The set has a center, and it is a library.
SocArXiv runs out of the University of Maryland Libraries, which means sociologist Philip N. Cohen serves at the pleasure of the dean of the libraries, as he puts it, and the dean pays for it, and the payment is small.
Around that center sit four groups that overlap.
The first is the open scholarship infrastructure world: the Center for Open Science, where Brian Nosek built the Open Science Framework that the archive runs on; MIT Libraries, where the Center for Research on Equitable and Open Scholarship hosted Cohen as a visiting scholar in 2018 and where Chris Bourg has been the most articulate dean in the movement; ASAPbio and Jessica Polka, who did for biology what SocArXiv attempted for sociology; Kathleen Fitzpatrick, who came out of the Modern Language Association arguing for open peer review before most sociologists had heard of a preprint. Cohen interviewed all four in 2018 and published the videos. That is the set’s canonical roster. It is librarians, a psychologist, a biologist, and a literary scholar, held together by a shared enemy.
The second is family demography: Joanna Pepin at Buffalo, Kate Choi at Western, Brandon Wagner at Texas Tech, Ge Gao, Kelsey Drotning, Jeehye Kang, Jaein Lee, Hao-Chun Cheng, the doctoral line. Reeve Vanneman upstream as adviser, Suzanne Bianchi as the mentor who died in 2013, Lynne Casper as the early coauthor, Matt Huffman and Jessica Pearlman from the Irvine years. Stephanie Coontz is the senior figure whose book he reviewed warmly in 2026 and whose function in the set is to be the historian who established that the traditional family was a mid-century episode. Philip Levine, Melissa Kearney, Sarah Damaske, Jennifer Randles, Jennifer Glass are the adjacent people whose work he engages, sometimes critically.
The third is the public sociology apparatus: Contexts, which he coedited with Syed Ali and Letta Page; The Society Pages and Sociological Images out of Minnesota, where Lisa Wade’s work sits and where Cohen’s blog posts circulate; Scatterplot and orgtheory, the discipline’s comment-section commons; the Council on Contemporary Families, which vets and releases research outside journals entirely and which he cites as a working model. Sociological Science, founded in 2014 by Jesper Sørensen and others precisely to publish fast with open access and a reproducibility policy, is the set’s demonstration project: the journal that does what he says journals should do.
The fourth is the reform wing that fights the American Sociological Association from inside. Committee on Publications members, the two hundred signatures on the 2019 petition against the association’s letter to the White House, the Publications Committee that passed a resolution the Council ignored.
What they value.
Access first, and it is a moral claim. The public paid for the research through grants and salaries, so the public owns it, and a paywall is a second sale of a thing already bought. Cohen says we do not aspire to have our work hidden from the people it most affects. The image that recurs is the researcher in a country whose libraries cannot buy the packages, reading the same paper as a researcher at Harvard.
Speed second. The set’s founding grievance is the twenty months between submission and publication at the discipline’s slowest journals, and the doldrums period in which a paper sits waiting for people whose arms are being twisted to read it while, as Cohen says, you hope there are people out there who want to.
Checkability third, and the rhetoric outruns the practice. Post the code, deposit the data, let a stranger recompute. Nobody recomputes. The value is held anyway, and Cohen has said in public that he cannot prove it.
Description fourth. The set prizes a fact a general reader can hold: divorce is falling, marriage is concentrating, the bottom tenth of Americans sits above half the world in income. The discipline treats descriptive findings as lesser, which Cohen says pushes authors into puffery.
The set’s exemplary man is the person who built the thing, who stood up a server, wrote the moderation policy, recruited the volunteers, and kept it running for ten years on a library line item. Cohen received a European prize in 2025 for that.
Second in the hero system is the man who pays a cost. The resignation letter published in public. The plaintiff’s name on the caption. The petition organized when the association’s leadership went the other way. Sacrifice is currency.
Third is the corrector who audits his own side. This is the set’s highest form of honor and its rarest performance, and Cohen’s standing rests on it more than on anything else: he went after Mark Regnerus, which was free, and then he went after Alice Goffman, which was not.
The villains. Elsevier, which bought SSRN in 2016 and gave the movement its founding trauma. Sage, which holds the association’s contract and, in the set’s most quoted line, throws the leadership a nice party at the conference. Springer, Taylor and Francis, Wiley: the five publishers who account for roughly eighty percent of recent sociology articles. And the association staff, who in this telling outlast the elected leadership, absorb reform proposals, and protect a revenue structure.
The status games:
Prestige-journal placement still counts and cannot be admitted to count. The set’s members publish in American Sociological Review while arguing that its selection function is a status device. Cohen’s own resolution is to publish in Socius and Sociological Science where he can, both open access, and to post preprints of everything, which converts a compliance problem into a display.
Download counts are the substitute currency. Three thousand downloads of a preprint before the journal ran it. Seventeen thousand papers in the archive. The counts are visible, public, and comparable, which is what a status currency requires.
Being cited by a journalist ranks high and is not admitted to rank high. Cohen keeps a media page and updated it in January 2024 with an anonymous comment calling him the best sociologist of his generation, said ouch, and observed that outreach time is research time forgone.
And the highest-status move in the set is the audit. Counting other people’s compliance is a service to the community, a display of one’s own compliance, and an assertion of standing to judge. The December audits do all three at once, which is why they are the set’s signature genre.
Now the normative claims:
Research funded by the public belongs to the public. Delay in dissemination is a harm with victims. A finding that cannot be inspected is not yet knowledge. A professional association exists for its members, and one that transfers money from teaching-heavy libraries to research-heavy ones through its publishing contracts has inverted its purpose. Scholars are citizens and owe the public engagement. Values shape the questions anyone asks, so disclosure of standpoint is an obligation.
The essentialist claims:
The first is about scientists: that a researcher, properly formed, wants his work checked, and that resistance to inspection indicates something about the resister. Under this assumption a person who will not share data is hiding.
The second is about institutions: that they follow revenue, and that a stated principle in conflict with an income stream will lose. It is usually right.
The third is about the public: that there is one, that it wants the research, and that access is the barrier. The evidence for this is thin. Removing a paywall does not create a reader who can evaluate an age-standardized rate, and the set has never addressed the gap between availability and competence.
The moral grammar:
The recurring word is owe. Researchers owe the public inspectability because the public funded the work. The association owes members an accounting. The journal owes readers the code. Obligation flows from having received something, which is why the argument is always about who paid.
Withholding is the characteristic sin and openness the characteristic virtue. In this world a boring correct paper posted publicly is better than a brilliant one behind a paywall, and Cohen’s willingness to say out loud that his platform contains a great deal of bad work follows. Quality is a lesser good than availability.
Confidentiality, human subjects protection, the ethnographer’s promise to a source, the reason a qualitative researcher cannot deposit a transcript: all of it enters as an exception to be managed. When Cohen was asked about it in February 2022 he improvised a restricted-access tier on the spot and said it was not his area. Nobody built it. Four years later it still does not exist, which tells you where the moral energy of the set is and is not.
Who is not in the set?
Ethnographers are not in it. Historians are not. Theorists are not. Anyone whose evidence cannot be deposited is structurally outside a movement whose central sacrament is depositing. The American Journal of Political Science required verification from 2015 and, three years in, had accepted no qualitative manuscripts at all under the policy.
The set reads that as a coincidence. Everyone outside it reads it as the point.
The Voice
Cohen learned to write for a clock. Two versions of every story, each carrying a different quote, same length whether the story was a fire or a zoning variance, plus one-sentence headline versions for the quarter hours. At the end of the local news he wrapped the market report and the weather while the second hand came around, because at six the network took the air. Nothing in that job rewards a subordinate clause. Everything in it rewards knowing how long a thing takes to say and stopping when the time is gone.
Forty years later the sentences still land on the beat.
The default sentence. Subject, verb, number. “Marriage is becoming rarer and more durable at once.” “The rent eats first,” borrowed. “There’s less of everything demographic happening.” He builds paragraphs by stacking short declaratives and then, at the end, dropping in one that is shorter than the rest. The rhythm is a news reader’s rhythm: statement, elaboration, statement, stop.
Numbers as sentences. His signature move is to convert an abstraction into a quantity a person can hold. Twenty-nine of the years between eighteen and fifty-five spent married in 1960, eighteen now. Only eight of a hundred and twenty-seven papers reproduced on the first attempt. The bottom tenth of Americans sits above half the world’s population. Four of fifteen, then eight of twenty-five, then twelve of eighteen. He states them and moves, and the absence of adjectival help is what makes them land.
Concrete over abstract. He will say that making banana bread is a short-term response and redesigning your home is not. Printing was expensive and you could decide what got shipped and bound and stuck on a shelf. The Gini index arrives as four named people holding stacks of cash. Homophily arrives as liking people like yourself.
Register-jumping. This is the most distinctive thing in the prose and the hardest to imitate. He runs a technically exact sentence and then follows it with something conversational, sometimes crude, and the drop is deliberate. Software developers were much less affected than working-class jobs, thank you Zoom. He proposes a great methods assignment: condoms prevent births, so fewer condoms should mean more births, no, therefore people are having less sex. He tells a room that Google is not really your friend, it is okay to use Google tools because they are awesome, but they will not love you back in the end. The joke always does structural work, marking the end of a section or puncturing a claim.
Self-interruption as a habit. He corrects himself mid-sentence in public more than any writer I have read. Speculative, or I should say suggestive. It’s a harsh way of putting it. I realize later maybe that’s porn searches, I’m not sure. Don’t hold me to this methodology exactly. My methods are squishy. The corrections are the same reflex that makes him check whether a survey instrument changed before he explains a trend, turned inward and left in the draft.
The concession as a rhetorical engine. His strongest passages are built by giving the other side its best version first and then turning it. Maybe people don’t care about marriage anymore. Or maybe they care more and have put it on a pedestal so high their own relationship doesn’t qualify. Crime seems bad, and maybe crime helps a society mark the boundary of acceptable behavior, and poverty motivates people not to starve, which has a logic. Then the turn. This is a debater’s structure and he uses it in writing and in lecture.
Negation as a favorite figure. Constant. It is not so much that people are not getting married ever, it is that they are getting married later. Reproducibility is not really the issue, but maybe accountability is. It’s not that they don’t want to be married, they have a higher standard. The construction lets him kill a familiar reading and install his own in a single move, and he leans on it hard enough that it is a fingerprint.
Two verbal tics. He says “sort of” and “you know” constantly in speech and neither survives into his prose, which means he writes better than he talks and knows it. And he begins a striking number of written arguments with an admission of what he cannot do: we don’t measure this well, any number here is an estimate, nobody knows the rate.
Then the other voice.
The prosecutorial register. Post titles are the purest sample: Brad Wilcox lies a lot, part whatever. It’s not bigotry to say Lyman Stone is a bigot. The American Sociological Association is collapsing and its organization is a perpetual stagnation machine. The syntax is the same as everywhere else, declarative and short, and the content has moved from what a thing is to what a person is. “Part whatever” is doing specific work: it announces a series, signals contempt for the necessity of the series, and preempts the charge of obsession by acknowledging it first.
He still links everything. He still quotes the other side at length before answering. He still puts the arithmetic in. The prosecutorial posts are as heavily sourced as the audits, which is exactly why they unsettle people who would find an unsourced insult easy to dismiss.
The teaching voice. A third mode, and the one with the largest audience. Patient, sequential, repetitive on purpose, and almost entirely free of the other two. He builds from a case rather than a definition. He tells students what he is about to do, does it, and says what he did. He addresses them directly, warns them when a slide has a lot of numbers, tells them the take-home message is written in red. And in that mode he takes the opposing theory seriously in a way the blog does not: functionalism gets its fever-fighting-infection version, the March for Life teenagers get the same analytic respect as the Women’s March.
Email. You have the sample. No, don’t know, not to my knowledge, no, with care, see Popper, asked and answered, I disagree with the premise, yes. Twenty-three words for nine questions. This is the news-desk economy with the warmth stripped out, and it is a mode, not a mood: he writes at length in his own comment section to people who ask him things there. The venue determines the length.
Now the two things nobody says about the style.
The first is that his prose is a public-facing instrument built by a man who works entirely in private data. He spends his days inside survey microdata that a general reader cannot open, and he writes as though the reader is standing next to him at the terminal. The whole style is a bridge over a gap he never mentions. This is why description is his real subject: description is the only output of quantitative social science that does not require the reader to have a license.
The second is that the strongest and weakest sentences he writes are the same shape. Consider two. “If the facts don’t support the generalization, then the anecdotes are not illustrations of a trend. They’re just little stories.” And: “Brad Wilcox lies a lot.” Both are compressed, both are declarative, both refuse hedging, both were written by a man who believes that a plain sentence is more honest than a careful one. The first is the finest thing in his corpus. The second is the thing that keeps colleagues who accept every one of his statistical arguments from citing him.
The style has no setting for a claim that requires trust. It renders those claims at the same volume and in the same syntax as the checkable ones, and the reader cannot tell from the prose which is which. That is the direct cost of the virtue: a man who trained himself never to soften a true sentence has no equipment for softening an unproven one.
If you want to imitate him, three rules and one prohibition. Put the number in the sentence. Give the other side its best version before you answer it. Break the register once per section with something funny. And do not let the prosecutorial mode into a piece that also contains arithmetic, because he does, and it costs him every time.
