Why Are the Humanities Missing the AI Moment?

Turns Out Reality Bats Last on Disciplines Too

Timur Kuran is a libertarian-leaning economist at Duke. His most-cited book, Private Truths, Public Lies, is about preference falsification: why people misrepresent their political views in public, and how those misrepresentations destabilize societies that depend on them. His second book, The Long Divergence, asks why the Islamic world fell economically behind Europe over six centuries and lands on an answer about institutional rigidity. He is not a man with a soft spot for the contemporary humanities.

Last month Kuran posted this:

“The humanities shine when they illuminate civilizational transitions, such as the spread of agriculture or industrialization. Now, as we navigate the uncertainties of AI-based existence, they should be drawing on their accumulated scholarship to help us understand the present.”

Worth taking seriously, because of who said it. Kuran is not a humanities partisan trying to defend his tribe. He is an economist/analyst who spends his career asking how institutions go bad and stay bad. When he points at a discipline that’s failing to do something he thinks it was built to do, the diagnosis carries a weight (at least it should, in my estimation) it wouldn’t from inside the humanities or from culture-war media.

What Kuran is pointing at is real. The humanities have done some of their most lasting work at moments of civilizational reorganization. Erasmus and the early modern humanists made sense of the printing press and the Reformation. The encyclopedists made sense of the industrial threshold. Hannah Arendt, Reinhold Niebuhr, Lewis Mumford, and Stephen Toulmin made sense of the post-war reckoning with industrial-scale violence and technological power. The work that came out of those moments is what most of us still read, decades or centuries later, when we want to understand how the present took its shape.

We are, by most reasonable readings of the evidence, inside another such moment. AI systems can produce coherent language, write working code, pass benchmark exams, and increasingly act as agents in the world. Whether the current trajectory is a normal-technology disruption or the beginning of something more discontinuous, the questions it raises — about labor, about authorship, about knowledge and authority, about what it means to be a person in conversation with something that is not one — are exactly the questions the humanities were built to engage.

From the academic humanities as an institutional whole, though, very little public-facing engagement with these questions is reaching the surface. Yes, as always, there are exceptions, and I’ll get to them, but the dominant public conversation about AI is being conducted by computer scientists, economists, technologists, and journalists. The disciplines best positioned to do Kuran’s work are conspicuously absent from the registers where the work would matter most.

Why?

That’s the question this piece is about. The short answer is that the disciplines best positioned to do Kuran’s work have, over the past half-century, built intellectual and institutional habits that make showing up structurally difficult. The longer answer involves three interlocking moves: how disciplines vary in their capacity to be corrected by external reality, what happens to sophisticated reasoners when they operate in disciplines without that correction, and how a particular meta-theoretical apparatus turned a legitimate philosophical project into something that functions as immunization against the kinds of critique that would normally pull a discipline back toward usefulness.

I’m a political scientist working in a different mode, heck in a different building, than the humanities. That said, my own discipline has spent decades thinking about how institutions sort themselves into reality-tracking and reality-flattening modes, and political science is itself sometimes the one doing the flattening. And, hopefully anyway, I’ll flag specific places where my position shapes what I’m arguing rather than waving at the problem in the abstract.

The Differential

Every academic discipline is staffed by people subject to the same cognitive and social pressures. Researchers want status. High profile publications are the coin of the realm. And worse, they want their preferred theories to be right.

For those outside academia, we academics argue inside guilds whose internal incentive structures reward cleverness and tribal loyalty alongside accuracy. These pressures are universal across academia. What varies across disciplines is whether something external to the guild pushes back when those pressures pull the discipline away from accuracy.

In engineering, sometimes the bridge falls down. In experimental physics, sometimes the apparatus does not produce the predicted reading. In medicine, sometimes the patient dies and the autopsy says why. In business, sometimes the model fails to forecast the market and money is lost. None of these external pressures are perfect, and none of them inoculate the discipline against ideological drift in its unchecked precincts. Finance has 2008. Medicine has the opioid crisis and the long shadow of psychiatric misdiagnosis. Engineering has its own decades of fashion-driven nonsense in software architecture that came and went without ever being tested against anything. But the pressures exist, and they create a background gravity that, over time, pulls those disciplines back toward the parts of reality their work is supposed to track. The bridge has to keep standing. The drug has to keep working. The model has to keep predicting. These are not abstractions; they are the brute facts a discipline cannot lie about indefinitely.

There’s a sharper way of saying this. Reality-correction shows up at the level of the individual scholar’s record. In disciplines with feedback loops, being wrong has reputational consequences that follow the person forward. The structural engineer associated with a bridge collapse carries it. The investment manager whose fund blew up has a Wikipedia page documenting how. The clinical researcher whose drug came back with a mortality signal gets remembered as the one who pushed it. The discipline keeps a running record of who has been wrong and how badly, and that record shapes how future claims from those people get received. There is professional cost to being publicly wrong, even when there isn’t legal cost. The reputational ledger is itself a kind of reality-correction, and arguably the most important kind, because it operates on the people rather than just on the institutions.

In the humanities, and in the precincts of the social sciences most affected by what I’ll call the constructivist turn, that gravitational pull has been (very) weakened, sometimes deliberately so. There is no straightforward analog to “the patient dies” for a discipline that has, in its most influential theoretical moments, talked itself out of the idea that there are external standards against which its claims could fail. The strongest constructivist arguments do not deny that the world exists; that’s a strawman of a real and more careful tradition. They make the more limited claim that the conceptual schemes through which we describe the world are themselves social products with histories, and those histories foreclose any clean claim to neutral description.

The reputational ledger does not work the same way in these disciplines either. The scholar whose theoretical framework gets superseded next decade doesn’t get publicly retracted. The literary critic whose canonical reading turns out to have been ideologically motivated rather than textually grounded can still be on the syllabus next semester. The sociologist whose foundational empirical claims fail to hold up under scrutiny keeps getting cited. There is no career cost to having been spectacularly wrong fifteen years ago, because there is no public standard to which “wrong” could be attached, and no shared institutional memory of who has and hasn’t earned credibility through accurate prior claims. Being wrong, in disciplines without reality-correction, doesn’t follow you.

And the citation side of this is just as striking, I think. In disciplines without reality-correction, what gets cited isn’t necessarily what’s correct or even useful. Citation patterns track ideological consonance and aesthetic fit with the citer’s own commitments much more than they track methodological accuracy, or so it seems to me. Scholars build on the work of other scholars whose conclusions they like rather than whose methods they trust. (And yes, before someone points it out, I’m including some of my own colleagues in that, and probably myself when I’m not paying attention.) The discipline’s running record of “who has been demonstrably right” gets replaced by a running record of “who belongs to my coalition,” and being on-narrative becomes, in practice, a stronger career advantage than being correct. Of course, citation networks in every discipline are influenced by social and ideological dynamics; the real question is whether accuracy-as-criterion gets a chance to push back when ideology pulls citation in another direction. In reality-correcting disciplines, mostly, it gets that chance. In the precincts under critique, the push-back is, frankly, weak enough that the ideological filter dominates much of the time.

The problem is that, well, once that’s in the water of a discipline, what survives is descriptive humility on the way in and a license to dismiss external correction on the way out. The next step, after a few decades, is a discipline in which approval-optimization inside the guild becomes the dominant selection pressure, because there’s nothing else to select on. That’s the disciplinary version of the RLHF problem in AI alignment, which is the thread Part 3 of this series ran. Disciplines that approval-optimize without reality-correction will always, over time, select for skillful rationalization over accuracy.

The picture I’m painting is too clean to be completely and literally true, and the soft version needs stating. Engineering and business and medicine are not fully inoculated against ideological drift, and the humanities are not fully without reality-correction. Historical scholarship gets corrected by archives. Linguistic claims get corrected by field linguistics. Translation gets corrected by parallel texts. The argument is about differential strength of feedback loops across disciplines, with the precincts under critique sitting at the weakest end of that continuum. The cluster is real even if its boundaries are fuzzier than a bright line would suggest. The disciplines this piece is about are the ones where the feedback gravity is weakest, where motivated reasoning has the most room to compound, and where a particular meta-theoretical tradition has spent fifty years arguing that external correction is itself a political move that should be resisted.

The Sophistication Amplifier

So far the argument has been about feedback loops at the level of the discipline. There’s a second mechanism layered on top, and it operates at the level of the individual scholar, and it inverts the intuition most readers bring.

You would think, and most people do think, that more knowledge, more careful reasoning, more years of training would make a person better at updating on disconfirming evidence. The folk model is straightforward: someone who knows a field deeply has more cognitive resources to spot mistakes, weigh counter-evidence, and revise their views when the data come in. By that logic, scholars and credentialed experts should be the most flexible, most calibrated, most readily-correctable believers in society.

The empirical literature on motivated reasoning has, for about two decades, kept landing on something close to the opposite. That qualification matters and I’ll get to it in a moment. The bare claim, in its strongest version, is that sophistication does not produce better updating. It produces more skillful defense of whatever the reasoner started out believing.

Three pieces of work are most commonly cited for this claim, all of which I’ve spent time with as part of my own academic research. Charles Taber and Milton Lodge’s “Motivated Skepticism in the Evaluation of Political Beliefs” (American Journal of Political Science, 2006) ran experiments showing that participants with higher political sophistication exhibited stronger biased assimilation and stronger disconfirmation bias than less-sophisticated participants. The more politically informed they were, the more time they spent finding ways to dismiss evidence they didn’t like. Dan Kahan’s work on “identity-protective cognition” generalized the finding into a framework: across multiple studies, higher numeracy and higher reasoning ability widened rather than narrowed partisan polarization on politically loaded empirical questions. The 2012 Nature Climate Change paper is the canonical demonstration. Hugo Mercier and Dan Sperber’s The Enigma of Reason (Harvard, 2017) offered a cognitive-evolutionary theoretical case for why this pattern would obtain at all: reasoning, on their account, evolved primarily to win social arguments rather than to discover truth, which means the cognitive equipment we have is well-tuned to producing justifications for whatever we’ve committed to socially.

Now the contestation. Kahan’s findings have been re-examined since 2017 and some of the strongest “polarization widens with sophistication” results turn out to be partly artifacts of question framing rather than robust general effects. Mercier and Sperber’s argumentative theory is exactly that: a theoretical proposal with evolutionary-psychology framing, not a tightly-tested mechanism. The lab findings on individuals don’t translate cleanly into claims about how disciplines collectively behave; that’s a level-of-analysis inference, not a direct empirical result. The careful reading of this literature would say: the directional pattern (sophistication amplifying defensive reasoning rather than corrective updating) shows up across multiple methodologies and is hard to make go away even if you discount the most aggressive readings, but the strong version “sophistication produces more skillful defense, full stop” is more contested than popularizers of the work imply.

I’m narrating it this way because it’s where my own position most affects the argument. I work in this literature. I have priors about which findings are robust and which are fragile. The way I just sketched the contestation is itself a reading; another political scientist working on the same material would tell the story differently. A reader who wants to push back on the strong version of the sophistication claim has plenty of material to push back with. The honest version of the section’s claim is that the directional pattern shows up consistently in good studies, the strongest readings of it are oversold, and the careful middle ground is enough to do the analytical work this piece needs.

The level-of-analysis inference is its own thing and worth flagging directly. Even granting the individual-level findings, getting from “individuals with high sophistication show stronger motivated reasoning in lab settings” to “academic disciplines collectively produce more sophisticated rationalization” requires an additional argument. The argument is roughly: a discipline is a population of individuals interacting under institutional incentives that reward winning intra-guild arguments. If the individuals are selected for and trained in the cognitive equipment that produces motivated reasoning, and the institutional incentives reward exactly the social task that equipment was built for, then the collective output of the discipline will lean toward more skillful rationalization than corrective updating. That’s a reasonable inference about a level the empirical work doesn’t directly speak to. I endorse it, with the acknowledgment that it’s an extension and could be wrong and would be worth more careful empirical work.

Combine the individual-level pattern with the structural account of the previous section and you get the diagnosis this piece is making. In disciplines with reality-correction loops, the motivated-reasoning machinery is still running, but it bumps against an external constraint. The bridge has to actually stand. The model has to actually predict. The drug has to actually treat the disease. The cognitive equipment is the same; the constraint differs. In disciplines without much reality-correction, the same equipment runs without that constraint. The smart people who have been selected for and trained in skillful argumentation are free to spend their entire careers building elegant defenses of positions that are never seriously tested against anything outside the guild. Academics tend to be smarter and harder-working than most people, which is precisely why their failure modes are so durable when the discipline lacks correction.

The argument doesn’t immunize the author. Everything I just said applies to me! Political science has more reality-correction than literary studies (formal models do get tested against elections, survey instruments have to describe and predict something) but less than physics or experimental medicine. I’m writing inside the same selection process I’m describing. So is anyone with a doctorate.

The specific place this matters most for my own argument is the next section, on the meta-theoretical apparatus of contemporary critical theory. That section makes claims about the intellectual culture of disciplines I do not work in and am reading from the outside. A humanist who works on that material from the inside might read the diagnosis differently than I do, and the careful version of what I’m about to say has to leave room for that possibility.

The Closing Door

The picture so far is concerning enough: disciplines without strong reality-correction, staffed by people whose training selected for skillful motivated reasoning, accumulating bad ideas that don’t get tested against much outside the guild. But the picture misses what makes the specific humanities precincts under discussion uniquely difficult to reform compared to other status hierarchies that also produce motivated reasoning.

Sales departments produce motivated reasoning. Legal teams produce motivated reasoning. Newsrooms produce motivated reasoning. None of these are inoculated against it. What distinguishes them from the academic precincts I’m pointing at is that they still treat external correction as legitimate. The salesperson loses the deal and learns something. The newsroom prints a retraction. The legal team loses in court. The cultures of these institutions have not produced a sophisticated argument for why being held to external standards is itself a form of domination.

In significant precincts of contemporary academic humanities and humanities-adjacent fields, that argument has been produced. It has been produced at high levels of theoretical sophistication. It has been refined and extended for decades. In the average case and in the popularizer-level reception, it functions as a resource readily available to deflect external critique. Any claim that the discipline has gotten something wrong can be reframed as a positioned reading, as speaking from privilege, as naturalizing the dominant view. The criticism doesn’t have to be addressed on its merits because the meta-theory has already explained what the criticism is really doing.

That’s the strong version of the diagnosis. The strong version needs to be immediately weakened, of course, because the strong version is unfair to the strongest contemporary work in the same tradition. Habermas spent his entire career arguing for normative foundations rooted in communicative reason and discourse ethics, which is the opposite of “tools without metaphysics.” Axel Honneth has done sustained work on recognition as a substantive normative framework. There are contemporary critical theorists doing serious, careful, falsifiable-in-principle work that resists the easy summary I’m tempted to give. The diagnosis applies most cleanly not to the deepest contemporary work in the tradition but to the popularizer reception, the seminar-textbook version, the social-media inheritance, and the everyday-discipline deployment of the apparatus by scholars who never deeply engaged with the foundational philosophical project.

Scott Alexander’s recent review of Martin Jay’s The Dialectical Imagination (itself a popular Substack engagement with the canonical history of the Frankfurt School) proposes one diagnosis of how the apparatus survived its original justification. The Frankfurt School itself, on this reading, was working from a mystical-Hegelian commitment to history manifesting something like God as Communism, and the apparatus of negative dialectics and critique-as-method was built to serve that telos. You didn’t specify the utopia positively because you couldn’t; you worked instead by negation, by noticing the cracks in existing arrangements, by holding open the space in which a not-yet-articulable better world could eventually emerge. The intellectual children inherited the tools and, in the average case, dropped the metaphysics. The negative-dialectical method survived without the eschatological purpose it was built to serve. What’s left, on this reading, is a sophisticated critique machine with no destination, available for deployment against any external claim and accountable to no positive vision.

The reading needs caution. It’s a popularizer’s diagnosis from a one-book review of Jay’s classic history, and a Frankfurt scholar would push back on several pieces of it. Adorno’s critique of instrumental reason was specifically a critique of instrumental rationality divorced from ethical orientation, the form of reason that built the gas chambers and systematized colonial extraction and produced the bureaucratic logic Hannah Arendt later called the banality of evil. It was not a generic argument against engineering or business or against the existence of feedback loops in disciplines. Reading the Frankfurt School as having simply had the polarity backward on reality-correction is a hot take rather than careful intellectual history. The Scott Alexander mechanism story is suggestive of something (of how a sophisticated philosophical apparatus can survive past its justifying purpose and become an unmoored defensive resource) but it is one possible story and not the verdict of scholarship.

With those caveats: the phenomenon the story is gesturing at, even if not delivered carefully, is real. In the average reception of contemporary critical theory, in the popularizer texts and the seminar applications, the apparatus gets deployed asymmetrically. It is invoked against external critics far more readily than it is turned reflexively on the discipline’s own commitments. That asymmetry is what makes the humanities precincts under discussion uniquely hard to reform compared to other status hierarchies. Other status hierarchies also generate motivated reasoning. What I think sets these precincts apart is the availability of a high-status intellectual resource for legitimating the refusal of correction.

The Same Failure on the Right

The structural critique I’ve been laying out cuts a particular direction, and it would be dishonest of me not to address why. Most of the load-bearing examples in the previous sections are from disciplines that lean culturally left, and the meta-theoretical apparatus under critique is one that grew out of leftist intellectual traditions. A reader on the political left who has been nodding along through the diagnosis has every reason to wonder whether the piece is performing the both-sides move I claimed in the opener while actually delivering a one-sided complaint.

The honest answer is that the structural diagnosis applies just as cleanly to a different set of institutions on the right, and the piece would be incomplete without naming them.

The clearest contemporary case, at least to my eye, is what’s sometimes called the dissident right intellectual ecosystem: the cluster of writers, podcasters, and political theorists around figures like Curtis Yarvin, Christopher Rufo, and Tucker Carlson, plus the donor-funded infrastructure that amplifies their work into policy. Like the academic humanities precincts the previous sections described, this ecosystem is a status hierarchy without much reality-correction. The bridge does not fall down when a Yarvin essay misreads the bureaucratic state. The drug does not stop working when Rufo asserts that CRT means whatever this week’s culture-war fight needs it to mean. The model does not fail to predict when Carlson narrates a conspiratorial worldview in the absence of evidence. The feedback gravity is approval-driven (engagement, donor satisfaction, intra-tribe status) rather than reality-driven. Dan Williams has made the corresponding case at length in his recent essay, which I won’t repeat here.

The sophistication amplifier from earlier applies to this ecosystem too, and arguably more strongly. The dissident right is staffed by smart, articulate, motivated reasoners whose institutional incentives are explicitly optimized for tribal-internal status. The reasoners get sharper at building defenses of in-group positions over time, the institutional structure rewards exactly that sharpening, and there’s no external feedback loop pulling the work back toward accuracy because the audience and the donor class don’t fund accuracy-checking.

The immunization layer is present too, in a different intellectual form. Where contemporary critical theory provides anti-foundationalist resources for deflecting external critique, the dissident right has built its own version: a worldview in which establishment institutions (universities, legacy media, civil service, regulators) are presumed to be a coordinated regime whose claims can be discounted as enemy propaganda by default. Any external critique can be reframed as evidence of the regime defending itself. The structural function is identical to the one I described in the previous section. Different intellectual content, same defensive mechanism.

I’m naming this case explicitly because the piece’s structural argument has to land symmetrically or it fails its own test. The diagnosis I’m making is about a type of institution: status hierarchies that lack external reality-correction, populated by sophisticated reasoners, equipped with intellectual resources for legitimating the refusal of correction. That type can be inhabited by left-leaning academic precincts or by right-leaning anti-establishment ecosystems. The political flavor varies. The structural failure mode is the same.

A reader will reasonably ask: if the failure mode is the same, why does the piece spend so much more time on the academic humanities than on the dissident right? Two reasons. The academic humanities have higher institutional prestige and longer-lived legitimacy claims to draw on, which means the failure carries further downstream than the analogous failure on the right. And the academic version of the failure is what’s most relevant to Kuran’s question, which is the question this piece set out to answer. The dissident-right case deserves its own piece. I’m noting it here so the structural argument doesn’t read as a partisan attack on academia by omission.

The Countercurrent

There’s something the piece would be unfair not to say. Not everyone has stopped showing up. The humanities, broadly construed, still contain people doing exactly the work Kuran’s tweet asks for. They are doing it from a wider range of intellectual and political coordinates than the previous sections might suggest, and a fair version of this section has to name that.

There’s a cluster of serious public-facing humanist work on AI from inside something close to my own sensibility. Shannon Vallor’s The AI Mirror takes up what AI does to the human capacities we used to think defined us. Antón Barba-Kay’s A Web of Our Own Making does the same kind of work on digital culture more broadly. Brian Christian’s The Alignment Problem is journalism, but humanities-trained journalism. Mary L. Gray and Siddharth Suri’s Ghost Work brings a sociologically rigorous lens to the labor structure underneath AI. Byung-Chul Han has been writing on attention and automation in a related register, with a European flavor, for over a decade. There’s a small but real revival of Iris Murdoch’s moral philosophy as a framework for virtue and attention in an age of distraction.

There’s another cluster doing equally serious work from intellectual coordinates I don’t share by default, and that the piece would be incomplete without naming. Ruha Benjamin’s Race After Technology and Safiya Noble’s Algorithms of Oppression do the work of making AI systems’ downstream effects visible in ways the technical literature doesn’t. Kate Crawford’s Atlas of AI traces the political economy of AI: the material, labor, environmental, and geopolitical substrate that the AI-as-magic-cloud narrative obscures. Joy Buolamwini and Timnit Gebru’s research on algorithmic bias, and Yuk Hui’s work on what he calls “cosmotechnics” (different ways of thinking about technology rooted in non-Western philosophical traditions), are both doing the civilizational-illumination work Kuran is asking for, from registers and concerns that don’t show up in the first cluster.

There’s a third cluster doing the kind of disciplined epistemology of public belief that this piece has been drawing on throughout. Dan Williams’s recent Conspicuous Cognition work, including the essay on speaking truth to power I cited above. Robin McKenna’s argument for a “strong programme in political epistemology”, which makes the methodological case for symmetric, reflexive analysis. Neil Levy on epistemic environments. C. Thi Nguyen on epistemic bubbles versus echo chambers. Joseph Heath on the dogmatism of contemporary critical theory. Scott Alexander on the intellectual history of the Frankfurt School, whose review of The Dialectical Imagination I’ve leaned on heavily above.

A fourth cluster is doing the reflexive work from inside the disciplines I’ve been critiquing, and this is the cluster the piece most needs to name. Ashley T. Rubin at the University of Hawaiʻi has been writing, both on her Substack and in Theory and Society, about how activism has been displacing theory in contemporary sociology and criminology. She is making essentially the diagnosis I am making, but from inside one of the disciplines under critique, with the specificity and methodological sensitivity the next-door view I’m offering can’t match. Rory Truex at Princeton is asking the adjacent question from inside political science specifically: in a recent essay he doubts whether humans will still be doing political science research in five or ten years, which is the AI-side mirror of what this piece is arguing about the institutional side. Joseph Heath belongs in this cluster too. Steven Brint at UC Riverside has documented the long-run hollowing-out of sociological theory. Jonathan Haidt and the broader Heterodox Academy network, whatever else one thinks of their politics, have been doing serious empirical work on the institutional dynamics this piece has been describing. And a generation of younger scholars are quietly continuing serious work without adopting the methodological commitments the loudest precincts have adopted.

These names share a pattern worth saying plainly. Some are at universities, but the work that defines them is happening at the edges of the official disciplines: in trade presses, on Substacks, in journalism, in independent institutes, in the loose connective tissue of the public-intellectual ecosystem that runs parallel to the academic mainstream. They are doing the work the official disciplines, in the configuration described above, no longer reliably reward. Academia is salvageable. But as currently configured, it isn’t the address you would look at first for the civilizational illumination Kuran is asking for. The work has moved to wherever the people who exited or routed around it have set up shop.

This raises a hard question about whether the existing structures can be reformed in time to re-enclose this work, or whether it has permanently migrated. The honest answer is that I don’t know. Academic institutions have absorbed insurgent intellectual movements before, and they could again. The current configuration of incentives makes that absorption harder than it has historically been. My uncertainty is real.

And Who Doesn’t Love a Good Coda?

Kuran’s tweet asked the humanities to do something they were built to do and that they are conspicuously not doing in the public registers where it would matter most. The piece I’ve just written argues that the reasons are structural: a discipline cluster with weak external feedback gravity, populated by sophisticated reasoners whose training amplifies motivated defense rather than corrective updating, equipped with a meta-theoretical apparatus that gets deployed asymmetrically to deflect external critique. The same structural failure mode shows up in other prestige reasoning communities when external correction is foreclosed, including the dissident right intellectual ecosystem. The political coloration varies. The mechanism is the same.

The argument establishes the structural pattern, with the caveats and contestations I’ve flagged along the way. It establishes that the work Kuran is asking for is happening, at the edges of the disciplines, in registers and from sources the dominant institutional apparatus does not reliably reward. It does not establish a clean reform program, because I don’t have one. It does not establish that the academy is beyond saving, because that’s an empirical question whose answer depends on what happens inside the disciplines over the next decade.

What I think it does establish is that the gap Kuran named is real, the reasons for it are deeper than the easy diagnoses on either side of the political spectrum, and the work of repair will be done, if it gets done, by people inside the disciplines who decide that the costs of the current configuration outweigh its comforts. The Kurans and the Williamses and the McKennas and the Rubins of the world can describe the gap from outside or from the edges. They can’t close it.

Reality bats last on disciplines too. The disciplines that were built to make sense of moments like this still exist. The people inside them have to decide whether the work they were built for still matters enough to defend.

---

This piece sits alongside the Reality Bats Last series. The four parts (Reality Bats Last, The Fifty-Year War, The Other Pillar, and Letting Reality Bat Last) build out the underlying philosophical scaffolding and stand on their own. If you’re new here, that’s where to start.

If the diagnosis lands, share it with someone whose discipline needs to hear it. Forwarding does more for these pieces than the algorithm does.

And if I got something wrong, tell me in the comments. The reflexive commitment in the piece is real; it doesn’t end at this piece.