Summary of Findings

The most consistent finding across all 121 entries is that institutional consensus forms faster than evidence warrants, and then becomes self-defending. The mechanism is not primarily fraud or stupidity. It is a structural problem: the same institutions that adopt a position control the journals, the funding, the training programs, and the professional consequences for dissent. Peer review, dietary guidelines, psychiatric diagnosis, intelligence theory, and COVID origin investigations all show the same pattern. A claim achieves early institutional endorsement, that endorsement is cited as evidence of validity, and critics are filtered out not because their methods are flawed but because their conclusions are unwelcome. The Lancet's COVID origin statement, the AHA's saturated fat guidelines, the AAP's gender care endorsements, and the APA's diagnostic expansions are all examples of the same failure: a professional body substituting its authority for the evidentiary process that authority is supposed to represent.

A second pattern runs through nearly every entry: the populations harmed were not the populations who held the assumptions. Elderly patients were deprioritized in vaccine allocation by committee members who were not elderly. Children were taught to read by guessing, a method designed by academics who were not the ones who would fail to read. Alzheimer's patients and their families bore the cost of three decades of misdirected amyloid research while the researchers collected grants and prizes. Working-class girls in Rotherham bore the cost of institutional decisions made by officials protecting their own professional and political comfort. This asymmetry is not incidental. It is structural. The people with the power to adopt and defend false assumptions are systematically insulated from the consequences, which removes the feedback that would otherwise accelerate correction. When the costs fall on people without institutional voice, the correction signal is weak and slow.

A third pattern is the corruption of the correction process itself. Several entries describe not just an original false assumption but a false correction that introduced new errors. Gould's reanalysis of Morton was itself methodologically compromised. The elimination of standardized testing, intended to reduce bias, may have removed the one instrument that allowed talented low-income students to signal ability past weak school records. The Theory of Multiple Intelligences arose as a moral reaction to eugenics-era IQ abuse, and that moral origin insulated it from the scrutiny that would have caught its empirical weaknesses. The pattern recurs: harm from Theory A produces emotional and political rejection, Theory B is adopted because it repudiates Theory A, and Theory B is then protected from scrutiny by the memory of Theory A's crimes. The correction inherits the epistemic vices of the original error.

What ties all of this together is a specific failure in the translation from qualified research finding to institutional policy. Academic claims carry error bars, sample limitations, and stated scope conditions. By the time a finding reaches a teacher training curriculum, a clinical guideline, a congressional hearing, or a newsroom style guide, those qualifications have been stripped out. The serotonin hypothesis of depression became "chemical imbalance" in pharmaceutical marketing. A small Dutch clinical protocol became the standard of care for a much larger and more heterogeneous adolescent population. A single Florida birth study became a presidential argument about physician demographics. Stereotype threat effects measured in controlled lab conditions became the explanation for national achievement gaps. In each case, the translation was not random noise. It was directional: the finding was enlarged, simplified, and made policy-ready in ways that served the institutional interests of whoever was doing the translating. The replication crisis, the dietary guideline failures, the reading instruction disaster, and the COVID origin suppression are not separate scandals. They are the same scandal, repeated across fields, because the incentive structure that produces them has not changed.

The 121 entries in the False Assumption Registry, read together, reveal a single dominant structural failure: institutions translate contested or outright false empirical claims into policy faster than they can detect the error, and then resist correction long after the error becomes visible. The pattern repeats across domains so different that the repetition itself is the finding. Gain-of-function research continued under redefined terminology after two moratoriums. Anti-bias training remained mandatory in most universities after controlled experiments showed it increased prejudice in some participants. Learning styles and multiple intelligences frameworks persisted in teacher certification programs for two decades after psychometric science rejected them. The 11.3 million undocumented immigrant figure shaped enforcement budgets and amnesty debates for years after researchers estimated the actual population was roughly double. Dietary guidelines built on the saturated fat hypothesis survived for forty years while the supporting data was quietly reanalyzed and found wanting. The common mechanism is not stupidity or malice but institutional inertia: once a false assumption is embedded in a consent decree, a professional association statement, a funding formula, or a certification requirement, reversing it requires overcoming the careers, budgets, and identities built on top of it. The Luevano Consent Decree of 1981 banned validated cognitive testing for federal hiring; forty-four years later, federal agencies still select employees using subjective methods that perform worse by every available measure. The assumption outlasted the generation that made it.

Several chains in the Registry converge on the same population as the primary victim of policies designed to help them. The clearest case runs through criminal justice. The assumption that policing disparities prove discrimination, rather than reflect crime-rate differences, produced court orders requiring demographic proportionality in stops and arrests. Proactive policing retreated. The national homicide rate rose 41.9 percent between 2014 and 2023. Black Americans, who constitute roughly 53 percent of homicide victims, absorbed the largest share of that increase. Baltimore, operating under a consent decree and a district attorney whose charging policies reflected the same underlying assumption, recorded a per-capita murder rate worse than its crack-era peak, with more than 1,500 killings over five years and clearance rates below 50 percent, meaning most families received no legal accountability. A parallel chain runs through education. The assumption that schools can equalize outcomes regardless of inputs produced No Child Left Behind's 100 percent proficiency target, never achieved, and decades of adequacy litigation that doubled real per-pupil spending without closing measurable gaps. Some Baltimore schools now report zero percent proficiency in math and reading despite near-top-decile spending. The assumption that standardized tests are biased instruments produced test-optional admissions policies and the Luevano Decree, both of which removed the one mechanism most likely to identify high-ability students from disadvantaged backgrounds who lack other signals. The equity-motivated intervention, in each case, produced the least equitable outcome.

A second cluster of chains connects epistemic suppression directly to compounding harm. When the grooming gang crisis emerged in Rotherham, Greater Manchester Police instructed officers not to record the ethnic backgrounds of perpetrators; the absence of national data made pattern recognition impossible for sixteen years, during which an estimated ten thousand children were abused across multiple cities. When the lab-leak hypothesis for COVID-19 emerged, a coordinated statement in the Lancet and resistance within the NIH foreclosed investigation during the period when evidence would have been most recoverable; no reformed biosafety protocols exist today, and gain-of-function research continues under definitional ambiguity. When researchers produced peer-reviewed findings on population genetics or group differences in cognitive test scores, professional associations issued statements, IRBs imposed higher evidentiary bars, and careers were ended, not by refutation but by sanction. The vacuum this created was filled by researchers outside institutional constraints, whose cruder claims reached larger audiences precisely because mainstream science had abandoned the territory. The suppression produced the discourse it claimed to prevent. The same logic applies to the Biden cognitive decline case: platform moderation labeled health concerns as misinformation, fourteen million primary votes were cast under false fitness assurances, and when the June 2024 debate made the assumption publicly untenable, the institutional credibility of both the press and content moderation systems took damage that has not recovered. Suppression of corrective information does not eliminate the underlying reality; it delays the reckoning while compounding the eventual cost.

What the Registry reveals about how societies correct course is mostly discouraging. Correction happens most reliably when harm is acute, attributable, and embarrassing to a specific institution: Critical Incident Stress Debriefing was abandoned after controlled studies showed it worsened PTSD in some patients; the UK introduced a phonics mandate after the Rose Review documented the damage from whole-language instruction; the Supreme Court struck down race-conscious admissions after decades of documented mismatch effects. Correction happens least reliably when harm is diffuse, when the assumption is embedded in legal structures requiring legislative action to undo, or when acknowledging the error requires institutions to implicate themselves. The chemical imbalance theory of depression drove a 400 percent increase in antidepressant prescriptions over two decades; the theory has been substantially revised, but the prescribing infrastructure, the direct-to-consumer advertising framework, and the defunded community psychiatry system it replaced remain largely intact. The 2020 inflection point is visible across a remarkable number of chains simultaneously: the homicide surge, the acceleration of test-blind admissions, the vaccine equity prioritization that CDC modeling estimated would produce seven percent more deaths than an age-based approach, the suppression of the lab-leak hypothesis, and the mass adoption of gender-affirmation protocols in pediatric medicine without the clinical trial restrictions that Sweden, Finland, and England imposed by 2022. The Registry does not show a society that learns from false assumptions. It shows a society that learns selectively, corrects when correction is cheap, and sustains error when correction requires someone powerful to admit they were wrong.

The most consistent finding across all 121 entries is that no single institutional type bears concentrated responsibility. Federal agencies, elite universities, professional associations, prestige media, international bodies, commercial actors, and activist legal organizations each appear repeatedly, and in most cases four to six distinct institutional types are simultaneously promoting the same false assumption. This is not a story of conspiracies or rogue actors. It is a story of interlocking incentive structures that happen to point in the same direction. The pattern that recurs most reliably is a three-stage chain: an academic originator produces a finding or framework; institutional adopters, including professional associations, government agencies, and medical bodies, formalize and mandate it; then commercial and media actors popularize and profit from it. Each layer adds a different species of authority while obscuring the evidentiary weakness of the original claim. The result is that by the time a false assumption reaches public discourse, it carries the simultaneous endorsement of peer-reviewed journals, professional licensing bodies, government statistics, and mainstream news coverage, making challenge feel like attacking all of them at once.

Federal agencies and professional associations are the two institutional types that appear most frequently as enforcers rather than merely promoters. Agencies such as NIH, CDC, EEOC, the Census Bureau, and the Department of Education do not typically originate false assumptions, but they formalize them through funding structures, regulatory mandates, official statistics, and advisory committee recommendations. Once an assumption is embedded in a federal grant program, a consent decree, a dietary guideline, or a national survey methodology, retraction becomes institutionally costly in ways that have nothing to do with evidence. Professional associations, including the AMA, APA, AAMC, American Anthropological Association, and WPATH, operate similarly: their policy statements and training requirements propagate assumptions across entire generations of practitioners before large-scale replication studies exist. The speed of adoption is itself diagnostic. The AAMC required unconscious bias training at 155 medical schools within days of George Floyd's death. The UC Regents voted to eliminate standardized testing despite their own faculty senate's objections. The Department of Education endorsed grit interventions the same year as the TED talk that popularized the concept. In each case, the compressed timeline made evidence evaluation structurally impossible, and the institutional weight of the adopting body then insulated the assumption from subsequent challenge.

Individual actors matter, but their role is less that of villains than of founding authorities whose institutional affiliations allow preliminary findings to bypass normal scrutiny. A researcher at Harvard, Stanford, or Johns Hopkins produces a paper; the university's prestige multiplies the claim's credibility; journals publish follow-on work within the established framework; and the founding theorist's personal authority makes challenge feel like personal attack. This pattern appears across entries as different as the marshmallow test, the amyloid hypothesis, the refrigerator mother theory, and the multiple intelligences framework. What these cases share is not fraud but a structural failure: elite institutional affiliation substituted for independent replication, and the founding theorist's continued defense of the claim, sometimes decades after contradicting evidence accumulated, delayed correction. The proponents of false assumptions rarely suppressed evidence through deliberate deception. They operated within incentive structures that rewarded supportive findings, discouraged null results, and made ideological alignment with prevailing moral frameworks a condition of career advancement. The suppression of dissenters, when it occurred, was itself institutionalized: formal university proceedings against researchers, journal gatekeeping, professional association enforcement mechanisms, and platform moderation policies all operated as bureaucratic processes rather than individual acts of censorship.

Responsibility is further diffused by the role of intermediary organizations that allow government and academic actors to maintain distance from direct accountability. Nonprofit intermediaries such as EcoHealth Alliance, Fusion GPS, the Global Disinformation Index, and the NAACP Legal Defense Fund each served as laundering mechanisms, converting government funding or politically motivated research into products that carried the appearance of independent scientific or legal authority. Prestige media outlets, particularly the New York Times, the Lancet, and Nature, amplified this laundering effect by treating institutional endorsements as substitutes for independent verification. The same outlets that published original claims issued corrections that were quieter, later, and less prominent than the original reporting. Perhaps the most durable enforcement mechanism across the entire registry is the legal system: Supreme Court citations, federal consent decrees, and activist litigation extended the lifespan of false assumptions long after their evidentiary basis was undermined, because challenging the assumption then required challenging a legal precedent rather than merely a scientific claim. The Brown v. Board citation of the Clark doll research is the clearest example, but the pattern recurs in stop-and-frisk litigation, gender medicine court briefs, and affirmative action Supreme Court records. Judicial imprimatur is the strongest known insulator of an empirical claim against revision.

The most fundamental difference between those who believed the false assumptions and those who suspected them is not intelligence, access to information, or even domain expertise. It is the question each group asked first. Believers asked: does this claim fit the story I already accept? Skeptics asked: what would have to be true for this claim to be wrong, and has anyone checked? This distinction sounds simple. Its consequences across 121 entries are enormous. The believer's first question is answered by narrative coherence, institutional prestige, and moral resonance. The skeptic's first question is answered only by mechanism specification, falsifiable prediction, and disconfirming evidence. These are not merely different methods. They are different definitions of what counts as knowing something.

Believers across every domain share a cluster of prior commitments that made them systematically vulnerable. The most pervasive is the blank-slate environmentalism: the assumption that outcomes are produced by inputs, that gaps reflect differential treatment, and that the right intervention closes any disparity. This prior appears in school spending policy, parenting research, gender medicine, IQ debates, and criminal justice simultaneously. It is not a finding; it is a premise so deeply held that contradicting evidence (the Coleman Report, twin studies, the sheepskin effect, the Flynn effect's hollowness) was absorbed without updating the underlying model. A second shared prior is teleological: history moves in a correct direction, and claims aligned with that direction carry a presumption of truth. This is explicit in "wrong side of history" rhetoric and Fukuyama's end-of-history thesis, but it operates implicitly whenever equity framing is treated as a near-decisive argument regardless of quantitative modeling. A third prior is the conflation of moral urgency with epistemic warrant. When the George Floyd moment produced simultaneous policy changes in vaccine prioritization, admissions testing, police deployment, and medical education, none of those changes followed evidence review. They followed moral consensus. The urgency compressed the normal skeptical interval to zero. Believers in these cases were not being dishonest. They were operating with a model in which moral clarity is a form of knowledge.

Skeptics share a different cluster of commitments, and the contrast is precise. Where believers accepted correlation as causation, skeptics demanded mechanism isolation: not just that two things move together, but a specified pathway that could be tested independently. Caplan's sheepskin effect, Pashler's crossover-interaction requirement for learning styles, Watts's representative-sample replication of the marshmallow test, Borjas and VerBruggen's birth-weight control in the Black-doctor study, Newman's birth-certificate audit of Blue Zone supercentenarians: each of these is the same move. The skeptic identifies the specific empirical condition that would distinguish the favored hypothesis from its rivals, then checks whether that condition holds. Skeptics also showed consistent attention to anomalies that believers explained away: adjacent Chicago blocks with identical demographics and wildly different shooting rates; homicide falling during the Great Depression despite high poverty; the DARPA rejection of the DEFUSE proposal; cohort-specific patterns in transgender identification invisible in aggregate data. These anomalies were not noise to skeptics. They were the data that mattered most, because they were the data the prevailing model could not accommodate. A third skeptic habit is conflict-of-interest detection. Matloff noticed that STEM shortage claims coincided with stagnant wages, not rising ones. Daszak's financial stake in EcoHealth Alliance was flagged before the Lancet letter's addendum acknowledged it. The Global Disinformation Index's funding structure predicted its rating patterns. Skeptics routinely asked who benefits from the claim being true, and treated the answer as relevant to how much scrutiny the claim deserved.

The pattern of who gets fooled versus who does not is not simply credentialed experts versus practitioners, though that distinction appears. It is more precisely insiders versus those with a vantage point outside the institution that benefits from the assumption. Amyloid researchers lost grants for challenging the hypothesis. Clinicians risked professional sanction for questioning gender affirmation. Journalists faced racism accusations for citing racial crime statistics. The social cost structure of these fields selected for false belief maintenance, and it selected most powerfully against those whose careers were most embedded in the relevant institutions. The skeptics who got it right were often outsiders by training (economists reanalyzing medical datasets, operations research professors modeling immigration demographics), by position (independent writers, NHS-commissioned reviewers rather than US medical establishment figures), or by willingness to absorb professional cost (Ferguson and Smith publishing a meta-analysis that contradicted the sentencing-disparity consensus; Inzlicht publicly recanting on ego depletion; Diamond and Rimland challenging autism orthodoxy from within clinical practice). The expertise paradox runs through the registry: domain experts were often more committed to false assumptions than informed outsiders, because expertise provided both the tools to construct sophisticated defenses and the institutional incentives to maintain them.

The deepest structural difference is in how each group handled the relationship between moral commitments and empirical claims. Believers consistently fused them. Challenging the doll-test methodology felt like challenging desegregation's moral legitimacy. Questioning the trauma-BPD causal chain felt like minimizing abuse. Applying base-rate reasoning to hate crime allegations felt like denying racism. Citing behavioral genetics felt like endorsing eugenics. In each case, the moral stakes of the empirical question were so high that falsification felt like betrayal. Skeptics, by contrast, maintained the distinction between what is true and what would be convenient to be true, even when the inconvenient truth complicated their own preferred positions. Lipsitch's modeling showed age-prioritization saved more lives within every racial group, accepting the equity framing while showing it pointed in a different direction. Ferguson and Smith acknowledged weak effects for drug crimes even while finding negligible effects for violent crimes. Flynn himself eventually concluded that the IQ gains bearing his name were "ersatz." This willingness to follow evidence into uncomfortable territory is not moral indifference. The skeptics in these cases often shared the underlying moral concerns of the believers. What they did not share was the conviction that moral urgency is a substitute for evidence, or that a claim becomes true because the alternative is distressing.

The most reliable early warning sign is not that a claim is wrong but that it has been insulated from the process that would reveal whether it is wrong. Across all 121 entries in the Registry, the single most consistent structural feature of a false assumption before its exposure is that the normal mechanisms of falsification, replication, and adversarial review were either absent or actively suppressed. The Clark doll study entered social psychology curricula without replication. The amyloid hypothesis consumed half of Alzheimer's research funding for decades while image anomalies in its foundational 2006 paper sat unexamined. The Lancet letter on COVID-19 origins labeled competing hypotheses "conspiracy theories" before any systematic investigation was possible. In each case, the question a careful observer should have asked is not "is this true?" but "what would have to happen for this to be shown false, and is anyone trying to make that happen?" When the answer is no, the claim is being maintained by something other than evidence.

The intellectual foundations of false assumptions share a specific and recurring architecture. A single study, a single case, or a single author's clinical notes bears weight that only a body of replicated, pre-registered research could legitimately carry. Kübler-Ross's grief stages derived from anecdotal notes on dying patients, not bereaved survivors, and were never tested longitudinally before entering medical school curricula. The marshmallow test's follow-up tracked fewer than ninety participants from an unusually stable Stanford population; the bivariate correlations were treated as causal evidence of child destiny. Kanner's 1943 paper described eleven subjects before refrigerator mothers became clinical orthodoxy. The pattern is consistent: the original researchers often noted the limitations themselves, buried in methods sections or footnotes, and the adopters simply ignored them. A practical heuristic follows directly: when a widely cited claim traces to a single study, ask whether the sample was representative, whether the design could distinguish correlation from causation, and whether independent labs have replicated the finding in different populations. If the answer to any of these is no, the claim is carrying more weight than its foundation supports.

The propagation patterns are as diagnostic as the intellectual foundations. False assumptions spread fastest when they are simultaneously convenient for multiple powerful institutional actors with different but aligned interests. The STEM shortage narrative served IT industry cost-reduction goals, university enrollment goals, and bipartisan competitiveness rhetoric; no single actor needed to coordinate because all benefited from the same story. The chemical imbalance theory of depression was propagated primarily through pharmaceutical advertising rather than peer-reviewed consensus statements, generating $2.8 billion in annual Prozac sales by 1998. The multiple intelligences theory achieved massive institutional adoption, curriculum redesign, and teacher training on a faster timeline than peer review and replication could possibly operate, because it felt egalitarian and validated what educators already wanted to believe. The specific warning sign here is velocity: when a claim spreads faster than the evidence cycle can operate, when temporary measures become permanent before outcomes are measured, when institutional mandates precede rather than follow validation, the consensus is being driven by something other than accumulated evidence. Rapid consensus formation is not a sign of a strong claim; it is a sign that scrutiny has been bypassed.

The structural incentives keeping false assumptions alive are often more visible than they appear. When the primary institutional voices promoting an assumption are also the primary financial beneficiaries of that assumption, independence claims require independent verification, not deference. The Information Technology Association of America, an industry lobbying group, originated and funded the STEM shortage narrative, then fed it to newspapers as independent reporting, which was then cited in congressional hearings as if it were independent research. The Global Disinformation Index operated on government contracts and ad-tech partnerships while claiming neutrality, and its methodology produced results that were one hundred percent directionally consistent with its funders' interests. The Fountas and Pinnell and Lucy Calkins reading curricula had massive commercial distribution that insulated them from scientific criticism about the three-cueing system for years after cognitive science had moved in the opposite direction. The practical test is to ask who profits if this claim is true and who loses if it is false, then ask whether those parties control the data, the funding, or the publication channels. When the answer is yes on multiple counts, the evidentiary record is being shaped rather than observed.

Early dissenters deserve disproportionate attention, and the manner of their dismissal is more informative than the dismissal itself. In case after case, credible critics with specific, falsifiable objections existed years or decades before public exposure. Arrow and Capron noted the lack of direct evidence for STEM shortage complaints in 1959. Jonathan Miller called Oliver Sacks a fantasist long before the books became famous. The Canadian Archaeological Association warned from the beginning that ground-penetrating radar detects soil anomalies, not graves, and that excavation was required before the mass graves narrative circulated globally. Bernard Rimland subjected the refrigerator mother hypothesis to methodological scrutiny it had never received in 1964. In every one of these cases, the dissenter was dismissed, ignored, or professionally penalized rather than engaged on the merits. This is the critical distinction: when a field responds to criticism with counter-evidence, it is functioning normally; when it responds with moral condemnation, career consequences, or simple silence, it is operating on social enforcement rather than empirical standards. The question to ask when a dissenter is dismissed is not whether the dismissal sounds convincing but whether it addresses the specific methodological objection the dissenter raised.

Several concrete smell tests work across domains. First, check whether the metrics used to evaluate the claim were selected from among those that confirm it, while metrics that would test it were not collected. Policing disparity studies compared stop rates to population shares rather than to offense rates; school integration studies measured demographic representation rather than academic preparation; anti-bias training evaluations measured diversity optics rather than discriminatory behavior. When the outcome metric has a known methodological vulnerability, such as selection effects, aggregation artifacts, or omitted variables, and that vulnerability is not prominently disclosed, the metric is being used rhetorically. Second, look for what the entry on income inequality and crime makes explicit: when publication bias has accumulated, the gap between published effect sizes and bias-corrected effect sizes is a direct measure of how much the literature has drifted from reality. A partial correlation dropping from 0.436 to a range of 0.007 to 0.123 after correction is not a minor adjustment; it is evidence that the field was not testing the hypothesis but confirming it. Third, watch for unfalsifiable expansions when evidence challenges the core claim. When controlled analyses showed the gender pay gap narrowed dramatically, advocates argued that occupational choices themselves reflected discrimination, making any outcome consistent with the discrimination hypothesis. When Putnam's trust data showed negative effects of diversity, advocates argued that segregation rather than diversity was the cause. When somatic therapy failed, non-improvement became evidence of insufficient somatic processing. A theory that can absorb any counterevidence by adding auxiliary hypotheses is not a scientific theory; it is a worldview, and it should be evaluated accordingly.

The final and most reliable heuristic is asymmetry of treatment between confirming and disconfirming evidence. The New York Times published over five hundred thousand articles during the Mueller investigation and offered no formal postmortem after the "did not establish" finding. The 1619 Project's founding date claim was quietly removed from the website without a correction notice after reaching millions of readers. The AAMC sent a clarification letter about the black doctors and newborn survival study after Supreme Court arguments had already concluded, after Justice Jackson's dissent had already cited the claim. In each case, the propagation was loud and fast; the correction was quiet and slow, or absent entirely. A thoughtful observer can apply this test prospectively: when a claim is being amplified, ask whether the institutions amplifying it have a track record of issuing proportionate corrections when they are wrong. If the answer is no, the amplification is not a reliable signal of truth. The Registry, read in full, suggests that the most dangerous moment in the life of a false assumption is not when it is first proposed but when it becomes what everybody knows, because at that point the institutions that would normally correct it have already staked their credibility on its being right.

The causal chain the thesis describes — expert claims hardening into common knowledge, which then drives bad policy, which then causes measurable harm — holds across a substantial majority of the 121 entries. The fit is strongest where the policy mechanism is discrete and the harm is quantifiable. The vaccine prioritization case is nearly perfect: CDC advisory committee members publicly argued an equity-first framework, that framework became media consensus, the CDC adopted it, and the committee's own modeling projected thousands of additional deaths. The grooming gang case is similarly clean: Home Office reports and academic "moral panic" framing gave police and councils intellectual cover to dismiss victims, and the result was 1,400-plus children abused in Rotherham alone. The 3-cueing reading instruction case, the facilitated communication case, the lobotomy case, the Lysenko case, and the Iraq WMD case all run through all four stages with named actors, named policies, and documented harms. Across the full registry, roughly 60 to 70 entries fit the chain well or very well. Another 30 or so fit partially, with one or two links weakened. Perhaps 15 to 20 fit poorly. The chain is a genuine and useful model, not a post-hoc imposition.

Where the chain breaks down, it usually breaks at one of two points. The first is the expert-claim node. Several entries originated not in credentialed expertise but in media activism, political pressure, or popular belief that bypassed expert gatekeeping entirely. The mass graves at residential schools story is the clearest case: the claim traveled from Indigenous activists and journalists to government policy (flags at half-staff, Canada Day cancellations, a papal apology) without meaningful expert validation. The "wrong side of history" assumption and the "honest race discussion is bad strategy" entry are similar: these are rhetorical and strategic postures, not scientific claims that filtered down through institutions. The thesis needs to acknowledge that the chain sometimes starts downstream of expertise, with media or political actors doing the laundering that experts do in the paradigm cases. The second weak point is the harm leg. Entries like Blue Zones, the five stages of grief, Myers-Briggs, and ego depletion produced real misdirection and wasted resources, but nothing comparable to famine deaths, sterilizations, or a homicide spike. The thesis would be more precise if it distinguished between entries where the harm is acute and attributable versus entries where it is diffuse and inferential.

The deeper pattern — moral convictions disguised as empirical claims — is present in a clear majority of entries, probably 65 to 75 of the 121. The pattern is not universal, but it is the dominant mode. The strongest cases are concentrated in areas touching race, gender, and immigration, which is itself informative. The vaccine prioritization entry is the purest case: the committee's own model showed the equity-first plan would kill the most people, yet the plan was still advocated, because the empirical optimization was subordinate to the moral commitment. The "trans skepticism causes suicide" entry is the most explicit weaponization: an empirical claim about causation was deployed as a moral cudgel ("blood on your hands") to enforce compliance, and the claim was never robustly established. The "race is entirely a social construct" entry, the "Black-White IQ gap is 100% environmental" entry, the "diversity is our strength" entry, the "structural racism causes health inequities" entry, the "grooming gangs are a moral panic" entry, the "airport profiling is racial discrimination" entry, and the Lysenko entry all show the same structure: a normative commitment generates an empirical claim, and the claim is then protected from falsification not by evidence but by the moral cost of questioning it. The UNESCO-era dynamic the thesis describes is visible in real time across dozens of entries.

But a meaningful minority of entries, perhaps 30 to 35, are primarily empirical errors without dominant moral disguise. The housing bubble denial was methodological overconfidence, not values-driven: economists used flawed models and mistook asset price trends for fundamentals. Peer review's quality-filtering assumption was institutional inertia and reasonable prior belief, not a moral commitment. The amyloid cascade hypothesis was scientific groupthink amplified by funding concentration, not a moral conviction about Alzheimer's patients. Ego depletion, the marshmallow test, and learning styles were laboratory findings that traveled too fast into policy, accelerated by commercial incentives (workshops, inventories, self-help books) rather than moral urgency. Murder rates as a crime proxy was a convenience and path-dependence failure. These entries fit the causal chain well but do not fit the moral-conviction pattern. They suggest the thesis's deeper pattern, while real and important, is not the only mechanism producing the failures the registry documents.

The thesis also misses or underweights several recurring patterns that the data makes visible. Financial incentives appear repeatedly as an independent driver, sometimes more powerful than moral conviction. The social media safety case is the clearest: platform executives had direct financial reasons to assert safety, and they funded research designed to produce ambiguous results. The pharmaceutical industry's role in the chemical imbalance theory, the testing industry's role in learning styles, and the consulting industry's role in Myers-Briggs and anti-bias training all show commercial interests sustaining false claims long after the evidence turned. Bureaucratic and institutional inertia is a second underweighted factor. The amyloid hypothesis persisted for two decades partly because NIH funding structures rewarded established research programs and punished heterodox alternatives. Peer review's persistence as a quality signal despite poor inter-rater reliability reflects the same dynamic: institutions built around the assumption cannot easily abandon it without dismantling themselves. A third pattern the thesis does not address is what might be called the Goodhart's Law failure mode, visible in the mental illness diagnosis entry and the murder rate proxy entry: a measure that was once a reasonable indicator becomes a target, and once it becomes a target it ceases to be a good indicator. This is neither a moral conviction nor a straightforward empirical error; it is a systemic property of measurement under institutional pressure.

There is also a pattern of motivated suppression that the thesis captures only partially. The thesis focuses on false claims being generated and amplified. But several entries document accurate findings being suppressed: Borjas's immigration wage data, Putnam's social trust findings, weekly CDC homicide data after Floyd's death, the DARPA rejection of EcoHealth's gain-of-function proposal. In these cases the causal chain runs not through a false expert claim but through the burial of a true one. The mechanism is related to what the thesis describes but is not identical to it: the moral conviction does not generate a false empirical claim so much as it creates a zone of epistemic prohibition around inconvenient true ones. The grooming gang case combines both: a false "moral panic" claim was amplified while accurate data about perpetrator demographics was suppressed. The thesis would be stronger if it treated suppression as a parallel mechanism alongside fabrication.

The overall verdict is that the thesis is substantially supported by the data, particularly in its first part. The causal chain from expert claim to common knowledge to bad policy to harm is the dominant pattern in the registry, present in recognizable form across roughly two-thirds of the entries. The second part, the moral-conviction-as-empirical-claim pattern, is also well-supported, present clearly in a majority of entries and especially dominant in the politically charged domains of race, gender, and immigration. The thesis correctly identifies the post-WWII moral urgency as a driver of this pattern, and the Lysenko entry, the eugenics-era IQ entry, the race-as-social-construct entry, and the UNESCO-adjacent entries all confirm that the dynamic has deep historical roots. What the thesis needs is qualification on three fronts: it should acknowledge that the chain sometimes starts with media or political actors rather than credentialed experts; it should recognize that financial incentives, institutional inertia, and Goodhart's Law failures operate independently of moral conviction and account for a significant minority of cases; and it should treat suppression of true findings as a parallel mechanism alongside amplification of false ones. The thesis is a genuine explanatory framework, not a complete theory of everything the registry contains.