An Election Can Defund a Field Like Sociology. It Can’t Replicate One.

SOC stopped correcting itself, so the correction came from outside. Some of it builds. Some of it bulldozes.

Here are two people describing the same discipline.

The first is Ray Rodrigues, chancellor of Florida’s public university system. In March he moved to strip Introduction to Sociology out of general education at all twelve of the state’s public universities, and he explained why: sociology is “social and political advocacy dressed in the regalia of the academy.” The board passed it on the spot, over the objection of its lone faculty member, on a motion that wasn’t even on the meeting’s agenda. A discipline that had been part of the American core curriculum for a century became an elective by surprise motion on a Thursday.

The second is Kevin McCaffree, a sociologist at the University of North Texas, quoted last weekend by Pamela Paul in a Wall Street Journal Saturday essay with the cheerful title “Is It Time to ‘Liquidate’ the Study of Sociology?” McCaffree edits a sociology journal. He is not trying to defund anything. And he says: “Trust in higher education is collapsing. That’s being driven in part by politicized degrees in social science, and in that respect, sociology is the belly of the beast.”

A former state legislator now running the university system and a scholar trying to reform it from the inside reached for almost the same sentence. That agreement matters, because most of the people fighting about sociology right now are treating it as a simple story with two sides. It isn’t. When the official who runs the system and the scholar trying to reform it reach for the same sentence, the thing they’re both pointing at sits underneath the fight. It’s a mechanism, and mechanisms don’t take sides.

The fight went fully public this month, but the machinery under it has been grinding a while. In June, ten scholars convened by a pair of university chancellors and chaired by the philosopher Paul Boghossian released a report arguing that the humanistic social sciences, sociology named among them, had come loose from the one thing that keeps any field in check: getting corrected by reality when you’re wrong. This past weekend the sociology version of that argument hit the front of the Journal, and by Monday two philosophers were posting essays that reframed academic freedom itself around it. The problem is old. The news cycle is new.

The mechanism is simple, and it’s the same one that decides whether a word like (cough) “liberal” keeps its meaning: a field keeps its standards only where someone can make being wrong expensive.

That’s it. That’s the whole thing.

Physics stays honest because a bridge falls down. Medicine, because the drug either works in the trial or it doesn’t. Accounting, because there’s an auditor who can end your career. In every one of those cases the discipline is chained to something outside itself that pushes back, and the pushing back is what disciplines the discipline.

Reality is the auditor, in the end. It always is. You’ve heard me rail again and again about this (sorry old readers, gotta set up the new readers). What changes from one field to the next is really the clock. Physics gets graded fast and cheap: the bridge is down by Tuesday, and there’s no arguing with the rubble.

Now ask what pushes back on a sociology paper about, say, the causes of a wage gap. Reality still bats last here too. But “last” can mean thirty years, and the verdict shows up blurred, filtered through a generation of policy and a hundred other things you can blame instead. On that kind of clock a field needs a stand-in, something to deliver reality’s grade early, and sociology’s stand-ins have mostly been captured.

There’s peer review, but peer review is other sociologists, and if the other sociologists share your commitments then review checks your citations and your methods section, not your premises.

There’s replication, but the incentive to replicate a politically congenial finding is roughly zero, and replicating an inconvenient one gets you called names.

There’s tenure, which is supposed to protect inquiry, and which in practice protects you from ever paying a price for being wrong about the world, as long as you’re wrong in the approved direction.

The field grades itself. And a field that grades itself, over enough time, starts maybe grading for the wrong thing.

You can watch it happen in the numbers. Musa al-Gharbi, a sociologist at Stony Brook who is himself on the left and himself critical of his discipline, notes that something like four in five recent articles in the field’s flagship journals touch on inequality, and that the discipline has gotten very good at measuring new axes of it while producing almost nothing a policymaker could use to fix any of it. Ilana Redstone, a sociologist at the University of Illinois, went further in that same Journal essay: within sociology, group disparity gets treated as de facto evidence of discrimination, which marks the discipline’s central causal question “settled” before the research even starts. “We’re treating questions around causation as settled,” she says. That is what a field looks like when the cost of being wrong has fallen to nothing. Not a conspiracy. Not stupidity. An incentive gradient that rewards arriving at the answer everybody already knows and punishes the person who checks.

Before anyone reaches for a word like “liquidate,” though, look at the actual audit, because one exists. Last spring a seven-year, DARPA-funded project called SCORE had 865 researchers try to reproduce thousands of social-science findings. The headline was grim: only about half held up. Sociology’s effects all but vanished on the retest: a correlation of 0.10 in the originals fell to 0.03 in the replications, which rounds to nothing.

So the reformers aren’t hallucinating. Something is wrong.

“Half the building is still standing,” as I said in my piece, but the more useful number was buried under the headline. Papers in journals that made you share your data and code reproduced at seventy percent. Papers in journals that didn’t, forty-one. The fields that looked best, economics and political science, didn’t look best because their scholars are purer. They looked best because their journals installed an auditor, a data editor who checks your code before publication, and made everyone submit to it. That kind of rot, the kind that comes from hiding your work, is fixable, and some fields fixed it. The ones in the worst shape are the ones that never built the machinery, and sociology is near the front of that line.

My own field, political science, replicates at about half too, and economics does worse. We’re better at showing our work. We are not better at being right. I’m reading another discipline’s numbers from across the hall, not defending my own and not anyone’s ally in the fight necessarily—but I am on the side of doing things right, damn it.

So the reform side is mostly right about the dx. But the diagnosis is the easy part, and it’s where almost everyone stops, both the people cheering the WSJ piece and the people furious about it. The hard part is what happens next, after a field stops correcting itself.

What happens next is that the correction comes from outside. It always does. A discipline that won’t audit itself doesn’t get to stay un-audited forever; it just loses the ability to choose who does the auditing. And several very different outside auditors are showing up at sociology’s door at once. Some of them build and some of them bulldoze, and telling those two kinds apart is the entire game.

Let’s start with the ones that build. They’re the more interesting kind, really.

Sociology’s reformers could not fix the discipline through its own institutions. They tried. The American Sociological Association ran a presidential election this spring in which one candidate, Melissa Wilde of Penn, campaigned on the radical proposition that the association should be politically neutral. She lost, roughly 51 to 37, on one-third turnout, to a candidate who studies right-wing movements. A month later a faction was boycotting the association’s own annual meeting. The theme of that meeting: “Disrupting the Status Quo.” The reformers looked at the guild, correctly concluded it was not going to reform from the inside, and did the only thing left. They built somewhere else.

Most of that building has happened at Heterodox Academy, which now runs a sociology community that functions as a shadow professional home for people who want to do the work differently. New journals, or old journals taken over and reoriented, like the one McCaffree edits, where the previous editorial board quit en masse after the publisher installed the new leadership over their objections. A magazine. Panels, working papers, a network. It is, in the most literal sense, a parallel institution, stood up because the original one was captured and the exits through the front door were all locked.

It’s the same move as the journals that installed data editors, just at the scale of a whole discipline rather than a submission portal. When the inside is captured, re-coupling to reality has to come from outside, and the constructive version of that is people leaving and building rather than people burning the building down. Sociology’s reformers did the constructive thing. They didn’t get the state to abolish their enemies. They started a venue and invited the discipline to compete with it on the merits. One of the report’s ten authors, the sociologist Ashley Rubin, has spent the months since writing the program out in plain terms: build alternatives instead of culling the old guard, reward rigor with incentives instead of punishing heresy with sticks, keep the whole thing voluntary. If that venue produces better work, and it might, the field re-couples the slow way, by bleeding prestige to a rival that’s chained to something real. That “if” is the whole wager, and it can lose. A rival only counts as a reality-check if it submits to the same test it demands of everyone else; a heterodox venue that rewards anti-woke priors as reflexively as the mainstream rewards the other kind would be a second tribe with better manners, not a correction. What separates the reformers from just another faction is not their politics but whether their sharpening ever points home.

Two philosophers made the theoretical version of this argument over the weekend, and both are worth reading. Hrishikesh Joshi posted the sharpest version. Academic freedom, he points out, is a weird privilege. Your plumber doesn’t have it. Your accountant doesn’t have it. A journalist can be fired for bad work; a tenured professor mostly can’t. That privilege was granted for one reason, “the free search for truth and its free exposition,” and it’s justified only so long as the free search is actually happening. If a field has become ideologically homogeneous enough that the free search has stopped, then the thing academic freedom was protecting isn’t there anymore, and the shield is guarding an empty room. Matt Lutz put it even more bluntly: academic freedom “is the freedom to keep on pursuing your ideas even if no one is buying them.” That can be a beautiful thing, the crank who turns out to be Galileo. It can also be a subsidy for work that failed in the marketplace of ideas and refuses to leave.

Both are describing the same auditor problem in the language of political philosophy. The market only disciplines producers if bad products can go out of business. Academic freedom, by design, keeps the producer in business regardless. Which is fine, and even noble, right up until the field uses that protection to wave off every correction as an attack.

Now for the kind that bulldozes.

While the reformers were building their parallel institution, the state showed up with a bulldozer. Florida didn’t start a rival journal. Instead, it deleted the course. The federal government, over the past year and a half, has frozen research funding at a string of elite universities, billions at Harvard alone, and pried a two-hundred-million-dollar settlement out of Columbia. Twenty-one state bills constraining what universities can teach or how they can govern themselves passed in 2025. This is not correction by competition. This is correction by appropriation and by statute, and its aim is not better sociology. Its aim is less sociology, or none.

And the public, it turns out, has already voted with its feet. The newest Gallup number, out two weeks ago, puts confidence in higher education at 38 percent overall. Republicans are at 23. But the striking figure is the Democrats, who fell to 50, a record low, down eleven points in a single year. The institution’s last loyal constituency is starting to leave too. And the young are pre-defecting: the share of high school seniors who don’t want a bachelor’s degree climbed from 18 percent in 2011 to 30 percent in 2024 on the Monitoring the Future survey, and the collapse is almost entirely on the right: disinterest among Republican boys roughly doubled, from 20 to 39 percent, while among Democrats it barely moved. The wrecking ball, it has a demographic tailwind.

Neither side in the WSJ fight will say the next part, because saying it means holding two thoughts that feel like they belong to opposing teams. Here is my attempt.

The reform side is slow to name the bulldozer for what it is, because it’s aimed at their enemies. Some see it and say so; Ashley Rubin’s whole program is building alternatives rather than letting the state do the culling. But the reflex, when Florida kills the sociology requirement, is relief, or at least silence, because the department that got hit is the department they’ve been fighting. But a discipline deleted by a board vote hasn’t been corrected. It’s been conquered. That doesn’t make the vote illegitimate: a public that funds these universities can withdraw its confidence, and the collapse in that confidence is real. But defunding a field and adjudicating its findings are different acts, and only the first belongs to the electorate. An election can shutter a department; it cannot replicate or improve a study. The wage-gap paper doesn’t get truer because the state cut its funding, and the politicization the reformers rightly diagnosed just gets swapped for a new one, this time wearing a suit and holding the checkbook.

You don’t fix a field that answers to the wrong master by handing it to a worse one.

And the activist side can’t tell the builders from the bulldozer at all, because both come from outside the discipline, and from outside the discipline everything looks like an attack. When Sociologists for Palestine and the Heterodox Academy reformers both get called enemies of the field, the field has lost the ability to distinguish a rival that wants it to do better from a state that wants it gone. That confusion is the one that guarantees the state wins, because a discipline that can’t tell its own reformers from its executioners has no one left to defend it whom the public will believe.

The courts, for what it’s worth, have so far leaned against the bulldozer. A federal judge ruled for Harvard last September, finding the funding cuts unconstitutional, and the appeal is still pending. So “the state is winning” isn’t quite true yet. But that court fight isn’t the real problem, and the real problem doesn’t wait on a judge. The vacuum was real, and a field that leaves a vacuum doesn’t get to pick what fills it.

Because that’s what all of them are, the data editors and the reformers and the bulldozer alike: answers to one question. If sociology won’t grade itself, who will? A rival chained to reality is one answer, the good one. A master chained to an election is the other: fine for withdrawing trust, useless for settling what’s true. And the discipline summoned the whole range by firing its own auditor decades ago and calling it academic freedom.

My own field is on that same list, which is why I try to bolt a small auditor onto my own work: use data, make reasonable inferences, run the arguments past people paid to attack them, check the claims against primary sources and flag the ones I can’t confirm for realsies. It’s not that hard to do things right. To be epistemically humble. To be methodologically conservative.

(And here, with the BBQ, when I make predictions, I try to make them falsifiable, so I can be wrong in public! And I keep a comment section full of people who’ll say when I’ve blown it. I love you all for it.)

None of that is insight. It’s machinery. And the lesson of the sociology war is that machinery is the only thing that ever works, and that a field which rips its own out will, sooner or later, get a crueler one bolted on by someone who was never on its side, especially when the higher ed environment is so multiply fraught as it is.

There’s a version of this that ends with AI, the one force at a scale large enough to change these incentives, which could become the cheapest reality-check a scholar ever had or the most sophisticated flattery machine ever built. I wrote about why many of our smart colleagues are so opposed to AI a while back.

The sociology fight is enough on its own for today though, because it’s the cleanest picture of the underlying rule: reality bats last, but only if you let it up to the plate. Sociology stopped letting it up. Now it has a whole line of auditors it never hired, and only some of them want the discipline to survive the audit.

The field still gets a say in which one it answers to. For a little while longer.

---

If I’ve gotten something wrong in here, the comments are open.

If you want these to keep coming, you can subscribe, and if you’d rather put a price on it you can actually see, there’s a coffee button.

Earlier in the stack:

• Why Are the Humanities Missing the AI Moment?, the piece this grew out of, on why whole fields stop getting corrected by reality.

• The Other Pillar, Part 3 of Reality Bats Last, where “approval-optimized versus reality-optimized” got its first run. That’s the engine under all of this.

• Everybody Wanted the Term “Liberal”, the same “nobody ever pays a price for being wrong” mechanism, pointed at a word instead of a discipline.

• Trust, Critique, and the Problem of Knowing Together, on what breaks when a field loses its shared standards for settling disagreement.

• Reality Bats Last, the start of the series all of this comes out of.