In December 2014, a visitor carrying the measles virus walked through Disneyland. Within weeks the outbreak spread across California and beyond, infecting more than a hundred people in a country that had declared measles eliminated in 2000. In the Bay Area, Renée DiResta (b. 1981), a former Wall Street trader turned venture capitalist, had a son approaching preschool age. She did what a trader does before taking a position. She pulled the data. California published vaccination rates by school, and the numbers stunned her. Some Bay Area preschools, filled with the children of engineers and executives, had immunization rates below those of South Sudan. She began looking at where the anti-vaccine message came from, and with the data scientist Gilad Lotan she mapped the networks. The maps showed that on Twitter about a quarter of the anti-vaccine content came from 0.6 percent of the accounts. A small, coordinated, passionate minority looked like a mass movement. She had found the subject that consumed the rest of her career.
Nothing in her training pointed toward public health. DiResta grew up in Yonkers, New York, the daughter of a family with no connection to Silicon Valley or Washington. At Stony Brook University she took five years to finish two degrees, computer science and political science, with two minors. During the summers of those undergraduate years, from 1999 to 2004, she interned at the Central Intelligence Agency. She decided against staying at the Agency, took the LSAT, and considered law school. Instead she took a job at Jane Street Capital, the quantitative trading firm in New York, where she started as a clerk writing code to scrape data from Bloomberg terminals in the days before the firm had data feeds. She stayed seven years and became an equity derivatives trader and market maker. The work rewarded speed, pattern recognition, and comfort with incomplete information. Prices moved on rumor before they moved on fact. A trader who understood how a story spread through a market before it appeared in a newspaper had an edge. She later said the common thread across her jobs was a love of high-intensity environments with big analytical problems and adversarial behavior.
In 2011 she moved west and became a principal at O’Reilly AlphaTech Ventures, a seed-stage fund, where she focused on hardware, manufacturing, and logistics. She co-authored The Hardware Startup in 2015 and joined the founding team of Haven, a supply-chain logistics company. She was, at that point, a competent and obscure figure in the technology economy, one of thousands of people in the Bay Area who moved between trading, investing, and startups. The Disneyland outbreak changed the trajectory.
In 2015 she co-founded Vaccinate California, a parents’ group that backed legislation to end California’s personal belief exemption for childhood vaccination. The fight over that bill taught her the lesson she repeated for the next decade. Her side had the medical establishment, the data, and majority opinion. The other side had the feeds. Anti-vaccine activists ran coordinated hashtag campaigns, flooded legislators’ social media accounts, and dominated search results. She called this the asymmetry of passion. Online influence requires no majority. It requires repetition, emotional intensity, and platforms whose recommendation engines reward engagement over accuracy. She noticed something else. When she followed anti-vaccine groups on Facebook, the recommendation engine began suggesting chemtrail groups, anti-GMO groups, and Pizzagate. The platform did not merely host conspiracy communities. It introduced people to them.
The 2016 election made that observation a national security question. After the intelligence community concluded that Russia had interfered in the campaign, the Senate Select Committee on Intelligence went looking for outside experts who could read platform data. DiResta had by then joined New Knowledge, an Austin-based firm that tracked online manipulation, as director of research. On August 1, 2018, at 9:32 in the morning, Chairman Richard Burr (b. 1955) called a hearing to order in room SH-216 of the Hart Senate Office Building. Poster boards displaying fake social media accounts stood on easels beside the witness table. DiResta sat with witnesses from RAND, Graphika, Oxford, and the German Marshall Fund, the emerging expert class of a field that had not existed three years earlier. She told the senators the country faced a defining threat of the generation and warned that future operations might use fake audio and video generated by artificial intelligence. She testified about Russian campaigns that pushed anti-fracking messages into oil regions and GMO fears into farm states. The senators, men who had grown up on network television, listened to a former derivatives trader explain how trending algorithms could be gamed.
At the committee’s request, DiResta and her New Knowledge co-authors then analyzed the datasets Facebook, Twitter, and Alphabet had turned over. Their report, “The Tactics and Tropes of the Internet Research Agency,” released on December 17, 2018, examined Russian Internet Research Agency operations against Americans from 2014 through 2017. The finding that stayed with her concerned race. The IRA ran fake pages aimed at Black Americans, built audiences around Black pride and police violence, and then pushed messages of alienation, including encouragement to sit out the election. The operation spent less effort converting voters than fragmenting communities and convincing people that participation was pointless. Burr said the data showed how aggressively Russia had worked to divide Americans by race, religion, and ideology.
Two days after the report’s release, the New York Times published a story that complicated everything. During the 2017 Alabama Senate race between Roy Moore (b. 1947) and Doug Jones (b. 1954), a small project funded through Democratic-aligned money had run a deceptive online experiment, including a scheme to make it appear that Russian bots backed Moore. New Knowledge’s chief executive, Jonathon Morgan, had participated. Public reporting centered on Morgan and others, and no clear public evidence shows DiResta directed the Alabama tactics. The association still cost her. A researcher who studied disinformation worked at a firm whose leadership had run a disinformation-style experiment. Her critics never let the detail go, and they did not need to prove more than proximity for it to work.
In June 2019 she joined the Stanford Internet Observatory as technical research manager, recruited by its founding director Alex Stamos (b. 1979), the former Facebook security chief. The Observatory studied abuse across information systems: state influence operations, election rumors, child exploitation, and later the effects of generative AI. Her team published research exposing covert Pentagon influence operations, a report that pushed the Department of Defense to reexamine its own propaganda practices, a fact her critics rarely mention.
In 2020 the Observatory joined the University of Washington, Graphika, and the Atlantic Council’s Digital Forensic Research Lab in the Election Integrity Partnership. During election week, students and analysts worked in shifts, logging viral claims about mail ballots, voting machines, and stolen votes into a ticketing system, the kind of workflow software a corporate help desk uses. A rumor about Sharpie pens invalidating ballots in Arizona would come in, an analyst would open a ticket, trace the spread, assess the claim, and in some cases flag it to a platform. Election officials, civil society groups, and platform trust-and-safety teams all touched the pipeline. To the researchers this was rapid-response scholarship, a public service in a year when the President of the United States was telling his supporters the election was rigged. To their later critics it was a censorship switchboard, a place where academics, government entities, and platforms sat in one reporting chain deciding which speech lived and which died. Both descriptions attach to the same ticketing queue. The fight that followed was over who gets to label a claim false, who gets notified, and what a platform does next. A companion effort, the Virality Project, applied the same model to COVID-19 vaccine rumors in 2021.
The reckoning arrived in December 2022, when Elon Musk gave internal Twitter documents to a handful of writers. Matt Taibbi (b. 1970), Michael Shellenberger (b. 1971), and others used the Twitter Files to argue that content moderation had fused with government pressure and elite preference, and they named DiResta as a central node in what they called the censorship-industrial complex. The undergraduate CIA internship, two decades old, became the load-bearing biographical fact. On podcasts and Substacks she became “CIA Renée,” a spy running a global censorship operation from a Stanford office. The House Select Subcommittee on the Weaponization of the Federal Government, chaired by Jim Jordan (b. 1964), subpoenaed Stanford’s documents in April 2023, enforced the subpoena in June, interviewed Stamos under oath, and included students, undergraduates among them, in its document demands. Stephen Miller’s (b. 1985) America First Legal sued DiResta, Stamos, and Kate Starbird (b. 1975) of the University of Washington in a case that a Louisiana federal court allowed into discovery in December 2024. Stanford spent millions on legal defense. DiResta answered her accusers in an Atlantic essay about becoming the main character of the fantasy-industrial complex. She had spent years studying how a rumor cascade selects a villain, strips away context, and hardens into a bespoke reality. Then she watched one do it to her. Her Substack biography compresses the experience into four words and a count of her children: Twitter Files bête noire, mom of three.
The constitutional question reached the Supreme Court as Murthy v. Missouri, a suit alleging that federal officials had coerced platforms into suppressing disfavored speech about elections and COVID-19. On June 26, 2024, the Court ruled six to three, in an opinion by Justice Amy Coney Barrett (b. 1972), that the plaintiffs lacked Article III standing. The ruling settled nothing underneath. It never decided when government contact with a platform becomes coercion. Her defenders read the decision as vindication. Her critics read it as a procedural escape from a First Amendment problem, and the private suits continued.
By then the Stanford Internet Observatory was finished as an election-research operation. Stamos had stepped back in November 2023. In June 2024 Stanford University declined to renew DiResta’s contract, other contracts lapsed, and remaining staff were told to look for jobs. The university disputed reports that it was dismantling the Observatory and said child-safety work, the trust-and-safety journal, and the annual conference would continue under a faculty sponsor. Jordan posted that free speech had won again. Shellenberger declared victory over a censorship operation. The Election Integrity Partnership announced it would not work on the 2024 election or any future one. Whatever Stanford called it, the outcome was the one the campaign’s architects wanted. Lawsuits, subpoenas, legal bills, and harassment had priced election-rumor research out of one of the richest universities on earth, and every other university watched it happen.
DiResta landed at Georgetown University. In October 2024 the McCourt School of Public Policy appointed her associate research professor, with positions in the Massive Data Institute and the Tech & Public Policy program. She became a contributing editor at Lawfare and kept writing for The Atlantic. That same year she received the Carl Sagan Prize for Science Popularization for translating propaganda research into public writing, and she published her synthesis, Invisible Rulers: The People Who Turn Lies into Reality, with PublicAffairs. The paperback arrives August 4, 2026.
The book argues that the modern information environment fuses two older systems, the propaganda machine and the rumor mill. Propaganda once moved downward through states, parties, and broadcasters. Rumor moved sideways through neighborhoods, churches, and workplaces. Social media collapses the two channels into one. A rumor becomes a meme, the meme becomes a movement, the movement becomes a news story, and the story becomes political reality. Her invisible rulers are no cabal. They are the interlocking forces of influencers, recommendation algorithms, and online crowds, operating where the old gatekeepers have lost authority. Small groups manufacture the appearance of consensus. Platforms reward outrage, certainty, novelty, and tribal belonging. Francis Fukuyama praised the book’s account of bespoke realities. Her opponents reviewed the author rather than the argument.
Her prescriptions frustrate both camps. She wants changed platform defaults, user control over algorithmic feeds, friction before virality, transparency about amplification, and civic education in propaganda literacy, rather than mass takedowns. Free-speech advocates see residual faith in expert moderation and institutional coordination. Anti-disinformation activists want harder intervention, faster. Her answer is that the information environment is already governed. Engagement algorithms and manipulation-for-hire govern it now. The choice is between opaque private rule and rules the public can see, contest, and revise.
At Georgetown her work has widened into a general theory. With Josh Goldstein of Georgetown’s Center for Security and Emerging Technology she published “Full-Spectrum Propaganda in the Social Media Era” in Security Studies, arguing that well-resourced states no longer choose between overt and covert operations. They run integrated campaigns across state television, diplomatic accounts, state news sites, covert persona networks, and influencers who may not know they serve one, with channels citing each other to build the appearance of independent confirmation. The framework sorts channels along two axes, overt against covert and broadcast against social, and treats the audience as a distribution channel the state could never build alone. She has applied the same logic to TikTok and ByteDance, arguing that the deeper risk lies past data collection in algorithmic control, since whoever owns recommendation can steer a society’s attention quietly over time. That claim remains her framework rather than a settled empirical verdict in each case, and she treats it as such. Her current research extends to AI-generated propaganda, scams, and privacy-preserving ways to verify humanness online without building a checkpointed internet.
The trajectory holds together better than it first appears. A woman trained on adversarial systems, at the Agency, on the trading floor, in venture capital, found in social platforms the largest adversarial system ever built and spent a decade mapping who exploits it. Her supporters call her the clearest analyst of propaganda in the platform age. Her critics call her the face of an expert class that decided its political judgments were science and used platform back channels to enforce them. Both descriptions draw on real material. She did help build reporting pipelines that connected researchers, officials, and platforms, and reasonable people can find that arrangement corrosive to free expression whatever its intent. She also produced some of the most rigorous public documentation of state manipulation campaigns in existence, and the movement that destroyed her research center relied on distortion, selective leaks, and harassment to do it. She studied how lies become social facts and then became one.
The problem she works on remains unsolved and might be insoluble. A self-governing people needs some shared account of reality, and the attention economy pays for fragmentation, paranoia, and spectacle. Every response so far has either done too little or created a new authority nobody trusts. DiResta’s career is the test case for whether a free society can defend the idea of shared fact without building a ministry of truth, and the returns to date suggest the question will outlive everyone now fighting over it.
Notes
Career history, CIA internships during the summers from 1999 through 2004, Jane Street, the LSAT detour, OATV, Haven, the concept of full-spectrum propaganda, the paperback publication date of August 4, 2026, and the Carl Sagan Prize all come from Renée DiResta’s own biography and the Niskanen Center interview, which also covers her Yonkers upbringing, five undergraduate years, clerk work, and the Disneyland origin story: Renée DiResta and Niskanen Center.
Vaccinate California, the finding by Gilad Lotan that 25 percent of anti-vaccine tweets originated from just 0.6 percent of accounts, her appearance in The Social Dilemma, and her 1981 birth year are documented at Wikipedia.
The August 1, 2018 Senate Intelligence Committee hearing, including the 9:32 a.m. start time, the SH-216 hearing room, Senator Richard Burr’s opening statement, and the witness list, is documented in the official transcript: U.S. Senate. DiResta’s statement describing online manipulation as one of the “defining threats of our generation,” along with the poster illustrating fake accounts, is covered by NBC News. Her testimony regarding anti-fracking and anti-GMO campaigns is summarized by CBS News.
The release of the Senate Intelligence Committee’s reports on December 17, 2018, together with Senator Burr’s statement, is documented here: U.S. Senate.
The dismantling of the Stanford Internet Observatory, including Alex Stamos’s departure in November 2023, the nonrenewal of DiResta’s contract, the timeline surrounding the House Judiciary Committee investigation led by Jim Jordan, and Stanford’s response, is covered by Platformer and NPR. Reactions from Jordan and Michael Shellenberger are reported by The Washington Times.
The December 2024 ruling allowing the America First Legal lawsuit to proceed is documented here: America First Legal.
Her appointment at Georgetown University’s McCourt School is documented at Georgetown University.
The paper “Full-Spectrum Propaganda in the Social Media Era,” coauthored with Josh Goldstein, together with Georgetown’s discussion of the project, appears here: CSET and Georgetown University.
The description “Twitter Files bête noire. Mom of 3.” comes from her Substack biography: Substack.
I added several pieces of self-evident background without separate citation. These include comparing the logic of a trading floor to the way rumors move markets before verified information arrives, likening the Election Integrity Partnership’s workflow to a help-desk ticketing system, using the Sharpiegate controversy as a representative EIP case because it is extensively documented in the final report The Long Fuse, describing senators of that generation as having grown up in the era of network television, and referring to the Pentagon report that, according to DiResta’s own biography, prompted the Department of Defense to reevaluate its approach. The details of Murthy v. Missouri, decided on June 26, 2024, by a 6-3 vote in an opinion by Justice Amy Coney Barrett on standing grounds, are matters of public record.
The Watchman of the Shared World: Renée DiResta’s Hero System
Renée DiResta’s life turns on two terrors, and each has a date.
The first arrived in December 2014. Her son had just turned one. A visitor carrying measles walked through Disneyland, and within weeks a disease America had declared eliminated moved through California. She pulled the state’s vaccination data and found preschools in the richest zip codes in the country with immunization rates below South Sudan’s. The terror was the oldest one there is, the body of a child who cannot yet defend against the world. Behind it stood a second-order version of the same terror, a society that had forgotten why the shots existed, a herd dissolving its own immunity because strangers on the internet told mothers a story.
The second terror arrived in December 2022. Writers with access to Twitter’s internal files named her a central node of a censorship operation, and within weeks a woman who studied rumor cascades for a living watched one select her. “CIA Renée” spread through podcasts and hearings and lawsuits. Strangers rewrote her biography while she held the original. This is the other death, the one Ernest Becker (1924-1974) says man fears as much as the grave. A person lives twice, once in a body and once in a name, and the name can be killed while the body walks around. DiResta has felt the cold of both deaths, the viral and the symbolic, and her hero system is built against the pair of them.
Becker holds that a man handles the knowledge of death by enlisting in a hero system, a cultural project that promises his life will count in something that outlasts him. The soldier has the nation, the monk has eternity, the founder has the company, the mother has the child. The system tells him what a hero is, and if he performs heroism by its lights, it pays him in the only currency that quiets the terror, the feeling of mattering permanently. DiResta’s heroism is watchfulness. The hero sees the machine that others cannot see, the troll farm behind the Facebook page, the 0.6 percent of accounts producing a quarter of the noise, the recommendation engine steering a bored mother from playground groups to Pizzagate. Having seen it, the hero warns the city. The project that outlasts her is the shared world, a public that can still agree on what happened, and her immortality is the immortality of the guard on the wall, invisible in the histories of peaceful years, present in every year that stayed peaceful.
The training reads like a preparation she never planned. Summers at the CIA as an undergraduate, seven years making markets in equity derivatives at Jane Street, where a trader learns that a price moves on a rumor hours before it moves on a fact, and the one who traces the rumor to its source eats the one who believes it. Venture capital, a hardware book, a logistics startup. Then the preschool spreadsheet, and the discovery that the skills of the tape reader worked on the feed. By 2018 she sat in room SH-216 of the Hart Senate Office Building telling senators, “This is one of the defining threats of our generation.” By 2020 she helped run a partnership that logged election rumors into a ticketing system and flagged some to the platforms. The watchman had a wall to stand on.
Every hero system runs on a subtraction story, the account of the self with the costs and appetites removed, and hers is the analyst’s. In the subtraction story she never sought power. She followed data, and the data kept leading uphill, from a parents’ group to the Senate to Stanford, and power kept arriving unrequested, the way a subpoena arrives. Her story subtracts that a ticket flagged to a platform is not observation, it is governance, a quiet participation in deciding which speech circulates. It subtracts the coalition, the foundations, universities, agencies, and one political party’s adjacent institutions that funded the wall and consecrated the watchman, and whose enemies noticed the pattern before her allies did. It subtracts the Agency summers, reduced on her website to a wry aside about what people on the internet love to discuss. And it subtracts the pleasure, because standing where the levers meet is a pleasure, and the analyst’s self-portrait allows curiosity as the only appetite. None of this makes her the villain of the caricature. It makes her a person, which the subtraction story is designed to prevent.
Her sacred values are reality, protection, and speech, and each word means what her hero system needs it to mean. Set the same words down in other systems and watch them change.
Take reality first. For DiResta, reality is a commons, like a water supply. It is the set of claims that survive method, the count certified, the vaccine trialed, the takedown documented, and it can be poisoned upstream by actors who understand the pipes. Defending it is public health. Monitoring is not an imposition on the commons, monitoring is how a commons stays potable. Now hand the word to a Soviet-born engineer in San Jose who left Kiev in 1979. For him reality is what remained after he subtracted the official version, and he performed that subtraction daily for thirty years as a civic discipline. He reads about an Election Integrity Partnership with a ticketing queue and feels the hair rise on his arms, because in his hero system the man who trusts the ministry dies stupid, and heroism is the samizdat instinct, the belief passed hand to hand beneath the notice of the certifiers. Her water department is his ministry. Hand the word next to a Hasidic diamond dealer on 47th Street. His reality was sealed at Sinai and transmitted through men whose names he can recite. The feed is noise from a world that was never going to include him in its consensus, and its collapse costs him nothing, because his hero system never banked at that branch. He trades stones worth millions on a handshake and the word mazel, which is to say he lives inside a high-trust reality of his own tribe’s manufacture, and it works. DiResta’s nightmare, the splintering of shared reality into bespoke realities, describes his people’s condition for three thousand years, except his tribe calls the bespoke reality a covenant and has buried its dead in it with honor.
Take protection. In her system the word points at the herd. Protection is the immunization rate, the pre-bunked rumor, the friction added before a lie goes viral, the child kept safe by the health of the whole. She came to the work as a mother, and the maternal charge under the analytic prose is what gives her writing its heat. Now give the word to a homeschooling Baptist mother in east Tennessee, and it points the other way with equal heat. Protection means the state’s needle stays out of her child’s arm and the school’s screen stays out of her child’s head, and she has read enough, in her own counter-canon with its own experts, to die on this. She is also guarding a child from death, and in her system from the second death, the eternal one, which the epidemiologists do not model. Each mother performs heroism at her own kitchen table, and each reads the other as the threat her heroism exists to stop. Give the word last to an emergency physician in Queens in April 2020, intubating patients whose families were still forwarding cures from WhatsApp. For him protection collapsed into triage, and misinformation stopped being a research topic the week it started arriving on gurneys. His system and DiResta’s are allies, but his runs on the body in front of him, and hers runs on the population curve, and the difference shows in what each will trade for control.
Take speech. DiResta’s formulation, echoed across her camp, is freedom of speech without freedom of reach, speech as an ecosystem to be gardened, the microphone distinguished from the mouth. Within her hero system this is a modest claim, since some editor always decides what amplifies, and she asks only that the deciding be visible and accountable. Hand the word to an old ACLU lawyer, the kind of Jew who defended the Nazis’ right to march through Skokie in 1977 and considered it the proudest wound of his career. For him speech is the individual’s shield against exactly the coalition DiResta assembled, the state, the university, the dominant press, and the platform, all agreeing on what counts as poison. He hears ecosystem and garden and smells the gardener’s boot. His heroism was defending the speech he hated, and a generation later the institutions that gave him his medals switched systems without holding a funeral. Now hand the word to a Salafi preacher in Cairo, and something stranger happens. He agrees with her. Speech must be governed, the feed corrupts, the young are led astray by influencers, and a righteous order curates what circulates. He commands right and forbids wrong, she moderates content and demotes harm, and the two systems, which share no god, no politics, and no century, converge on the premise that the information environment is too dangerous for laissez-faire. The disagreement is over which clerisy holds the pruning shears. The ACLU man notices the convergence and rests his case.
There is also the tribalist, and he deserves his full turn, since his is the oldest system on the field. The tribalist, nationalist, and traditionalist holds that men do not live in an information commons, they live in peoples, and every people that survives curates its story. The Passover Haggadah is curated. The Gettysburg Address is curated. Grandmothers are moderation systems. From inside this hero system, DiResta’s error is not that she governs speech, everyone governs speech, it is that she claims to govern from nowhere, in the name of a species-wide public that has never existed, with method standing in for a god. Her shared world, the tribalist says, is the tribal story of one tribe, the credentialed, the mobile, the institutionally employed, and the revolt against her was other tribes recognizing a rival priesthood and treating it as one. Yet the tribalist grants her more than her libertarian critics do. He honors watchmen. He agrees the feed is a weapon and that someone must stand on the wall, and he respects that she stood there under fire and paid. His correction is one sentence long. Know whose wall you stand on, and say the name of your people, because a watchman who claims to guard everyone is either lying or lost.
What lifts DiResta’s case above the usual run of these essays is that she is herself a professional student of hero systems and came within one step of Becker without citing him. Invisible Rulers describes bespoke realities, influencers who sell belonging, crowds that manufacture consensus, ordinary people who join online movements for identity and status and the feeling of fighting a great battle. This is Becker’s material wearing a lanyard. Becker wrote in The Denial of Death that culture is a shared illusion that makes the terror of mortality livable, and that men will kill to defend the illusion because the illusion is what stands between them and the abyss. DiResta documents the supply side of modern illusion with more empirical care than anyone alive, the troll farms, the recommendation engines, the engagement payouts. Where she stops is the demand side. Her account explains why a lie reaches a man. It does not explain why he grips it like a rope over a drop, why correction reads to him as attempted murder, why the anti-vaccine mother and the election-fraud believer defend their claims with a ferocity all out of scale with any policy stake. Becker explains it. The claims are load-bearing walls in immortality projects. The mother who believes the shot is poison has organized her heroism around protecting her child from it, and to accept the correction is to have been, for years, the danger in her own house. No fact-check offers her a way to survive that. Remove every troll farm on earth and the hunger for the saving lie remains, because the hunger comes out of the grave, and the platforms did not dig the grave, they only sold advertising on the way down. This is the ceiling on DiResta’s entire field, and she has spent a decade pressing against it with better and better instruments.
Her own ordeal proves the point on her body. In Escape from Evil Becker argues that groups purge accumulated death anxiety by loading it onto a victim whose destruction lets the group feel its world cleansed. The movement that made “CIA Renée” was not doing analysis, it was doing hero work. It had a cosmology, the regime of censors strangling the people’s voice, and a cosmology needs a devil with a face, and a woman who had interned at Langley, traded at Jane Street, worked at Stanford, and flagged tweets was a casting director’s gift. Killing her name paid her accusers in the same currency her watchman’s post paid her, the feeling of defending a world. She understands this in outline. Her Atlantic writing on becoming the main character of what she calls the fantasy-industrial complex is controlled, ironic, and wounded in the right places. What her published work has not yet said is that the machine that processed her runs on the same fuel as the machine she serves, that watchman and mob are both terror-management, and that her side’s certainty of guarding reality feels, from the inside, exactly like the other side’s certainty of exposing it.
How self-aware is she of the trade-offs? More than most subjects of this series. She concedes the central point her honest critics make, that the information environment will be governed by someone, and she argues in the open about who and how, which is candor of a kind the platforms never offered. She has admitted the wall cost her, the harassment, the subpoenas, the security consultations a mother of three should never need. The blind spot sits where Becker predicts it, at the foundation. She writes as if a baseline reality waits underneath the manipulation, recoverable once the pipes are cleaned, and the possibility she does not entertain in print is that the appetite for bespoke reality is constitutional, that her preschool parents in Palo Alto were not tricked into fearing the needle so much as they were shopping for a heroism, and that her own coalition supplies its members the same product in a different wrapper. A watchman can see every enemy outside the wall. The one thing the post does not let him see is the wall.
She earns empathy, and the empathy should be said plainly. She did the work. The reports on the Internet Research Agency and on the Pentagon’s own covert operations were real contributions to public knowledge, and the second one cut against her supposed masters, which her enemies never mention. She absorbed years of organized cruelty without becoming cruel in print. She kept writing under her own name while strangers rewrote it. Whatever one makes of the ticketing queue, the woman standing behind it was braver than the institutions that abandoned her, and Stanford’s lawyers should have to read that sentence twice.
Her hero is the sentinel, the one who stays awake over the sleeping city and accepts that the city will never know which nights the watching saved it, and whose reward is the sight the sleepers are spared. The rival her writing never names is the church, the original shared-reality machine, which governed what circulated for centuries and could do what no trust-and-safety team can, forgive the believer his saving lie while slowly trading it for a larger one, because it had something to offer the man on the rope besides a correction. And the cost her ledger cannot price is what a decade of studying belief as manipulation does to the student, because a woman who has traced ten thousand convictions back to their engines can no longer take a conviction whole, including her own, and the watchman’s final wage is a city she can see with perfect clarity and no longer live in as a citizen.
The Expert Without a License: Renée DiResta and Stephen Turner’s Problem
During election week in November 2020, an analyst on the Election Integrity Partnership rotation opened a ticket. The claim under review said Sharpie pens invalidated ballots in Maricopa County. The analyst traced the accounts spreading it, wrote an assessment, and the ticket moved through a queue that connected university researchers, election officials, civil society groups, and the trust-and-safety desks of the platforms. Downstream, a voter in Mesa saw a label under a post telling him the claim about his own ballot was disputed. He never learned who wrote the ticket. He had no way to weigh the analyst’s evidence, no vote over the analyst’s appointment, and no procedure for appeal that he could name. He experienced the analyst’s judgment the only way a citizen can experience judgment he cannot inspect, as power.
Stephen Turner (b. 1951) has spent a career on the question that loop poses. In “What Is the Problem with Experts?” and in Liberal Democracy 3.0: Civil Society in an Age of Experts and The Politics of Expertise, Turner argues that expertise creates a standing embarrassment for liberal theory. Liberal democracy rests on the premise that citizens can discuss public claims and judge them, and that officials answer to that judgment. Expert knowledge breaks the premise, since the citizen cannot assess the claims of the epidemiologist or the actuary on their merits. Democracies have lived with the break through settlements. Experts advise, officials decide, citizens judge the officials, and the expert’s cognitive authority gets validated over time in use, the way the lay passenger validates aeronautical engineering every time the plane lands. Turner’s warning concerns the cases where the settlement fails, where expert judgment enters the decision loop at a point the citizen cannot see, and where nothing in the citizen’s experience ever tests it. There the expert rules without a license, and the political system stores up a legitimacy debt that someone eventually collects.
Renée DiResta poses Turner’s problem, and her career supplies the case study his books lacked.
Begin with how the expertise came into existence, because Turner insists on the question most commentary skips. Cognitive authority has a supply side. Some expertise wins acceptance through results any layman can check. Some wins acceptance only within a sect that already believes. And some gets called into existence by subsidy, by foundations and agencies that want a class of knowers to exist, and that certify the knowers they fund. Misinformation studies belongs to the third kind. Before 2016 the field barely existed. After the election, money arrived at speed. Foundations wanted grantees, platforms wanted researchers to receive their data sets, the Senate wanted outside analysts, and universities wanted centers. The Stanford Internet Observatory launched in 2019 with platform cooperation and philanthropic funding, and DiResta became its technical research manager without a doctorate, a former derivatives trader whose credential was a body of work the new field certified because the new field had no older standard to apply. None of this says the work was bad. Much of it was careful. Turner’s point cuts elsewhere. The field’s authority was conferred by its patrons before it could be validated by its public, and a discipline whose peer reviewers were summoned by the same grants that summoned the authors reviews in a circle.
Watch the circle from inside, through her eyes, because from inside it looked like duty. A measles outbreak had shown her that coordinated minorities could capture the feed. The Senate had handed her platform data and asked what Russia did with it. She answered with the most detailed public accounting then available and told the committee the country faced a defining threat of the generation. When the 2020 election approached, election officials had no capacity to monitor viral rumors and platforms had no appetite to coordinate. Someone had to stand in the gap. The researchers stood in it. From inside, the ticketing queue was a public service performed by the only people equipped to perform it, and the talk with platforms was speech, citizens petitioning companies, protected like anyone’s.
Now watch from the other side, through the eyes of a House staffer in 2023 reading subpoenaed emails in a windowless room, because Turner requires this view too. The staffer sees a federal agency that cannot censor speech under the First Amendment. He sees that agency in contact with a university consortium. He sees the consortium flagging posts to platforms, and the platforms acting on some flags. He does not need a conspiracy for the pattern to alarm him. He needs only the observation Turner supplies, that expert judgment had been wired into an enforcement circuit at a point no voter could reach, and that the wiring let each node disclaim the power the circuit as a whole exercised. The agency only shared concerns. The researchers only shared findings. The platform only enforced its own policies. Authority without an author. The staffer’s boss, Jim Jordan, put the conclusion on a poster. DiResta’s camp answered that the committee misread routine research correspondence, leaked fragments, and defamed scholars. Both descriptions fit the record, and that both fit is the finding. The settlement between knowledge and power had never defined where advice ends and rule begins, so each side could describe the same emails in good faith and reach opposite verdicts.
Turner’s typology sharpens the diagnosis. The misinformation expert differs from the aviation engineer in the audience for his claims. The engineer’s audience includes the public, which validates him in use across millions of uneventful flights. The misinformation expert’s audience was never the public. It was platforms and agencies, bureaucracies that acted on his findings, and the public met the findings only as outcomes, a label, a demotion, a vanished account. Turner names this configuration as the most corrosive one available, expertise exercised on the public through intermediaries rather than accepted by the public through experience. The citizen in Mesa cannot check the analyst. He can only obey the label or resent it. Multiply him by fifty million and the resentment becomes a constituency.
The revolt, when it came, followed Turner’s script so closely one could teach the script from the clippings. He argues that expertise which outruns its license does not get refuted, since the public lacks the means to refute it. It gets revoked. The revocation arrives as politics, crude, opportunistic, indifferent to the merits of particular studies, because revocation is the one instrument a democratic public retains over knowledge claims it cannot assess. The Twitter Files, the Weaponization subcommittee, the America First Legal suits, and the Murthy litigation made a single motion in different registers. They did not engage the field’s findings about the Internet Research Agency or rumor cascades. They attacked the field’s standing to sit in the loop. Matt Taibbi and Michael Shellenberger built the case in installments. Jordan enforced subpoenas and read student names into the record. Stephen Miller’s lawyers sued. The methods ranged from journalism to harassment, and the merits ranged from real questions to fantasy, and Turner’s frame holds through the whole range, because a legitimacy crisis does not select its collectors for fairness. The debt gets collected by whoever shows up.
One episode deserves its own paragraph, because it was the exception that proved the license problem. In October 2020, when the New York Post published the Hunter Biden laptop story, platforms suppressed the link, and figures across the expert class treated the story as probable foreign disinformation. Here, for once, the public got what Turner says the field otherwise never provides, a claim it could validate in use. The laptop was real. The story checked out. Every citizen could run the test himself, and millions did, and the field failed the one lay-checkable test it ever faced in public. An engineer survives a thousand landings and earns deference. A field survives on deference and loses its one landing. No committee hearing damaged the enterprise as much as that single verifiable miss, because it converted the skeptic’s suspicion from theory into experience.
The Supreme Court had the chance to write the missing settlement and declined. In Murthy v. Missouri, decided June 26, 2024, the majority found the plaintiffs lacked standing and left unaddressed the question underneath, when government contact with a platform becomes coercion. Turner might have predicted the abstention. Liberal constitutional doctrine has categories for the state and for the speaker, and thin ones for the commissioned intermediary, the expert consortium that is neither state nor citizen and carries messages between them. The Court looked at the circuit and could not find the node where the constitution attaches. So the circuit remains unadjudicated, the researchers remain exposed, and the next administration of either party inherits the ambiguity intact.
Stanford, facing millions in legal costs and no constitutional cover, ran the calculation a patron runs. By June 2024, Alex Stamos had stepped back, DiResta’s contract lapsed, staff were told to look for work, and the university insisted nothing was being shut down while the election work stopped. Turner’s supply-side analysis explains the collapse without any reference to who was right. Subsidized expertise lives at the pleasure of the subsidy. A field created by patrons in 2017 could be uncreated by patrons in 2024, and the personnel could do nothing about it, because the field had never acquired the independent base that validated disciplines hold, a public that misses them when they go. Aviation engineering cannot be dissolved by a nervous provost. Election-rumor research could be, and was.
DiResta’s response to the collapse is where her case turns instructive rather than merely illustrative, because her remedies read as an attempt to renegotiate the license on Turner’s terms, whether or not she has read him. The program in Invisible Rulers retreats from the configuration that destroyed her. Takedowns and flagging pipelines put expert judgment at an invisible point in the loop. Her proposals move judgment to points the citizen can inspect. Changed platform defaults are visible rules. User control over algorithmic feeds hands the lever to the layman. Friction before virality operates on all claims alike, without an analyst adjudicating each one. Transparency requirements let outsiders audit what the deference once concealed. The phrase she coined years earlier, freedom of speech, not freedom of reach, already contained the retreat, since it concedes the speech and fights only over amplification design. Read through Turner, the program amounts to converting Type-of-expertise, from judgment exercised on the public through bureaucracies toward engineering the public can validate in use, from the priest toward the civil engineer. It is the most serious attempt at relicensing the field has produced, and it came from the person the revocation hit hardest, which is no accident, since she alone among her peers has felt the full price of ruling without a license.
The relicensing remains incomplete at the point Turner would watch. Media literacy sits in her program alongside the design remedies, and literacy curricula are expertise teaching citizens whom to trust, a catechism for correct deference, drafted by the same class whose deference collapsed. And the design remedies still require someone to set the defaults and define the friction, which reopens the appointment question one level up. Turner’s deepest claim is that the problem admits no solution, only better and worse settlements, and that the standard fixes reproduce the problem at a remove, transparency reports written by experts, oversight boards staffed by experts, fact-checkers ranked by other fact-checkers. Her program improves the settlement. It does not escape the regress, and nothing can, which is Turner’s conclusion and should be the field’s.
What her case adds to the Turner literature is a limiting case the typology implies without developing. Every prior expertise claimed a competence the citizen lacks in some domain, bridges, drugs, monetary policy. Misinformation expertise claims competence over the citizen’s own weighing of testimony. Its founding premise holds that laymen cannot reliably judge which claims to believe, and must have the judging environment managed for them. But the layman’s capacity to weigh testimony is the one competence liberal democracy cannot delegate, since it grounds the vote, the jury, and the discussion Turner’s liberals stake everything on. The aviation engineer says, you cannot build the plane, and the citizen agrees and boards. The misinformation expert says, you cannot be trusted to decide what to believe, and the citizen who agrees has conceded the premise of his own self-government. This is why the field drew a revolt no economist ever drew, and why the revolt reached for constitutional language rather than technical rebuttal. The expertise did not sit inside the liberal settlement awkwardly, the way nuclear physics does. It contradicted the settlement’s first axiom, and the public heard the contradiction before the field did.
DiResta now works at a policy school, writes at Lawfare, publishes on full-spectrum state propaganda, and argues in public about defaults and friction and audit. She has become an expert on the terms of her own license, which no one in her field was in 2019, and the education cost her the field, the post, and for a season the name. The citizen in Mesa still cannot check her work. But her current program asks less of his deference and more of his inspection than anything her field produced in its funded years, and Turner’s framework suggests that this, rather than any hearing or ruling, is what the beginning of a legitimate settlement looks like, experts bidding for a license from the public instead of billing the public for one it never signed.
DiResta: ‘Corrections Are Censorship: Jacob Siegel’s Latest Fiction’
Renee DiResta writes on her Substack:
Jacob Siegel misleads readers in his new book. When I asked for corrections, he cried censorship in The Free Press—writing as accuser, investigator, and fact-checker all in one. A lie-machine exposé.
Jacob Siegel published a book last week called The Information State. It is a sweeping grand theory of how a deep state censorship machine was assembled to control American public opinion. It’s really just an expansion of his Tablet essay on the “censorship industrial complex” from 2023, in which he declared Frame Game Mike Benz a State Department “whistleblower” and echoed the allegations of the Twitter Files; if you read that, you’ve read the book. This is actually the problem: tons of evidence has come out since, in the courts and in Congress, and Siegel chose to ignore it and just repeat the same allegations. It’s since come out that Benz worked at State Department for two months, but no matter…same sources as scaffolding, etc.
In the book, Siegel insinuates the usual bullshit: I ran a wing of this censorship complex out of the Stanford Internet Observatory, I may be a secret CIA agent, etc. There’s a lot of guilt-by-association and straight-up errors. He describes a project I worked on as possibly the largest censorship initiative in existence.
Siegel’s book relies on innuendo, and then he pretends to be shocked when readers fall for the innuendo; that game is what we’re going to break down here. His reviewers keep doing what he wants them to do: reading between the lines and coming to (false) conclusions. The problem is that his fictional character — me — has leapt from the page, and dared to request factchecks over this, from three publications. After they’ve looked over the evidence, pubs have issued corrections in line with their editorial standards. The Brownstone Institute did nothing. The Free Beacon Issued a correction. The Baffler pulled their review.
On the merits, DiResta wins the narrow dispute. The 22 million figure comes from a post-election research dataset, and the actual flagging numbers, roughly 4,800 URLs with about 65 percent receiving no action, are in the public record: reports, amicus briefs, congressional testimony. Even the Jordan subcommittee, which had every ticket under subpoena and every incentive to inflate, never claimed 22 million flags. When your most hostile investigator with full document access declines to endorse a number, that number is dead. Her strongest structural point is the closing question: if the mass flagging did not happen, what was the mass censorship? Siegel’s book needs the big number because without it EIP shrinks to an academic tagging project that platforms mostly ignored.
Her account of Siegel’s method is also accurate as a description of how innuendo works in this genre. He never writes “EIP flagged 22 million tweets for censorship.” He builds a character sketch, describes round-the-clock monitoring and takedown requests, then drops the number in that context and lets reviewers complete the inference. When they do, he bears no responsibility for the false sentence because he never wrote it. This is a real technique and she names it well. Her Free Press complaint also lands: an accusatory email from the accuser is not a fact-check, and “she did not deny it” after stripping her reply down to one sentence is a construction any editor should catch.
Now the weaknesses. “Crazy bitch shit” and “lose his mind” cost her. She predicts in the piece that she will be cast as an unhinged woman, then hands Siegel the quote to do it with. When you know the trap, walking into it reads as either lack of discipline or a calculation that her audience rewards the register. Probably both. Substack pays for combat, and her subtitle (“lie-machine exposé”) shows she knows it.
The maternity leave passage is her weakest ground, and her April 16 update concedes as much. She says she did not lead EIP, then says she absolutely was a leader of it and proud of it. Her distinction, singular founder-villain versus one leader among several, is defensible, but she needed two paragraphs and an update to make it, and Siegel needed one screenshot. In a fight about precision, that is an unforced error. She calls his move a motte-and-bailey; a reader could say her original sentence was the bailey and the update the motte.
There is also a category question she skates past. She frames corrections requests to publications as pure counterspeech, and mostly that is right. But when the person requesting the correction spent years as the country’s most prominent scholar of platform moderation, and one outlet responds by pulling a review entirely rather than fixing it, the optics do some of Siegel’s work for him even though she asked only for a correction and the screenshot backs her. The Baffler’s overreaction is not her fault, but “the review vanished after she emailed” is the kind of fact pattern that feeds his narrative machine regardless of intent, and she knows how those machines run better than anyone.
The larger structure is the one you have seen before: both parties now occupy positions where the fight is the product. Siegel sells a book whose thesis requires a censor, so a correction request is the best marketing he could ask for. DiResta writes a Substack called Calling Bullshit on the Bullshit Industrial Complex, so his attack is her content. Each accusation funds the other side’s next post. That does not make them equally right. On the checkable facts she is right and he is not. But it explains why the dispute escalates rather than resolves, and why neither party has an incentive to let it die.
The Great Delusion
If John J. Mearsheimer’s anthropology is right, it provides a structural validation of DiResta’s empirical observations while reframing the ultimate stakes of her work.
DiResta’s research documents how ordinary digital tools can be used to launder conspiracy theories, elevate niche beliefs into mainstream opinions, and distort consensus. In a traditional liberal framework, the solution to this problem relies heavily on education, fact-checking, and the critical faculties of individual citizens who, when presented with accurate information, will logically reject falsehoods. Mearsheimer’s anthropology suggests that this individualist, reason-based defense is a fundamental misunderstanding of the human animal. If reason is the least important of the ways we determine our preferences, and if individuals develop deep attachments to their group long before critical faculties form, then no amount of fact-checking will dissolve a tribal narrative. When an online crowd rallies around a specific rumor or bespoke reality, they are not engaging in a detached intellectual exercise. They are engaging in tribal signaling and consolidation. The facts are irrelevant because the group attachment dictates the moral code, rendering individual critical reasoning secondary.
DiResta often writes about the incentives that drive online systems, noting that when attention, money, status, or political power reward manipulation, we see more of it. She tracks how platforms allow coordinated networks to mimic real communities, creating artificial amplification that feels like grassroots consensus. If Mearsheimer is right that humans are tribal at their core and designed to survive by being embedded in a society, then digital platforms do not create polarization out of thin air; they act as a supercharged accelerator for man’s innate tribal instincts. Human beings seek out social groups to escape the atomistic void of individualism. Social media algorithms simply automate and optimize this search, herding individuals into hyper-socialized digital tribes with unprecedented speed. The “invisible rulers” DiResta describes—the influencers who game algorithms to turn lies into reality—are simply exploiting the deep human need for group belonging and collective identity.
DiResta’s work often looks for structural interventions, such as design changes to platforms, decentralized social media, protocol-based governance, or “bridging-based” recommendation systems that promote content appealing across ideological lines.
However, if Mearsheimer’s view is correct, the splintering of reality into fragmented, adversarial factions is not a technical glitch or a malfunction of the information ecosystem that can be engineered away. It is the natural, inevitable expression of human nature when freed from the artificial constraints of centralized, institutional authorities. For decades, elite institutions and a centralized media ecosystem enforced a superficial, liberal consensus that masked man’s tribal core. By decentralizing communication, the internet did not corrupt a rational public; it merely stripped away the institutional filters, allowing human beings to revert to their primary, tribal state. If Mearsheimer’s anthropology holds, the bespoke realities DiResta studies are permanent fixtures of the human landscape, because the drive to protect and defend the tribe will always supersede the pursuit of an objective, universal truth.
‘A Big Misunderstanding’
To David Pinsof, DiResta represents the misunderstanding myth. Her entire career rests on the premise that societal fractures like political polarization and vaccine skepticism are malfunctions of our informational infrastructure. She views the public as gullible targets infected by a digital virus. If tech platforms simply deployed the right interventions and experts raised public consciousness, the country might return to a shared reality.
Pinsof offers a colder alternative. The citizens sharing memes, resisting institutional mandates, and participating in online tribalism do not suffer from a technical bug or a cognitive blind spot. They understand their incentives perfectly.
From this perspective, the online ecosystem is not a broken information utility; it is a highly efficient arena for high-stakes, zero-sum competition over the coercive apparatus of the state. Partisans do not share hyper-partisan propaganda because they are too stupid to spot fake news. They share it because confirmation bias helps them win arguments, demonize political enemies, and forge alliances with their peers. Stupidity online is strategic.
DiResta frames her work as a neutral effort to protect democratic consensus and restore institutional trust. Pinsof invites us to look past these stated motives and consider the actual motives. Defining misinformation and deciding which narratives require intervention is an instrument of immense social power. By setting the boundaries of acceptable discourse, elite institutional researchers create a framework that they happen to be uniquely qualified to police. It transforms local political preferences into objective standards of truth, allowing the credentialed class to dismiss their rivals not as competitors with conflicting material interests, but as irrational actors who need a nudge from their bethers.
The conflict between online factions does not stem from bad beliefs that better algorithmic engineering can fix. It stems from deeply conflicting motives over status, power, and resources. No amount of fact-checking or media literacy can bridge that divide. The only misunderstanding in disinformation research is the belief that political warfare is just a big misunderstanding.
The Vigilant Public: Renée DiResta Through Hugo Mercier’s Not Born Yesterday
In January 2020, Princeton University Press published Not Born Yesterday: The Science of Who We Trust and What We Believe by the French cognitive scientist Hugo Mercier. The book arrived weeks before a pandemic that would make “misinformation” the most funded word in American media criticism, and it argued that the entire panic rested on a false premise. Humans are not gullible. We evolved a suite of cognitive tools that Mercier calls open vigilance, and these tools make us too hard to influence, not too easy. The masses did not need protecting from lies. The people who believed they did were repeating an error as old as Plato.
Renée DiResta built her second career on the opposite premise. A former Jane Street trader with a Stony Brook degree in computer science and political science, she moved from Wall Street to venture capital to the study of anti-vaccine networks on Twitter, then to a Senate-commissioned report on Russian influence operations, then to the research manager’s desk at the Stanford Internet Observatory, where she helped run the Election Integrity Partnership in 2020 and the Virality Project in 2021. Congressional subpoenas, lawsuits, and online harassment followed. Stanford dismantled the Observatory in 2024. She landed at Georgetown’s McCourt School of Public Policy and published Invisible Rulers, a book whose title concedes nothing to Mercier. Lies, in her account, get turned into reality by influencers, algorithms, and crowds. In his account, that hardly ever happens, and when it appears to happen, the causation runs the other way.
Reading DiResta through Mercier is not a matter of scoring one against the other. It is a stress test. Mercier’s book supplies a body of evidence about how persuasion works, and DiResta’s career supplies a decade of interventions premised on a theory of how persuasion works. Where the evidence and the interventions meet, sparks come off.
Start with Mercier’s core claim, because everything else follows from it. Communication between organisms with divergent interests survives only if receivers benefit from listening. A gullible receiver gets exploited until he stops listening. Evolution therefore built humans to check every message against prior belief, to weigh the speaker’s competence and incentives, to demand arguments, and to track who has been right before. These checks run in infants. Twelve-month-olds resist testimony that contradicts what they have seen. Three-year-olds trust reporters over guessers. The checks never turn off. They get sharper with experience.
From this Mercier derives a prediction that most educated people find hard to swallow: mass persuasion should almost always fail. He then shows that it does. The Nazi propaganda apparatus, the most notorious persuasion machine ever built, moved anti-Semitic sentiment only in the districts that were already anti-Semitic before 1933, and in districts with low prior anti-Semitism, radio propaganda backfired. Ian Kershaw (b. 1943), the historian Mercier leans on, concludes that Nazi propaganda succeeded only where it could “build on existing consensus.” Soviet propaganda fared no better. Chinese citizens who consumed more state media trusted the government less. The Chinese Communist Party eventually gave up on persuasion and shifted to friction and flooding, making inconvenient information hard to find and burying the rest under celebrity gossip.
Democracies show the same pattern with better data. The 2018 meta-analysis by political scientists Joshua Kalla and David Broockman found that the net persuasive effect of campaign contact in American general elections, all the flyers, calls, canvassers, and ads, is zero. Cambridge Analytica, the firm the Guardian credited with hijacking democracy, Mercier calls a scam, and the Republican operatives who watched it work agreed, remembering “pop psychology B.S.” and no evidence of results. The wild swings in campaign polls turn out to be largely artifacts of who answers the phone.
Now set DiResta’s threat model beside this. From her earliest work she framed the problem as one of amplification reaching vulnerable minds. Foreign governments pilot memes to see what sways opinion. Extremist groups exploit an asymmetry of passion to shape the reality of viewers. The 2018 New Knowledge report she led for the Senate Intelligence Committee treated the Internet Research Agency’s Facebook and Instagram output as a serious assault on the American mind. The Election Integrity Partnership flagged election rumors to platforms in 2020 on the theory that viral falsehoods, left standing, would corrode democratic behavior. The Virality Project extended the model to vaccines in 2021 and, in its most criticized move, advised platforms that even true stories of vaccine side effects could fuel hesitancy and deserved attention.
Each intervention assumes that exposure drives belief and belief drives action. Mercier’s chapter on fake news attacks that causal chain at both links, and he names DiResta’s founding events while doing it. The Washington Post headline “Fake News Might Have Won Donald Trump the 2016 Election” and the Independent’s Brexit equivalent appear in his text as specimens of the very misconception his book exists to correct. His counter-evidence is specific. During the 2016 campaign, fewer than one in ten Facebook users shared any fake news, and 0.1 percent of Twitter users accounted for 80 percent of the fake news on that platform. The people visiting fake news sites were not persuadable moderates but the ten percent of Americans with the most conservative information diets, intense partisans scouting for material to justify a vote they had already decided on. When Brendan Nyhan (b. 1978) and colleagues corrected Trump’s false statements to his supporters, most supporters accepted the corrections and none wavered. When political scientists Jin Woo Kim and Eunji Kim tracked the Obama-is-a-Muslim rumor across survey waves, they found the rumor changed stated beliefs among people who already disliked Obama and changed no votes at all. Disliking Obama made people accept the rumor. The rumor made no one dislike Obama.
Mercier generalizes the point through bloodletting. Galen (129-c. 216) wrote the theory that justified opening European veins for seven centuries, yet a quarter of the world’s cultures bled their sick without ever hearing of humors. The theory did not produce the practice. The practice, and the need to defend it against competitors, produced the theory. Blood libels follow the same grammar. The Kishinev pogrom of 1903 was not caused by the ritual-murder rumor, because the same rumor circulated every Easter without a pogrom, and the violence, when it came, bore no relation to the accusation. Who avenges a murdered child by looting liquor stores? Scholars of ethnic riots concur, in the line Mercier quotes, that crowds seek justifications for a course of action already under way. Wanting to commit the atrocity comes first. Believing the absurdity comes second, as cover.
If Mercier is right, DiResta’s decade of supply-side intervention targeted the wrong side of the market. Flagging tickets, labeling posts, and pressuring platforms all attack supply, and the demand for justification routes around suppression the way Chinese conspiracy theories route around the most heavily policed information environment on earth. Mercier says so in as many words: attempts to shut off the channels through which conspiracy theories spread cannot eradicate them.
The point of the book that cuts deepest, though, concerns stakes. Mercier draws a line between beliefs that touch our vital interests and beliefs that float free of them, and shows that vigilance tracks the line. Workplace rumors, where being wrong costs money and standing, run 80 to 100 percent accurate; employees at one downsizing firm knew the layoff list a week before the announcement. Wall Street takeover rumors are right nearly half the time and markets price them sensibly. Global rumors about presidents, celebrities, and popes are junk, and people hold them the way they hold beliefs about the shape of the earth, reflectively, nominally, at no cost. The Pizzagate believers left one-star Yelp reviews. The 9/11 truthers who thought the CIA could demolish towers never feared it could silence a blogger. People are not gullible about their vital interests. They are careless about beliefs that cost nothing, and they are careless because it is rational to be. The misinformation field thus concentrated its fire on exactly the class of beliefs that Mercier’s evidence shows to be inert, while the beliefs that steer actual behavior, the local, the occupational, the material, were never in danger.
There is a reflexive turn here, and Mercier does not spare the people on his own side of the education gradient. The belief that the masses are gullible is itself, he argues, a culturally successful misconception, and it spreads by the same demand-side logic as the rumors it condemns. From Plato through the Enlightenment, elites who benefited from the status quo cited popular gullibility to argue against democracy. Elites who sympathized with the people, like Rousseau (1712-1778), cited gullibility to explain why the people had not yet revolted, or why they voted wrong. The masses “are never corrupted, though often deceived.” In 2016 a class of American professionals suffered two verdicts they experienced as inexplicable, Trump and Brexit, and fake news supplied the explanation that spared them a harder accounting. The misinformation field was, on Mercier’s model, a justification market. It sold the losing coalition a story in which the voters had not rejected them, the voters had been hacked.
DiResta was the market’s most talented supplier, and here the frame turns on her with some force, because her own audience believed her through open vigilance, not despite it. Editors, senators, foundation officers, and platform trust-and-safety teams checked her claims against their priors, found the fit excellent, noted her credentials and her lack of obvious commercial motive, and bought. The Senate report’s reception, the standing ovations for the “hacked democracy” narrative, the funding that flowed to Stanford, all of it followed the demand curve. When the political scientist Gregory Eady and colleagues published their analysis of the Internet Research Agency’s Twitter campaign in Nature Communications in 2023, they found exposure concentrated in one percent of users, nearly all strong partisans, and no measurable effect on attitudes, polarization, or voting. Mercier’s framework had predicted that result in advance. It also predicts why the finding changed few minds among those invested in the threat.
Mercier adds a sly coda that reframes what the Russian operation accomplished. Trajan’s column spirals its victory reliefs a hundred feet up where no Roman could read them, because the message was never the reliefs. The message was that the regime could build the column. Putin’s hockey team wins by cheating in front of everyone, and that is the point. Propaganda often works as a display of power rather than a vehicle of content. By this light, the Internet Research Agency’s product was not persuaded Americans, of whom there were roughly none. Its product was the American belief that Russia had reached into the national mind, a belief that made Russia look ten feet tall at a cost of some millions of dollars and a St. Petersburg office building. The misinformation field did not counter the operation. The field was the operation’s delivery system. Every Senate hearing, every report, every book about hacked democracy carved the column higher.
What does the frame concede to DiResta? A fair amount, and her later work moves toward the concessions. Mercier grants that beliefs cheap for their holders can be catastrophic for their targets. The Kishinev rumor cost its believers nothing and cost Jews their lives. Mao Zedong (1893-1976) believed Lysenkoist agronomy at no personal risk while forty million peasants starved. Where rumor functions as coordination for violence or harassment, the harm is real even though persuasion never occurred, and DiResta’s attention to brigading, mob targeting, and rumor-to-violence pipelines survives the Mercier test intact, though it survives as a public-order concern rather than an epistemic one. The mission changes. You are no longer protecting minds from false belief. You are tracking crowds that have already chosen their target and are shopping for a pretext.
Invisible Rulers also shows DiResta half-converting to the demand side. She now writes of participatory propaganda, of audiences as distributors, of bespoke realities that people assemble for themselves, and she tells interviewers that the misinformation crisis is really a crisis of trust. Mercier could sign most of those sentences. Her lineage runs through Edward Bernays (1891-1995) and Jacques Ellul, theorists of propaganda as environment rather than injection, which sits closer to Mercier than the hypodermic-needle model her early institutional work operationalized. The remaining disagreement is about remedy, and it is not small. DiResta wants better design: algorithms that reward accuracy over engagement, changed defaults, institutional capacity to watch the adversary. Mercier’s closing chapter locates the remedy elsewhere. Institutions earn belief by being trustworthy, and the anti-vaxxer’s failure is a failure of openness with real grounds behind it, pharmaceutical companies that bury failed trials and buy doctors. Clean up the conduct and the trust follows. Conspiracy theories recede in Norway and flourish in Pakistan because Norwegian institutions give citizens less to work with.
On Mercier’s account, the Virality Project’s decision to treat true stories of vaccine side effects as a moderation concern was therefore worse than a tactical error. It handed a vigilant public documented grounds for the suspicion that health authorities would shade the truth for their outcomes, and open vigilance never forgets a demonstrated incentive. The intervention meant to protect the chain of trust corroded a link, and links, Mercier writes on his last page, are the game. Science spread its counterintuitive claims through society on the strength of chains of trust and argument, fragile, centuries in the making, and every institution that spends credibility to manage a news cycle is drawing down the only account that matters.
DiResta’s fall reads through the same lens as her rise, which is the frame’s final courtesy to her, since it declines to treat her enemies as any smarter than her friends. As a mother organizing for California’s SB 277 in 2015, she had aligned interests and local stakes, and her audience’s vigilance vouched for her. As the research manager of a Stanford lab flagging citizen speech to platforms while in contact with government agencies, she presented a different incentive profile, and half the country read it and priced her accordingly. The rumors that then engulfed her, that her long-disclosed undergraduate CIA internship made her a spook, that she ran a censorship regime, spread among people who wanted a justification for a fight they had already joined, exaggerated by the same demand-side logic that inflated Russian bots for her own coalition. Jim Jordan’s (b. 1964) subpoenas and Matt Taibbi’s (b. 1970) “censorship industrial complex” persuaded no one who was not already enlisted. When the Supreme Court disposed of Murthy v. Missouri on standing in June 2024, the vindication changed nothing, because the beliefs about her had followed the coalitions, not the evidence. She became the subject of a titillating rumor, held cheaply by millions, at devastating cost to its target. No one in the story was gullible. That is the terror of Mercier’s book. The machinery worked as designed, on everyone, from the start.
What survives of her campaign, weighed on Mercier’s scales, is the part that was never about belief: the tracking of coordinated harassment, the rumor-to-violence pipelines, the forensics of fake account networks as counterintelligence rather than mind-protection. What does not survive is the founding premise, that a vulnerable public required a professional class to filter its information diet. The public was not born yesterday. It read the filterers the way it reads everyone, by asking who they were, what they wanted, and who paid them, and it filed the answers where it files everything else.
The Alliance Map of Renée DiResta
Take her Atlantic corpus as a dataset. Between March 2020 and November 2025, Renée DiResta published pieces on Russian interference, Chinese conspiracy diplomacy, anti-vaccine influencers framed by a Tucker Carlson image, QAnon, the right’s disinformation apparatus preparing for Trump’s loss, right-wing social media divorcing from reality, the one-sided misinformation of 2020, Elon Musk fighting for attention, the Twitter Files as missed opportunity, Arizona’s voting-machine rumors, rumors on X becoming the right’s reality, Musk’s soap operas for conspiracy buffs, and the right-wing attack on Wikipedia. Two pieces advise public-health officials on communicating better. One praises Wikipedia as a model for the CDC. Now ask what predicts this list. A principle predicts poorly. A principled anti-falsehood beat covering 2020 through 2025 would have produced entries on the lab-leak dismissal, the laptop suppression, the collapse of the Steele dossier, and the years of official assurances about an aging president. None appear. An alliance map predicts the list almost line by line.
David Pinsof, David O. Sears, and Martie G. Haselton supply the map-reading method in “Strange Bedfellows: The Alliance Theory of Political Belief Systems” (Psychological Inquiry, 2023). Their claim runs against the whole self-understanding of political actors. Belief systems do not derive from values. They derive from alliance structures, historically contingent coalitions of groups thrown together by similarity, transitivity, and shared rivals, and the beliefs are ad hoc justifications generated to mobilize support for allies in conflicts. The generating tools are what the authors call propagandistic biases. Perpetrator biases minimize an ally’s transgressions, stress mitigating circumstances, and embellish good intentions. Victim biases maximize an ally’s grievances and attribute the rival’s motives to malevolence. Attributional biases credit an ally’s advantages to talent and effort and its setbacks to external attack. The authors stress two things the casual reader misses. The biases run symmetrically across all coalitions, and the partisan applying them is sincere, because inside a coalition, motivated reasoning functions as an honest signal of loyalty. Distrust your allies’ side of the story and they stop treating you as an ally.
Read DiResta through the map and start with her coalition, because Alliance Theory forbids starting with her ideas. Pinsof and his co-authors trace the late twentieth-century split of the American upper class into rival elites, intellectual and business, knowledge workers on one side and corporate wealth on the other. DiResta belongs to the intellectual-elite alliance: research universities, foundations, legacy media, public health, election administration, and the platform trust-and-safety departments those institutions staffed and trained. Her rivals follow from the alliance structure rather than from any philosophy: the populist right coalition, foreign state rivals, anti-vaccine networks that defied public health, and, from April 2022 forward, one man.
Musk gives the theory a test. Through 2021 he ran the most subsidized electric-car company on earth, and the knowledge class treated him as an eccentric ally. Her Atlantic file contains nothing on him. He bids for Twitter in April 2022, the terrain her alliance had spent six years learning to govern, and her first Musk piece appears that month. Three more follow across three years, tracking his migration into the rival coalition, from fighting for attention to soap operas for conspiracy buffs. Nothing in Musk’s epistemic conduct changed categories in April 2022. His alliance did. Pinsof’s account of interdependence and rivalry detection predicts the timing of her attention better than any account of her stated values, because the stated values existed for a decade while Musk drew no fire.
The same logic explains her final entry in the dataset. In November 2025 she defends Wikipedia against a right-wing campaign. Wikipedia is the knowledge class’s commons, written by its members, governed by its norms, feeding the AI models on which its next institutional bets ride, and she had already held it up as a model for the CDC in 2021. An attack on Wikipedia is an attack on alliance infrastructure. The defense follows as a corollary of the map. So does the earlier praise: within her coalition, Wikipedia counts as authority repaired, while a crowdsourced encyclopedia governed by the rival coalition’s editors might have appeared in her corpus under another name.
Now run the vocabulary through the theory, because the vocabulary is where her case extends Pinsof rather than illustrating him. The paper argues that partisans frame conflicts as morality to recruit third parties, creating common knowledge that one side is moral so that neutral observers can join at low cost. Each coalition derogates in the idiom of its own capital. Business elites call their rivals lazy and parasitic, an economic idiom, since wealth is the asset they hold. Religious coalitions call their rivals sinful. The intellectual-elite alliance holds epistemic capital, degrees, journals, data access, and its derogation idiom is epistemic: the rival coalition is misinformed, the rival’s claims are disinformation, the rival’s media diet is a pathology requiring intervention. The idiom performs the recruiting function on the referees who count for this alliance, platforms, advertisers, agencies, and courts, none of whom can be recruited with sin but all of whom respond to accuracy. On this reading, misinformation is what the knowledge class calls the other side’s propaganda, and information is what it calls its own, and the terms feel like measurements from inside because the coalition’s members staff the measuring institutions. DiResta coined a refinement in 2021, ampliganda, propaganda amplified by real people, and the refinement keeps the asymmetry, since the corpus applies it to rival networks and never to the amplification cascades her own coalition ran through its newsrooms.
The propagandistic biases sort her hard cases with uncomfortable ease. Perpetrator bias: New Knowledge, where she directed research, housed the Project Birmingham operation, a disinformation-style experiment in the 2017 Alabama race, and the episode enters her camp’s account with full mitigating apparatus, a rogue executive, a small budget, no proof of her participation, an aberration. The Virality Project flagged true vaccine stories that might fuel hesitancy, and the episode enters as a defensible judgment call under emergency conditions. Rival transgressions receive the victim-bias treatment in mirror image: systemic, coordinated, malevolent, and escalating. Attributional bias: her alliance’s authority derives from method and rigor, internal causes, while its collapse in 2024 derives from lawfare and harassment, external attack. The rival coalition’s beliefs derive from manipulation by grifters and algorithms, external causes, never from persuasion, since crediting a rival with persuasion concedes it an internal strength. Her 2021 piece stating that 2020’s voting falsehoods came almost exclusively from the right makes the attributional move, we flag them more because they lie more, and Alliance TheoryAlliance TheoryAlliance Theory predicts that both camps will read this essay as vindication. That prediction, at least, seems safe.
The Emergency Register: Renée DiResta Through Securitization Theory
At 9:32 on the morning of August 1, 2018, Chairman Richard Burr (b. 1955) gaveled open a hearing of the Senate Select Committee on Intelligence in room SH-216 of the Hart Building. Poster boards stood on easels beside the witness table, blown-up screenshots of fake social media accounts, the props arranged before the witnesses spoke. Renée DiResta, then director of research at a threat intelligence firm in Austin, told the senators that online manipulation was one of the defining threats of our generation.
Read the sentence slowly. It names a threat. It scales the threat to a generation, which is to say, to everything. It issues from a witness whose firm sells detection of the threat, whose datasets measure the threat, and whose field exists only if the threat does. The senators, men formed by network television, could not check a claim about recommendation algorithms any more than they could check a claim about centrifuge cascades. They could accept or refuse. They accepted, and the acceptance built a field.
There is a body of theory built for that sentence. In the 1990s, Barry Buzan (b. 1946) and Ole Wæver (b. 1960), working at the Copenhagen Peace Research Institute, developed what became known as securitization theory, formalized with Jaap de Wilde in Security: A New Framework for Analysis (1998). Their insight was that security is not a condition out in the world waiting to be measured. Security is a speech act. A securitizing actor declares that a referent object, the state, the nation, the environment, faces an existential threat. If the relevant audience accepts the declaration, the issue leaves normal politics, where claims get debated, budgets get contested, and losers get another round, and enters the emergency register, where speed beats deliberation and measures forbidden in ordinary times become licensed. The theory stays agnostic on whether the threat is real. It studies the move. Wæver added a normative preference that most readers skip: he thought desecuritization, the return of issues to ordinary politics, was usually the healthier direction, because the emergency register is where democracies store their exceptions.
The Copenhagen School catalogued the conditions under which the move succeeds. The grammar must be right: a threat, a referent object, a point of no return, a way out. The speaker must hold a position of authority with the audience. The threat should carry features that make it easy to imagine. All three conditions stood ready in SH-216. The intelligence community had already certified Russian interference, lending the witness borrowed state authority. The 2016 result had left half the political class searching for an account of what had happened to them. And the props supplied the imagination: here is the fake account, here is the Black activist page run from St. Petersburg, here is the machine reaching into the American mind. DiResta did not have to build the emergency from nothing. She had to name it, scale it, and stand beside it as its interpreter.
What sets her case apart from the theory’s standard specimens is the referent object. Classical securitization defends borders, regimes, currencies, identities. DiResta’s referent was shared reality, the epistemic base on which self-government is said to rest. No one had securitized that before, and the novelty carried a structural consequence the Copenhagen authors never had to model. When a defense minister securitizes a border, the audience already funds armies and intelligence services that can assess the claim against the speaker’s interest. When DiResta securitized the information space, no independent assessment capacity existed. The instruments that measured the threat, the network maps, the platform datasets, the account attributions, belonged to the same small circle that announced it. The witness named the emergency and held the only thermometers. She securitized a domain and chartered its priesthood in one appearance, and for the next four years the people who assessed whether the threat was growing were the people whose budgets grew with it.
The emergency register then licensed its extraordinary measures, and their form deserves attention because it explains the constitutional dead end that came later. A state emergency runs through the state, where courts and elections can reach it. This emergency ran through private platforms. The Election Integrity Partnership logged viral election claims into a ticketing queue in 2020, with university analysts, election officials, and platform trust-and-safety desks sharing one workflow. The Virality Project extended the model to vaccine rumors in 2021 and advised platforms that even true stories of side effects could fuel hesitancy. Analysts worked election week in shifts. Speed beat deliberation, which is the signature of the register: a rumor about Sharpie pens could not wait for a peer-reviewed rebuttal, so it got a ticket, a trace, and in some cases a flag. When the plaintiffs in Murthy v. Missouri asked the Supreme Court to find the emergency, the Court could not locate a state actor to hold, and on June 26, 2024, it dismissed the case on standing. The exception had been engineered to run outside the constitutional register that supervises exceptions. Securitization through platforms evades the checks that securitization through states must face, and the evasion, which looked like cleverness in 2020, meant that no court ever ruled the measures lawful, so the legitimacy question stayed open for whoever wanted to collect it.
The Copenhagen School explains the move. It takes the Paris School to explain the mover. Didier Bigo, writing against Copenhagen’s focus on dramatic speech acts, argued in “Security and Immigration: Toward a Critique of the Governmentality of Unease” (2002) that securitization in modern states is less a speech than a trade. A transnational class of professionals, police, intelligence officers, border technicians, risk analysts, private security vendors, competes for budgets and jurisdiction by defining threats their own skills happen to fit. Bigo calls them managers of unease. They circulate among agencies, firms, and academies, carrying their techniques with them, and the competition among them selects for threat inflation, because the market for assessments punishes deflation. No analyst ever lost funding by overstating a danger he was hired to watch. The understaters lose the contracts, the hearings, and the field, and they lose them before anyone learns they were right.
Set DiResta’s biography against Bigo’s portrait. Summers at the Central Intelligence Agency from 1999 to 2004. Seven years on a trading floor, where the job is pricing risk from incomplete information. Then New Knowledge, a private firm whose product was disinformation defense sold to corporations and, after 2016, to committees. New Knowledge was an unease vendor in the strict Bigo sense: its revenue was a function of the perceived size of the threat it detected. The firm’s chief executive went further than the model requires, participating in Project Birmingham, the 2017 Alabama experiment that manufactured the appearance of Russian bot support for Roy Moore, a vendor fabricating a sample of the danger it sold protection against. No public evidence shows DiResta directed those tactics, and the distinction should be kept. The structural point survives the distinction. She rose through a firm whose commercial logic was the logic Bigo describes, then carried the toolset into a university lab funded by foundations that wanted the threat watched, staffed by analysts whose careers required its persistence. The techniques of counterterrorism, the actor attribution, the network mapping, the fusion of state and private data, migrated into the civilian information space inside the résumés of the migrants.
Bigo also predicted the form the work would take. The Paris School holds that mature securitization stops announcing itself and settles into routine, into software, watchlists, and queues, where the emergency no longer needs declaring because it lives in the workflow. The EIP ticketing system is that settlement. No one stood up in November 2020 and proclaimed an existential threat over each Arizona rumor. An analyst opened a ticket. The form had a field for the claim, a field for the spread, a field for the assessment, and the emergency ran through the form the way water runs through pipe, invisible, procedural, and continuous. A voter in Mesa met the register as a label under a post. He could not see the queue, and the queue could not see him, and Bigo would say that this mutual blindness is what the governmentality of unease looks like from both ends.
Run the counterfactual the frame demands. Imagine DiResta testifying in August 2018 that the Internet Research Agency’s reach concentrated among committed partisans, that its persuasive effect on the election was likely near zero, and that the greater danger lay in overreaction. The testimony might have been closer to what later measurement found. It would also have ended the hearing, the field, the Observatory, and the career, because a threat priced at its measured size funds no apparatus. The selection pressure runs upstream of any individual’s honesty. The field recruited people who found the threat vast because people who found it vast were the ones the field could use, and each of them, sincere, credentialed, and diligent, then produced the assessments that confirmed the recruitment. She has been more careful with numbers than her field’s median, correcting the 22 million figure down to some 4,800 flagged URLs when a hostile author inflated it. The care is real and it operates inside the ratchet, not against it, because the ratchet turns on the size of the mission, not the size of any statistic.
The mission grew the way Bigo says jurisdictions grow. Anti-vaccine networks in 2015. Russian operations in 2018. Election rumors in 2020. Vaccine rumors in 2021. Then AI-generated propaganda, deepfake text, TikTok’s recommendation engine as an instrument of algorithmic control, and the problem of verifying humanness online. Some of the expansion tracks technological change that any honest observer might track. The frame reads the pattern, and the pattern never once contracted. No cycle ended with the finding that the threat had shrunk and the watchers could stand down. Emergencies staffed by professionals do not file reports recommending their own demobilization, and Robert Higgs (b. 1944) built a book, Crisis and Leviathan, on the observation that the apparatus never returns to baseline.
Then came the part of the theory that almost never gets written, because securitizations rarely fail in public. The Copenhagen School holds that the move lives on audience acceptance and dies on its withdrawal. DiResta’s securitization needed three audiences: Congress for legitimacy, platforms for enforcement, and the press for moral cover. Between 2022 and 2024 all three broke. Elon Musk bought the enforcement layer and handed its internal files to writers who ran the securitizing move in reverse. The Twitter Files, the Weaponization subcommittee, and the America First Legal suits declared an existential threat to a rival referent object, free speech, named the anti-disinformation field as the threat, and licensed their own extraordinary measures, subpoenas that swept in undergraduates, discovery aimed at researchers, legal costs designed to price the work out of existence. Note what this was not. Wæver’s desecuritization returns an issue to ordinary politics, where people argue and vote. Nothing returned to ordinary politics. The information domain stayed in the emergency register and changed hands. The field that had processed rumors got processed as one, by the same grammar, before a new audience, with a new priesthood standing beside its own poster boards. Jim Jordan (b. 1964) announcing that free speech had won again is a securitizing actor closing a successful operation, and the symmetry is the finding, not an irony to savor.
Score the frame where it strains, because a frame that explains everything explains nothing. Securitization theory cannot say whether the threat was real, and parts of it were: the Internet Research Agency existed, rumor-to-violence pipelines kill people in India and Myanmar, and coordinated harassment destroys lives, including hers. The theory reads the register, not the facts filed in it. The Pentagon report strains the Bigo portrait, since her team exposed covert American influence operations at no funding advantage and some cost, conduct an unease manager might have skipped, though a defender of the frame can answer that policing all intruders, including one’s own state, enlarges jurisdiction too. And her conduct after the collapse runs against the ratchet. The program in Invisible Rulers, changed defaults, user control over feeds, friction applied to all claims alike, transparency that outsiders can audit, moves the domain’s governance out of the emergency register and into design politics, visible, contestable, and slow. Whether she has read Wæver or not, the program is desecuritization by his definition, offered by the person the register consumed. It asks less deference than the ticketing queue asked and it surrenders the priesthood’s monopoly on the thermometers.
The offer has no takers yet. Her old coalition wants the emergency back, staffed and funded. Her new enemies have an emergency of their own and no reason to close it. Both camps now hold watchtowers, and the theory’s last lesson is the earliest one Wæver taught: the emergency register is easy to enter and hard to leave, because everyone who enters it acquires a post, and the post comes with a view of every threat except the register. In room SH-216 a witness told the senators the danger would define a generation. Eight years on, the sentence holds, though not as she meant it. What defined the generation was not the manipulation. It was the emergency declared over it, the professions chartered to manage it, and the discovery, made twice from opposite directions, that a society which governs its speech from the emergency register will be governed by whoever holds the register that year.
The Oldest Law: Renée DiResta and Folk Deterrence
In April 2023, the House Select Subcommittee on the Weaponization of the Federal Government subpoenaed Stanford University’s documents on the Election Integrity Partnership. The demands swept in the names of students, undergraduates among them, who had worked shifts logging election rumors during their junior year. The subcommittee never published a refutation of the partnership’s findings. It did not need one. The students’ names in a congressional docket did the work a refutation cannot do. Every graduate student in America weighing a career in this field read the docket and understood the terms. So did every provost. By June 2024 Stanford had let Renée DiResta’s contract lapse, the partnership had announced it would work no future election, and the field had lost its flagship at the richest university on earth without a single finding being overturned.
The internet has a name for this. FAFO, fuck around and find out, the phrase that turns up on gun-range t-shirts, in police-blotter comment sections, and in every schoolyard that ever existed. It reads as a joke and functions as jurisprudence. FAFO holds that conduct carries prices, that the prices get set by the offended rather than by referees, and that the collection teaches the audience more than any argument. It is folk deterrence theory, and it descends from the oldest written law there is. The Code of Hammurabi and the lex talionis of Exodus did not invent retaliation. They found it running and tried to cap it, an eye for an eye as a ceiling on the exchange rate, because the feud predates the court by longer than the court likes to remember.
The formal literature caught up in the twentieth century. Thomas Schelling (1921-2016) separated the power to hurt from the power to win in Arms and Influence and argued that hurting is bargaining, a language spoken to audiences rather than targets. Donald Black (1941-2024), in “Crime as Social Control,” showed that most of what states prosecute as crime is self-help justice, punishment executed by people who believe no referee will hear their case. Robert Axelrod (b. 1943) ran the tournaments in The Evolution of Cooperation and found that tit-for-tat, retaliate once, then match the other side’s last move, beats both saintliness and savagery, provided the players can see each other’s moves and expect to meet again. Read together, the three supply the grammar of FAFO: deterrence prices conduct instead of judging it, the addressee is the crowd rather than the culprit, and the system stabilizes only when signals are legible and the game repeats.
Run DiResta’s decade through that grammar and the disinformation wars stop looking like a dispute about truth. They look like a feud between coalitions, each running a deterrence campaign dressed as an epistemics campaign, each experiencing its own campaign as hygiene and the rival’s as terror.
Her coalition struck first, and the frame requires saying so without flinching. Consider what content moderation does, as conduct rather than as doctrine. A takedown removes one post from one feed. Its effect on the poster’s beliefs runs somewhere near zero, and its effect on the audience is the point. A takedown teaches every observer what gets a man removed. A label prices a class of claims. A deplatforming prices a career. The Election Integrity Partnership’s ticketing queue, whatever its analysts intended, operated as a pricing bureau: this rumor drew a flag, that account drew a strike, and the platform’s enforcement published the tariff schedule to everyone watching. The Virality Project went further in 2021 and advised platforms that true stories of vaccine side effects deserved attention, which drew a red line across a class of speech that included accurate speech. Deterrence campaigns do that. They price by category, since categories teach faster than cases. Half the country read the line as drawn across its own mouth, and in deterrence the reading is the reality, because deterrence lives in the audience’s head or nowhere.
The coalition did not experience any of this as a strike. It experienced it as public health, which is how first movers in feuds always experience their opening, and the sincerity changes nothing about the signal received. From 2017 through 2021 the deplatformings ran in one political direction often enough that the target population stopped parsing individual cases and priced the pattern. By the folk law’s logic, a coalition that governs a rival’s speech while calling the governance neutral has fucked around. The finding out took five years to arrive because the aggrieved coalition believed the referees were captured. Here Black’s essay earns its place in the essay. Self-help justice arises where law is unavailable, and the right’s information wing surveyed the referees, courts that move in years, universities that housed the flaggers, platforms whose trust-and-safety desks trained under the flaggers, a press that cited them, and concluded that no tribunal it trusted would hear the case. So it built its own tribunal, outside all of them.
The counterstrike arrived in the structure Black predicts, a trial conducted through channels the plaintiff controls. Elon Musk bought the evidence room in October 2022 and handed the files to writers. The Twitter Files were discovery. The hearings of Jim Jordan were prosecution. The America First Legal suits of Stephen Miller , naming Renée DiResta, Alex Stamos, and Kate Starbird and surviving into discovery in a Louisiana federal court in December 2024, were the civil docket. The harassment was the sentence, executed by the crowd, as folk sentences are. And through the entire proceeding, the merits sat untouched. Matt Taibbi and Michael Shellenberger never overturned the Internet Research Agency attribution. The subcommittee, holding every ticket under subpoena, never showed the 2020 flags were wrong. Deterrence does not argue. It prices, and the price list ran: millions in legal fees, security consultations for a mother of three, undergraduate names in dockets, a research center wound down, a field’s hiring pipeline poisoned at the source. Murthy v. Missouri failed at the Supreme Court in June 2024 and the failure cost the campaign nothing, because the campaign’s court was never in Washington. General deterrence had already succeeded. Ask the provosts.
The frame explains the feature of the episode that the legitimacy frames explain worst, the disproportion. DiResta flagged tweets and absorbed a sentence sized for a regime. Folk deterrence works that way by design. It punishes faces rather than culpability, because the audience learns from faces, and she was the face on the poster board, the witness from the Senate hearing, the researcher with the Agency summers that made the story write its own villain. Schelling taught that effective threats leave something to chance and that excess signals resolve. A proportionate response, a stern op-ed answering her op-eds, teaches nothing. A response that wrecks a career for conduct a court never adjudicated teaches every onlooker that the coalition’s lines cost more to cross than anyone can pay. The cruelty was the syllabus.
Now run the exchange backward once more, because tit-for-tat has no first round that both players accept. Ask her coalition when the feud began and it answers 2016, with a foreign operation and a flood of lies that licensed emergency measures. Ask the rival coalition and it answers earlier, with a decade of institutional capture that made the emergency measures possible, and behind that another grievance, and behind that another. Feuds have no agreed opening. Each side holds a ledger on which its own last strike appears as retaliation and the rival’s appears as aggression, and the ledgers never reconcile because reconciling them is the sovereign’s job and the sovereign is missing. This is the condition lex talionis was written for. An eye for an eye reads as savage until you see what it replaced, which was a head for an eye, a clan for a head. The code capped the exchange rate because someone with authority over both parties existed to cap it. The information wars run uncapped. No institution holds authority over both coalitions, the courts declined the case on standing, and each camp reads every cap proposal as the rival’s attempt to freeze the feud at a favorable line.
The current equilibrium looks like what deterrence theorists call mutual deterrence and feud scholars call exhaustion. Her coalition’s enforcement apparatus is dismantled or in retreat. Meta ended third-party fact-checking in January 2025, X runs on community notes, the partnerships are dissolved, and researchers have exited the field or rebranded their work. The rival coalition’s campaign wound down for lack of targets. Both sides now patrol a demilitarized information zone where almost nobody flags anything, which the right calls freedom and the left calls abandonment and Axelrod might call a fragile truce. His tournaments carry a warning for this one. Tit-for-tat stabilizes when moves are visible and the players expect repeated rounds. The information feud runs on murky attribution, anonymous mobs, and asymmetric visibility, and under those conditions tit-for-tat degrades into vendetta, since each side punishes its estimate of the other’s move rather than the move. The next round is loaded whenever an administration changes and the enforcement machinery changes hands. Both camps prefer the feud to the code because each expects to win the next round, and that expectation is the engine feuds run on for generations.
DiResta’s own conduct since the sentence reads, through this frame, as an attempt to move her disputes back under referees, which is what a party does when self-help has priced her out. The fight with Jacob Siegel over his book ran through corrections requests to publications, editor by editor, with documents. One outlet corrected, one did nothing, one pulled a review. Measured tit-for-tat, proportionate, addressed to institutions that still hold a shred of authority over both parties. Her Substack register tells the other half of the story. The combat subtitles, the profanity, the promise to break down every smear, are counter-deterrence signaling, a target advertising that she is expensive to attack, that every innuendo draws a documented reply. A woman who absorbed a folk sentence has learned the folk law, and her current posture, referee where possible, retaliation where necessary, is what the literature says a rational player does in a system with weak courts. The register costs her. She predicted her enemies might cast her as unhinged and then handed them quotes for the casting, which is the known tax on deterrence talk: credible threats require heat, and heat photographs badly.
Score the frame where it strains. FAFO assumes a transgression, and the transgression is the disputed question. The frame reads the punishment and stays agnostic on the crime, which means it cannot say whether the EIP’s flags were right, whether the IRA report was sound, or whether the Virality Project’s advice was defensible under emergency conditions. Those questions belong to other tribunals, and this frame’s finding is that no tribunal both sides accept has heard them, or will. The frame also fits the field better than it fits her margins. Her team’s Pentagon report exposed covert American influence operations, a strike against her own coalition’s security wing, and feud actors do not strike their own patrons. The report is evidence that a researcher lived inside the operative, and the frame, honest, records the evidence against its own reading. And the frame cannot price truth at all, which is its limit and its lesson. A deterrence system settles who may speak at what cost. It never settles what is so. Societies invented courts, journals, and juries because the feud answers the first question while starving the second, and a country that lets its information disputes run on folk law will get what folk law delivers, order of a kind, priced in faces, with the truth left standing outside the tribunal, uncalled.
The oldest law is old because it works and older than the code because it needs no one’s permission. DiResta spent a decade inside institutions that believed they had replaced it, universities, platforms, committees, partnerships, all of them staffed by people who thought the exchange of consequences had been civilized into procedure. The feud came through the procedure like water through a fence. Her field fucked around in the currency of flags and found out in the currency of subpoenas, and the finding out taught the watching country nothing about Russian bots, vaccine rumors, or stolen elections. It taught the price of standing between Americans and their speech, posted where every provost could read it, which was the message, which was always the message, because deterrence has only ever had one.
The Voice
Her voice phases track her institutional position.
The early voice, 2015 to 2019, is the swashbuckler. Read “The Digital Maginot Line,” the 2018 Ribbonfarm essay, and you find war metaphor run to the horizon: an Information World War, combatants, theaters, arsenals. She writes “we are immersed in an evolving, ongoing conflict” and means it. The register is apocalyptic briefing, a threat analyst who has seen the future and needs the room to wake up. This is also her coinage period. She mints phrases built for citation: asymmetry of passion, ampliganda, the 2018 Wired formulation that free speech is not free reach. The coinages share a design, compact, chiasmic, brandable, the work of someone who studied how phrases travel before she wrote any. A trader learns that the market buys the crisp formulation, and she prices her language accordingly.
The middle voice, the Stanford years, is bureaucratic natsec. The reports speak of actors, operations, narratives, inauthentic coordinated behavior, tradecraft. Passive constructions, hedged attributions, numbered findings. The diction did political work she may not have priced: she sounded like Langley because the field borrowed Langley’s lexicon wholesale, and when her enemies later needed a spook, the transcripts read like depositions from one. The vocabulary that gave her authority with senators became evidence against her with everyone who distrusts senators.
The current voice, post-fall, is the most alive and the least disciplined. The Atlantic essays run controlled irony over a wound: she narrates her own destruction in her professional vocabulary, the rumor cascade, the main character, the fantasy-industrial complex, which is a counter-coinage riffing on her accusers’ censorship-industrial complex, itself riffing on Eisenhower. Her Substack drops the control. Bullshit in the masthead, lie-machine exposé in a subtitle, crazy bitch shit in the text. She knows the price of the register, predicts in the same piece that she will be cast as unhinged, and pays anyway, because Substack pays combat and because a woman who absorbed what she absorbed has earned some heat and knows her audience agrees.
Across all three phases, constants. She argues from numbers first: 0.6 percent of accounts producing a quarter of the content, 4,800 URLs against the claimed 22 million. Quantity is her ethos, the trader’s tell, precision offered as credentials. Her root metaphor is epidemiological: virality, superspreaders, inoculation, pre-bunking, information as contagion. The metaphor smuggles in its politics, since contagion implies a public health authority, quarantine powers, and patients who cannot consent, and half her troubles trace to readers who understood the implication before she conceded it. Her second metaphor field is infrastructure: pipelines, channels, amplification, friction, defaults, design. She talks about speech the way an engineer talks about water systems, and the talk carries the same assumption, that someone competent should be maintaining the pipes.
She is a taxonomist. Two-axis frameworks, overt against covert, broadcast against social, typologies of actor and channel. She fights over definitions the way lawyers fight over jurisdiction, moderation is not censorship, misinformation differs from disinformation differs from malinformation, because whoever holds the definitions holds the field. Her concessions come first and function as armor: legitimate questions exist about researcher-platform contact, and then the pivot. In speech she is fast, fluent, paragraph-length, low on hedges, with the podcast softeners, sort of, right?, riding on top of sentences that were never soft. She anticipates objections and answers them before the interviewer finishes, a habit from adversarial rooms that reads as brilliance to allies and as slipperiness to enemies.
What the voice avoids tells as much. Almost no moral vocabulary: harmful, corrosive, and manipulative where another writer might say wrong or evil. System-talk where a preacher might use sin-talk. Almost nothing confessional; the one personal anchor she returns to is maternal, the preschool data, the mom of three, and she deploys it with discipline, the sole warm object in a cold lexicon. Her self-descriptions compress to deadpan, Twitter Files bête noire, four words carrying two years of subpoenas, and the compression is the style: irony as load-bearing wall, the analyst refusing her enemies the satisfaction of a visible bruise.
The costs. The wonk density stacks abstraction on abstraction until only the guild can follow. The contagion metaphor concedes her critics’ case in her own imagery. The natsec diction feeds the spook story. The irony reads as condescension to everyone outside the room. And the analyst persona has no off switch: she processes her own persecution with the same instruments she ran on troll farms, which produces superb essays and suggests a woman who can no longer encounter anything, including her own suffering, except as a case study. The prose never breaks. Whether that is strength or armor grown into the skin, the prose declines to say.