
> No mater how many dimensions you plot politics in, the two parties will each have a discrete point location in that space, and you can always draw a straight line that passes through them.
I think this is building in too much intuition from Cartesian geometry. It seems very unlikely to me that the shortest line that passes between the parties in such a space is the same straight line that you'd get from linear interpolation.
Let's think non-parametrically. For each of the two candidates in the race, order all of the voters by winnability, starting with those requiring the smallest change to win over or alienate.
The median voter theorem would assert that *these lists are the same*. If Alice and Bob are the candidates, the set of marginal voters that Alice and Bob might win or lose with small positioning changes are the same set.
I submit that in multiple dimensions, these lists differ. By making small changes, Alice can win over Bob-voters that would otherwise be safe for him, and vice versa. In wibbly multidimensional space, this is how the candidates rotate around each other.
Take the current US presidential race as an example. One issue where the candidates differ is in their stance on whether presidents should profit from business ventures that attract foreign money. Can Harris really win over any marginal voters by saying she'll accept *a little bit* of Saudi investment in her nascent t-shirt business?
> If so, how come when Biden was replaced with the less-demented Kamala and the Democrats’ betting odds went way up, Trump didn’t change any of his positions AFAICT?
It seems to be widely accepted that Trump spouts meaningless syllables and one cannot take anything he says at face value. His supporters regularly make claims to this effect in response to people asking their opinion on Trump's latest utterance; including here in these comments sections.
It is unclear to me how people who claim that position actually determine what Trump's policies are, but this process cannot be related to the statements he makes if no statements he makes are to be taken at face value. IDK, perhaps they read tea leaves or something. In any case, if nothing he says or does can change what people think of him, why change what he says or does?
How has the LDP (Japan's Liberal Democratic Party) won consistently since the war? In free and fair elections?
It has an electoral system (SNTV) which combines both "choosing a candidate within the party" and "choosing a party over the other party" into a single vote, and it has a bunch of formalised internal factions, which means that any voter wanting to vote for (or against) one faction has to vote for the party in order to pick between the factions.
This lets it target multiple different positions on the electoral spectrum at once.
Japan hasn't used SNTV for their more powerful lower house since 1993. They use the majoritarian parallel voting there, and a mix of SNTV & party-list PR for their senate. I believe Taiwan and South Korea also experimented with & ultimately discarded SNTV.
In the Japanese case before the 90s, they found that it lead to corruption- politicians within the same party didn't have much to compete on ideologically, so they competed on pork barrel spending and dubious payoffs to local power brokers
I hear that part of it's that because the LDP's always in power, and it's got only weak ideological goals, if an opposition party comes up with a popular policy idea the LDP just implements it immediately and takes the credit.
I think there was a bit where they bribed the owner of 2ch to covertly shape discussions in their favour, too.
2ch?
A Japanese-language general-purpose BBS. 4chan is somewhat descended from it as an English-language imitator (technically, it's an imitator of a splinter of 2ch that added image support).
2ch is/was huge, with millions of posts per day at its peak (and almost all of that from Japan), so it's significantly more relevant than any single English-language BBS ever was (though still not on par with the social media titans).
If, for presumably cultural reasons, the ruling party is more capable of rapidly altering positions (or equivalently, the voters shift relatively slowly), they can stay close enough to the median that no one cares enough to switch. (This assumes some level of hysteresis, which I think is a reasonable model.)
"As per the Median Voter Theorem, most people prefer the (centrist) Democrat to the (more extreme) Republican. But the far left doesn’t care enough to vote, so the Republicans carry the day."
You've just explained 2010. It's more complex than that, everything is, but those are the blurry contours of 2010.
What’s going on right now is just Trump. He’s extremely unpopular but it’s not because of his policies. It’s mostly the things he says, the way he acts and J6. The policies he actually enacted in office were moderate. But the left pushed further to the left during and after his presidency, which is why Trump is still competitive. The median voter hates Trump but does not like the Democrats either, hence the stalemate.
You seem to have gotten the causality backwards. The only reason that Trump was seen as remotely viable in 2016 was because the Dems had already been going leftward full-speed for years, making the notion of "owning the libs" very attractive in some quarters.
Exactly. A lot of people think that "Trumpism" is all about Trump, but Trump is a businessman first and foremost. He saw a market full of unmet demand, decided to meet it, and built a successful brand.
In other words, it's the precise opposite. Trump is about the "Trumpists."
It’s both. Trump won the primaries because of immigration. The Republican Party had been moving towards amnesty for years beforehand, which was unpopular for Republican voters. Trump promised to stop this and gained supporters because of it.
But because Trump was so unpopular with Democrats, they decided afterwards to pivot towards an extreme and incoherent position where they opposed any enforcement of immigration laws but claimed they didn’t support open borders.
So Trump was filling an unmet demand among Republicans and then the left reacted by opposing that.
I don’t think Hillary lost because she was seen as an extremist on policy. She lost because she was seen as corrupt and held many Americans in contempt.
Obama also pretended not to be extreme at first, but didn't do anything to curb extremists on his side, and eventually started fanning the flames himself, particularly on race. I don't think anybody expected Hillary to be any different.
I'm always up for a good "the Democrats started it first" but Trump's big breakout hit was the fact that neither party wanted to speak against immigration and he was able to capture that market. Also the mainstream Republicans refused to drop out splitting the mainstream vote.
More generally, much of Trump's support is a reaction against the perception by Republican primary voters that conventional politicians of both parties had been colluding to enact unpopular policies favored by "elites", particularly with regard to immigration and free trade. There was a similar reaction among Democratic primary voters, but a combination of different primary systems and different candidate environments lead to Trump winning and Sanders losing their respective nominations. One key difference, as you allude to, was the other candidates: the Republican field was wide open, while the pro-establishment Democrats were united behind Clinton. Trump's celebrity status probably made a difference as well. And there was a huge impact from the way the parties allocated delegates: Democrats allocated proportionately subject to a 15% vote threshold, plus a large number of current and former officeholders able to act as ex officio unpledged delegates, so Sanders probably would have needed a solid majority of primary votes (or at least a very strong plurality) to win even in a three or four way race. And Republicans mostly used "winner take all" or "winner take most" allocations so Trump could run up large delegate tallies with 25-40% of the vote in any given state.
Tangentially, Sanders and Trump used to have quite a bit more in common policy-wise before moving towards the centers of their respective parties in order to win the primaries. Sanders has always been protectionist on trade issues, and was fairly strongly nativist (on economics/class-warfare grounds) up until early 2016. And Trump had been strongly pro-choice and had advocated for a wealth tax when he'd previously flirted with running for President in 2000.
Can you tell me more about Trump flirting with a wealth tax? That sounds so weird. I can't google anything because google is flooded with how Trump wants to not tax the wealthy.
The following search gets useful results on DuckDuckGo, and presumably would work equally well on Google:
"wealth tax" Trump 2000
The key is to place "wealth tax" inside double quotes, which means you only get results containing that phrase.
https://www.vox.com/2019/1/31/18203999/donald-trump-wealth-tax-14-5-percent
Specifically, the proposal was for a one-time special assessment at a 14.5% rate on all wealth above $10MM, intended to pay off the national debt.
This would seem to fit the argument though, Obama won by pretty large margins (biggest since Reagan iirc). By the MVT that would suggest there was room for activists to push the Dems left (which I think we saw over Obama's terms) and perhaps overshoot.
That is at most half the reason Trump was viable in 2016. The other half, at least, was Hillary Rodham Clinton being so monumentally anti-charismatic that to many voters it didn't matter what her policies were if she was the one mouthing them.
In 2016 Trump claimed that we would protect medicaire and social security which opened a lane for people who care a lot about those things but dislike other democrat policies to consider voting for a republican.
Candidates pretend to be close to center right before the election. You can see that now.
Wait, shouldn't median voter theorem work *less well* in a highly polorized society?
When the polarization is low, we actually have this "median voter personality" - someone who could realistically vote for either democrats or republicans. And, therefore, appealing to them is a great idea.
But when the polarization is high such median voters cease to exist. Moderate republicans will keep voting republican, no matter how far right the republican candidate goes, because the thought of voting democrat is just unconceivable due to the polarization. And likewise with modern democrats.
It seems logical that in highly polarized societies the radicals will have more and more power making the "centrist Democrat’s nightmare scenario" more of a reality. Do we have a more up to date version of this analysis that showed that such situation almost never happens? I'd like to check the dynamics and therefore put my hypothesis to the test.
The relevant change is how much "you can reduce everything to a single straight line." I agree "polarization" is the wrong word for it: "dimensional reduction" perhaps?
No, the issue is that Scott is describing a greater reliance by voters on party labels, and saying this should make them more willing to vote for a guy with the wrong label who's closer to their personal position. This makes no sense; greater reliance on labels makes people less likely to vote for someone with the wrong label.
Moving from a situation where the median voter theorem has no meaning because there's more than one political issue, to a situation where the median voter theorem has no meaning because there's only one political issue and it isn't continuous, is not actually an improvement in the explanatory power of the median voter theorem.
No matter how many dimensions your politics has, a two party system will always consist of two points somewhere in that space, and you can always draw a line between them. That line them becomes 'left vs right' for the current moment, regardless of its contours.
The reason we're stuck with a two party system in the US is that we use first past the post voting, which means spoiler effects (and some other technical stuff) drives out any third parties. It's also why median voter theorem doesn't work well here - median voter theory only applies to Condorcet voting methods (like ranked choice or Approval voting), and teh US doesn't use a Condorcet method.
Revisit The Myth of the Rational Voter by Bryan Caplan. Nearly 20 years later it’s still an excellent guide to critically assess the idea of a rational voter. To the extent politicians are rational, they know the voters are biased and become rational in exploiting it.
> Instead, Democrats are often pretty far left, and Republicans pretty far right.
That might be narcissism of small difference? From the outside looking in, there's a remarkable consensus on many topics in American politics. You guys just don't talk about the things you agree on, because it's boring?
Eg a while ago Germany privatised their snail mail and removed any monopoly rights. In contrast, you'd be hard pressed to find any mainstream politician in the US who would contemplate such a move. Even suggesting that is far outside the overton window of polite conversation in the US.
Similar for the German railways (Deutsche Bahn): in that case, they were about to privatise is but then the Great Recession hit, and they never got around to it again. But it's organised as a private company that just happens to be 100% owned by the government. And there's plenty of other companies competing with them for passenger rail traffic.
In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties.
Those are examples where Germany looks a bit more 'right' than the US. There's plenty of examples where Germany looks more 'left'. Eg, if you want to open a computer repair business in Germany, you need to do 3 years of government approved training, then work for a few years, and then do another two years of government approved training. That's nuts, and wouldn't fly in the US. (Though occupational licensing is becoming worse and worse in the land of the free, too.) See Meisterzwang, if you want to read more. Germany also taxes software engineers differently depending on what kind of software they write, but the difference is only something judges made up, nothing any real software engineer understands or supports. (See 'Freiberufler'.)
For a different topic: German has regulations that all pork that's sold must be safe to be eaten raw. Because Germans like to eat raw minced pork on breadrolls. Trying to introduce such regulation in the US would mark you as quite the weirdo. That wouldn't really fall into a standard left-right spectrum.
The greatest trick that the US establishment ever pulled was somehow transforming politics from being about policy to being about the culture war.
It's often about culture war in other countries too.
"In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties."
Would it, though? Maybe in the Acela Corridor, but in the rest of the country "Defund Amtrak" wouldn't be even remotely controversial for a Republican.
Where do people have cocktail parties?
In the quiet car on the Acela.
"In the US, even suggesting that their government passenger rail service should be privatised will get you uninvited from any future cocktail parties."
On the other hand, suggesting the rail infrastructure should be nationalised - as is the case in pretty much every European country - is well outside the Overton Window, and you don't get rail traffic competition without that.
" And there's plenty of other companies competing with them for passenger rail traffic."
Well, it's more accurate in most cases to say that there are other companies competing with them for government contracts to operate local rail services - which is the case in several US cities as well.
" But it's organised as a private company that just happens to be 100% owned by the government."
Which is also the case with Amtrak...
Is the postal monopoly something a strong consensus of Americans strongly support, rather than something few people care one way or the other strongly enough to make waves about it? Beyond the few hot button issues in politics, there are thousands upon thousands of accumulated laws and regulations that stick around mostly because few people care strongly about them one way or the other, as voters will only pay attention to a limited number of issues at a time. Often also there are small interest groups that care strongly about preserving them, while few others care to oppose them. In America it's further complicated by a bicameral legislature, presidential veto and Senate filibuster making it hard to change laws; if there is no prospect of an overwhelming support to change a law, it's not even worth starting to push the issue. Is the USPS monopoly something the vast majority of Americans strongly support, rather than one of these thousands of regulations few people pay attention to?
> Is the postal monopoly something a strong consensus of Americans strongly support, rather than something few people care one way or the other strongly enough to make waves about it?
There isn't really a difference between those two ideas. People won't vote for anything they don't care about.
Every so often someone tries to legalize self-service gas stations in Oregon, and it gets defeated by a scare campaign, despite the fact that believing anything the scare campaign said would require you to be far too stupid to be able to feed yourself. It's not like this is an untested policy.
It doesn't matter for the argument as long as you can still veer towards the center and get more popular approval. Harris could become less favorable to abortion, though still more than Trump. The question is why doesn't she. For this, it doesn't matter the size of the spectrum.
There does seem to be one unspoken assumption here. This states that (hypothetically at least) there is a one-dimensional line of positions from right to left with voters having preferences somewhere on this line, but does not state (and seems to simply take for granted) that the voters are evenly distributed along that line.
This is obviously not the case. Intuitively, you'd naively expect there to be a regular distribution with the bulk of voters clustered around the center. Of course, this would only make the naive Median Voter Theorem's predictions that much stronger, which is clearly wrong.
A better explanation would be that there are *multiple* regular distributions. Research in Moral Foundations Theory supports this idea: there are five fundamental, inborn "axes" used by human beings to evaluate questions of morality, and there's a strong tendency for liberal people to only reason with two of them at the expense of the other three, while more conservative people use all five more or less equally. So with two fundamentally different ways of looking at the world and evaluating the issues, you would expect to see two regular distributions of voters along the left-right line, with the largest groups clustered near the local maximum of each distribution, and the two parties staking out positions pretty close to those two local maxima.
Which sounds a lot like what we actually have.
>This states that (hypothetically at least) there is a one-dimensional line of positions from right to left with voters having preferences somewhere on this line
No matter how many dimensions your political model has, a two party system has just 2 points within that space as viable options on the ballot, and a straight line can always be drawn between those 2 points. Coalitions are then forced to form themselves relative to that line, and that line becomes what we mean when we say 'left vs right' for the span of that election.
Every individual voter has one point on that line which is closest to their actual position in higher dimensional space. That is their point of preference on that line.
>but does not state (and seems to simply take for granted) that the voters are evenly distributed along that line.
Not at all. The median voter theorem says elections that use Condorcet voting methods (which we do not) will converge on the weighted average of public preference, ie the spot that minimizes total regret across all voters (as measured by the distance between their personal point on the line and the point which we call the center).
Basically there's an optimal frontier at which moving left to satisfy people on the left would create more dissatisfaction than that on the right, and moving right to satisfy people on the right would create more dissatisfaction than that on the left. That equilibrium point where you can't minimize total regret any more by moving along the line is the center of public opinion, and it exists for any possible distribution of preferences across the line.
This all assumes that there isn't a "knowledge problem", and that both parties know what the exact center is so they can make their perfect plans. More importantly, it assumes that the center doesn't change.
But as you know, coalitions change, the electorate changes, and parties are constantly making electoral guesses about which combination of factors they can compromise on to build a winning platform. This can swing dramatically from one election to the next in ways that might never have been predictable ahead of time.
For example, in 2004 when Republicans were running on finishing the job in Iraq and Afghanistan, you'd never have predicted that they'd be running on "no new wars" two decades later, or that Democrats would be running on building up a stronger military and support from Dick Cheney.
In 2022, Republicans found out that abortion was a devastating electoral issue for them. Trump reacted accordingly, but soon found that he couldn't escape the issue because his donors demanded he push it in the opposite direction his polls suggest is a winning strategy. Yet, just 2 years ago there was a serious national attempt by Republicans to push a national ban through ... before the issue beat them up. Sure, there are still some Republicans whose constituencies favor a ban, but the party saw big losses from it in 2022 and recalibrated accordingly. The point is that they didn't recalibrate until after the losses, suggesting there was a knowledge problem prior to the election.
If this hypothesis is true, we should expect election-winning issues from midterm elections (especially in swing states) to be adopted in presidential platforms two years later.
This was more of an issue before the digital age, which answers Scott's question of 'why were the margins so much bigger back then, but way closer now?'
These days information gets distributed and integrated pretty quick; even if normal people can't name specific policies and pieces of legislation, changes in their party's 'vibes' get transmitted to them through elite culture production in a million little channels pretty rapidly and effectively.
Also, it's not too important for voters to have highly accurate knowledge on an individual level. Basic experimental power calculation - we average across tens or hundreds of millions of voters. As long as there is *some* signal by which actual changes in the parties affect some people's understanding of those parties, it doesn't matter if the average voter is extremely noisy and unreliable, we will extract teh signal when averaging over millions of data points (votes).
But the underlying problem remains: the sampling mechanism is infrequent (once every 2-4 years) and therefore all reasoning is based on a very small n. The glut of new methods doesn't change this. You still have to figure out what's relevant to the metric you care about, or how to adjust your sampling method to more accurately reflect the 'true' outcome, just like pollsters do. The fundamental question isn't "who's posting what to social media?" That only matters to the extent it tracks with the more important questions, "Who is going to show up on election day?" and "How are they planning to vote?" The answers to the questions we care about are still just as infrequent as they've ever been.
Whether it's polls, cell phone tracking, app usage, social media, combined aggregation, etc., you still have to grapple with the same question: how well does this new metric predict the outcome of the election?
All these new methods suffer from the same knowledge problem that polls suffer from: you can't calibrate statistical results from a single data point. So they do past-casting from old data points ... just like polling organizations have been doing for decades.
But which ones? How much do mid-terms matter? What about off-cycle elections, recalls, run-offs, etc.? Will it be relevant to incorporate old data, and how far back does an election result become irrelevant to the current cycle? Were important issues from last cycle fads, such that we should expect the electorate to revert to their old habits, or are they enduring trends that will continue moving forward? Are there new fads we're not picking up on, or are the new fads something that reflect in the news/social media but won't impact polls?
The answer to each of these questions is not "Yes" or "No", but somewhere between. The amount a pollster should recalibrate on any of these questions is always a judgement call, whose validity can only be confirmed in retrospect. Except how much can you 'confirm' a statistical model from a single data point? These problems are not 'solved' with new prediction methods.
"But we have so many more sources of information! Surely that improves our accuracy." Aggregate a dozen data sources calibrated from the same one or two hard data points and it's garbage-in/garbage-out.
Off-topic: This problem bears a surprising similarity to the Fermi Paradox and the Drake Equation. How much should you calibrate your estimate of how much each factor matters (e.g. presence of a gas giant, or a moon, or a magnetic radiation shield, or even presence of liquid water) in your estimate of intelligent life developing? We only have one data point, so it's impossible to calibrate your results, but always possible to generate new variables (e.g. Great Filter).
One additional factor to consider is the distribution of voters isn't uniform - it's bimodal. There aren't very many voters in the median
Doesn't need to be for the theorem, there's still a center of public opinion no matter what distribution those opinions are in.
Mathematically, the median voter theory doesn't require a single axis, like left vs right, to drive the analysis. You can calculate a multidimensional distance that takes into account many issues or even how intensely voters care about them.
I’d think one suffices. It’s common to hear “I vote straight line red in every election”
One suffices right now, but not always historically.
All models are wrong, but some are useful. A one-dimensional model is useful to some extent, but it's also interesting to have a look at higher-dimensional models and see if anything new pops out.
A two dimensional model (say, "social issues" vs "economic issues") gives you interesting new behaviours that you don't get in one dimension. If the views of voters on these issues are not perfectly correlated then you'll have voters in weird corners and you can expect to see the parties circle each other a bit as they try to sweep up these weird-corner voters without losing touch with their base. This explains why the direction of the Republican-Democrat axis in the US feels like it's currently rotating, something you don't get in a one-dimensional model.
Going beyond two dimensions is a more realistic model but probably doesn't give you enough new insights to justify the extra complexity.
It's more that in a two-party system, you can always draw a line between the position of the two parties, no matter how many dimensions those points are plotted in.
It's more relevant in systems with more than two choices, that the theory still works with an arbitrary number of dimensions and choices.
But the "median voter theorem" does usually require a single axis, see https://en.wikipedia.org/wiki/Median_voter_theorem#Extensions_to_higher_dimensions
Thanks for the link.
You and John are talking about different sets of dimension, I think.
John is talking about dimensions of preference, like economic vs social.
The single axis you are referring to is about the space of candidates/choices; it shows what happens when you cannot draw a single axis through all of the parties.
You can indeed have an arbitrary number of dimensions of preference/politics/issues/etc. As long as you only have two candidates, they will each be at some point in that higher-dimensional space, and you can draw a single axis through them. The median voter theory will then work on that singular axis.
You can calculate a multi-dimensional distance, but then the distribution of voters’ preferences is not guaranteed to have a unique median.
Yes, it depends more than my original comment suggested. Pazzaz provides a link to more detail above.
Re: control of the House, for most of the last couple decades Republicans have had an inherent House advantage. This reflects 1) more and better gerrymandering. Ballot initiatives and judges blocked Dem gerrymanders in CA and NY, nothing stopped the Republicans in TX and FL. 2) The inherent propensity of democrats to live in hyper concentrated pockets. There are urban House districts that vote 90 - 10 Dem, but even extremely republicans areas (TX panhandle, NW Georgia) don't reach that level. Those are basically wasted votes in a first past the post electoral system.
Interestingly 2022 broke this trend somewhat, with a narrow Republican popular win leading to a narrow Republican House majority. I think this reflected greater Republican support in black and particularly Latino ghettos (where they still lost anyway), some rural areas becoming more lopsidedly Republican, and suburbs moving Democratic.
I think #2 is a larger factor than #1.
What is important about collusion is what is at stake. If you're running the Conservative Party or worse, you are Israel's Yair Lapid, you need to worry about never losing elections, because there's little constitution and little courts, Labor or Likud can just win and make all the stuff you care about illegal.
But because of three distinct powers, a reasonably strong constitution, and the filibuster, if one party loses, the other party won't just undo everything they have done. What you're optimizing for is getting a trifecta, maybe a 60+ trifecta like Obama had. In this sense, "Democrats can have school funding" and "Republicans can have military funding", the way you described.
But you can't run Likud, Labor Party, Yesh Atid, or the Conservative Party this way!
In my country, Brazil, we have a really big constitution and even more judiciary oversight. If Israel is in one extreme for how much power one given election has, U.K is still bad, and the U.S. is in the side where it isn't too bad, Brazil is in the other extreme. The Supreme Court could just run the country without any oversight because as the Constitution is so extense (the 2nd most extense in the world!), there's little you can do once elected. And indeed, Congress is an amorph group of centrist people with little ideologic beliefs. (See the wikipedia article: https://en.wikipedia.org/wiki/Centr%C3%A3o )
And indeed, all presidential candidates that run since the down of the 6th republic and 2018 were very firm in saying they weren't right-wing. Indeed, Bolsonaro big contribution to Brazil politics is the discovery that you can win elections by being right-wing.
Somehow now politics resemble more U.S. politics. The first-round is expected to be a primary, and if you can't pander to the median voter of your coalition, you're toast. Why? I don't know. Maybe politics has become an sport where people really like winning but for little policy subtantiation? Or people don't realize there's little they can achieve once elected. Idk.
But the US is a country with a very powerful Supreme Court.
How are Brazil's Supreme Court justices appointed and how much latitude do they have to reinterpret or emphasize parts of the constitution over others?
So on the theory that parties try to decide just how crazy they can be:
Parties are coalitions of many interest groups. Those interest groups are sometimes pointing in sort of opposite directions, sometimes they are vaguely pointing the same direction, but they're all competing for the same amount of oxygen.
So each individual interest group would accept an increased 0.5% chance of their party losing the election (c'mon, will it really be that close?) if it meant locking their party into their interest group's positions. Repeated many times this leads to some real stupid things, stupid things you can see in the other party even if you can't see it in your own.
(This isn't even bringing in the Iron Law of Institutions.)
I’m a fan of this, but it doesn’t explain how it stays so close to 50/50. Surely one party has interest groups which are more willing to make that bargain, right?
How does a preferential voting system ( as we have in Australia ) mitigate the median voter theory ?
So one of the big things missing from this article is that the median voter theory only applies to Condorcet voting methods, which the US does not have. That's the main reason the US doesn't align well with the theory.
Australia uses Instant Run-Off voting, which is also not Condorcet. In practice, the first-past-the-post voting method used in the US, and the IRV method used in Australia, are nearly identical in terms of outcomes in most real-world cases.
There are Condorcet methods that use preferential ballots, an any system that uses them will create outcomes in line with the median voter theorem. But in practice I don't think any countries currently use these methods at scale, because the elites who decide the voting method are rarely advantaged by outcomes that match the center of overall public opinion.
thank you 🙏🏻. Voted in our state election today. Numbered all candidates as I preferred from 1 to 7. For a valid vote, we have to number ALL … can’t leave a blank against the absolutely not wanted at all candidates/parties. Which nutter to put last is the challenge!
Here’s my hypothesis:
The collusion theory where both parties trust each other to run on unpopular things is on the right track in that it acknowledges that parties care about things other than winning. Usually both parties are really OK with losing, especially since fundraising goes up when a party is not in power.
But they can’t lose horribly or repeatedly because they want to be taken seriously; they both want special interests to court them and special interests don’t want to bribe a side that doesn’t have a chance of winning.
Scott noted that the Republicans don’t seem like they are putting much effort into calibrating to balance out the Democrats. Well, you actually only need one mover to play the game and find balance, and Democrats have been the ones doing that recently.
Replacing Biden with Harris was the result of these incentives playing out pretty perfectly: someone just likable enough to run a tight race but not so likable as to abandon the party’s interests and beat Trump easily.
To be clear, I don’t think Democrat politicians are literally thinking through this exact logic in their heads per se; rather, at scale this is how the incentives play out in the system.
I think that the reasons the Democrat chose Harris rather than "a better candidate" was based on Harris being the obvious Schelling point candidate just as Biden was the obvious Schelling point candidate in 2020.
Yeah, that's probably a big part of it as well.
Much like "professionals think logistics" politicians think turn out to vote. There is no "median voter". Professionals look at the spectrum as "your mother on one side and your ex-spouse on the other". Your mother will always vote for you, your ex will never vote for you. There are hard limits in each district on the persuadable voter. There is also a hard limit on how much fraud will occur in each district (yes it occurs). Each median voter is an amalgam of multiple single issue concerns. There is no "median voter" in reality, just as no one is "average" in all dimensions. The overriding concerns will vary depending on the immediate concerns at that election. Your candidate just needs to be less of a turn off compared to the opposition in a 2 way race. As explained by a national consultant " A company thinks a 45% market share is great. In a 2 way race it is called loser". Where US politics went bad is when the courts ruled that majority election, ie 50% +1 needed to win was discriminatory. 50%+1 meant the more mainstream candidate won the primary and then the general. Mere highest vote total leads to winning candidates outside the mainstream. (Ranked choice is worse. remember no method of voting satisfies all criteria). If we want more "mainstream" median politicians we need to have party primaries where the winner needs 50%+1 to win, run offs of the top 2. The election winner needs 50%+1, run offs as needed. Fraud can be kept as low as possible with same day in person voting, Government issued IDs, and paper ballots in transparent boxes, ie just like Europe.
Wave elections occur when one sides voters can not bring themselves to vote for "their sides candidate" and just stays home. No one actually shifts parties other than very few. A 10% shift in a groups preferences are historic and rare.
"When the courts ruled that majority election, ie 50% +1 needed to win was discriminatory."
What is this a reference to? I don't think that the Courts handed down the rule that pluralities count as wins, but I could be misinterpreting something.
>There is also a hard limit on how much fraud will occur in each district (yes it occurs).
I’m not sure I understand. Are you claiming voter fraud in the U.S. plays a significant role in election outcomes, or that due to the “hard limit” it doesn’t? If you think it does, can you link to the data that suggests that?
> Fraud can be kept as low as possible with same day in person voting, Government issued IDs, and paper ballots in transparent boxes, ie just like Europe.
The assertion you make here about European countries requiring same-day in-person voting is false. Many European countries do allow mail-in voting (Germany, UK, Sweden, Spain to name just a few).
I don't believe you're a professional on this issue. Scott discussed turnout vs swing voters, and you haven't actually corrected anything he said (instead you acted like you hadn't read it).
I think the easy explanation for the 55-81 Democratic control of Congress is a lot of southern Democrats were elected Blue but voted Red. If you look at the voting record of those congressional years it will look a lot more 50/50 than party affiliation alone would predict. (I suspect the same is true of the other years that don't look balanced either)
They voted red on one big issue but not in general.
Yeah, they voted red in general.
https://www.vox.com/2015/4/23/8485443/polarization-congress-visualization
That doesn't say anything about Southern democrats.
All of the blue dots in the red section of the scatterplots are Dixiecrats
No, they didn't. Per DW-NOMINATE the primary axis of politics was economic and regional/social issues were the secondary axis of politics. Your link still shows the parties primarily in separate clusters (even if less cleanly so) in the past, and doesn't mark out southerners at all.
No, my link doesn't show separate clusters - it shows lots of blue dots on the right side of the scatterplot. As someone who was alive and very much politically active during that time period I can confidently state that many southern Democrats were essentially Republicans in all but name. Many even went so far as to switch parties.
Some did switch parties. But merely being alive doesn't mean you were counting Congressional votes.
OK - when you offer some evidence disputing the evidence that I've brought I'm more than ready to listen.
Is it possible that elections were farther from 50/50 in the past not due to information scarcity? Today's politicians don't need to guess how popular they are, they know what voters say about themselves, they can just open Twitter and get more information about the state of the election than anyone could have dreamed of 40 years ago.
It seems intuitive to me that if both candidates had perfect information about the electorate, they would find some kind of strategic equilibrium that made elections tend towards a toss up, but I can't really formalize that intuition.
Yes, this is very likely correct. Information abundance has made both parties more 'efficient' in terms of ability to measure public opinion and respond to it quickly.
When both parties are doing this at maximum efficiency, the natural equilibrium is 50% (or 50/50 in the electoral college). The median voter theorem shows why - they should both want to be very close together at the center of public opinion.
And absent some culture war issues where the public is divided 50/50 and therefore it is safe for the parties to signal their allegiance by disagreeing with each other loudly, this basically is what has happened. Despite culture war over whether immigrants are scary or inspirational, both parties are currently touting their tough-on-the-border credentials. Despite culture war over whether cops are bastards or heroes, both parties are running tough-on-crime fund-the-police-more platforms. Despite culture war over whether China should be hated and feared or if hating and fearing China is racist, both parties are advocating trade tariffs and strategic industry subsidies to fight China. Etc.
I don't think Guam would be a lock for 2 democratic senators. The current non voting representative is a Republican, as was the previous-to-current governor.
The median voter usually isn’t someone whose opinions cluster in the center of two ideologically sorted poles. They are someone whose views don’t sort ideologically. They might think we should deport all immigrants and nationalize healthcare; or that we should end the Fed and subsidize organic farms. They are just as likely to think a centrist candidate doesn’t reflect their values as the other candidates.
What might explain the 1955-1981 Democratic congressional dominance is lack of polarization combined with a huge incumbency advantage during that period. One party happened to be in power when this incumbency advantage suddenly took hold. Why might such an advantage suddenly take hold in 1955? Well, it corresponds with the first year the majority of Americans owned TV sets. If congressmen were advertising on TV back then -- and I don't know that they were -- it may have created a positive feedback loop of winning -> gaining name and face recognition -> raising more money to advertise on TV. This inertia worked for Democrats because they had first mover advantage until Reagan came along and changed the game with more TV charisma than anyone had had yet. After Reagan, party polarization gets real.
What about state level politics? How does the Median Voter Theorem work in, say, Wyoming, where both the state house and senate have been Republican since 1967? Or Maryland where they have been Democratic since 1920? Surely the Maryland Republican party has an incentive to move to the left, and the Wyoming Democrats to the right, but this doesn't seem to have happened despite decades of effectively one-party rule.
David Schleicher has actually studied the issue of local partisan competition:
https://www.econlib.org/archives/2008/12/schleichers_mod.html
Basically, national politics dominate.
The GOP and DNC have a lot of control over who runs in state races, you can't put your name down as a Republican on the ballot if the GOP doesn't affirm you as a member of their party.
They're not going to help a Republican senator get elected if he's going to vote against them in the Senate all the time, even if that's what would be needed to win his district.
And without an R or D next to your name and with zero funding and resources from the national party, you can't win anything larger than a small-town mayorship.
So local politicians are beholden to the national parties, and can't diverge too far from them.
One obvious answer is that primary voters consider both ideological purity and ‘electability’ and that median voter theorem comes in through the second channel in a thermostatic wisdom of crowds kind of way. So if you run Dukakis and get annihilated then next primary you weight electability much more heavily (and conversely if you run Romney and still lose then next primary you say ‘fuck it, let’s roll the dice.’
All of this assumes voters sum all their interests together, build an average and vote based on that. This is not how it feels lately. Here in Germany, the latest elections were mostly about the stance on migration (even local elections where the elected political bodies have no influence on this matter), less so on the cost of climate change politics and whether or not the new right is electable. For the US, it seems a lot is still on abortion.
And everywhere, the split seems to be around wokeness
> Here in Germany, the latest elections were mostly about the stance on migration (even local elections where the elected political bodies have no influence on this matter)
That doesn't make sense. It's like saying that women couldn't influence politics before getting the vote, or that merchants couldn't influence politics... and still can't.
Having or lacking statutory authority over something just doesn't have much to do with how much influence you have over that thing.
Brexit was determined by a non-binding referendum. How non-binding was it?
What I am saying is that city, county and state elections were about things like migration issues. But that is a federal issue, a county or even state parliament cannot change the asylum or migrant benefits laws
I don't see how that responds to my comment at all. County officials definitely can change the asylum or migrant benefits laws. The fact that there is no legal channel for them to do so doesn't mean that they can't; it doesn't even mean that they aren't the primary means by which such changes happen.
And... that's what my original comment says. What were you responding to?
>County officials definitely can change the asylum or migrant benefits laws
Not in Germany. These are federal (and partially EU) laws that cannot be changed locally.
Again, I'm going to have to ask you to respond to something in my comment. Why do you believe that "These are federal (and partially EU) laws that cannot be changed locally" is a relevant way to reply to "Having or lacking statutory authority over something just doesn't have much to do with how much influence you have over that thing"?
Do you think that I've overlooked part of your point? If so, what is it?
Do you think that you've overlooked any part of my point?
Why do you assume the opposite? Why would a county assembly have any influence over the federal immigration policies? They deal with building permits, infrastructure maintenance and whether (big topic here) the freshwater supply can sustain the expansion of the Tesla factory. All they can do about immigration is to push back on the state and federal government that they cannot afford to house the next batch of immigrants. The federal government can safely ignore that and just reply "but you have to, it's the law"
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”. Doesn’t feel like the kind of thing that would get lots of votes. Maybe after decades of humiliation and failure you could convince someone
Look at the last two competitive primaries. In 2020, most of the Democratic candidates were running left, but the winner was the guy who said “actually we should veer right”.
And in 2016, most of the Republicans were running right, but the winner was the guy who said “actually we should protect social security and Medicare, maybe even increase them, and cancel all wars”. He didn’t *spin* it as veering left, but on the traditional political substance Trump is definitely to the left of most Republicans in the past few decades. He just has an angry culture war persona that makes him *feel* right wing, while campaigning on undoing the core Republican platforms of cutting social security and Medicare and expanding wars.
> In 2020, most of the Democratic candidates were running left, but the winner was the guy who said “actually we should veer right”.
...and then didn't. And did anyone *really* believe he would?
I think people are starting to wake up to the common pattern that Democrats these days always seem to run as reasonable moderates, but then when elected they invariably govern as extreme-left as they can get away with.
I'd say that a lot of the "seem to run as reasonable moderates" is entirely due to the heavy media spin that the "mainstream" democrats inherit by right. Because the main media outlets are *openly and blatantly* in the tank for the Democratic party establishment. It doesn't matter what they say or do, the main media outlets are going to paint them as saints who are moderate as apple pie and will paint the Republican as the second coming of Hitler-Satan.
And the right-leaning alt-media will paint the Democrat as being the second coming of Stalin-Mao, so that part is totally bipartisan. The right-leaning media is a bit more fractious in not *always* lock-step supporting the Republican--GOPe candidates get tarred as Democrat-lite these days.
I am highly amused at the idea of Biden as somehow a died-in-the-wool far-leftist as opposed to a milquetoast centrist.
By their fruits ye shall know them. His administration's policies, and the results it produced, are significantly to the left of even Obama.
The main things where he is left of Obama are ones he inherited from Trump, like his harsh anti-free-trade policies, and his big cash giveaways at the beginning of his term. There’s also a lot of mid-level appointees that he got from the Warren team who implemented lots of small left things, that I agree are left of anything from Obama or Trump. But the main Biden policy bills are Manchin-approved centrism. He didn’t do anything as big as Obamacare or same sex marriage.
> But the main Biden policy bills are Manchin-approved centrism.
The ones that *passed* are. Now look at the ones that he wanted but that didn't get past Manchin and Sinema.
> He didn’t do anything as big as Obamacare or same sex marriage.
- Open borders, and all the gigantic messes downstream from it.
- Title IX abuse that goes far beyond anything Obama did.
- Criminal prosecution of political opponents, a tactic straight out of the USSR and Communist China. (For all that Democratic scaremongering claimed President Trump wanted to do this to Hillary Clinton, he never did it. But Biden did!)
- The repeated, entirely lawless attempts to nullify student loans
Heck, here's the Washington Post rejoicing over how extreme-left Biden's administration has been, far beyond anything Obama did: https://www.washingtonpost.com/opinions/2023/07/13/why-progressives-winning-inside-democratic-party/
I haven’t seen any evidence of open borders, much as I would like them.
And there is no universe in which he is prosecuting political opponents. Who are you even thinking of? Bob Menendez? Donald Trump? Rod Blagojevich? Hunter Biden? These are criminals, who should be prosecuted, and any competent government would prosecute them. You would have to be Berlusconi’s Italy to *not* prosecute people like this.
So, I have a question in the form of a chart: https://i.imgur.com/Cb7PTNS.png
In other words, if you plot voter preferences in one dimension, the median voter is a point, but if you plot them in higher dimensions, wouldn't you get a higher-dimensional interface with lots of points? That is, aren't there actually tons of very distinct policy platforms that would win half the voters?
If you only have two candidates to choose between, then no matter how many dimensions your preferences are plotted on, you can draw a single line between the position of those two parties in the higher-dimensional space, and every voter will have some point on that line closest to their actual position in the higher-dimensional space. Median voter theorem will then apply to that line and people's proxy location on it.
If you have more than two candidates to choose between, then yes, it is possible to construct arrangements where you cannot reduce the higher dimensions down to a single line, and public opinion can have multiple balancing ('center') points which each point at a different party.
This happens more often in thought experiments than in reality, though. Political parties don't actually choose random positions in the higher-dimensional space, those dimensions are typically correlated to each other in a lot of ways that constrains the space of internally-coherent parties. And those parties tend to converge towards center points in order to get more votes. You need fairly particular configurations to get multiple center points which each favor a different party, it doesn't happen by chance too often in practice.
Also relevant
https://scholars-stage.org/culture-wars-are-long-wars/
But two parties cannot occupy the exact same position of median voter, it would make them indistinguishable. So there is a force that pushes them apart, at least on "branding" issues. If it pushes them equally, we get the observed result, no?
I know my fellow residents on the left and right wouldn't agree but you can see a real 'median voter theorem' effect in some sense in Canada where the parties are very very powerful and top-down - where we have more than 2 parties (usually ~2.5).
Liberals have done well since the 1900s as the 'centre' party (not centrist) and they often get booted out to crash and burn extremely painfully once people get tired of economic mismanagement.
The Tories come in focusing on the 'basics' but being the sum of various anti-Liberal factions across the country results in them imploding in the effort to appeal to median voters.
The UK seems to be pretty similar in this regard too, and it seems its a result of the stronger parties.
The US has not only weak parties but various 'backbenchers' (in that they are not part of Cabinet or top ranks of the deparment) like Rand Paul or Cortez who appear to be more popular and visible than their own partie - simply for being expressive virtue signallers in Congress.
"Parties and candidates seem to do a suspiciously good job staying equidistant from the median voter, far beyond the pathetic amount of effort they explicitly put in. I still can’t tell if it’s all coincidence, or whether there are deeper currents at work."
Baseball managers, before the analytics explosion, naturally adjusted to the run scoring environment. They used one run strategies more in low scoring seasons and less in high scoring seasons. I think very few of them thought of it in those terms, they were simply rewarded with wins if they adjusted and losses if they didn't.
"If some people dislike Trump because he committed 10,000 felonies and an attempted coup, do the Democrats enter that into their calculations and veer slightly further left?"
FWIW, this is in basically Matthew Yglesias's position on what happened — Trump is a below-average quality candidate, because he's an offensive nutcase, so this gave Democrats latitude to shift left on policy (relative to Obama, let's say) while maintaining about a 50% chance of winning. (Separate question is whether this was a good idea.)
Yes, and I think the primaries essentially ensure that democratic leadership always *wants* to push as far to the left as they can, because they were selected to represent the median democrat.
Vance's pick as VC also seems to have been done under the expectation that Trump would face a weak democratic candidate. Why didn't he switch later? I'm assuming because Trump has this whole personal loyalty thing going on.
That, and he *is* facing a weak Democratic candidate. As the burst of Democratic euphoria from the abrupt switch dies down, that's becoming more and more obvious, and Trump continues to gain in the polls a little bit more every day. The latest RCP polling has him (narrowly) ahead in every one of this year's swing states.
"Vance's pick as VC also seems to have been done under the expectation that Trump would face a weak democratic candidate. Why didn't he switch later?"
Switch to who?
> Why didn't he switch later? I'm assuming because Trump has this whole personal loyalty thing going on
There's no real mechanism by which he could switch, after the convention. Vance is the Republican Party's nominee for Vice President, and being the nominee for President doesn't give you the right to change that.
Some of the questions around “Why don’t the parties simply X?”, where X is some action that would definitely get the more votes, boil down to the fact that what appears to be a spectrum is actually composed of specific issues, many of which are extremely emotionally charged and polarizing. Some of these issues are so deeply psychologically motivating that some voters actually think more in terms “punishing” the other side for having such a terribly immoral and evil stance in on their specific issue.
Other commenters have brought up the way in which attitudes on gay marriage shifted substantially in the past decades as an example of the parties successfully maneuvering around the central axis. But this was possible because gay marriage was not one of these hypercharged emotional issues for very many people. It was never the sort of issue that would make a preponderance of republicans say, “I could never, in a million years, vote for someone who endorsed this policy. I would rather vote for higher taxes than vote for gay marriage.” Very few republicans felt that way about gay marriage, to the consternation of many conspicuously religious conservative pundits, who felt confused and betrayed by their own party’s abandonment of this issue.
The thing that confuses me about the present moment is that both parties seem to commit what appear to me to be frequent own-goals by willingly adopting policy positions that are widely loathed by huge numbers of Americans. In the frame of this article, they’re pandering to the most extreme elements of the base in a way that actively infuriates and enrages pretty much everyone who isn’t an extremist. And both sides do this! Why doesn’t one party simply choose to be the one that doesn’t uphold *any* policy positions that the median voter thinks are insane/stupid/evil?
Because as Scott mentioned, motivating your base, your donors and your volunteers/staff are three disparate activities. The most successful compromise is apparently to make a platform with mostly normal positions, then mix in a few insane ones and hope it doesn't spoil the voters' appetite.
You can see this working by contrasting with mayoral platforms. When elections are small enough that a candidate can furnish all the funds and labor themselves, platforms tend to be all sane positions or all insane positions, not this inorganic mix.
The parties also aren't unitary actors who can take actions that would clearly benefit them.
> If so, how come when Biden was replaced with the less-demented Kamala and the Democrats’ betting odds went way up, Trump didn’t change any of his positions AFAICT?
I guess there wasn't a huge change between June and August, but Trump this time around has made moves towards the middle. He is going to specific interest groups and promising policies. He went to the Libertarian convention and promised to free Ross Ulbricht, a super niche issue. He basically told the GOP party platform they were banned from mentioning abortion.
That said, Trump IMO is a result of the primary system. He's not a very popular general election candidate, has high net negative favorability, but he has a strong base in the primary. I think the Median Voter theorem is true, but within the constraints of the current political system, which are numerous. In a parliamentary system, I suspect Republicans would have replaced him.
By the time Biden dropped out it was too late to change anything, but I think that if Biden had dropped out a month earlier then Trump wouldn't have picked Vance as a running mate.
Vance was the pick of a Trump who is confident in his poll lead and can afford to pick someone for alignment rather than electability. If Biden was out and Kamala was in the middle of her media-puffed bounce then Trump would have "tacked to the centre" by picking for electability.
But it would still have to be someone loyal enough to contest election results.
It’s also assassination insurance
I think the simplest explanation for parties not having identical platforms in the middle of the policy space is that a single dimension of policy is distortingly idealized. There are many different issues that actually mattered to different voters and many voters have constellations of preferences that are not perfectly aligned with one another. Consequently, coming up with a platform to appeal to a maximum number of voters is not a simple problem. You end up with some things that the parties agree on and a lot of things that they disagree on. All in order to maximize their chances of appealing to a majority of voters the right combination of policies can in theory appeal to substantially more than 50% if the other party does, a bad job of calibrating their own policies.
half-baked thought:
So if there is unofficial colluding and you wanted to moderate the extreme positions of the opposing party, you should moderate your own policies?
That seems to be the opposite of the Overton Window-shifting strategy of pushing more radical policies to move the median voter toward you.
sortition, the formation of legislative bodies by random selection, wouldn't have this problem :)
The two American parties have wildly different rhetoric, but isn't it a common complaint from their bases that they don't govern all that differently when they're in power? That could be because of inability (split government, filibuster, bureaucratic inertia, other checks and balances), but also because they cater to the median voter. Though I don't see why the median voter theorem would apply more to policy than to rhetoric; the opposite would make more sense.
Total post-hoc speculation: perhaps voters intuitively know that parties won't shift policy all that much one way or the other if they get elected, so the radical rhetoric doesn't put them off. At the same time, while they govern similarly, the exaggerated rhetoric helps voters know how the parties are positioned relative to each other (as well as helping candidates win primaries). If, say, Republicans govern in a centrist way as do Democrats, then if Republicans didn't have distinctly right-wing rhetoric either, right-wing voters wouldn't know they have to vote Republican to signal to politicians that they want more right-wing governance.
--------
One more reason for major parties to not be perfectly centrist is to keep minor parties at bay. If the major parties were indistinguishable, people wouldn't worry about wasting their votes on minor parties, which could allow them to grow until they can actually challenge the two major parties; or at least until they can spoil the major party closer to them.
If, say, the major parties were both centrist, but Republicans were still recognized by the public as more right-wing, and a more radical right-wing minor party started gaining awareness, then some right-wing voters would turn to that party, while left-wingers would stay with the Democrats, so Republicans would lose (like how Reform spoiled the too centrist Conservatives in the recent UK election). This possibility incentivizes Republicans to be somewhat more radical.
--------
Maximum collusion between parties wouldn't push them to be maximally extreme in different directions. If they both want unpopular policies, they can also collude to both support both unpopular policies (in the example, both school and military funding), so voters have no choice. They may not collude to be perfectly equivalent to keep third parties at bay and to win primaries, but there's no incentive to be maximally different.
Indeed, if they were maximally extreme in one direction and the other, that would presumably also include maximally undoing the other party's policy, e.g. one party would increase military funding, but slash school funding well below where a centrist compromise or perhaps even the median voter would put it, and vice versa, so in expectation there would be no benefit to the parties over colluding to stay moderate (though not indistinguishable).
--------
It's not exactly right that candidates in the primaries have to cater to the median of their party, since presumably some primary voters are tactical enough to care about their candidate's electability in the general election, even though many primary voters don't do that. So in the primaries candidates' target should be somewhere between their party's median and the general electorate's median.
To begin with, the 'wildly different rhetoric' is ussually restricted to issues where the country is already divided close to 50-50, and ussually issues that are not very consequential. They tend to actually be close together on most important issues, and just not talk about those very much.
Can we safely assume that the overwhelming majority of voters vote on candidate positions, rather than on something else?
We know people vote based on other things as well - candidate personality, media coverage, the demographic of the voter, voter inertia, in some cases the demographic of the candidate. Is it fair to assume that all of this is insignificant in the picture?
The median voter theorem applies to voter preferences, not voter policy preferences. Anything that makes voters prefer one option over another is a dimension on which the theorem operates, including everything you listed.
Trump did say something like, "If it was popular vote it would have been a different campaign"
I think Canada has some good things going on with campaign finance. The Liberals banned corporate and large donations and instituted a per vote party subsisdy
subsidy
The Conservatives banned union donation. I really like that median voters can have an effect on party finances without directly donating, which is a harrowing experience of all caps paper mailouts that are hard to unsubscribe from, that appear to be designed to activate very intense supporters
I agree that the evenness of the splits is surprising but I don’t find the median voter theorem plausible. How do trump’s felonies or biden’s senility fit into that? When you factor in trends like, declining trust in all kinds of institutions, and massive increases in debt, I think you’re watching some long term play itself out, whereby a polity splits in two as it dies, and something in that process seems to equilibrate the split along roughly even numbers.
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”. Doesn’t feel like the kind of thing that would get lots of votes.
This strikes me as a serious failure of imagination. Trump is way to e.g. Romney's "left" on entitlements, gay rights, marijuana, and trade policy. He's successfully captured the support of more minority voters and averted the "permanent Democratic majority" scenario that Republicans were panicking about during the Obama years. Also, as this example demonstrates, politics is not one-dimensional—Trump brought his party's platform closer to the median voter by moving "to the right" on immigration. Primaries have functioned pretty effectively at producing electable candidates recently; I don't see any reason to expect that to suddenly change.
Get out of one dimensional thinking. Think of a two dimensional political map where equality is the left-right axis and amount of liberty is the vertical axis. (see: https://holisticpolitics.org/NewParty/ ) The party positions are rotating counterclockwise. Trump has pushed the Republican Party kicking and screaming into the long neglected upper left quadrant. Lower-right Republicans (Liz Cheney) are now voting Democrat.
But no matter where the parties are in higher-dimensional space, you an always draw a straight line between them.
That straight line becomes the 'left vs right' axis for that election.
And yes, it might be very different from the 'left vs right' line from 20 or 10 or 2 years ago! But we are so polarized and stereotyped that we will call it the 'left vs right' axis no matter where it actually goes between, and form coalitions in line with it.
All part of the madness of a two-party system, which is the inevitable side effect of first-past-the-post voting.
"But no matter where the parties are in higher-dimensional space, you an always draw a straight line between them."
You've said that at least half a dozen times. It is past tiresome, and even more tiresome if it incites people to try and rebut you in half a dozen different subthreads. Maybe focus your attention on one or two, and otherwise let it go that there are people who are Wrong On The Internet.
Why did we see Dems win in Alaska and Arizona, Republican majority states, over the last four years? Because the Republican choices were too far right for moderate Republicans and Independents. And the Dem choices were not far-left. Seems pretty straight forward.
Voting science nerd here:
The "reasonable assumptions" of the median voter theorem assume a reasonable voting method. Literally anything better than the mess the US is currently using would qualify. Plurality voting and its variants suffer from the *spoiler effect*. The spoiler effect leads to *center squeeze* and polarization.
There is nothing secret or complicated going on.
See for example:
https://psephomancy.wordpress.com/2022/09/15/some-election-simulation-results/
https://arxiv.org/abs/2306.07147
http://zesty.ca/voting/sim/
Some better voting methods (self promotion):
https://hiveism.substack.com/p/four-levels-of-voting-methods
Also, you will have a hard time finding a political scientist who is educated in this regard. The topic of voting methods is a kind of niche one even within science. Most experts are from various fields: economics, statistics, math, self-educated activists, etc. Only a few are political scientists.
zesty.ca is the link I always give people to try to give them a quick intuitive understanding of these issues. Love the visualizations.
Thanks for these awesome links. I suspected this whenever I start to talk to political science people and they get this "I hope he stops talking" look when I mention Arrow's theorem.
This is especially striking in the Senate. Of the 50 states, 31 are to the right of the nation as a whole (see here <https://en.wikipedia.org/wiki/Cook_Partisan_Voting_Index#By_state>), so naively you'd expect 62/100 senators to be Republicans on average. Instead it's closer to 50. That's because Democrats have done a really impressive job of getting high-quality moderate candidates in swing states (and sometimes red states). Democrats have the vast majority of swing-state Senate seats (11 out of 14) and a few more senators in red states than Republicans do in blue states (though that could change this year).
Also -- quick fact check: in the six presidential election cycles you mention in Part II (2000 through 2020), the average electoral college advantage for the GOP has been 0.6 points, not 2 points. It was:
0.5 points in 2000
-0.4 points in 2004
-1.7 points in 2008
-1.5 points in 2012
2.8 points in 2016
3.8 points in 2020
(See here: <https://en.wikipedia.org/wiki/Tipping-point_state>)
What does ES mean? It's prefacing the bit in italics at the top of the post.
Epistemic status. (https://forum.effectivealtruism.org/posts/bbtvDJtb6YwwWtJm7/epistemic-status-an-explainer-and-some-thoughts)
Thank you.
I now remember I've come across this term before, but so infrequently I'd forgotten about it.
All this chicanery and maneuvering regarding trying to strategically appeal to different hypothetical voters seems to serve as a pretty good argument against electoral democracy in general as a desirable system.
Median Voter Theorem is like the Efficient Market Hypothesis: It's about equilibrium situations. Elections are far more noisy and slow feedback loops than the market, so it can take a long time to reach equilibrium: Republicans losing the house a lot is a PART of that feedback process that gets us to where we are now. We've only been having elections for a couple hundred years, and we've also not been in a stable political, economic, or social equilibrium for most of that time.
You also keep talking about parties being coordinated centrally but that's not necessary. Marginal new members of each party (whether from independence or from the other party) can gradually shift the positions of a party over time without any central plan.
Political scientist here... :-). Mostly just want to say how much I enjoyed your thoughtful take on this important concept within our discipline. But, I did feel compelled to point out one considerable error. The electoral college had a strong R bias in 2016 and 2020 but that is most definitely not a feature over all elections post 2000. In fact, in 2008 and 2012 there was a clear D bias. So, there's no way you should simply be subtracting 2 percentage points from the D margin every year since 2000. Actual figures nicely laid out here: https://centerforpolitics.org/crystalball/a-brief-history-of-electoral-college-bias/
The Median Voter Theorem fails IRL because the initial assumption of a one-dimensional political spectrum is false. It's not even a decent approximation. It's not close to true.
Politics is not even 2-dimensional, as depicted in the common square diagrams beloved of libertarians. You probably need 5 or 6 dimensions to get even a decent model.
One way to see this is that it is common for two candidates with very similar positions, perhaps even the same positions, to perform very differently based other factors. Those other factors are simply manifestations of dimensions other than the left-right dimension.
The missing ingredient of the Median Voter Theorem is "political capital." Regardless of the electorate's preferences, professional politicians are obsessed by the concept. To summarize quickly, it essentially means "popularity points" and leaders love to have them but their parties hate to accumulate too many because they want their core constituencies happy by spending their political capital on them. That's why parties often spend political capital, always trying to just keep enough to win the next election, by 0.1% if possible. That's the strategist's real dream.
This was most evident to me during Bush II's second term. Having won clearly in the 2004 presidential election, he started off with a lot of political capital, which his aides at the time urged him to spend in core rightist concerns (at the time, mostly law and order, school vouchers, stuff like that) but Bush essentially tried to maintain that political capital longer than many thought wise, and it was eventually consumed by negative newsflow, things like Iraq and Katrina -- as his aides feared.
I don't think "we should spend our political capital, because that is the point of having it" is the same idea as "we should spend our political capital, because it deteriorates over time".
What if our switch to voting the "lesser of two evils" and fine tuned national campaigns is the culprit? In the past we would, I think, vote for the candidate we liked, with less regard for party affiliation. Then national campaigns got better at scaring us about our opponents, but no better at making us like their candidate. So over time we find ourselves agreeing with the negative campaigning and feel pressure to not vote in the worst candidate. Using wedge issues (immigration, abortion, police/crime, whatever works) to make the opponent's look as bad as possible. Positions that are too unpopular disappear, leaving just the wedgiest of wedge issues, pushed as hard as possible. Over time this creates the culture war, because these topics have been optimized to find the hardest edge to push against an opponent.
It seems that negative campaigns are far more successful than positive, as much as I hate that. Jeb Bush was probably a very competent administrator and would likely have been a good president. But Trump calls him "low energy" and suddenly he's toast.
>But we may still find this surprising. The parties - especially the Republican Party - don’t feel like masterminds executing a complicated dance where they determine the exact amount of extremism the voters can tolerate.
They may not be doing math, but politicians (and parties) are doing this all the time. Why do you think the RNC scrubbed abortion completely from their platform this year? Did they all have a change of heart? No, they figure the median voter is currently more worried about abortion restrictions right now than they are concerned about fetuses, so goodbye pro-life, we can’t afford you this year.
I think you're leaving out at least two important factors:
1) Polls exist, giving parties information continuous information about where the center is, and parties must publicly signal positions to voters which also signals positions to the other party. As a result, no collusion (as commonly considered) is required for triangulation to occur. Also note that the US holds *very* long elections by global standards so there's a lot of time for the triangulation to occur.
2) polls are quite noisy, so it's difficult to identify an exact median position on any one issue. This uncertainty is amplified by the multidimensional nature of political positioning: combining positions on (eg) immigration, taxes, and abortion) there's theoretically *several* median cumulative positions available that a candidate might aim for based on their priorities.
Example: most Poli sci people I read thought (based on lots of polling!) That Democrats would need to tack to center on abortion to take advantage of the post-Dobbs situation, but... That's basically been wrong.
Half baked, but here is my off the cuff take:
1. The median voter is an emergent feature of the political spectrum.
2. The political spectrum has become highly polarized.
3. The political parties are not particularly ideologically coherent.
4. The culture war plus economic sentiment spits out an ever changing ideologically incoherent median voter.
5. We are just watching homeostasis ticking up and down like a thermometer set to 70 degrees.
Adam Mastroiani of Experimental History just published a study today on how Dems and Reps pass the "idealogical Turing test"; basically they can't tell when a statement was written by a real Dem/Rep or someone who is pretending.
His take away was that each side actually understands the other side's point of view, at least to a degree. But I think it more likely supports your half-baked idea that the political ideologies are emergent and incoherent, and that we Americans are all trained on the same political data...so we mostly regurgitate the same nonsense that everyone already knows.
That's not the Turing test. The ideological Turing test is whether you can write a statement explaining the views of the other side that is accepted as accurate by the other side. My recollection is that the studies I have seen that tested this found Republicans more able to pass this test than Democrats
I was just doing a quick summary of someone else's post. If you must correct someone, I'd suggest reading the post I'm referring to and correcting the author. Thanks!
No it's ok, I have a permit https://www.youtube.com/watch?v=E2kJtIIYmXc
The experiment actually does match up with what Bryan Caplan defined as an Ideological Turing Test.
Single-axis left-to-right as a model of political positions is an oversimplification, and enough people deviate from "choose the candidate who can win whose closer on this axis" to throw a bit of a wrench in the predictions that model would make. Notably, swing voters sometimes have a mix of left-wing and right-wing policy preferences (which makes "closest" hard to calculate"), make decisions on the basis of something that the parties don't particularly select for (which changes their preference from somewhat arbitrarily), have a bias towards change versus status quo (which changes their preference depending on who's currently in charge), or have other structural preferences that outweigh the usual left-to-right concerns (at least sometimes; e.g. Bernie/Trump voters, the Cheneys apparently).
I wish you had bothered to examine whether the median voter theorem is even true at all. The available data indicates that it is not. Most swing voters are economically left and socially center-right. And then you have a small sliver of people who are economically right and socially left. But the truest swing voters do not have a consistent ideology. This is why rhetoric and marketing are far more important than party positions.
https://fivethirtyeight.com/features/the-moderate-middle-is-a-myth/
Voters do not fall on a single left to right spectrum. There are multiple spectra on different issues. Also multiple interest group conflicts that do not fall on the left-right spectrum. Parties are complex coalitions. The median voter theorem is correct but too simplistic to account for more than a part of .the behavior of national political parties.
There's an interesting dynamic, too, in American politics (which is maybe basically what Scott referred to, just from another POV): very frequently, some party tries to inch closer to the median voter, then loses, and all their base say, "See?! We told you so! America will only support us once we go full-bore right / left!"
The point is that I think unsuccessful attempts to deploy the median voter strategy can lead the median *primary* voter to a more extreme place.
>In 1964, LBJ beat Goldwater 61-39. In 1972, Nixon beat McGovern 61-37. Even as recently as 1984, Reagan beat Mondale 59-41.
>But we already know that partisanship was weaker in those days. The Median Voter Theorem only works if you can reduce everything to a single straight line. If voters don’t care about the right-left spectrum, they might judge based on criteria like “Reagan is more charismatic”, and then if everyone agrees that he’s charismatic you can get 59-41 or 90-10 or whatever numbers you want.
Alternate possibility: Maybe polling has gotten more accurate over time, and the parties are more precisely following optimal tactics than they were before 2000.
Each party is trying to win while spending the minimum amount of money in the campaign. So if a party is ahead in the poles, they start spending less, and vice versa.
This can by itself explain the near 50-50 outcomes, though reality is probably more complex.
Why does it make sense for them to do that? If a party wins while having lots of money left over, surely that will signal to the donors to donate less next time? If I were a campaign manager I'd try to win while spending all of the money I have.
I think the donors are smarter than that. They probably follow the same pattern: donate more if their party is behind in the polls and vice versa.
I'm noticing a few factors that can lead to heterogenous candidates.
One, I saw Majromax mention: the measurement can be off. If I'm a candidate, the MVT says I want to adopt the median position; but that doesn't mean I know what it is. That's why I run polls. Polls are imperfect, though, and maybe my polls tell me something different from what my opponent's polls tell him. The catch here is that the R and D polls would need to consistently tell each candidate that the median position is R or D respectively, when in reality, one would expect them to update toward each other and be nearly identical by now, and they aren't.
Two, in the US, it's not just a median position, but a median position altered by an electoral college effect. If the median position is against tariffs on apricots but nearly all of those people live in four states, then the non-median position might get more electoral votes anyway. And this is also a reason each party might get different poll results; they can't afford to poll everything everywhere. Though again, one would expect them to have learned which issues are most important and probably agree on that by now, and again, they don't. So all this explains is why the true median might not be the one sought by candidates - in other words, why party platforms, even if you average their positions, will be consistently off from whatever a pure population poll would tell you - but I suspect this is a minor effect even so.
Three, the median position isn't just one issue, but many (as many comments have noted). A candidate can be median on five positions and still lose because they weren't the most important positions. Yet again, though, one would expect each party's pollers to know this by now and be equilibrated.
Four, each position waxes and wanes in importance relative to the others, and this can change in a matter of weeks. Interestingly, general positions matter, on the economy, civil rights, defense, education, and so on, and sure enough, most candidates *do* take the same position on the basics, which tells me that they *do* conform to the MVT! They only differ on more topical issues such as a war somewhere or bioresearch reform or ethanol subsidies or which metal should back up the dollar or whatever, and these will froth around so chaotically that the topicality-adjusted median position (TAMP) will drift, and each party *will* turn up different results depending on when and how they poll.
It might even explain why each party will appear to be pro-war one decade and anti-war the next; it depends on why each war started, and which party is out of power, and therefore able to flex fastest toward the TAMP. It also applies to other issues; if a party's position is sort of virgin and untested, it's also pliable, and able to adapt to new polls. So imagine the Median Voter as having a general position that both parties adopt, with a bunch of variations on it that the parties can only pivot to when they're not too committed.
Five, is the sort of flip side of One: each party can exert pressure on the median position. This works to the extent that voters often don't even have a position on some issue, but will go along with whatever they think the median position is, so here's this handy party machine to tell them, if it dares. Wellp, it dares. But only so far. No Democrat is going to say, know what, most of the US is pro-gun after all, so let's repeal all the gun laws except for felons! Each party is still anchored to its previous commitments (you could say the median position is strongly against waffling), and can't change them without a plausible justification. But that still leaves a lot of room to wiggle, for anyone who isn't a purist extremist.
And this helps each party. If the TAMP is against changing major positions and slightly tolerant of changing minor ones, then it's easier for a party to convince voters that the median position is the one that party has long favored, than for that party to move itself to the TAMP, which drifts from week to week anyway. And since it's easier to stick with the same messaging than to change it every week, it explains why parties have consistent, heterogenous depictions of the median.
In this frame, one can view the MVT as producing a median window, and each party trying to persuade voters that that median is narrower and closer to its side of that window.
If most voters are fairly content with the status quo then I've no doubt the Median Theorem is pretty much true.
But, as the Tory party in the UK recently found out the hard way (although, incredible as it sounds, most of the fools still haven't grasped this and learned their lesson!), centrism is a fair weather creed.
Just as beach goers enjoying the sun will scatter for cover in all directions when storm clouds gather and the rain starts, if the economic climate darkens or causes of discontent proliferate then votes will polarise and for all practical purposes one median splits into two, on left and right, like a cell dividing!
A few points:
1) Discussions of rational intent for parties fall down when they turn to Trump specifically. Trump is not a rational actor.
2) California Democrats' distaste for the national Republican party (or, mutatis mutandis, other such distastes) matters less in a gubernatorial race than in one for Congress, because a governor does not become part of a legislative caucus. That's a reason Arnold could get elected governor at a time when no Republican could any longer be elected senator from California.
3) Any consideration of why the Republicans couldn't win Congress in 1955-81 has to take into consideration the fact that, for most of that period, Congress was in fact controlled by "the conservative coalition," between most Republicans and conservative Southern Democrats. (This coalition was actually founded around 1938.) The erosion of the tradition that Southern conservatives had to be Democrats reached a tipping point around 1980, and that's one reason the Republicans started to win the body, because now they had Southern conservatives on their side of the aisle, and not just in informal coalition with them.
Political science is no more a science than economics.
A much simpler explanation: Both US political parties are primarily focused on being the 2 halves of the rulers of the United States.
It doesn't matter who wins for most federal apparatchiks so long as both parties defer to said apparatchiks.
This is also noticeable in the way both parties are against populists whether right populist like Trump or left populist like RFK Jr and Tulsi Gabbard.
The oligopoly of political parties dynamic describes Europe very nicely: for all the nice talk about diversity of representation - the machinations against AfD and Le Pen make it clear that the mainstream parties simply just don't want anyone else horning in on their action.
This also explains why both legacy Democrat and legacy Republican parties have paid exactly zero attention to what Americans overall want. These parties are not there to serve the people, they are there to ensure they can oligopolize the levers of power.
But all good things come to an end - the right and left populists have now merged under Trump, RFK Jr and Tulsi Gabbard. The dinosaur neocon warmonger Republicans are in the process of migrating to the Democrat party, which in turn is becoming ever more transparently the tool of the Federal Bureaucracy including literally dozens of ex-military, ex-FBI, ex-CIA types in office or running for office under the Democrat shingle: https://covertactionmagazine.com/2024/10/21/class-of-2018-cia-pentagon-democrats-continue-to-advance-hawkish-policies-in-congress/
The median voter theorem assumes that in a one-dimensional two-party setting, both parties will glue themselves to the median voter (known as Hotellings law).
However, in this situation the median voter has no incentive to join either party. Since s/he will get their preferences served without having to lift a finger.
Those who have an incentive to lift a finger (spend time and other limited resources on political activity) are those whose preferences are different from the median voter. Let’s call them firebrands. Both parties have them, and the further away they are from the median, the more firebrand they are.
Now introduce the concept of the median party member. The party members are those choosing the candidates that face the voters. (Simplifying, but that is what models are for.)
The median position within the party depends of the relationship between the firebrands in the party (who want to drag the party toward their own non-median preferences) and the pragmatists, who want power above anything else. (“A politician without power is an abomination” ). The latter will fight the firebrands to appeal to the median voter.
However, since the pragmatists have to compromise with the firebrands within each party (and sometimes the firebrands sweep the floor- in particular in gerrymandered districts, where there is no risk of losing to the opposition party anyway), the political profiles of the candidates in both parties will diverge from the median voter. In particular in non-marginal districts & constituencies & states.
And from there on you can do your own combinations.
There will be a long-run tendency back toward the median voter in two-party systems, but there will be a lot of hiccups along the way. As the political history of countries like Great Britain, New Zealand (before NZ changed to a version of proportional representation in 1993) and the US show in abundance. Edit:
One type of hiccup occurs when one party is conquered by a firebrand faction (defeating the pragmatists), which then pulls the party toward an extreme. In theory (following Hotelling's law) this should prompt the other party to move in, to occupy the median ground. However, sometimes this instead fuels/provokes the firebrands in the other party to move their party further in the other direction (defeating their pragmatists, who want to move center-wise).
An example was the radicalization of Labour in Great Britain after Thatcher and her neoliberal faction conquered the Conservative Party in 1979. It took until 1994 before the Labour pragmatists were able to defeat the Labour firebrands, and elect the moderate Tony Blair as their leader. Resulting in victory a few years later.
…An even more dramatic type of hiccup in two-party systems occurs when a faction of firebrands conquers one of the parties, and then move it across the aisle to the extreme of the opposition party! This happened in New Zealand between 1984 and 1988, when a neoliberal faction conquered the NZ Labour party from within and moved it to the right of the National (conservative) party. This resulted in a “I am more neoliberal than thou” competition between the parties in the following years, until exasperated voters in a referendum voted to change the whole system to a version of proportional representation. Which resulted in a shift away from extreme neoliberalism and back toward the median.
The US has arguably been through a light version of something similar. The Democatic party under Clinton moved toward a neoliberal position (Clinton’s welfare reform of 1996 being a high water mark in this regard), and from then on until 2016 the two parties competed about who was most consistently neo-liberal. When Trump, who saw an opening for a political entrepreneur, conquered the Republican Party from within, and shifted the party profile toward immigration control, protectionism and disengagement from non-vital wars – very far from neoliberalism. And understandably more popular with the US working class, where the median voter is located.
...Which in its turn has resulted in the Democratic Party now scrambling back toward the median to regain their working class constituency, who now suddenly have another option. This process back toward the median was evident under Bidens’ “build back better” renewed industrial policies. And regardless of who wins November 5th , this “return to the US median voter” is likely to continue.
This theorem is an interesting angle to look at the elections and democracy in general. But I think some points are missing here.
1. I think/hope democracy is about power to the people. Insofar it makes sense for a politician to adapt their opinions on certain topics to better cover the opinions of the majority. But on the other hand, politics and elections especially should be a competition of ideas, shouldn't they? A credible politician is the one who can convince the voters of the superiority of their (own, not the voter's) stance in the political spectrum, or not? Isn't it a bit of a misery that election seems to be the voting for the best weather vane, at least according to the median voter theorem? I mean doesn't it look like deception when power-hungry politicians change their position just to get elected? Anyone here who wants to be ruled/represented by such a person?
To say the least, the median voter theorem assumes a static public opinion. To what extent is this actually the case?
2. I also think that media is playing an important role here. The center/mainstream of accepted opinions/major consensus narrative has become narrower and shifted more and more to the right in the last thirty years (at least in Germany, how about the US?). I think everywhere as a reaction (Covid has been a catalyst) people retreated to their respective filter bubbles. As an example, in Germany, the main topic in the media is (illegal) immigration. This is however, in my opinion, pretty detached from the problems this country actually has. Therefore, it is a question, in which way the actual policies are still influencing the voters' opinions/votings because of the media's influence and how much it is just the making of clever spin doctor's. I remember the "Dancing Queen" Theresa May, it only extended her time as prime minister for one year.
People are flooded with impressions on a daily basis, they are weather vanes themselves. On the other hand, I am to young to compare with the "good old times".
3. I also want to make the point that it pays off to care about people. In Graz, Austria's the second largest city, the mayor is a communist. And this in a country where the neo-fascists have gained the majority in the last nationwide parliamentary election. I guess the local communist party doesn't have particularly communist positions rather social-democratic ones, I would say. Still they had to overcome a pretty high threshold, but their persistent actions and support for common people in health care and affordable housing made them credible and electable. But I'm afraid this is limited to be effective in local/regional politics.
Are there common sense voters in this age of heightened partisanship and what seems to be strong tribal affiliation? I for one do not fit neatly in to the Dem or Repub Parties, but will not vote for Harris and if i vote will vote for Trump. I noted somewhere that Martin Gurri who sat out the last two elections said he will vote and it will be for Trump.
Interesting, as his son Adam seems to be decidedly anti-Trump.
Isn’t it difficult to veer left or right credibly? If the Republicans started supporting all the Democrat positions on hot issues, which party would the voters that care about those issues vote for? And the same in reverse.
The Republicans have lately tried to paint themselves as pro-union defenders of the working class. It has sort of worked, but at the cost of basically purging the Republicans elite. And the Democrats have actually been assisting, by clearly supporting policies and displaying attitudes that favor the elite.
I'm wondering whether this eventually results in the Democrats supporting upper middle class tax cuts.
Trump won the Republican nomination in 2016 despite his lack of support for Republican positions.
He won the election, too.
Yes, not having any baggage of a history as a Republican politician helped him in the general election. But we were talking about how one can win the primary while tacking to the center.
Sometimes “moderating” makes them a maverick, sometimes it makes them a RINO.
I thought we were talking about the general. Why would tacking to center even be a temptation in the primary?
Was Trump tacking to center, or rotating the axes of the left/right spectrum? He tried to get support from traditionally Democratic voters, but not by adopting Democrat policy ideas. Trade and immigration policy was barely a spectrum, more of a consensus, before Trump trampled it.
Trump got a lot of support in the primaries from people who had never voted in a Republican primary before.
Okay, but was he tacking left or realigning the axis, redefining what the left-right divide means (or maybe taking off the mask that hid what it means)? The median voter theorem is not about redefining the terms of debate. It is about trying to occupy the center of a stable and well defined spectrum of voter opinion. Trump was more disruptive than that. He did not try to make his ideas sound like a more sensible version of what the Democrats have been saying.
How about multiparty systems and non-first-past-the-post voting systems?
In the aftermath of the 2001 elections here in Italy I would often hear the opinion that the mainstream left was only defeated because the extremist Communist Refoundation party received 5% of the popular vote.
That sounds like a clear-cut case of what Scott says doesn't happen (extreme voters punishing the mainstream parties for being moderate).
This happened during a period in which Italy was experimenting with a mixed first-past-the-post and non-first-past-the-post system. It's possible that people were still voting out of habit under the assumption that smaller parties had a chance, as if there were no first-past-the-post.
"Maybe after decades of humiliation and failure you could convince someone, but I don’t see the history of parties suffering decades of humiliation and failure before finally agreeing to turn around."
I don't know enough but weren't the Republicans on the left until about 1920? And didn't they veer right in the 1970s to get votes from Southerners discontented with civil rights reforms?
No, per DW-NOMINATE the Federalists are the original right-wing party before the Republicans (the primary axis of politics is economic), and indeed John Adams is recognizably conservative (if not as far right as Hamilton) while Jefferson was a comparative radical who was enthusiastic about the French revolution and thought we needed periodic revolutions rather than being bound by the dead hand of a Constitution written by previous generations. Nor were the Republicans left of the Democrats prior to 1920, McKinley was definitely to the right of Bryan. Woodrow Wilson, who gave us the income tax and tried to bring us into the League of Nations, was on the progressive left (part of why he's ranked so highly by historians despite being a segregationist).
Mapping current labels onto historic parties leads to poor analysis. The Federalists were a pro business, pro big government party. In 1820 Monroe won every state basically unopposed. In 1824 there was a new split in the Democratic Republican party due to the contested election.
In 1900 Republicans were to the left of Democrats but in 1912 everyone was a variant of progressive (by early 1900s definition). Because Teddy Roosevelt ran against Taft, his followers drifted from the Republican party and eventually became aligned with the followers of FDR in the same way that many of the Anti-Trump republicans are becoming Democrats.
The Federalists were electorally viable when the US was a timocracy where only property-owners could vote. Democrats dominated after property restrictions fell away. The government at that time relied on tariffs for revenue, so prior the 16th amendment it couldn't tax the rich to buy off voters the way big government could later.
> In 1900 Republicans were to the left of Democrats
No, as I said, McKinley was on the right.
The federalist were not viable because the major issues that defined the difference between the Federalists and Democratic Republicans ceased to matter (the end of the French Revolution). The Democrats split from the Democratic Republicans (the Whigs also split from the Democratic Republicans). The Democrats dominated until the civil war because there was not clear ideological groupings.
https://en.wikipedia.org/wiki/Second_Party_System
In 1900 Republicans were the party of big government, while the Democrats favored ending the gold standard and were opposed to imperialism (the American occupation of the Philipines). Republicans were still the party of civil rights. Ideologically the Republicans of this time were probably most similar to Late Obama and Hillary's campaign (pro business pro government) while the Democrats were closer to Trumponomics.
https://en.wikipedia.org/wiki/Third_Party_System
https://en.wikipedia.org/wiki/Fourth_Party_System
William Jennings Bryan was the Democratic candidate in 1900. He had advocated a graduated income tax, just as the Socialist Labor Party and Populist Party (which also nominated him) had done. The Populists also advocated that the government create deposit-taking banks so people wouldn't have to use private ones, and that the telegraph system be nationalized, and that the government employ laborers for public works during times of economic depression.
> The common-sense answer is probably right, but one thing still bothers me: how did the Republicans keep losing both houses of Congress every year from 1955 to 1981? Wouldn’t they have shifted left during this time? I don’t know, this was before our current partisan era, and there were lots of weird deals around civil rights going on.
I think the way to understand this is that US political parties were not meaningfully coherent entities. They were brands, but lacked sufficient overt hierarchical control or more decentralized perceived-unity-of-interest-within (by voters, influence, politicians) to model as a single interested actors making strategic choices. Today, the two parties still mostly lack overt hierarchical control (although Donald Trump may be changing this for Republicans), but perceived-unity-of-interest-within is strong, so they do approximate strategic actors.
Partisan outcomes before 1980 were more side effects of an "all politics is local" (or factional) world. Shifting coalitions within (and across) the parties were the strategic entities. Whether the result was a "win" for Democrats or Republicans was a by-product of all that, not orthogonal by any means, but not strongly correlated either with the players that mattered more in those times.
Some good political science that describes this shift is Lee Drutman, "Breaking the Two-Party Doom Loop".
The parties have gotten WEAKER over time, as primaries have displaced conventions.
The parties were stronger in the sense of being able to choose their national candidates. But that's all they could do. All those delegates at all those conventions were pursuing local power, primarily over local concerns, often in ways that copartisans in a different geographies would find abhorrent. Dixiecrats and North Eastern liberals shared little but a party. Goldwaterites and Rockefeller Republicans the same.
I just linked to an actual data graphic on close the votes of northern & southern Democrats were to Republicans, and now I have to do so again:
https://en.wikipedia.org/wiki/NOMINATE_(scaling_method)#/media/File:NOMINATE_polarization.jpg
What do you mean to suggest? The graph shows the two parties were lots more similar to one another than now, when you average through them. (It'd be fun to see a breakdown of factions for Republicans as much as Democrats).
Party differences were less than now, cross-partisan factions more frequent. Party membership was much less predictive of behavior and alignments than now, even though yes, national parties could select their candidates more strategically.
It's not the case that they "shared little but a party". They really did vote similarly, even if the parties have gotten more distinct since.
"So if candidates are rational,"
A huge percentage if voters are not. Politics is the mind killer. Blue team vs red team, etc.
I think what this misses is the median voter does not have median *opinions*. The median opinion on abortion is that it should be legal, probably up through some threshold time like viability or sometime in the 2nd trimester, and legal after for the health of the mother. And the median opinion on climate change is that it's real and the US should try to greenify energy production but also make gasoline as cheap as possible.
The median *voter* thinks abortion should be illegal always and we should outlaw coal tomorrow. Or that abortion should be legal in all cases no questions asked but also wind should be illegal because it kills birds. And while sure there's some theoretical multi-dimensional hypercube median voter you can reduce to a number line, it's not clear that adopting the median opinion on every issue is the way to get the median voter (even setting aside the areas you discuss).
I think one of the big thing that has changed, Ezra Klein's Why We're Polarized is great on this, is that previously, political parties and people were less aligned across issues than they are now. We know that abortion being legal is the left-coded opinion and that cutting taxes on the rich is the right-coded opinion, but there's nothing necessarily connected between those two issues. You could imagine a country with a far-left economically, liberation theology Catholic voting bloc with strong pro-life views but also strong income inequality & progressive taxation policies. It's just not the society we happen to live in.
Polls used to show that voters opinions on diverse range of issues were less ideological and less correlated than they are now. One fascinating factoid I remember is in the 90s, Democrats and Republicans (I don't remember if controlling for race or not) had indistinguishable views on the OJ Simpson trial. Compared to now (or recent history), and there are huge polling differences between the parties on whether American Sniper or Selma should win Best Picture. A person's opinion on abortion used to tell you very little about their views on immigration; now it tells you a lot.
I think that Daniel Ortega has rebranded himself as a pro-life Catholic while still being far on the economic left.
Abortion was never broadly legal in Nicaragua, as far as I know, including during the Sandinista era in the 1980s, but yes they toughened up the laws in 2006 or so (banning it without exception, including in cases where the mother's life was at risk), and yes the Sandinistas did support that.
Abortion is also illegal in communist-ruled Laos, although as a general rule, worldwide, "Left" political parties have been more aligned with feminist values, and thus more likely to favor abortion rights at least to some degree.
I think immigration is actually a better example: questions of nationality and ethnicity really do seem to me to be completely orthogonal to the economic left vs. right, whereas you can at least draw a reasonable case why someone who cares about socialism should also care about women's rights (starting with the historical note that 19th century socialists often were critical of traditional gender roles and expectations).
Doesn't Catholicism prescribe economic leftism? What's the conflict?
> I said above that the real world doesn’t follow the Median Voter Theorem. But is that true? Certainly both parties don’t have indistinguishable and perfectly centrist platforms. But here are the last twenty years of election margins of victory
This is not well stated. As the image itself states, it shows the share of the national popular vote going to each party.
But the margin of victory isn't the difference in number of votes you get. That would only be true if you could win by getting more votes. The margin of victory is the number of votes that would need to change in order to change the winner.
This drew a lot of commentary in 2020, when Biden got 5 million more votes than Trump did at the same time that his margin of victory was 0.08 million votes. (See e.g. https://www.washingtonpost.com/graphics/2020/elections/vote-margin-of-victory/ ; take a look at that URL.) Differences in popular vote totals are just an epiphenomenon. Winning by 80,000 votes out of 158,000,000 votes cast is an absurdly thin margin.
> The common-sense answer is that it would take a while, there would be a lot of negotiation, but five or ten or thirty years later the Republicans would get their act together, shift left, and start winning elections again.
You know, one of my favorite bits of video game dialogue comes from a platformer in which the protagonist leaves a conversation by pleading "pressing business far, far to the right of here".
Left and right, with their total lack of any objective difference, are good labels for political parties.
This example is a case where the assumption of one-dimensionality makes examining the problem pointless. You need a different model. There is zero chance, in the scenario you describe, that the Republicans respond by "shifting to the left". Rather, policies will move around between the parties, and "left" and "right" will then be defined by whatever state the party platforms end up in. Juicing the political power of Guam, Puerto Rico, and Washington DC should end up pulling both parties to the left (as defined today) on economic redistribution at the same time it pulls them both to the right (...) on social conservatism.
Polarization might have fucked with this. I have a relative who was as 10% off median as you can get; 60/40 split D/R. Real traditional god bothering farmer type but also very concerned with environmental issues.
In 2000, he was very excited to vote for Mccain, but history took the stupid path and we got dubbya. Next comes 2008, and after 8 years of dumb, venal administration here comes Mccain again! And his running mate is who it was.
At this point he hard committed to only ever voting D until team R de-Palinized. He made a solemn vow: If St, Francis (R) was running for dog catcher, he was voting for Caligula (D).
I think the 2008 election was an inflection point where the parties stopped sharing a common reality, and any median voters that still exist only exist because they are low information. Unless there is some sort of correction, it will probably get worse over time instead of better.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
This isn't true. California did elect a Republican Governor, twice, while being a solid blue state. It's fairly common for solidly red or solidly blue states to elect governors of the party opposite their overall lean. This happens less frequently with Senators, but still happens a nontrivial amount (with Democrats generally being more successful at getting elected Senators in red states than the other way around, related to your observation about how the Senate should be stacked against Democrats, but doesn't actually have a Republican lean in practice).
Schwarzenegger being a super famous movie star is sort of the exception to the rule.
Schwarzenegger is the only recent example in California only because one state doesn't go through Governors that often. The current Governor of Vermont is a Republican, and the Governors of Kansas and Kentucky are Democrats. Go back to some random year in recent history, and you'll probably find more examples like this that were in office at the time.
Fair enough.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
In previous decades, southern Democrats could run under a national label while adjusting their campaigns to match the local tendencies. Why can't Californian Republicans adjust in the same way?
Perhaps it's that state-level Republicans want to "graduate" to the national party.
But I think it's mostly tribalism. Republicans in Cali would rather maintain their alignment with the national Republican tribe then take power by severing that connection. Nobody wants to be Joe Manchin or Larry Hogan.
This question was pondered by David Schleicher and Bryan Caplan:
https://www.econlib.org/archives/2008/12/schleichers_mod.html
The conclusion is that it wouldn't work, as Californians would still vote for the official Democrat even if the California Republican party took to nominating Democrats.
I think it's because the large cities outweigh the rest of the state, which tends (citation needed) to vote Republican. To capture the urban vote, the Californian Republicans would need to move so close to the Democrats' positions that they'd probably lose the rural/non-urban vote, and wouldn't be very Republican by national standards.
In which case, why vote for ersatz Democrats when you can vote for real Democrats?
I suspect the answer to the civil rights era question about Republicans shifting to gain congressional control is that because there was less partisanship, they cared about *conservatives* controlling congress, not *Republican conservatives.*
I believe DW-NOMINATE still showed that the primary axis of political disagreement at that time was the one that roughly mapped onto party labels rather than regionalism. So they most likely cared more about having Republicans in charge even if they were northern moderate Republicans rather than having a southern Democrat.
Part of the tricky thing about applying this theorem, I think, is that there is a qualitative difference between the median voter and the strictly partisan voter. The median voter is not a guy who has an equal balance of right/left positions or is equidistant on the issues from each extreme. The median voter is a guy who maybe has an elementary, abstract understanding of like one or two issues (usually acting as signifiers of moral character in a particular candidate rather than a schema for policy) and is voting based mostly on how much money he has in his pocket after a 2-4 year run. Strictly partisan voters in America tend to be more engaged.
I think that the main reason that the median voter theorem does not apply is that nothing is ever one-dimensional in politics.
Consider people voting on how spicy a pasta sauce should be. The MVT applies, you have exactly one axis, "concentration of capsaicin" and both the spicy and the bland party compete for the median spiciness voter, so he gets his wish.
Even for a sauce, this would not be realistic. People will want different amounts of garlic, mushrooms, animal products and all kind of stuff, so you quickly go to a high-dimensional space where the MVT does not apply.
I think politics is better modeled as building a coalition of various special interests. In two-party systems like the US, the coalition building happens before the election, while in proportional representation systems the special interests are kinda represented by parties, which can negotiate coalition agreements after the votes are cast.
The dynamic which pushes the coalition size towards 50% for either party would be that if you have a secure majority, strive within the coalition becomes the main political battlefield. After all, the other members of the coalition do not exactly share your interests, and if you managed to kick some of them out, you could enact more of your policies. If you are polling at 75%, that will increase the expected amount of political impact your special interest group has. If you coalition is polling at 45%, you actually need their votes to win, so it is in your best interests to make concessions to them.
I think that you can sometimes see similar effects in interstate anarchy international relations as simulated by Paradox games. It would be in the security interest of every polity to conquer everyone else, so they can feed on their riches and don't risk getting conquered themselves. However, they generally lack the military power to accomplish that. So they form (and sometimes, break) all forms of political alliances. While I am not an expert here, I think that a typical outcome is that you get blocks of roughly equal size.
For example, if you have an island with three polities with powers of 50, 30 and 20, the natural blocks would be 50 vs a coalition of 30 and 20. 50 would have little to gain from entering an alliance with either, its security interests are met adequately, and if its enemies were to war among themselves, why would it stop them? Likewise, 20 will recognize that an alliance with 50 is foolish: after 50+20 defeat 30, 20 will find itself in a very unfavorable situation. (Of course, election politics are only vaguely similar. In particular, while a military defeat can critically weaken a country for decades and thus shift the power balance, defeating a special interest group at the voting booth will generally not shift the power balance. Sometimes, the opposite is true: while Roe was the law, being pro-choice just meant defending the status quo of 1973, which did not seem to hinge on any election outcome. With Dobbs, moderate pro-choicers are more likely to align their votes along that issue because they want to change things back to the way they were.)
Or consider two local polities (A and B) in a long-running conflict with two large blocks (0 and 1) expanding their areas of interest to cover them. If A gets support from block 0, then B would be foolish to try to also get support from 0. After all, if the block has done its homework, it likely has sound strategic reasons to prefer an alliance with A, and to convince them that B would make the better partner is bound to fail. The better strategy would be to form an alliance with block 1, which has a direct interest in seeing the efforts of 0+A fail. In election politics, this is analogous to an issue becoming partisan. For whatever reason, the pro-lifers (A) become part of the Republicans (0). Then the logical step for the pro-choicers (B) is to become part of the Democrats (1).
Of course, real life in both international relations and domestic politics is obviously more complex than all that, with shared history, cultural, religious or ideological ties gating which alliances are feasible and which are not.
What is capsaicin doing in pasta sauce? These American innovations! 😀
https://en.wikipedia.org/wiki/Arrabbiata_sauce
A perfectly Italian pasta sauce, which contains red chilies which in turn contain capsaicin. :)
Looking it up, the origins seem to be obscure. Wikipedia says it was popularised in Rome in the 1950s-60s when spicy food was popular. Another site claims it was invented in Sicily in the 1920s.
Since most "this recipe was invented by X in year Y" lore is invented or muddled, I'm going with "this was a new recipe invented in the 50s to cash in on a fad".
Depending on how long we want something to be in existence before it becomes 'traditional', then Arrabbiata may or may not be 'traditional'. Is a Hawaiian pizza traditional Italian cooking? 😁
"It's an older sauce, sir, but it checks out"
Regarding your last paragraphs, it's possible for one party to get stuck in a bad equilibrium where they can't win running on their current platform, so everyone who's pragmatic/a hack leaves for the other party, leaving only people who are too ideologically tied to the current platform to be willing to change it to make the party more competitive. This persists until the party collapses and is replaced by another, a coherent group of outsiders who like a weaker version of the party's ideas do a hostile takeover and make the platform more pragmatic, or there is some major change in the political environment which suddenly makes the previously hopeless platform viable again.
Democrats veered right under Clinton, after having lost the last three presidential elections.
The recent convergence is probably driven by improvement in data quality. In reality, everyone's making very approximate guesses at where the median voter is and where potential party nominees are relative to that median voter. In the last 30 years, compute has become vastly cheaper, allowing ever-more-powerful analysis of fine-grained data to determine exactly where that median voter is and how to target optimally. A lot of the lopsided elections of the 20th century simply stemmed from bad estimation of where median voter is.
There's also a dynamic of investing in the future: trying to move the median voter to you, rather than the reverse. A party will implicitly or explicitly lose an election taking a principled stand in order to grow the movement and win in future elections. Goldwater in 1964 can be understood this way.
Your diagram shows 50% Republicans and 50% Democrats leading to a candidate at the 25th percentile along the partisan axis against a candidate at the 75th percentile. Unfortunately, things are worse than this. A whopping 50% of American adults aren't members of either party. If we assume these people are mostly in the center and closed partisan primaries we get candidates at the 12.5th percentile and the 87.5th percentile. There's a whole political advocacy organisation focused on allowing unaffiliated voters to choose a primary and vote because having 25th and 75th percentile candidates would be SO MUCH BETTER than where we are now.
I don't think much more than 50% of American adults even vote in most elections. Unless you mean 50% of registered voters aren't members of either party?
> Democrats and Republicans don’t have platforms exactly identical to each other and to the exact most centrist American.
Sure they do!
They both think that slavery should be illegal. They both think that capitalism is better than communism. They both think we should be a democracy rather than a monarchy or theocracy. They both think that women should be allowed to own property and that marital rape is a crime. They both believe in closed borders. Etc.
The two parties *feel* radically far apart to someone stuck living under them, but this is just the standard emotional reaction to inter-factional conflict (see the old Emo Philips joke).
Sure, they disagree on... check notes... which sports team trans kids should play on, I guess? And a few other narrow wedge issues that are specifically chosen to be things that are close to 50/50 splits in the population, bimodal distributions where the 'centrist' position is itself unpopular or incoherent.
But out of the space of all possible platforms, or even just all police positions that actual political factions have taken across human all human cultures and history? The venn diagram of the two party's policy platforms is indistinguishable from a circle, at any level of zoom larger than an electron microscope.
(another way of saying this is - the theorem implies we should have a narrow Overton Window, and we in fact do. Whether the parties are close together or far apart depends on what scale you are using, and is not well correlated to how close or far they *feel* to a partisan.)
The voters in the two parties disagree on the distribution of scarce resources in society. One party want most of it to go disproportionally to "traditional Americans" and the other party wants it to go disproportionally to marginalized communities.
Most of the rest is detail.
Median voter theory applies to the position of the parties/candidates, not the voters.
Yes, that's true. However, I think most political parties attempt to adopt platforms that appeal to the most voters. Hence, the platforms of the Republicans and that of the Democrats broadly reflect the priorities of their base.
> Instead, Democrats are often pretty far left, and Republicans pretty far right. What’s going on? I think at least three things.
None of these three points are wrong/objectionable in their own right, but I think you are missing a central point here, which the WIkipedia entry mentions at the end of the first paragraph: the theory assumes the use of a Condorcet voting method.
First past the post is not a Condorcet method.
From the intro to Duncan Black's original paper on the theory:
>To develop our theory, we must make some further assumptions. Our major assumption will be that each member of the committee ranks the motions in a definite order of preference, whatever that order may be.
The theory is premised on ranked-choice voting. It does not straightforwardly apply to first past the post voting, as we use in the US.
This is why, as the wiki says, modern political scientists often talk about the median voter PROPERTY rather than the median voter THEOREM. The median voter property is possessed by any voting system which selects the candidate closest to the center of public preference. Many voting systems achieve this property, but the system we use in the US does not.
The general idea is that this is one of many properties it is desirous for a voting system to have, and it is used as a point to argue for or against different voting methods in political science.
I disagree with the framing of the parties' actions as product of some rational thought process informed by the median voter theorem. It's much simpler than that: politicians' most important skill is to accurately read the room, and they are doing it constantly. It's just a constant feedback process.
Take Roe V Wade being overturned: to argue whether it's a R+0.1%, D+0.1% or a D+2.0% is necessary to do the calculus, but that's not what went down in the GOP's elites' minds. Instead, it was more like "polls say it's unpopular, media coverage is terrible, even Fox isn't defending it much... unequivocal support only by our crazies. Well, that went worse than expected. Ok, drop "fetus is a human life" and go with the "states' rights" angle"
Roe v Wade was decided, and then overturned, not by politicians who face re-election but instead unelected judges.
Who were appointed by a man seeking re-election.
Roberts & Alito were both appointed during GWB's second term.
Why do you mention only them? All three Trump appointees voted to overturn Roe.
Incidents are individually consistent with a theory do not prove it. Seeking them out is just an example of confirmation bias. Instead, scientifically we seek out examples that can DISPROVE a theory. This idea of falsification goes back to Popper.
I'm disagreeing with *your* theory.
Which was all hunky-dory when the decisions were favouring the popular causes of the left. Once the court stopped being an arm of the Democratic Party, suddenly it was a national disgrace that it didn't represent the demographic makeup of the nation and there should be term limits for judges etc.
I see it as an artefact of neoliberalism: both parties essentially agree on all the important factors (i.e. being neoliberal). This explanation works even better in other countries, none of which have the divided two-stage (primary) election system. This is also why the median voter theory breaks down prior to the neoliberal era.
Edit: I reckon my argument doesn't need to define neoliberalism to make sense, but since that's what people seem to want:
Neoliberalism is return of liberal ideas, previously in vogue from about 1860 to 1930, when socialism became the predominant ideology of those seeking to improve society with *new* ideas (conservatism always exists and seeks to improve society with *old* ideas, and isn't automatically wrong to think so).
Liberalism is based on the idea of the maximisation of the right of the individual, chiefly through two mechanisms: the removal of laws the restrain individuals from taking actions that do not obviously harm others (in the short term); and the strengthening of individual or corporate property rights as opposed to government appropriation or some common good.
Is "neoliberal" just defined as "all the stuff that the major parties agree on" though?
No, because they didn't agree on this stuff prior to the 1980s.
One way to define "Neoliberal" today is "Whatever preserves the status of local elites in the West.
I disagree and have updated my post to define neoliberalism.
In the US, before the coming of Maga populism, conservatism and liberalism were largely the same thing, because the US was established more or less on liberal principles, so that is the political tradition here.
Yes, the US has always stuck a lot closer to its founding liberal principles than most countries. But FDR's New Deal era did hew much closer to socialist principles while it lasted. The US wasn't immune to international trends.
Trump probably did move left when Biden withdrew, but did so by not announcing any policies as extreme as they would have been. It’s a bad look to shift course too visibly. But my sense is that people’s platforms evolve over time, as they reveal their policies a bit at a time and see the electorate’s reaction. And those changes would be shaped by how popular a party is. A politician would be more scared to follow an unpopular principle when they’re doing badly in polls. But I think this looks like doing more vs fewer unexpected things, not like doing unexpected things in different directions
Katherine Gehl and Michael Porter seem to support something along the lines of the collusion hypothesis, describing America’s two party dynamic as a duopoly benefiting the existing players in the multi-billion dollar politics industry. See https://gehlporter.com/
I came across this argument 5 or 6 years ago via this episode of the Freakonomics podcast and it stuck with me: https://freakonomics.com/podcast/americas-hidden-duopoly-2/
So I guess the question is how does the Democratic party gain from a challenge by Trump? Does he help them mobilize their core, thus preserving the Blue status of Blue states?
If that's the answer, will they ride this horse all the way over the cliff?
Yes, and yes?
Currently I'm seeing something bruited about online that an ex-chief of staff (or something) from the Trump administration is saying that Trump expressed admiration for Hitler, and at this point I'm going "You know what? Sure, whatever"and I'm not even reading the stories because I can't muster up the energy to be outraged.
I had an exchange with a person on here about "What will Harris do once elected?" and their main (and only) point was "She won't be Trump. Vote for her because Not Trump. Whatever she does or doesn't do if elected, will be What Trump Wouldn't Do".
There's been eight years of constant whipping up hysteria about him, and I've always thought that ignoring him after 2020 would have been the best response: "Donald who? Oh, that has-been? He's not relevant to anything now". But no, apparently the best thing the Democratic Party can think of is "Vote for us, we're Not Trump, because did you know? He's the Devil!"
Right over that cliff all the way down to the bottom and then keep on going down till you hit China.
RE: Good political science on electoral systems.
Classical economics is built on the 'rational economic agent' that moves always in the direction (in choice space) which maximises its utility. While this model has some predictive power, Kahnemann and many others showed that people hardly ever act like this. And hence economists cannot predict recessions etc. because their models don't reflect reality.
Social choice theory has the same deficits. Arrow's Impossibility theorem is fairly trivial mathematically (and might show up in an undergrad course on discrete topology or similar, but is otherwise never mentioned). It gets talked about a lot and somehow entered the popular consciousness, but it has hypotheses that are never satisfied, and so doesn't tell you much about what's going to happen in the world. The median voter theorem is a response to Arrow: under still unrealistic assumptions, you can satisfy the median voter. But only if the median voter is a mathematical abstraction (insert the old joke about the mathematician predicting the winner of a horse race by assuming the horses are identical spheres).
The US system is particularly archaic, complex and unrepresentative. (I notice that most of the responses to the post are pointing out details of the US electoral system that they feel were glossed over in the post.) There's an understandable desire to reduce it to a mathematical formalism which can be analysed. But this type of analysis throws all of the details that matter away - I don't see how it could lead to meaningful insights. The type of analysis that Nate Silver did in 2016 and 2020 was, to my mind, a more sensible way of understanding and predicting outcomes for the election. The voting reform movements in the US, particularly at the time of the last redistricting used some interesting mathematics to show that some districts were gerrymandered. But that's slightly different than this discussion.
I guess it's important to work out what the goal is of the political scientists. I think it's less to build models that describe or predict real world electoral systems (particularly that of the US). It's more like mathematical research: here are axiom and here are conclusions. The applications of that to the real world are problematic... but where the media interest is.
This is interesting as I have thought for years that when one party moves to a more extreme position, the default response seems to be for the opposition to move in the opposite direction, when it would make so much more sense to move either to the centre or to cross the centre into enemy territory a little. That way you keep all your own voters and pick up voters from the other side who aren't happy about the new extremist policies. Let's take an example just for the sake of argument. Let's say the Democrats promise completely open borders. Would the Republicans be better off promising closed borders and mass deportations, or better off promising limited immigration for well-qualified and vetted applicants? You can think up lots of similar examples.
1) Both parties in the US are indeed quite similar from an outside perspective. From a French point of view, they offer both slightly different flavour of ultra-libertarian economic policy for example. There is extremism and polarisation in the US but there is also a broad agreement around an extremely narrow overton window on many issues.
2) As is often the case, your viewpoint is very US centric - not your fault, but you would greatly benefit from a deep dive into the political system and history about half a dozen other democracies. And indeed the median voter theorem seems pretty applicable to countries like France (where the two main parties were so convergent that Macron basically replaced both of them with his own), Germany or the UK. Don't mistake the specific of the US system for deep laws of political science.
2) There is obviously a feedback loop between what the voters want and what the parties propose. Parties move toward the median voter but voters also gravitate toward the issues that are emphasized by their preferred party.
Blue states and red states create large constituencies that are against their party moderating because those specific constituencies have little to gain from it. What I mean is... if you're a California Democrat, why would you moderate? If you're a Kansas Republican, why would you moderate? Your party dominates the state you live in. Yes, during federal elections, this lack of moderation can hurt. But it tends to cancel out to some degree, with both parties suffering some from it. Republican governors of red states often have no real reason to moderate, and likewise with Democrat governors of blue states. And these are influential figures even at the national level (Newsom and DeSantis are good examples of this). The Republican politicians and voting blocs that determine the overall direction of the federal party are more likely to arise out of red states, and even when they don't (like Trump), the primary process forces these blue state Republicans to shift some to the right. Similar situation for the Democrats.
If you look through the history of American politics, you'll see many similar periods to what we have today here, particularly on a North vs. South basis.
Now, the median voter theorem is undermined in the opposite direction as well. Democrats in red states can feel a strong need to signal that they're good progressives even though they live in a relatively conservative part of the country. Republicans in blue states can be similar here as well.
So many charts and number lines that put Democrats on the right and Republicans on the left. 😔
Makes me wonder if subconsciously the people creating these charts are going “well of course democrats are positive and republicans are negative! Of course democrats are right!”
Something this simple isn't going to work very well in describing a recursive system. You have to factor in that voters are watching the polls, talking to other voters, and strategically calculating how best to organize their votes. You see this in the primary period discussion we always have over "electability" vs. doctrinal purity. If some set of voters perceive the other side as "going too far" (this describes both sides today) then there will be a tendency to match that with their own veer toward more extremism. If, however, they perceive that the other side is playing it safe for strategic reasons (presumably so they can veer more extreme after the election, those scoundrels) , then there will be more pressure within their own party to do that too. Note that these are popular *perceptions*, not realities, and so are susceptible to the full range of propaganda and misinformation that bubbles up every election cycle.
I am old enough to remember the era when everyone was complaining that there weren't real differences between the two parties. No matter who won, they used to say, you ended up with more or less the same policies. That ended with Reagan and Gingrich.
We are still playing out what happened in the 1990's.
“Isn’t this weird? In a world with so many numbers - 55-to-45, 80-to-20, 99-to-1 - all of the past three elections have been approximately 49-51. So maybe the Median Voter Theorem does work.”
I don’t see why this result suggests that the median voter theory works.
If both parties are trying to win the election, and they can respond to each others actions, we could expect that the margins of victory would be small, and would get smaller over time as parties get better at strategizing, responding to information, etc.
The top baseball two teams going into the world series tend to be roughly as good as each other, because a competition for who’s best at something is likely to give those results.
So sure, we see close results. This seems true in a world where the median voter theorem works, and in one where it doesn’t. If, say, the best strategy to win elections was to make the most delicious cookie, we would expect both parties to get better and better at cookie-baking until they got pretty close to a 50-50 match. If the solution was to run the tallest candidate, we’d expect them to find the tallest and second-tallest candidate.
If the solution was, say, a really complicated mix of getting votes out, and persuading people, and catering to median votere, and also fringe voters, in a careful balance - it does not seem hard to imagine that both parties might work equally hard at that very complicated strategy, and we’d see very close elections, no?
Close elections seem compatible with *any* strategy, not just median-voter.
Am I missing something?
So though I understand everything you're saying here. None of it makes sense to me in the current political climate. There is no middle position anymore. It's all extremes, and to try and live in some middle position, is to get attacked from both sides. And trying to have some middle position is mostly not allowed.
Sometimes it seems like we're being played by the politicians. It's the midterms and we need to raise more money, so you member from the other party start some abortion bill, and I'll start a gun bill, and then you can fund raise trying to stop my gun bill, and I'll do the same with yours. We both win... more money. The perfect play for this is when it's all split 50/50. Then 'everything' is always close to going the other way. 'everything' is in quotes because it doesn't include everything, just those things that most divide us. We never try and solve the important things... the budget/spending, big business, pharma, defense, too much money in politics!
I mean Donald Trump was the guy who said "hey we should veer left". He dropped the most conservative and least popular republican policy: cutting social security and medicare in order to balance the budget. It worked pretty well!
TLDR: As the US demographically splinters, voting behaviour decouples, big electoral alliances form that for game theoretic reasons each encompass around 50% of the electorate and the presidential nominee is one factor in this alliance formation.
--------------
Let me explain in more detail! It's not the parties that choose a candidate that appeals to roughly 50% of the population.
The candidates are downstream (and interdependent) with the culture war dynamic that splits the American public roughly in half.
Earlier in the 20th century the largest bulk of the electorate were white, protestant middle class, whose voting behaviour was very correlated. All candidates had to at least somewhat appeal to them if they wanted to be politically viable. Often, one candidate's message just resonated somewhat more with that demographic and this candidate landed a huge landslide.
As American society grows more racially diverse, marriage rates go down (decorrelating men and women), people move to cities and form subcultures, the electorate fractures into different alliances.
And these alliances have a habit to form about 50% each. Why? Because if you are a group in an alliance (e.g. black women, LGBT, Mormons), your ideal alliance has 51% of the power, so to win the spoils but share them amongst the smallest group. And in a two-party democracy, political power often just means voters.
When one group becomes too influential, the members at the lower ranks can feel the heat. You can see this in the progressive alliance who started questioning the merit of the "white, male cis gays" or the "straight black men are the white people of the POCs". Why are black men moving towards Trump?
So the culture war just preselects the electoral alliance. The pres. candidate of each party is a part of that culture war. Their selection cements but also shapes the alliance that the party represents.
So why are there so many undecideds? Because a lot of people don't neatly fall into one category and they're not actively part of the culture war. They just get served two unappealing candidates through the dynamics of the culture war and terminally online people who then wonder how there can possibly be any undecideds.
> I wouldn’t want to be the guy in the Republican primary who says “actually we Republicans should veer left”.
I think that Trump did this in 2016. Not on the left/right spectrum but obviously politics is more complicated than a one dimensional spectrum. But Trump was significantly less culturally/religiously conservative and significantly more economically populist than previous Republican candidates.
Seems relevant that polling has improved a lot (in particular finding more representative samples and correcting for ways in which a sample is not representative) and many previous landslide elections were surprises (often to everyone involved I think). This suggests that the weak form of the median voter theorem where every election is pretty close, because the candidates are about equidistant from the median is holding more true now because the candidates now have a better idea where the median actually is. In the past there was a lot more uncertainty on this, some people guessed where the median was wrong (due to say bad polling) and got destroyed in the election.
> Parties and candidates seem to do a suspiciously good job staying equidistant from the median voter, far beyond the pathetic amount of effort they explicitly put in
I suspect the real answer here is something akin to Adam Smith's "invisible hand" - which is to say, democratic elections are a free market (of sorts), and free markets are very good at producing results that appear coordinated and well thought-out, but are actually just the result of all of the actors individually acting in their own best interests.
RE: research on the topic, I'm not sure *precisely* what you're looking for, so it could be that you're already doing this, but one reason that you might have had trouble could be because most literature on the topic, especially when modeling party competition, cites Downs instead of Hotelling or the descriptive model names ("median voter", "spatial"), so if looking for "median voter" specifically you'd probably only get less than half of the available research. You'd get things like Jones et al. (2022), of course, but the search would probably not find papers like Dodd (2015), which as far as I can tell do not have the words "median voter" together. Grofman's work is a bit older (2004 to 2006) but it also covers a lot of relevant scenarios (and just due to length, a 300 page book is going to cover much more than a 30 page journal article). Though, I believe Grofman does mention "median voter", I'm not sure how well that would show up in search results.
Jones, Matthew I.; Sirianni, Antonio D.; Fu, Feng (2022-02-02). "Polarization, abstention, and the median voter theorem". Humanities and Social Sciences Communications. 9 (1): 1–12. https://doi.org/10.1057/s41599-022-01056-0. ISSN 2662-9992.
Dodd, Lawrence C. (April 2015). "Congress in a Downsian World: Polarization Cycles and Regime Change". The Journal of Politics. 77 (2): 311–323. https://doi.org/10.1086/680041. ISSN 0022-3816.
Grofman, Bernard (2004-05-17). "Downs and two-party convergence". Annual Review of Political Science. 7 (1): 25–46. https://doi.org/10.1146/annurev.polisci.7.012003.104711. ISSN 1094-2939.
Owen, Guillermo; Grofman, Bernard (2006-05-11). "Two-stage electoral competition in two-party contests: persistent divergence of party positions". Social Choice and Welfare. 26 (3): 547–569. https://doi.org/10.1007/s00355-006-0087-1. ISSN 0176-1714.
Adams, James F.; Merrill III, Samuel; Grofman, Bernard (2005-03-21). A Unified Theory of Party Competition: A Cross-National Analysis Integrating Spatial and Behavioral Factors (1 ed.). Cambridge University Press. https://doi.org/10.1017/cbo9780511614453. ISBN 978-0-521-83644-9.
That weird period between 1955 and 1981 is unlikely to ever repeat itself again. The Democrats had been the party of the South and the Republicans the party of the North. The parties really didn't represent progressives and conservatives, respectively - rather each had their own progressive and conservative wings.
That meant that Republicans couldn't win in the South, no matter how conservative. Additionally, conservative Democrats in the South wouldn't get kicked out of the party, no matter how conservative.
Political scientists at the time decried this state of affairs - without some degree of partisanship, you can't have the median voter theorem. Or, put differently, if I can't tell whether the Democrat is on the left or the right, how am I going to vote? Thus political scientists in the 1950s started calling for *more* political polarization. And we are living in the world they imagined!
Several people have brought up that the distribution is actually bimodal and I think the argument is not getting enough traction / is shot down too readily.
If the population falls into two disjunct clusters of roughly 50% it is absolutely possible to be the closest candidate (of two) to roughly 50% of the population without being in the middle.
Now this only works with what we might call perfect polarisation but I find that to be congruent with every observation we make. Viewed from the outside the US seem to have managed to perfectly polarise itself.
In fact intuitively it would seem easier to polarise the population than to be at the exact center of opinion. Getting marginally closer to the center when your opponent follows the same strategy appears to have quickly diminishing returns. Polarising the marginal voter you way when your opponent has the same strategy by contrast is a positive feedback loop.
I can see the counterargument that this would allow one candidate to move to the other side of the center and getting votes from both clusters. But that is completely unrealistic. Real people, when sufficiently polarised, would rather not vote or give their vote to an outsider candidate than to vote "technically closest candidate with chances of winning but wrong side of the center", no?
To loop around, I don't really understand why the bimodality argument is so quickly shot down in the comments. Wasn't this the blog that had the "scissor statement" post? Am I missing something?
I've always thought this phenomenon of "both parties approaching the center" was true, but I've never written it up, given it a name, or read too deeply into the existing literature about it. But I think it is true due to something most of us rationalists have trouble understanding: Politicians and activists are just going on feelings, or to use a more recent term, "vibes." They don't consciously move toward the center. They just start to get a vague, collective feeling that, if they don't, they'll lose. Think of religious Republicans and gay marriage. They probably still hate the idea of it, but they've just sort of vaguely given up on fighting that battle because they know they won't win.
> [ES: I am not a political scientist. I’ve tried to find good political science on these topics and failed; maybe you can point me in the right direction]
I would suggest starting with the Wikipedia article you linked, which states in the opening that the assumptions used aren't very reasonable—namely, the median voter theorem is a theorem about how majority-choice (Condorcet) rules act, and is limited to this class of fancy voting rules. There are generalizations that extend this to include strategic approval voting or honest score voting, but see the line that says:
> It cannot be applied to systems like ranked choice voting (RCV) or first-past-the-post at all, even in two-party systems.
In particular, both RCV/IRV and FPP have a polarized equilibrium, because the center is "too crowded" to hang out around there.
Neither party normally has a unified platform for congressional candidates.
The organizations which we call parties in the USA are long term coalitions of factions which would be separate parties in a parliamentary system.
The northern branch of the Democratic Party threw the southern branch under the bus when the Great Migration of blacks from the south to the north threatened their decades of single party rule. They found other ways besides segregation to neuter the newcomers.
Upward mobility among their constituencies threatens their coalition. They have doubled down on the grievance politics in an attempt to retain minorities while not alienating their biggest single constituency; government workers.
https://www.theringer.com/2024/10/25/24279042/why-is-every-recent-presidential-election-so-close
One thing I feel is missing in a lot of these discussions is that there are different goals for politicians and while public choice/rational voter are useful models they are not complete. Many politicians have sincere ideological commitments (e.g. https://www.vanityfair.com/news/2017/03/paul-ryan-medicaid, I can't find a link but the Democrats who voted for the ACA knowing they would lose re-election). A party wants to be the largest coalition that can accomplish the goals of it's members and the smaller the coalition the less conflict there is between different groups (e.g. there is no policy that the Democrats can take on Israel/Palestine that will not cause them to lose votes from part of their coalition and both parts of the coalition would prefer for the Democrats to kick out the other part).
Also because people can't remember anything before 2016 in 2008 and 2012 the electoral college favored the Democrats.
my theory:
The bubble and bias theory on why parties don't pursue the median voter efficiently.
many in the party apparatus are too cloistered and blind to honesty understand the median voter.
additionally, many of the party faithful are even more blind, making it much harder to pursue Vs the party supporters.
source: I've witnessed discussions in the opposition in the following countries: UK 2016 to present. US left. Israel anti Netanyahu block 2015 - present.
it's quite obvious that the opposition in all those cases were ridiculously oblivious to the median voter views and preferences, and multiple times gave up on practically open goals due to such blindness.
I don't have time to read all of the comments so someone may have already mentioned this, but I believe the problem is in the voting system (first past the post). There are types of voting systems deployed in the real world that have resulted in all candidates being essentially centrist mirrors of each other and far less adversarial.
I feel like an analysis of this topic without digging into why the mechanism of voting seems to have such a big impact will result in potentially incorrect conclusions.
Two-party systems sometimes result in hiccups that delay the return toward the median voter. I gave examples in a comment above of such delays from Britain 1979-1997, New Zealand 1984-1993 and the US 1993-2020 (but still ongoing).
Multi-party systems can show more stability across time, due to the stabilizing influence of “center parties”, who control the crossover vote in parliaments. A center party is a party that can make an allegiance both to the left and the right, without alienating its core of loyal voters. Center parties can be small but none the less control the political game, since they control the median vote in the national parliament.
But proportional representation (which leads to multi-party systems) is no guarantee for median-voter politics. That is only the case if the center parties are moderate parties - as they are, for example, in Scandinavia. If the center parties instead are religious extremist single-issue parties, multi-party systems may instead give a boost to religious extremism – as seen in Israeli multi-party politics.
Another possible source of collusion: it's not the parties per se that collude, but the political campaign industry. Most of the campaign messaging I see is along the lines of "this election is really really close, we desperately need your money". Some of that money presumably ends up in the pockets of campaign employees. And because it's an easily measurable signal, those employees are probably in large part *evaluated* on how much money they bring in.
Basically, the campaigns are trying to raise as much money as possible, and only secondarily to win the election.
> In theory, the Republican candidate for Governor in California could appeal to the median Californian; in practice, Republicans almost never win in California because Californians hate the national Republican Party.
But it does happen. Schwarzenegger was governor of California. Christie was governor of New Jersey. Giuliani was mayor of NYC.
I was bothered by exactly this issue a few years ago, and write about it on LessWrong:
https://www.lesswrong.com/posts/EdoCrSGM8FzJhez3C/why-are-politicians-polarized
I reached some of the same hypotheses, but also including irrationality of candidates/parties, and donations/activism declining with a veer towards the center, even if individual voting behavior does not.
I also had a different but similar visual choice for the graphs. :)
The House is an unusual case, because of gerrymandering, and there are some rather extreme gerrymanders in some states. In principle, a gerrymander can give you a majority of the representatives with only 34% of the popular vote. It also helps that Democratic voters tend to be more geographically concentrated in urban areas, making gerrymandering easier. An individual politician in a gerrymandered district, especially a "safe seat" held by the minority party, is going to win the general election anyway, so they only have to appeal to the median primary voter, making the national party seem more extreme. Additionally, oarties also have ways (some underhanded) to discourage primary challengers against incumbents. (For example, according to what I've read, if your company or organization works for a primary campaign against an incumbent Democrat, the rest of the Democratic party will blacklist you and will never hire you or your organization to work for a campaign ever again.)
The way I model it is that candidates are typically pulled between the base and the center. When they start to look like they're losing, the base gives them permission to move towards the center. But the second they look like they're winning, they start facing pressure to slide back towards the preferences of their base (which in practice means many of their friends and family). Thet aren't actively seeking equilibrium, but the natural result is both candidates will sort of bob around 50% all the time
I know nothing about political science, it is the first time I read about the median voter theorem, but still I'm arrogant enough to strongly dislike it.
Firstly, it is one-dimensional. Multi-dimensional spaces are surprising, you cannot rely on intuition tuned for one-dimensional space to reason about them, even 3-dimensional intuition can fail you badly with n-dimensional spaces. But the issue is deeper, I don't see easy way to generalize the theorem to N-dimensional space because dimensions will not be equal by their strength. People will prioritize them somehow, they will stop considering some of them because some other are more important for them.
Secondly, I generally do not believe in "average" or "median" people, because attempts to rely on them before failed spectacularly. Mathematically speaking issues relying on "average person" stem from multidimensionality, but even without mathematics people are so vastly different, that "average" or "median" person just a non sense. But I suspect with the median voter there will be more complex problem, because they are not form a normal distribution in n-dimensional space. I think there are a lot of clusters, which tend to remain clusters. The drift from one cluster to the other is possible, but mostly people stick to their cluster. I don't know will it break the median voter theorem by itself, but read the next paragraph.
And thirdly, this theorem ignores completely that there is no fixed median voter. What is important for people depends a lot on what they hear, on what important for other people around them, it depends on external factors which are not fixed for a lot of people. Yes, some deeply partisan voter would live in their informational bubble and would not hear anything that can shift their priorities. But some people would read news, and each piece of news will shift their priorities, maybe just temporarily, only for five minutes, but it will. Candidates can _change_ median voter by shifting their priorities of what is important and what is not. Now if people are grouped in clusters and clusters tend to drift on a "political spectrum" (or rather in the political multidimensional space) as a whole, then what?
All in all, my uneducated opinion that the median voter theorem looks more like "it is the only mathematical thing we can tackle" not like "it is the model that grasps all the important facets of the reality". Social sciences use such models sometimes, and one of the signs that this is probably the case is the impossibility to apply this theorem to the reality. You cannot know the multidimensional space of preferences of voters, you cannot measure politicians and place them in the multidimensional space, so the question is: how did scientists set up the experiments to validate the theorem? In my experience it pays to be vary every time when you see a mathematical quantitative model that cannot be tested on real data.
I wouldn't want to underestimate the role of simple miscalculation. For example, it sure looks like Harris thought she was winning and was just afraid of messing it up somehow. Turns out this was very wrong. There's going to be a lot of that in every election. Everyone has their theory of where the median voter is, but we don't really know until we see who showed up to vote. Often those theories are wrong.
what do you think of catholics in the US?
(1) catholics won 12/20 of the US elections, lost 2/20 and then 5/20 are ambiguous/split (i think 2020 is actually ambiguous/split. fox says 49D/50R as opposed to others that says 52D/49R.)
(2) assuming the 5/20 are wins and assuming 2000 & 2016 count as wins considering the popular vote vs electoral vote thing, catholics have won nearly almost all of the US elections post ww2 namely 18/20 as predicted by the median voter theorem - the remaining 2/20 are 1968 & 1952. this means catholics have a 14 win streak starting 1972
(3) catholics are 20% of the US but bipartisan. protestants as a whole are 40% and always republican. atheists are 30% and always democrat.
non-orthodox jews, jews as a whole, atheists, muslims, non-abrahamic religions are always democrat.
haredi orthodox jews, mormons, evangelicals, christians as a whole are always republican.
i think mainline protestants, modern orthodox jews and maybe orthodox christians are bipartisan. idk.