voting

Same Engine, Different Gearbox

Picture two rooms. (I made them up. You have been in both.)

The first is a faculty meeting at a decent university. Fifteen people, a plate of supermarket biscuits, an agenda item about a new hiring policy. A visiting lecturer (she, or he, if that makes you happy), new this term, asks whether the policy has ever been tested against the outcomes it promises. Nobody gasps. Nobody raises a voice. The chair says "good question, let's take that offline," and the room does something small and perfect. Eyes drop to the biscuits. Pens move. The agenda advances. She will not be asked to the next lunch, and she may never work out why.

The second is the back bar of a footy club after the AGM. Eight blokes are arguing about the council, the price of diesel and whether the new coach has any idea what he's doing. They agree on almost nothing. Then a young fella who has just moved to town says something kind about the council's new housing plan. Nobody gasps. Nobody raises a voice. Somebody says "yeah, nah," and the room does something small and perfect. A round gets bought for everyone but him. The conversation turns to the weather. He will not be asked to the next barbecue, and he may never work out why.

Two rooms. Two sets of manners. One mechanism.

And each room is quite sure the other one is where the tribalism lives.

The Engine

For most of human history, a person without a tribe was a person without a future. No protection, no shared food, nobody to raise your children if you died. Natural selection does not need to be told twice. The minds that sorted the world quickly into us and them, and punished anyone who wobbled, got to be our ancestors.

You can watch the machinery start from nothing. In the early 1970s the psychologist Henri Tajfel took strangers, sorted them into groups on trivial grounds (a coin toss, or a preference for one painter over another) and asked them to hand out small rewards. They favoured their own group, even though nothing was at stake and they would never meet the others. No history, no grievance, no shared interests. A label was enough, and the engine turned over.

If that is the machinery, there is no reason to think it was installed in one half of the political spectrum only. Cory Clark and her colleagues made exactly this point in 2019: selective pressures shaped every human mind to be tribal, so the sensible starting position is that left and right are equally afflicted. They called it the "evolutionarily plausible null hypothesis." The evidence they leaned on was a meta-analysis by Peter Ditto and colleagues, covering 51 experiments and more than 18,000 participants. Partisan bias was robust, and it was no larger on one side than the other. Give liberals and conservatives identical information, flattering one side's beliefs in one version and challenging them in the other, and both sides bend it by about the same amount.

So if anybody tells you that only the other lot are tribal, you now have the data to inform them that they are being tribal.

You know where I sit from my earlier pieces (I wrote about cloud people and dirt people, and I was not subtle about which one I am). Discount me accordingly. But look at what that result does to the argument I am about to make. It cannot be that one side has the engine and the other does not. Both do. The difference has to live somewhere else.

I think it lives in the gearbox.

The engine is the reflex to belong, to defend the group and to punish defectors. The gearbox is what a tribe uses to turn that reflex into agreement, the glue that gets thousands of strangers to say the same thing at the same time. I call the first one the engine and the second one the gearbox. Left and right run the same engine through different gearboxes, and the gearbox decides what agreement looks like, how far it travels and what it costs to break.

What the Glue Is Made Of

What are the two tribes glued together with?

My answer, which I hold with moderate confidence and a fair amount of intuition, is that they use different materials.

The right's glue is mostly things you can stand in: a place, a family, a church, a flag, a set of customs older than anyone present. The left's glue is mostly things you can read: ideas, categories, moral principles meant to apply to everybody, and the credentials that certify who is qualified to say what those principles mean.

There is a framework that points the same way. Jonathan Haidt and his colleagues, in Moral Foundations Theory, found (as I understand it) that liberals draw mostly on two moral foundations, care and fairness, while conservatives draw on those two plus loyalty, authority and sanctity, more or less evenly. Loyalty, authority and sanctity are tribe-and-place materials. Care and fairness, applied to everyone, are idea materials. It is a theory built largely from questionnaires, and a questionnaire is a thin instrument for measuring a tribe, so treat it as a lens and not a law.

I have argued before that the left's gearbox runs on deference to credentialed consensus, on whatever the people with the right letters after their names have agreed. That is a claim about glue, and glue makes predictions.

If your tribe is held together by place and custom, you should agree hard with your neighbours and loosely with strangers in other towns. If it is held together by shared training and a shared reading list, you should agree wherever the training reaches, which is everywhere. So does the national shape of agreement differ between left and right?

It does.

The Choir

In 2018 Peter Ondish and Chadly Stern published what has become the standard test (Social Psychological and Personality Science). They took two large, nationally representative American datasets, the General Social Survey (53,081 people, 1974 to 2016) and the American National Election Studies, which together cover more than 80,000 people and over 400 political issues, and measured how much each group agreed with itself. Liberals agreed with one another more than conservatives did. Moderates agreed least. Pick two liberals at random and their views will sit closer together than those of two conservatives picked at random.

It travels, mostly. A European follow-up using the European Social Survey (38 countries, more than 376,000 people) found leftists agreeing more than rightists on average. A second dataset, Eurobarometer (18 countries, about 375,000 people), did not show it. So the effect is real, patchy, and not a law of nature.

The researchers floated an explanation, and it is a good one. Conservative conformity is local. Each community conforms hard to its own neighbours, but neighbours in different places conform to different things, so the national picture comes out thin. The footy club is strict, and the footy club two states over is strict about something slightly different. A thousand villages with a thousand firm rules do not add up to one national consensus. One choir trained on the same hymn book does.

That is the gearbox showing. The faculty meeting's rule travels with the lecturer wherever she goes, because the training travels. The footy club's rule stays at the footy club.

There is a twist in the data that I enjoy more than I should. In 2014, Stern, West and Schmitt reported that conservatives overestimate how much their own side agrees with them (a "false consensus" effect), while liberals tend to see themselves as more unique. A preregistered replication with 1,005 people found that everyone overestimates, so nobody is exempt, but the effect was strongest among conservatives and weakest among liberals. Liberals also scored higher on need for uniqueness, which accounted for part of the gap.

The left agrees more than it thinks it does. The right agrees less than it thinks it does. The choir believes it is a thousand soloists.

Ondish and Stern drew a practical lesson from this, as the Pacific Standard summarised them: liberals' attachment to individualism can make effective group activism harder. A tribe that prizes its own originality finds it awkward to act in unison. I will come back to that.

Moralisation, Not Hatred

The second place the gearbox shows is in how each side treats the other.

The popular version says the left is the tolerant side and the right is the hating side, or (from the other camp) the reverse. The data are messier than either. A 2020 study in PNAS (Moore-Berg and colleagues, with a representative sample of 1,056 and a longitudinal sample of 2,707) found that Democrats and Republicans dislike and dehumanise each other about equally, and that both sides wildly overestimate how much the other side hates them. Most people on both sides thought the other lot was nastier than the other lot reported being.

Then there is behaviour, and there the picture tilts. Krishnan Nair and colleagues ran five preregistered studies in 2022 and 2023, including two social media field experiments with 10,000 participants, and published them in the Journal of Personality and Social Psychology in 2025. Democrats were more likely than Republicans to block a counter-partisan follower. They rated Republican neighbours more harshly and Republican job candidates lower. There was no partisan difference in following someone back. A 2025 conjoint experiment (Norman, in Political Psychology) found the same direction.

Why? The Nair studies point at a belief, not a feeling. The gap was driven by Democrats' conviction that Republicans pose a harm to disadvantaged groups, particularly racial and ethnic minorities.

So hatred is the wrong word. Hatred is a feeling, and the surveys say both sides carry about the same amount of it. What the field data picks up is a belief that the other side hurts the vulnerable. If you hold that belief, shutting a Republican neighbour out is not spite. It is duty. You are not being unkind. You are protecting someone.

I call this moralisation, and it is exactly what the engine has always done. Tribes have always policed their defectors in the language of right and wrong, because enforcement is cheapest when the enforcer feels righteous doing it. The gearbox only decides which moral language is available. On the left it is harm and fairness. On the right it is loyalty and sanctity. Ask a conservative what should happen to a man who burns a flag at a veterans' funeral and watch how quickly the same machinery spins up. The difference is not that one side moralises and the other does not. The materials differ, and so does who ends up cast as the villain.

Education Sharpens the Tribe

The third place is education, and here I have to correct my own first instinct. I started this thinking that more education makes people more conformist. The data does not quite say that. It says something more interesting.

Ines Lee and Eileen Tipoe studied 18 OECD countries between 2010 and 2018 and found that in most of them the gap between liberals and conservatives on inequality, immigration and gay rights widens with education. The gradient is steepest in the United States. In Europe, Matthijs Rooduijn found the same kind of split on immigration across 23 countries: the less educated tend to be negative, the more educated positive.

That is sorting, not conformity. Two people with PhDs disagree more than two people who left school at sixteen. It is not a story about intelligence. It is a story about how much of your side's syllabus you have been taught.

My reading (I believe, and it is an inference, not a finding) is that education puts people inside institutions, and institutions teach, reward and enforce the boundaries of the tribe. Education teaches you the position. Institutions teach you the price. By the time you hold the degree you know exactly where your side's edges are and what it costs to step over them.

And the price is not paid evenly. In late 2024 the Foundation for Individual Rights and Expression surveyed 6,269 faculty at 55 American colleges and universities. Forty-seven per cent of conservative faculty said they felt unable to voice their opinions because of how others might react. Among liberal faculty the figure was 19 per cent. Nobody is free, and one in five liberal professors is holding their tongue too. But the gap is roughly two and a half to one, and it points the way you would expect from a gearbox where the credential is the glue and the faculty meeting is the room.

As an aside - backed by the gravitas of a thousand TikTok videos and zero research - I cannot name a single person who lost their job for being too woke. I am sure someone will write in with a list. I will read it with interest.

I read the faculty figures as a measurement of who is more afraid of whom. One survey in one setting does not prove it. The counter-argument is coming.

Nobody Changes Their Mind for Free

The claim I expected to make here is that the left is less likely to change its mind. The best study I can find says the opposite, barely.

In 2025 a team of researchers from opposing camps (an adversarial collaboration, with Philip Tetlock among the authors) ran two preregistered studies with 6,181 Americans to test whether ideology predicts how readily people update their beliefs when shown evidence. Conservatism was weakly linked to less updating, an effect of about 0.05, which is the kind of number you need a statistician and a magnifying glass to see. Extremism, of either flavour, was linked to less updating. Centrists and moderates updated the most.

So the line does not fall between left and right. It falls between the extremes and the middle. And I think the better explanation is not psychology but price.

Whether you can change your mind depends on what it costs you to say so out loud. The bloke at the footy club whose whole social life sits within a kilometre of his house cannot change his mind in public. Neither can the professor whose friends, career, reading list and last forty articles all sit inside one set of institutions. Both are rigid. Neither is stupid. Both are paying rent on a belief, and the rent rises with the share of your life that lives inside the tribe.

The engine punishes defectors. The gearbox decides who does the punishing and how far the punishment reaches. At the footy club it is exile from the barbecue. In the faculty it is the career. Same engine. Different reach.

People are not rigid because of the side they are on. They are rigid in proportion to what it would cost them to move.

The Best Case Against

If I were arguing against myself, this is where I would start.

The engine is symmetrical. Nothing here shows that one side is more tribal than the other, and the Ditto meta-analysis says the bias is about equal. If you came for a verdict on whose tribe is worse, there isn't one here, and I would distrust anyone who claims to have one.

The effects are modest and mostly American. The consensus result failed to appear in one of the two European datasets. The education gradient is strongest in the United States. Who goes to university, who watches which media and who lives where all sort people into tribes before the gearbox gets to them, so a good deal of what I have called gearbox could be selection.

Enforcement runs both ways, wherever a tribe holds the lever. After Charlie Kirk was killed in September 2025, roughly three dozen workers were reported suspended or fired over their responses, including staff at MSNBC, Clemson University, the Federal Emergency Management Agency and the Carolina Panthers. A Washington Post columnist, Karen Attiah, said the Post had dismissed her over her posts. A website went viral by publishing a searchable list of thousands of people accused of posting critical messages about him. That pressure came from the right, and it worked. The faculty meeting is not the only place a tribe can punish a defector. It is just the place where the left happens to hold the pen.

I do not read that as an objection to my argument. I read it as the argument. The engine is the same, and each tribe pulls whichever lever it holds. The gearbox is not a fixed trait. It is a map of who controls what.

The faculty numbers I leaned on come from FIRE, a free speech organisation, which makes it an advocate, and a survey of academics tells you about academia. Self-censorship also cuts across the spectrum. One in five liberal faculty report it, and a Boston Globe editorial reading FIRE's student survey concluded that some students want controversial views kept off campus whatever their politics.

And the rigidity result cuts against what I have written before. Conservatives came out very slightly more rigid, not less. Slightly is the operative word, but I will take it on the chin.

What survives is narrower than what I started with, and I think harder to knock over. Not a verdict. A shape. Left and right run the same engine. The gearbox differs in what it is made of, how far agreement travels and who pays when someone breaks it. I might be wrong about the details. I have been wrong before (1994, a ute, long story). The engine I am less worried about.

What It Means for the Machinery

Parties are gearboxes of a second kind: machines for turning tribal reflex into votes and policy.

A tribe held together by credentials should be very good at capturing the institutions that set a society's norms (universities, professions, newsrooms) because the same training reaches all of them. A tribe held together by place should be very good at being everywhere at once and agreeing with itself nowhere in particular, which is hard to organise and hard to dislodge. And the choir that thinks it is a thousand soloists has the problem Ondish and Stern spotted. It agrees in its heart and organises badly, while the village that distrusts the centre has the opposite trouble. I would even argue that this maps onto why each side so often looks baffled at how the other one wins.

That paragraph is inference, not data, and I hold it more loosely than the rest.

Back to the Rooms

Put the visiting lecturer and the new bloke from the footy club in the same car park and they would not need to compare notes. Same look. Same silence. Same long walk to the car. Each has been told the other room is where the tribalism lives, and each room is very sure of it.

That is the engine. Everything else is gearbox.

Next time somebody in your room asks the wrong question, watch what your own face does.

Enjoying our content?
Please subscribe to get email updates

Related Post