Conspiracy Theories
in a Hurry

The Whole Thing in One Page

The public picture of a conspiracy theorist is a man in a basement joining photographs with red string. It is reassuring because it places the error inside somebody unlike you. The research offers less comfort. Conspiracy belief is not a separate species of mind. It recruits ordinary suspicion, ordinary trust and ordinary efforts to explain events, then can arrange them so that the conclusion becomes harder to lose than to gain.

Begin with the necessary distinction. A conspiracy is secret coordination by two or more people. Conspiracies occur. Watergate was a conspiracy. COINTELPRO was a covert state programme. The Tuskegee syphilis study involved deception and the denial of treatment. A conspiracy theory is an explanation that gives hidden coordination a central causal role. The label identifies the form of the claim, not whether it is true. Evidence must decide whether the allegation is documented, remains a live hypothesis, is poorly supported or has been protected from revision.

Hidden agency is attractive because events arrive as noise. Institutions fail through divided responsibility, bad incentives, error and chance. Large outcomes can have small or accidental causes. A plot turns this mess into intention. Someone wanted the result, someone controls it, and scattered anomalies now belong to one design. The explanation may be frightening, but it is usable.

No single need produces belief. Across hundreds of studies, associations with uncertainty, lack of control, social threat and identity are modest and uneven. Some people seek an explanation. Some seek a villain. Some defend a group, recover status or enjoy belonging to a minority that claims to see what others miss. Intelligence and education can improve scrutiny; they can also supply more skill for defending a preferred conclusion.

Claims are supplied as well as chosen. Politicians, broadcasters, influencers and communities select enemies, frame coincidences and decide which source deserves trust. A theory that flatters a coalition or excuses its leaders can travel even when another conspiracy, aimed at the same audience's own side, does not. The believer is therefore neither a solitary defective mind nor a passive victim of a feed. Belief is built in relationships.

The decisive change concerns evidence. Weak clues count for the theory. Missing evidence suggests concealment. Contrary evidence discredits the institution that produced it. Failed predictions become proof that the enemy changed plans. This is not absolute immunity, and all explanations can survive some awkward facts. The danger is escalating rescue: each failure adds another hidden actor or exception while the central claim loses no probability.

The internet accelerated the process and made investigation participatory. It did not invent it, and polling does not show every conspiracy belief rising in one direction. Printed forgeries, secret-society panics and political accusations travelled long before feeds existed.

The practical answer is neither blind trust nor ridicule. State the allegation precisely. Separate an anomaly from the explanation attached to it. Trace sources to their origin. Ask who is promoting the claim, how many people must coordinate, what independent traces should appear and what evidence would lower confidence. Genuine conspiracies leave records, money, witnesses, contradictions and participants who do not all control one another.

Suspicion protects a society from power. Suspicion without a stopping rule protects the theory from reality.

That is the book.

Why You Should Care

On 4 December 2016, Edgar Maddison Welch entered a Washington pizzeria carrying an AR-15 rifle and a revolver. He had driven from North Carolina to investigate an online claim that children were being abused through a political trafficking ring centred on the restaurant. He fired several shots into a locked door. Customers and staff fled. He found no imprisoned children because the story was false.

The episode is often filed under internet madness, which misses its instruction. Welch did not think he was acting without evidence. He thought he had moved beyond passive belief and was checking the evidence for himself. His mistake lay in the route from suspicion to action: anonymous claims became clues, clues became a coherent plot, institutional denial became expected, and the remaining uncertainty licensed an armed search.

You should care because the same route appears at lower intensity in ordinary life. A delayed medical result becomes proof of suppression. A coincidental pattern in prices becomes coordination. A clipped video establishes a larger plan whose details are supplied by comments. An organisation's evasive statement is treated as confirmation, though the statement may also be evasive because its lawyers wrote it. The mind dislikes leaving a cause blank, and public systems often supply enough opacity to make a hidden cause plausible.

You should also care because some suspicions are right. Governments have spied on political groups, companies have concealed harms, police forces have protected misconduct and officials have lied. Telling citizens to trust authority because conspiracy thinking is dangerous would be a cure worse than the disease. Institutions earn confidence by making decisions inspectable, correcting errors and allowing independent inquiry. A society unable to suspect power is defenceless against secrecy.

The problem is therefore harder than separating sensible people from cranks. It is learning how suspicion stays answerable to evidence. Watergate did not become true because the burglars looked sinister or because a president denied wrongdoing. It became established through arrests, money trails, testimony, documents, recordings, court processes and conflicts among institutions with different incentives. The case narrowed as evidence accumulated. It did not grow by treating every absence as proof.

Conspiracy theories matter before they reach violence. They can lower stated vaccination intentions, intensify prejudice, corrode cooperation and make every shared institution look like one coordinated enemy. Yet they can also mobilise political participation rather than producing apathy. Their effects depend on the theory, the audience and the action it recommends. A private belief about ancient aliens is not equivalent to an accusation that named neighbours are trafficking children.

By the end of this hour, you should be able to see four separations that public argument often destroys. A conspiracy is not the same as a conspiracy theory. Distrust is not the same as error. An anomaly is not the same as an explanation. Correction is not the same as submission to authority.

That final separation changes conversation. Ridicule can turn an evidential disagreement into a contest over dignity. A fact dump can fail because the person does not grant the source standing. Endless patience can also become collusion with a claim that shifts whenever tested. Better responses preserve respect while fixing the terms of inquiry: what is being alleged, which evidence is independent, what alternatives fit, what result would count against the claim, and when the investigation is finished.

The reward is larger than protection from strange stories. The subject teaches how people reason when evidence is incomplete, trust is damaged and belonging is at stake. Those are the conditions of much public life. Conspiracy theories are an extreme case of a general problem: how to remain open to concealed wrongdoing without granting secrecy the power to explain everything.

The Core Ideas

Suspicion Is a Human Capability

A conspiracy requires secrecy, coordination and more than one actor. It need not be grand. Two bidders agreeing to fix a price, officers coordinating false statements and executives hiding a known defect all qualify. Some conspiracies are exposed quickly. Others remain hidden for years. Many fail because one participant talks, records survive or the conspirators are less competent than the plan requires.

That matters because the phrase conspiracy theory is often used as a verdict. It can mean that a claim alleges hidden coordination, that it is poorly supported, that it is politically disreputable or that the speaker should be ignored. Those are different judgements. A claim does not become false because powerful people could have coordinated, and it does not become credible because governments have lied before.

A conspiracy theory is best treated first as a kind of explanation: it gives covert coordination a central causal role in an event or condition. Scholars disagree about whether the actors must be powerful, their purpose malevolent or the account opposed to an official version. Those features are common, not universal. None settles the evidence. The same allegation can begin as a live hypothesis, become documented through converging records, remain unresolved or harden into a protected belief.

For practical inquiry, those evidential states matter more than the label. A documented conspiracy has survived serious attempts at checking. A live hypothesis states who coordinated, what they did, by what mechanism, across which period and with what expected traces. A protected theory can move its actors, mechanism or evidential rules whenever a test threatens it. The boundaries are imperfect because inquiry is a process. That is the advantage: claims can move in either direction as evidence changes.

Suspicion itself is protective. Human beings depend on testimony but cannot accept every speaker at face value. We inspect motives, compare accounts and notice when words and interests diverge. This is epistemic vigilance: the ordinary work of deciding whom and what to trust. Children learn that people can deceive them. Adults know that institutions can present self-serving accounts. A person who never suspects hidden motives would be easy to exploit.

History supplies enough reasons to keep the capacity alive. The FBI's COINTELPRO operations used covert methods against political organisations. The Watergate cover-up involved White House officials, campaign money, false accounts and pressure on investigators. In the Tuskegee syphilis study, participants were denied informed consent and later denied effective treatment. The common retelling that doctors deliberately infected them is false, but the documented misconduct is severe enough without invention.

The problem is calibration. Every system of vigilance balances two errors. Trust too readily and concealed wrongdoing passes. Suspect too readily and coincidence becomes design, disagreement becomes collusion and innocent people acquire invented motives. No rule can remove both risks. The question is whether suspicion becomes more discriminating as evidence arrives. Whistleblowers, inspectors, courts, journalists and rival agencies matter because they create routes by which hidden conduct can leave the control of the people hiding it.

A sound investigation can rise or fall in confidence. It distinguishes what is known from what is inferred and what remains possible. A protected theory moves in one direction. New material adds actors and mechanisms; contrary material expands the cover-up. The difference lies less in whether the first suspicion was socially approved than in whether inquiry can still change its destination.

This distinction also protects dissent. A critic should not have to prove a complete plot before questioning conflicts of interest or secrecy. Inquiry can begin with a narrow concern and remain narrow. The pressure to choose between total trust and total conspiracy is itself an error.

Events Become Plots When Agency Beats Accident

Events do not arrive labelled with their causes. They arrive as outcomes, fragments and delays. A plane disappears. A candidate survives a scandal. A disease spreads unevenly. A camera fails at the worst moment. The mind has to build a cause from traces, and hidden agency is one possible answer.

Agency has an explanatory advantage: it joins facts. If several events came from one plan, their timing and direction belong together. An intention also gives the story a centre. Someone wanted the result. Once that actor is supplied, accidents become moves, omissions become decisions and beneficiaries become suspects. The world may become more threatening, but it becomes less shapeless.

This is why conspiracy explanations often compress systems. Modern outcomes emerge from institutions containing competing offices, contractors, regulations, informal habits and people working with partial information. Failures can be produced by no single commanding mind. A regulator misses a warning, a company delays, a minister avoids ownership and a database does not connect. The combined result can look coordinated because the harms align. Sometimes it is coordinated. Often common incentives and institutional blindness produce similar conduct without a secret meeting.

Several proposed biases help explain the pull. People detect patterns in noisy information. They infer agents behind events. They sometimes assume that great consequences require proportionately great causes. Yet these tendencies are easy to sloganise and harder to measure. Experimental links between illusory pattern perception and conspiracy belief are smaller and less consistent than popular accounts suggest, especially when compared with paranormal belief. A laboratory task involving random coin sequences is not a miniature political crisis.

The stronger claim is functional. Agency explanations become attractive when they improve coherence relative to the alternatives available to the person. An official report may list procedural failures but leave motive, injustice and responsibility unsatisfied. A conspiracy account gives all three at once. It can therefore feel like a deeper explanation even when its factual basis is poorer.

The proportionality intuition adds force. When an event changes millions of lives, a minor cause can feel insulting. A leader's assassination by one gunman, a market crash emerging from distributed incentives or a pandemic beginning with one transmission chain can seem too small for the outcome. A larger plot appears to respect the scale of the consequence. Nature and complex systems have no obligation to respect that symmetry.

Agency also clarifies blame. A systemic explanation can sound as though nobody is responsible, even when many people made culpable choices. The choice is not always plot or accident. Distributed actors can conceal, evade and reinforce one another without one central command. A good account can assign responsibility at several levels while refusing to invent a director for the whole result.

Coincidence is also emotionally expensive. Any large archive contains anomalies. Thousands of photographs, calls, movements and statements guarantee apparent alignments after the event. Searching backwards makes rare combinations look designed because the search began with the outcome. The number of unrealised patterns disappears from view.

A clue therefore needs a comparison. How unusual would this fact be if the alleged conspiracy were false? A blurry frame, deleted email or changed timetable may look suspicious, but complex organisations produce missing records and altered plans every day. The relevant quantity is not how striking the clue feels in isolation. It is how much more likely the clue would be under the plot than under ordinary failure. Does the theory predict it before discovery, or merely absorb it afterwards? Could shared incentives, error, selection or chance generate the same pattern? Hidden agency is sometimes the best answer. It earns that status by beating alternatives, not by making the fragments feel like a story.

Uncertainty Creates Demand, Not Proof

Conspiracy theories flourish around questions that matter and remain unsettled. Sudden death, war, epidemic, economic loss and rapid social change create information gaps at the moment people most want control. Authorities speak before evidence is complete, revise early claims and disagree in public. The revisions may be signs of honest learning. To an already distrustful audience, they look like a script coming apart.

The obvious explanation is that anxiety and lack of control cause conspiracy belief. The evidence is less tidy. A large 2025 synthesis combined hundreds of studies and found modest average associations across epistemic motives, existential concerns and social motives, with wide variation among measures and settings. Experimental findings were weaker than correlational ones. People who feel threatened may become more receptive, but the same data can also reflect the reverse direction or a third condition that produces both distrust and distress.

This matters because a crisis is not a psychological switch. Most people experiencing uncertainty do not adopt a conspiracy theory. Many seek reliable information, defer judgement, help neighbours or focus on immediate tasks. Some believers held the relevant worldview before the crisis and used the new event as material. Others endorse one claim while rejecting the rest. The content must match an available enemy, source network and grievance. Explanations need conditions, not a universal panic mechanism.

Uncertainty can still change what a theory offers. It provides closure by naming a cause. It restores a form of control by turning passive confusion into investigation. It can preserve moral order by making suffering the product of wicked intention rather than indifferent chance. Each benefit is psychological rather than evidential. Feeling less lost after adopting an account does not make the account more likely.

Distrust complicates the picture because it can be earned. Communities remember deception, neglect and unequal treatment. The Tuskegee study is an obvious American example: United States Public Health Service researchers concealed the study's nature and withheld treatment after penicillin became standard. Such history changes the prior plausibility of official misconduct. It does not establish any later allegation, but it makes demands for unearned trust unreasonable.

Institutional behaviour can then feed the loop. Secrecy justified by security, commercial confidentiality or legal caution leaves spaces that outsiders fill. A delayed correction looks reluctant. A polished statement that admits no uncertainty sounds managed. Officials may fear that acknowledging error will damage confidence, yet refusal to acknowledge error damages the process by which confidence could return.

Trust is therefore neither a personality trait nor a blank cheque. It is a judgement about competence, honesty, incentives and correction. People may trust a hospital on surgery and distrust a ministry on procurement. Treating all distrust as one variable hides the institution, history and subject that gave it shape.

The correct inference is conditional. Uncertainty raises demand for explanation. Low control can make agency and action more appealing. Distrust lowers the authority of official correction. Timing matters as well. Early in a crisis, evidence is thin and revision is normal; later, a community may remember only that experts changed their story. None of these conditions supplies proof of a plot. They describe the market into which a theory enters and help explain why the same claim lands differently across audiences.

That distinction prevents condescension. A believer may be responding to a genuine information failure, a history of exclusion or an institution that speaks badly. The response can still be wrong. Respecting the conditions that made a claim attractive does not require accepting the claim. It makes possible a better question: what would a trustworthy process need to show, and what evidence should move us meanwhile?

The Theory Defends a Social Position

A belief can describe the world and locate the believer inside it. Conspiracy theories often do both. They divide people into those who see, those who sleep and those who deceive. Knowledge becomes a social position: awake, independent, brave, loyal or morally clean.

That position can repair status. A person ignored by experts may become an investigator among peers. A group that feels humiliated can reinterpret defeat as evidence that hidden powers feared it. Political failure then proves neither poor strategy nor limited support; it proves sabotage. The theory preserves competence and innocence by moving causation outside the group.

Research on collective narcissism captures one part of this. Groups can combine a grand view of their importance with a conviction that outsiders fail to recognise it. Across many countries, that pattern has been associated with endorsement of conspiracies against the nation or group. The association does not mean that strong group attachment is pathological. Secure identification can coexist with criticism. The dangerous form makes recognition permanently owed and opposition inherently suspect.

Partisan fit changes which theories attract belief. In six studies conducted in the United States and Israel, a general tendency towards conspiratorial thinking predicted endorsement more strongly when a political theory accused the respondent's opposing camp than when it accused their own. The finding is bounded to two political systems and survey measures, but it exposes the interaction. A general appetite for plots does not choose the villain alone. Coalition and content work together.

A different reward is uniqueness. Some people value possessing knowledge that the majority rejects. Experiments and surveys have found a modest link between this need and conspiracy endorsement. Minority status can make a claim more appealing because disbelief by others becomes part of its value. Yet uniqueness is no master cause. The same person may follow consensus in medicine and reject it in politics, or adopt a popular conspiracy inside one community where it no longer feels dissident at all.

Identity determines which villain fits. Conspiracy theories are not politically interchangeable. They draw on available histories of enemies, hierarchies and sacred values. The forged Protocols of the Elders of Zion portrayed an invented Jewish council directing world affairs. Its contradictions were useful because Jews could be blamed for finance and revolution, cosmopolitanism and tribal separation. The document did not discover prejudice. It organised and licensed it.

Scapegoating converts a diffuse problem into a responsible out-group. Economic disruption, demographic change or institutional failure becomes the work of a population portrayed as coordinated. Individual difference disappears. Every member can be read as an agent or beneficiary of the plot. Violence then presents itself as defence.

The social function can make correction costly. Abandoning a claim may threaten friendships, political identity and a story of personal awakening. A request for evidence is heard as a demand to rejoin the duped majority. The factual proposition has acquired moral cargo. Political entrepreneurs exploit this by making doubt a loyalty test, while leaders can use conspiracy claims to deflect blame or present themselves as uniquely able to see hidden enemies. Conspiracy is therefore not confined to people challenging power. It can be an instrument of power.

This does not mean believers secretly know the claim is false and choose it for comfort. People can sincerely believe an account that also protects identity. They can be uncertain, share material for discussion or perform certainty before possessing it. Surveys that force a yes-or-no answer flatten that range. Motives need not be conscious, and benefits do not prove causation. The practical lesson is narrower: changing a belief may require a dignified route out. A person needs to be able to revise without accepting that they were stupid, cowardly or disloyal. Where a group rewards escalating certainty, that route becomes hard to find.

Belief Is Built With Other People

Few people construct a large conspiracy theory alone. They receive fragments through friends, broadcasters, influencers, activists, relatives and strangers who appear to share their concerns. Trust attaches first to people, then travels to claims.

A government report may carry technical authority and little relational authority. A nurse, veteran or parent may have no direct knowledge of the event but sound like someone who would not lie. Social credibility is not irrational because much knowledge must be borrowed. The weakness appears when trust in a person substitutes for evidence that the person could possess.

Belief also has a supply side. Somebody chooses an enemy, selects the anomaly and puts the first interpretation into circulation. The supplier may be sincere, commercially rewarded, politically strategic or all three. In one US study, members of Congress were more likely to promote conspiracy theories before contested primary elections and less likely afterwards; individual-donor fundraising was higher after promotion. The setting is narrow, but the result shows why public demand is only half the market.

Power can use conspiracy as well as attract it. After the 2023 earthquakes in Türkiye, analysis of social-media activity found that supporters of the incumbent president preferred earthquake conspiracy accounts. A nationally representative face-to-face survey found a correlation between endorsement and support for the governing coalition. Because the design is observational, it cannot establish that the beliefs altered voting behaviour. It does show that conspiratorial blame can protect authority rather than merely attack it.

Digital conspiracy cultures turn reception into participation. QAnon offered cryptic posts rather than a complete doctrine. Followers interpreted clues, compared timestamps, searched records and circulated screenshots. The activity resembled collaborative investigation and gave participants a role in producing the account. They were making connections and teaching one another how to see.

Participation changes commitment. A claim discovered through one's own search feels earned. Hours spent assembling a case become investment. Public predictions create reputation. Senior interpreters gain status by connecting new material to the shared framework. A theory can persist because a community is good at generating work, even when it is poor at producing discriminating evidence.

The internet changes speed, scale and memory. Search retrieves confirming fragments from different years and places. Likes, shares and comments make attention look like agreement. Recommendation can lead towards adjacent material, while moderation can move communities to spaces where opposition is rarer. The user also chooses, follows and interprets. The feed is neither sole author nor neutral pipe, and exposure does not automatically produce belief.

Older media performed related work. Pamphlets, sermons, newspapers, political clubs and radio connected suspicious publics long before feeds. Late eighteenth-century writers blamed the French Revolution on secret philosophical and Illuminist coordination. The Protocols moved through print, translation and political sponsorship. The internet did not invent hidden knowledge or the economics of alarm.

Communities edit theories. Claims mutate after failed predictions, merge with older enemies or split into factions. One member supplies a motive, another a symbolic clue, another testimony. Distributed authorship makes responsibility hard to locate and lets participants disown the weakest version while retaining the wider suspicion. Visible agreement can also exaggerate consensus: dissenters leave, lurk or are removed, while prolific contributors supply most of the apparent crowd.

Belief is therefore a relationship among a claim, its suppliers, a source network and an audience. Correcting one sentence may leave that network intact. A durable response asks what the community supplies: explanation, companionship, status, action or a language for injury. Facts remain necessary. They work inside a social structure that decides who may deliver them and what accepting them will cost.

Evidence Enters an Asymmetric Machine

Every explanation selects evidence. A doctor weighs some symptoms more than others. A historian distrusts a self-serving memoir. A scientist protects a theory from one faulty measurement. Revision does not require surrender at the first awkward fact.

Conspiracy theories become distinctive when the selection rule is asymmetric. Evidence that fits can be weak, anonymous or indirect. Evidence that conflicts must meet a higher standard, and its failure is attributed to corruption. The claim receives the benefit of possibility; the alternative must deliver certainty.

The machine often begins with accumulation. Ten suggestive items feel stronger than one. Yet quantity helps only when the items are independent and each has evidential value. Ten websites can repeat the same original rumour. Three experts can cite one disputed dataset. A video compilation can combine unrelated anomalies selected from thousands of ordinary moments. Search itself selects: typing the suspected plot into a search bar asks for material that fits it. Source laundering turns repetition into the appearance of corroboration.

Motivated reasoning adds direction. People scrutinise hostile claims and accept friendly ones more readily, especially when identity or moral stakes are high. Education can improve this scrutiny without making it even-handed. A skilled reasoner may find more objections to unwelcome evidence and more sophisticated defences for a desired account. Intelligence is a tool; the goal decides where it is applied.

Disconfirmation then changes category. If a predicted arrest does not occur, the operation was delayed. If a document is absent, it was destroyed. If insiders deny the plot, their denial shows discipline. If institutions agree, coordination explains the agreement. If they disagree, the confusion is staged. Each repair is possible. Their cost lies in how many new assumptions are added without creating new tests.

This is why self-sealing should be treated as a matter of degree. Ordinary theories also contain auxiliary assumptions. A failed experiment may reflect broken equipment rather than a dead theory. Investigators judge whether the rescue is independently supported and whether it yields further predictions. The defect is not surviving one contrary fact. It is surviving every possible outcome at no loss of confidence.

Claims that believers endorse mutually contradictory conspiracies illustrate the need for care. Positive correlations can arise because respondents reject the official account while giving low endorsement to several alternatives. The shared belief may be “the public story is false”, not confident acceptance of two incompatible plots. That is still epistemically important, but it is not the comic irrationality often advertised.

A theory improves when it risks something. It specifies what should be observed if the claim is true, what would be unlikely if it were false and which result would reduce confidence. It distinguishes the central allegation from disposable details. It allows sources to be credible on one point and wrong on another. This is a demand for comparative evidence, not mathematical decoration. A hundred weak clues do not become strong if they all arose from the same selection process. One reluctant witness, authenticated ledger or independently obtained recording may carry more weight because the source path differs and the evidence was costly to produce.

Without those rules, research becomes collection rather than discrimination. Every search can add material because the theory controls the meaning of whatever appears. The investigator feels increasingly informed while the range of possible conclusions shrinks to one.

Suspicion Needs a Stopping Rule

Watergate is useful because it shows what justified conspiracy belief looks like. Five burglars were arrested inside the Democratic National Committee headquarters. Investigators traced money. Officials gave conflicting accounts. Witnesses testified. White House recordings existed outside the control of the journalists who first pursued the story. Courts, Congress, prosecutors, newspapers and participants had different incentives, yet evidence converged. The conspiracy became more specific as it became better supported.

A stopping rule does not mean inquiry must end at official reassurance. It states what would raise, lower or settle confidence. It identifies the decision being made and the evidence worth waiting for. It prevents the inquiry from becoming a permanent identity whose only output is another clue.

Coordination creates expected traces. People communicate, transfer money, move objects, create schedules and make mistakes. Large, long conspiracies face leakage, turnover and conflict. Secrecy can be strong where the group is small, the task narrow, records limited and penalties severe. Compartmentalisation can hide a larger operation from many participants, though somebody still has to connect the compartments and the operation still produces effects. A theory requiring thousands of independent clinicians, laboratories, judges and journalists to maintain one precise deception for decades has a larger burden. The number alone does not refute it, but it changes what evidence the claim owes and how much independent convergence should appear.

The cost of weak rules appears when belief authorises conduct. Edgar Welch treated an invented trafficking story as sufficient reason to enter a restaurant armed. Anti-vaccine conspiracy material has reduced stated vaccination intentions in experiments. Surveys often find links between specific conspiracy beliefs and support for political violence, although such evidence cannot establish that belief alone caused later action. Many theories remain conversational or recreational. Harm depends on target, instruction, intensity and context.

False accusation also has unequal victims. Claims about hidden paedophiles, ethnic control or medical murder attach danger to named people and groups. The accused cannot disprove every secret act, since the allegation defines missing evidence as concealment. A theory that began as distrust of power can become permission to punish someone with less power.

Correction works best when it restores discrimination. Across misinformation research, clear factual rebuttals usually improve accuracy; the famous claim that correction routinely makes people believe falsehood more strongly has been overstated. Improvement may be incomplete, short-lived or limited to the measured item. Technique-based prebunking can improve immediate recognition of methods such as scapegoating, false dilemmas and emotional manipulation in controlled studies. Those trials do not establish permanent immunity in public life. Conversation can identify the source or value at stake. No method guarantees conversion.

Institutions carry a separate duty. They should publish inspectable evidence, admit uncertainty, explain revisions and permit independent challenge. Trust cannot be demanded as the alternative to conspiracism. It has to be made rational enough to extend provisionally.

Core Idea 1 began with a defence of suspicion. The loop closes here. Genuine conspiracies are exposed through shared standards that let independent evidence converge and let weak allegations fail. A self-protecting theory corrodes those standards by making every institution part of one actor and every result compatible with the plot. Suspicion then defeats its own purpose.

Keep the capability. Add the stopping rule. Power should know it can be investigated, and the investigator should know what could prove the investigation wrong.

How It Actually Works

A gun in the restaurant

Edgar Maddison Welch carried three firearms from North Carolina to Washington on 4 December 2016. He entered Comet Ping Pong with an AR-15 rifle visible in his hands and a revolver. Customers and employees fled. At a locked room he tried a butter knife, then fired the rifle through the door. After twenty minutes he left the weapons inside and walked out unarmed. Police arrested him. Nobody had been shot.

The conspiracy theory Welch had come to investigate claimed that political figures were connected to a child-trafficking ring centred on the restaurant. Online investigators had read ordinary references in leaked emails as code, joined them to photographs, art, social connections and older stories of elite abuse, then circulated the result as a case assembled from public evidence. The restaurant did not contain the operation described.

Welch's conduct exposes the full mechanism because it crossed every boundary. A morally urgent subject supplied the stakes. Genuine elite wrongdoing made the category plausible. Leaked documents supplied authentic raw material. Ambiguous words became coded signals. Users linked separate domains into one design. Repetition and collaborative searching created the appearance of corroboration. Denials came from institutions already placed inside the suspected network. The theory implied that official delay left children in danger, so private action acquired moral permission.

He did not need certainty. He needed enough confidence to drive, arm himself and open a locked door. That is where belief becomes visible. Surveys record agreement. Conduct reveals which uncertainty a person is prepared to impose on somebody else.

The gap opens

A conspiracy theory usually enters through a gap between an event and an account people accept. The gap may be factual: nobody yet knows why a system failed. It may be moral: the available explanation does not name anyone proportionate to the harm. It may be social: the speaker offering the explanation is not trusted.

Crises enlarge all three. Early information is fragmentary. Officials correct estimates, change guidance and disagree. News rewards novelty before verification is complete. Images circulate without dates or location. A person watching from outside sees the revisions but not the meetings, measurements and arguments that produced them.

The first theory does not have to explain everything. It needs to make one unsatisfied feature legible. Why was a camera off? Why did one official sell shares? Why was this word used? Why did the policy resemble an earlier proposal? A question can be responsible. The conversion begins when the absence of an immediate answer is treated as positive evidence for one hidden answer.

Real institutional opacity helps. Governments classify records. Companies settle without admitting liability. Police protect colleagues. Scientific work contains specialist language and provisional results. Legal advice strips public statements of candour. The official account may therefore be incomplete, evasive or wrong without one central plot controlling it.

The gap is an opening, not proof. What happens next depends on which fragments are available, who interprets them and whether the inquiry permits more than one destination.

The claim is supplied

A complete plot rarely appears from nowhere inside one listener. Somebody names the hidden actor, selects the suspicious detail and offers a story in which the pieces belong together. The supplier may be an activist, politician, broadcaster, anonymous account or friend. Their incentive can be conviction, attention, money, coalition-building or protection from blame.

This machinery predates digital media. After the French Revolution, Augustin Barruel and John Robison argued that philosophical networks, Freemasons and the Bavarian Illuminati had coordinated the upheaval. The Illuminati had existed. The leap was making one secret society the commanding cause of a revolution shaped by fiscal crisis, war, institutional breakdown and popular action. Real secrecy gave an excessive explanation its anchor.

Political supply matters because conspiracy theories can challenge authority or defend it. After the 2023 earthquakes in Türkiye, one study combined social-media analysis with a nationally representative face-to-face survey. Supporters of the incumbent president showed greater preference for earthquake conspiracy accounts, and endorsement was associated with support for the governing coalition. The evidence is observational and tied to one event. It cannot show that the theories caused votes. It shows that hidden-enemy stories can redirect responsibility away from those already in power.

A leader cannot make every audience believe every accusation. The story must fit existing identities, enemies and source relationships. Studies in the United States and Israel found that conspiratorial thinking predicted partisan conspiracy belief more strongly when the allegation targeted the opposing camp. Claim and coalition met halfway.

Digital media lowers the cost of the offer and lets audiences modify it. That is an acceleration, not an origin. The question is no longer only why a person wanted an explanation. It is who supplied this one, why now and what they gain if the audience adopts it.

An anomaly becomes a clue

An anomaly is an observation that does not fit the current account. A clue is an observation interpreted as evidence of a particular cause. The passage between them is where many theories are built.

Consider a missing frame in a video. It may result from editing, transmission, storage failure, a damaged file, a privacy decision or deliberate removal. Calling it an anomaly preserves the alternatives. Calling it a clue to suppression assigns intention. The observation has not changed; its role has.

Online archives make anomalies abundant. A person can search thousands of photographs for gestures, dates, logos and background objects. Every public figure has misspoken, changed position, met an unpleasant person and appeared in an image whose context is unclear. Large datasets guarantee coincidences. Selection removes the ordinary material and presents the rare alignment alone.

Retrospective search adds a second distortion. After an outcome, investigators know what to seek. A phrase written years earlier can be read as prediction. A policy paper becomes a plan for the later event because some words overlap. The many proposals that were never enacted vanish from the comparison.

A discriminating clue should alter relative probabilities. It should be more expected if the alleged plot exists than if it does not. A politician knowing a donor is common under both possibilities. A secret payment tied to a documented request is different. A changed timetable may be routine. A timetable changed by someone who then lies about the reason deserves more weight.

Conspiracy research often keeps the intensity of the anomaly and skips the comparison. The locked door looks sinister because the theory has already supplied what lies behind it.

A villain compresses the system

Complex causes are cognitively expensive. A financial collapse can involve leverage, incentives, regulation, modelling errors, herd behaviour, liquidity and political choices spread across years. A pandemic can involve biology, transport networks, hospital capacity, unequal exposure, changing data and local policy. No actor controls the whole field.

A villain reduces the number of moving parts. The same group caused the event, hid the evidence, paid the experts and benefits from the response. Apparent contradictions can be assigned different functions inside one plan. What looked like institutional fragmentation becomes division of labour.

This compression supplies moral clarity. Systems can distribute responsibility until every participant says the decisive choice belonged elsewhere. A conspiracy account refuses that escape and insists that somebody intended the harm. The insistence can identify real collusion. It can also turn shared incentives into secret command. Firms may copy profitable conduct without meeting. Officials may repeat one another because they depend on the same data. Journalists may converge because sources and deadlines are similar.

The chosen villain comes from culture. Anti-Jewish conspiracy theories did not emerge from neutral analysis of finance or revolution. They reused an available out-group and portrayed difference as coordination. The Protocols, a fabricated antisemitic text, supplied fictional minutes of a fictional council, giving prejudice the appearance of documentary proof. Exposure of the forgery did not remove its political use because the text's function was accusation, not investigation.

Other theories target intelligence agencies, pharmaceutical companies, bankers, immigrants, communists, globalists, secret societies or a vague elite. Some of those actors possess power and sometimes conspire. The error lies in allowing the category to explain each event before the case-specific mechanism is shown.

Once the villain is in place, beneficiaries become authors. Anyone who gained is presumed to have planned. Yet outcomes create winners who did not cause them, and planners often fail to benefit. Motive begins an investigation. It cannot complete one.

The investigators arrive

A modern conspiracy theory can offer more than an explanation. It offers a job.

QAnon began with cryptic posts by an anonymous figure claiming privileged access. Followers called fragments “crumbs” and their interpretive work “baking”. They searched speeches, maps, timestamps, news reports and earlier posts, then assembled “proofs” in collages, documents and videos. The corpus invited endless close reading because vague questions and compressed phrases could connect to many later events.

Ethnographic research on one QAnon community found active source evaluation rather than passive acceptance. Participants criticised clickbait, rejected material they considered too fringe and demanded links. They also treated selected texts as privileged and interpreted apparent failures within a political framework that protected the source. Research technique and conclusion were joined by community rules.

That combination explains why generic advice to think critically can fail. Participants already understand themselves as critical thinkers. They compare sources, search archives and detect manipulation. The problem lies in what receives scrutiny, which sources begin with authority and whether the method can return a conclusion the community dislikes.

Participation creates rewards. A newcomer can bring a screenshot. An experienced interpreter can connect it to older material. Recognition arrives through replies and sharing. The work supplies companionship and purpose. It can convert spectatorship into a heroic role: ordinary users are helping expose an enemy that formal institutions refuse to confront.

The process is not identical across theories. Some spread through family conversation, religious networks, wellness communities or partisan media with little collaborative decoding. Others are consumed with irony or partial belief. QAnon is useful because its knowledge-making machinery was unusually visible, not because every believer becomes a baker.

The investigator role deepens the commitment. A claim received from another person can be dropped. A case built through one's own labour feels owned.

The case file grows

Accumulation feels like investigation. A folder contains fifty screenshots, six videos, three interviews and a map of connections. Its weight seems to measure the weight of the case.

The missing question is independence. A news article may cite a social post that cites a video whose creator read the article. Ten accounts may repeat one anonymous allegation. Several photographs may come from the same selected event. The apparent sources form a circle rather than ten lines reaching the claim from different directions.

Digital presentation hides this ancestry. A copied sentence loses its first date. A cropped image loses its caption. A compilation places weak items beside stronger ones, allowing credibility to bleed across. The viewer sees variety of format and mistakes it for variety of origin.

The file also grows through domain crossing. A restaurant logo, a charity, an email phrase, an artwork and an old criminal case may have no causal relation. A theory gives them one. Research on Pizzagate has contrasted this sprawling alignment with documented conspiracies that remain tied to a narrower domain of actors, communications and acts. Genuine schemes can be wide, but width creates more specific connections to prove.

Search results strengthen the asymmetry. Searching the exact allegation retrieves people discussing the allegation. Their number looks like evidence that something happened rather than evidence that a phrase circulated. Recommendation systems supply adjacent material because similarity predicts attention, not truth.

A healthy case file should shrink as well as grow. Duplicate sources are collapsed. Claims that fail are removed. Alternative causes receive their own evidence. Confidence changes when a central prediction misses. Without deletion, the archive becomes a machine for turning time into certainty.

Denial changes the source

The restaurant denied the allegation. Police called it fictitious. Journalists found no evidence of the claimed operation. Within an ordinary investigation, several independent failures should lower confidence. Within a protected theory, they may alter the status of the institutions instead.

This is the self-sealing turn. The police are compromised. Journalists protect their class. Missing records were destroyed. A witness is frightened or paid. The lack of arrests proves the network's reach. Each explanation is possible in some case. Together they ensure that no public outcome can count against the theory.

Failed predictions receive similar repairs. The date was symbolic. The operation was postponed. Exposure forced the conspirators to change course. The prediction was designed to confuse enemies. The interpreter, rather than the source, misunderstood. A movement may quietly discard one claim while preserving the authority that generated it.

All inquiry uses auxiliary explanations. Watergate investigators did not abandon the case whenever an official denied involvement. Scientists do not discard a strong theory because one instrument malfunctions. The question is whether the rescue has independent support and creates new risk. A documented broken instrument can explain one result. An invisible saboteur invented after every result explains nothing discriminating.

A useful audit therefore records rescue costs. How many new actors, exceptions and hidden mechanisms have been added? Did the change produce a test? Has any result lowered confidence? If a theory survives both an event and its opposite, prediction has become decoration.

The social network can enforce the seal. Members who question a central source may be labelled infiltrators, shills or people who have not researched enough. Exit becomes further evidence that the enemy can reach anyone. The theory now protects the community that protects the theory.

Belief leaves the screen

Most conspiracy belief does not end with a rifle. It changes smaller decisions: which doctor to trust, whether to vote, where to send money, whom to avoid and which institutions deserve cooperation.

Health claims show the route. A theory may begin with justified concern about corporate incentives or regulatory failure. It then attributes adverse events to one concealed plan, treats safety monitoring as theatre and reads correction as evidence of capture. Experiments have found that exposure to anti-vaccine conspiracy material can reduce stated vaccination intentions. That result concerns an intention measured under study conditions, not automatic refusal in a clinic. The distinction matters because belief, intention and action are separate outcomes.

Political effects also vary. Conspiracy belief can reduce confidence and conventional participation, but it can motivate campaigning, protest and voting. A theory that portrays an election as already controlled may encourage withdrawal in one group and urgent mobilisation in another. There is no single behaviour called the consequence of conspiracism.

The target and instruction matter. A theory about a remote historical mystery may function as entertainment. A claim that a named ethnic group controls the state can organise exclusion. A claim that officials are concealing a dangerous treatment can change medical conduct. A claim that children face immediate danger can licence rescue fantasies against real people.

Observational studies have linked endorsement of many conspiracy theories with self-reported support for political violence. Such correlations cannot show that the theory caused the attitude, and different claims carry different risks. They do show why content cannot be dismissed as private eccentricity when it supplies enemies, urgency and moral exemption.

Belief can also damage relationships. Relatives may experience repeated correction as contempt; believers may read disagreement as proof that loved ones are asleep or captured. The theory becomes a loyalty test. Losing the claim can then mean losing a social world.

What a real conspiracy leaves behind

A real conspiracy is hidden by people who face limits. They need to communicate, move resources and coordinate action. Participants possess unequal information and loyalty. Rivals, auditors, journalists, courts and accidents can expose traces.

Watergate began with burglars caught in an office. The case expanded through campaign finance, testimony, institutional conflict and White House recordings. Evidence did not depend on one interpreter. Separate actors authenticated pieces, challenged them and compelled disclosure. The account distinguished the break-in, cover-up, obstruction and degrees of participation instead of making every political opponent part of one timeless cabal.

COINTELPRO offers another lesson. The programme's existence is established through FBI records, later investigation and documentary archives. Knowledge of it should increase prior suspicion that security agencies may target political groups. It should also improve standards: which office, operation, target, date, authority and record are alleged now?

The Tuskegee study shows why factual correction and moral seriousness must coexist. Researchers did not deliberately infect the men. They recruited men who already had syphilis, failed to obtain informed consent and withheld effective treatment after it became available. Correcting the infection claim does not defend the study. It protects the documented wrongdoing from a false detail that can be used to dismiss the whole history.

Genuine cases often become narrower as they strengthen. Named actors replace “they”. Dates replace permanent plots. Documents separate knowledge from participation. Some suspected people are cleared. The account becomes less total and more actionable.

A conspiracy theory tends towards the opposite shape. It absorbs domains, makes the actor harder to locate and interprets failed detail as proof of deeper power. The practical difference is not that one sounds respectable. It is whether evidence reduces the set of possible claims.

How we know

Research on conspiracy belief combines surveys, experiments, repeated polling, content analysis, interviews, ethnography and historical work. Each sees a different object. Agreement with a questionnaire item may express suspicion, partisan loyalty or possibility rather than settled certainty. It is not the same as sharing a claim or acting on it. Experiments can isolate effects but often use artificial material, restricted samples and immediate outcomes. Correlations among distrust, anxiety, identity and belief do not establish direction of cause.

The strongest motive synthesis pooled 971 effects from 279 independent studies and found modest average associations with epistemic, existential and social motives, alongside substantial heterogeneity. That scale supports plural explanation. It cannot identify one person's reason.

Cross-national studies now cover dozens of countries, and recent work in Türkiye and Israel extends a literature dominated by the United States and western Europe. Translation, sampling and political context still limit comparison. A scale may not measure the same social act everywhere, and one country's elite cues do not become a universal mechanism.

Digital traces overrepresent people who post. Ethnographies reveal communities. Historical prevalence depends on changing questions and uneven archives. The evidence is strongest that conspiracy belief is heterogeneous, socially organised and capable of consequence. It is weaker on durable intervention effects and on how the balance among motives, suppliers and institutions travels across cultures.

What People Get Wrong

“Only stupid people believe conspiracy theories”

This is attractive because it turns disagreement into diagnosis. It is also a poor model. Conspiracy endorsement has been associated with lower analytic reflection and some measures of cognitive ability, but effects vary and explain little about any individual. Education does not confer immunity. It can improve source checking, statistical reasoning and knowledge of institutions; it can also provide vocabulary and argumentative skill for defending an identity-linked conclusion.

Belief is usually specific. A person may reason carefully at work, trust engineering standards and accept one political conspiracy shared by people they respect. Another may reject several popular plots and endorse a medical one after personal mistreatment. Calling both stupid erases source, grievance and topic.

The insult also creates a practical failure. Revision now requires accepting a damaged identity as well as a factual correction. The person can defend dignity by defending the claim. Better analysis asks which evidence rule failed, what the belief supplies and what would permit change. Intelligence affects reasoning capacity. It does not choose the goal towards which reasoning is directed.

“Conspiracy theories are false by definition”

If this were true, Watergate would have stopped being a conspiracy theory only on the day enough people accepted it. The language would classify claims by reputation rather than content and evidence.

A conspiracy theory gives hidden coordination a central causal role. That description does not settle whether the actors coordinated. A claim may be documented, remain a live hypothesis, fail for lack of evidence or become protected from every possible test. These are evidential states, and one allegation can move among them as records, testimony and alternatives change.

Some scholars reserve the term for explanations with further features such as powerful actors, malevolent goals or opposition to an official account. Others build poor evidence into the definition. No taxonomy removes the need to inspect the case, and a loaded label can hide the judgement that requires support.

The burden remains claim-specific. A narrow allegation supported by records can strengthen while a wider story built around it fails. Evidence that officials concealed one act does not establish that they directed every related event or silenced every independent institution. A theory should become more precise as it becomes better supported.

The correction matters in both directions. Dismissing every allegation enables secrecy. Treating the existence of past conspiracies as support for a new one commits the reverse error. Governments have lied; therefore this government may be lying. The conclusion begins inquiry. It does not settle which claim, by whom, through what mechanism and with which traces.

“The internet created the problem”

Conspiratorial explanation travelled through sermons, pamphlets, newspapers, books, clubs and broadcast media for centuries. Secret-society accounts of the French Revolution, antisemitic forgeries and accusations around presidential assassination did not require recommendation algorithms.

The internet changed the machinery. Publication costs fell. Archives became searchable. Strangers could assemble evidence collages, teach interpretive rules and maintain communities across distance. Platforms made attention visible and allowed one theory to absorb another at speed. Bots and commercial actors can enlarge reach.

Those changes do not mean exposure equals belief or that every theory has become more common. Repeated polling of named claims has found increases, stability and declines rather than one universal rise. The evidence is geographically and historically limited, but it defeats a story of automatic technological escalation.

Blaming the internet alone also hides demand. People select sources, follow trusted figures and seek accounts that fit prior grievances. Institutions create opacity. Political entrepreneurs supply enemies. Platforms shape circulation inside that wider system. Removing a post can reduce reach; it cannot by itself repair the distrust and identity that made the account useful.

“A crisis automatically makes people conspiratorial”

Threat and uncertainty can increase the appeal of intentional explanations, but crisis is a condition, not a button. Most people facing epidemics, war or economic shock do not adopt conspiracies. They may defer judgement, follow authorities, rely on family or focus on immediate survival.

The strongest synthesis of psychological motives finds modest average relationships and substantial variation. Experimental effects are weaker than correlational ones. This is what one would expect from several routes operating together: pre-existing distrust, identity, information gaps, threat, source networks and political opportunity.

Crises do create fertile material. Evidence changes quickly. Officials contradict one another. Early estimates fail. The public may see revision without seeing the data that required it. A theory can preserve certainty by interpreting every change as concealment.

The correction matters because crisis communication should not assume one frightened public. Some audiences need clear uncertainty ranges. Others need proof that harms and conflicts of interest are being investigated. People with justified historical distrust need more than commands to trust. The same event lands inside different relationships, and those relationships help decide whether uncertainty becomes inquiry, patience or plot.

“Believers accept every conspiracy theory”

Research often finds a general readiness to suspect hidden coordination. That tendency does not mean one person endorses every available claim. Specific theories are shaped by ideology, group identity, prior knowledge and the source presenting them.

A famous claim says believers will accept mutually contradictory accounts, such as two incompatible causes of the same death. Later analysis shows why that result needs care. Positive correlations can be driven by people who reject the official account while giving low or uncertain endorsement to several alternatives. Their stable position may be distrust, not full belief in both plots.

Contradiction still occurs. Large theories can hold enemies who seem ideologically opposed, or shift mechanisms after predictions fail. Communities also disagree internally and police some sources as too extreme. Treating the whole audience as indiscriminate misses this source evaluation.

The practical consequence is important. Do not argue against a bundle the person has never endorsed. Ask for the central allegation and degree of confidence. A person who says the official account is incomplete may be much closer to ordinary scepticism than someone who claims certainty about an alternative. Precision creates somewhere for evidence to work.

“Facts alone will cure the belief”

Facts are indispensable. The myth lies in alone. A correction has to be noticed, understood, retained and granted authority. It may repair one claim while leaving the source network, identity and general suspicion untouched.

The opposite myth, that correction usually backfires, is also overstated. Across misinformation research, clear rebuttals normally improve factual accuracy rather than strengthening the falsehood. The improvement can be partial and may decay. A person can update the detail while retaining the broader worldview.

Effective correction states what is false, supplies a coherent replacement and shows the evidence route. It avoids repeating a sensational claim more than needed. Where trust is damaged, the messenger and process matter. A concession of uncertainty can be stronger than theatrical certainty.

Prebunking takes a different route by teaching recurrent manipulation methods before exposure, such as scapegoating, false dilemmas and emotionally loaded language. Controlled studies show promise on recognition and immediate judgements, while durable real-world effects remain less certain. No intervention removes the social cost of leaving a community. Factual repair and a dignified route to revision have to meet.

“They are harmless private opinions”

Many conspiracy beliefs remain low-stakes. People discuss ancient mysteries, hidden technologies or celebrity rumours without reorganising their lives. Harm should not be inferred from the label alone.

The content changes the risk. A theory can lower medical intentions, identify an ethnic or religious group as a coordinated enemy, delegitimise an election or accuse a named person of abusing children. It may recommend vigilance, withdrawal, harassment or direct rescue. The distance from proposition to conduct is part of the theory.

Evidence on consequences needs the same discipline demanded elsewhere. Experiments often measure stated intentions, not behaviour. Surveys linking conspiracy endorsement with support for violence are observational. They establish concern, not a single causal pathway. Edgar Welch's armed entry into Comet Ping Pong is a documented case, not an estimate of what most believers do.

Private belief also has public spillovers below violence. It can make cooperation costly, force relatives into loyalty tests, flood innocent people with accusations and consume institutional time. The correct response is proportionate. Do not criminalise eccentric thought or inflate fringe material through panic. Attend to target, reach, urgency, instruction and the believer's movement towards action.

Use It

State the claim in full

Do not investigate a mood. Turn “something is off” into a proposition with actors, acts, timing and mechanism.

Who is alleged to have coordinated? What did each person know? What outcome did they seek? Which communication, payment, instruction or concealment connected them? Over what period? Which part is documented, inferred or possible?

Precision often reveals several claims hiding inside one. “The company knew the product was dangerous” differs from “the company designed the harm”, “regulators were bribed” and “every independent study was falsified”. One could be true while the others are false. Evidence for the narrow claim should not silently support the widest.

Record confidence as possible, plausible, probable or established. Suspecting incomplete disclosure is different from certainty about a named plot. A testable allegation should fit in a paragraph without relying on “they” or “everything”. If the actor expands whenever challenged, the object of inquiry has not been fixed.

Separate anomaly from explanation

Write the observation first, stripped of motive. The video has a missing interval. The official sold shares before the announcement. Two reports use the same wording. A witness changed one detail.

Then list several mechanisms that could produce it. Technical failure, routine practice, a shared source, legal editing, memory error, conflict of interest, negligence and deliberate concealment may remain open. They need not be equally likely. The exercise prevents one cause from becoming the observation itself.

Ask for the relevant base rate. How often do cameras fail in comparable settings? How often do officials trade under pre-arranged plans? How much repeated wording comes from templates? An event can be unusual in daily life and common inside the system that produced it.

Now ask what the anomaly establishes. Does it connect named actors or show capacity, motive, communication or action? Or does it remain unexplained? Missing explanation justifies further inquiry. It is not evidence for whichever theory supplies the most dramatic answer.

Keep the grammar clean: “the record is missing” before “the record was removed”. One preserves the question. The other spends the answer.

Ask what would count against it

Before searching, write the result that would lower confidence. Do this while the theory is still cheap to revise.

The result might be a failed prediction, an authenticated record, independent laboratories converging or a proposed mechanism proving impossible. Specify how much it should matter. A peripheral error may remove one detail. Failure at the central causal link should cost more.

Test symmetry. Would the same standard satisfy you if it supported the official account? If an anonymous witness counts for your theory, why does a named witness not count against it? If incentives discredit one study, inspect the incentives of the channel defending your view.

Keep a revision log. Record predictions when made, collapse duplicated sources and remove failed claims. A case file that can only grow is an archive of commitment.

Stopping rules can be practical. Evidence may justify a low-cost precaution without supporting certainty, publication or an accusation that would damage someone. Belief and action need not share one threshold.

Count the coordination burden

Estimate the operation required, then look for the traces an operation of that size should produce.

Count planners, informed participants, contractors, enforcers and people likely to discover the conduct through ordinary work. Separate those who need to know the goal from those who can perform one task without understanding it. Compartmentalisation reduces knowledge but adds managers, interfaces and failure points.

Ask how long secrecy must last, how often people leave, what rewards or penalties sustain silence and whether rivals can inspect the same evidence. A six-person bribery scheme lasting a month differs from a decades-long fabrication maintained across competing laboratories.

Large scale is not automatic disproof. States have run extensive secret programmes, and organisations can normalise misconduct without telling every worker the whole purpose. Scale sets the evidential debt. A larger theory should generate more communications, records, witnesses, inconsistencies and material effects outside one source network.

Compare the burden with the alternative. Distributed negligence requires no central meeting but still needs a causal path. “The system failed” should not become an escape from responsibility. Compare mechanisms rather than granting either secrecy or complexity immunity.

Trace independent provenance

Follow each important item backwards to the earliest accessible source. Do not count the same allegation several times because it appears in a podcast, article and video.

For a document, ask who created it, when, why, how it was obtained and whether the whole record is available. For an image, find the earliest file, date, location and uncropped context. For testimony, separate observation from hearsay and later inference. For an expert, check whether the expertise covers the question.

Draw a source tree. Several branches may collapse into one anonymous post, press release or dataset. Agreement among outlets using one wire report is repetition. Agreement among a physical record, a reluctant participant and an independently obtained financial trail is convergence.

Notice incentives without turning them into verdicts. Governments, companies, activists, creators and whistleblowers can gain from a claim. Incentives change scrutiny; they do not make every statement false. Apply the same rule to preferred sources.

Replace “the media”, “experts” and “researchers” with named institutions, methods and documents. Good provenance restores the disagreements, dependencies and separate routes hidden by those compressed categories.

Repair the process of trust

When another person holds the belief, find the live claim and the source carrying most weight. Do not unload every correction available.

Ask what convinced them and what would change their mind. Listen for the theory's function: unexplained harm, institutional betrayal, group defence, moral urgency or the status of independent inquiry. Understanding the function does not concede the facts. It shows why one rebuttal may miss.

Correct clearly. State what is false, show the evidence route and supply a replacement account with enough causal detail to satisfy. Distinguish documented misconduct from invented additions. The Tuskegee study was unethical because consent and treatment were denied; correcting the claim that researchers deliberately infected the men protects the documented wrongdoing.

Preserve a dignified exit. Praise the better standard now being applied rather than demanding a confession of stupidity. Admit remaining uncertainty. Do not promise that institutions never lie.

Institutions need the same repair. Publish methods and conflicts, explain changes, separate known facts from provisional judgements and permit independent review. A demand for trust is weaker than a process that can survive distrust.

Set boundaries around conduct. Respect does not require endless debate, harassment or tolerating accusations against innocent people. A conversation can remain open while an unsafe action is blocked.

The limits

These lenses cannot guarantee truth. A disciplined conspiracy can destroy evidence or exploit institutions that prefer quiet. Some questions remain unresolved because decisive records never existed or remain inaccessible. A stopping rule may end inquiry provisionally, not forever.

The method can also be performed theatrically. A partisan investigator may list alternatives, cite sources and announce falsifiability while protecting the conclusion in practice. Method is visible in revision, not vocabulary.

Power is unequal. A citizen cannot verify a security service, pharmaceutical trial or financial network alone. Dependence on expertise remains. The answer is plural, inspectable institutions with incentives to challenge one another, not personal verification of everything.

Do not turn anti-conspiracism into social conformity. Whistleblowers, minorities and reporters have been called paranoid while exposing abuse. Test the evidence and revision rules, not whether respectable people approve.

The one thing to keep

Keep the stopping rule.

Whenever hidden coordination is alleged, ask what could make the claim smaller, less likely or finished. That question does more work than asking whether you trust the speaker. Trust is graded and fallible. A stopping rule forces the theory to meet the world.

Suspicion has two stages. The first permits inquiry: institutions can deceive, incentives matter, anomalies deserve explanation and official accounts can fail. The second imposes discipline: actors must be named, mechanisms compared, sources traced, predictions recorded and confidence allowed to fall.

Without the first, power receives immunity. Without the second, accusation receives immunity. Both remove correction.

A real conspiracy can survive this demand. Evidence from independent routes converges, the allegation becomes more precise and some possibilities close. A self-protecting theory cannot afford closure. It treats every outcome as another entrance to the same room.

The permanent change is small and severe. Do not ask only, “Could they have done it?” Many things are possible. Ask, “What should I observe if they did, what else could produce it, and what would make me stop believing they did?”

Suspicion earns its value by risking an answer.

Terms

A glossary of the distinctions used to keep suspicion, evidence, motive and social transmission from collapsing into one vague explanation.

Conspiracy. Secret coordination between two or more actors to pursue an aim, often unlawful or deceptive. Real conspiracies vary from narrow agreements to covert state programmes and leave different traces.

Conspiracy theory. An account in which covert coordination does much of the explanatory work, often assigning it to powerful or malevolent actors. The label identifies the claim's structure; evidence must decide whether it deserves belief.

Conspiracy hypothesis. A provisional proposal that secret coordination may explain an event. It remains a hypothesis while rival causes, expected traces and disconfirming evidence can change its probability or scope.

Conspiracy mentality. A general readiness to interpret important events through hidden plots. It is more stable than endorsement of one named theory, which changes with politics, knowledge, identity and exposure.

Monological belief system. A proposed network in which one conspiratorial belief supports others through shared distrust. The term is useful, but coherence and contradiction must be measured rather than assumed.

Self-sealing belief. A belief protected by interpreting contrary evidence as part of the concealment. Self-sealing comes in degrees; the key issue is whether rescue assumptions create new tests or remove them.

Falsifiability. The capacity of a claim to conflict with possible evidence. A falsifiable allegation states what would count against it, though practical testing may still be difficult or ethically impossible.

Epistemic motive. A need connected with understanding, certainty or accurate explanation. Conspiracy accounts can satisfy it by joining fragments, even when the resulting coherence exceeds the evidence.

Existential motive. A need connected with safety, control and managing threat. Perceived powerlessness can make intentional explanations and investigative action appealing, but average associations are modest and uneven.

Social motive. A need connected with identity, belonging, esteem or group position. A theory may protect the self or group, grant special knowledge or identify outsiders responsible for loss.

Uncertainty. Incomplete knowledge about what happened, why or what follows. Uncertainty raises demand for explanation but cannot choose among explanations without comparative evidence.

Source laundering. Repetition that disguises a claim's common origin. Several accounts may appear to corroborate one another while all depend on the same post, witness, press release or dataset.

Distrust. Reduced willingness to accept a source's claims or intentions. Distrust can be justified, generalised or misplaced; it should be tied to a particular institution, history and subject.

Epistemic vigilance. The ordinary human monitoring of testimony for competence, honesty, motive and coherence. It protects against deception, but can become asymmetric when friendly sources escape the scrutiny applied to hostile ones.

Illusory pattern perception. Seeing structure in data produced by chance or selection. It may contribute to some beliefs, yet experimental links with conspiracy endorsement are less uniform than popular summaries imply.

Agency detection. Inferring that an actor caused an event. Agency is often the right explanation, but large outcomes can also emerge from accident, feedback, shared incentives and distributed failure.

Partisan congruence. Fit between a political claim and the interests or identity of a person's preferred camp. Conspiratorial thinking can predict endorsement more strongly when an alleged plot targets an opposing group.

Proportionality bias. The intuition that events with great consequences should have causes of comparable size. It can make lone actors, minor errors and cascading systems feel inadequate as explanations.

Confirmation bias. Seeking, interpreting or recalling information in ways that favour an existing belief. The practical problem is not noticing supportive evidence; it is applying unequal tests and ignoring base rates.

Motivated reasoning. Reasoning directed partly by desired conclusions, identities or interests. Greater skill can improve accuracy when accuracy is the goal, or improve defence when belonging or status dominates.

Identity-protective cognition. Processing information in ways that defend membership, values or standing within a group. Revision becomes costly when accepting a fact threatens loyalty or social connection.

Collective narcissism. Belief that one's group is exceptional yet insufficiently recognised by others. It can combine grievance and grandiosity, making hostile conspiracy explanations attractive under some conditions.

Need for uniqueness. Desire to possess uncommon knowledge or stand apart from the majority. Minority endorsement can increase a claim's appeal for some people, though it is neither necessary nor sufficient.

Rumour. Unverified information circulating because it matters to a group. Rumour may be true, false or mixed; conspiracy theories can organise rumours by supplying hidden actors and intention.

Misinformation. False or misleading information shared without requiring an intention to deceive. A person can transmit misinformation sincerely, and correction should separate the claim from assumptions about the sharer's motive.

Disinformation. False or misleading information deliberately produced or distributed to deceive. Proving disinformation requires evidence about intention, not merely showing that a statement was false.

Echo chamber. A communication setting where compatible voices dominate and opposing claims lose reach or legitimacy. Chambers can emerge through selection, networks, moderation and platform design rather than one algorithm alone.

Algorithmic amplification. Increased visibility produced by automated ranking or recommendation. Algorithms optimise selected outcomes such as attention or relevance; exposure may increase without producing belief in every viewer.

Inoculation or prebunking. Preparing people to recognise a misleading technique before exposure, often through a warning and weakened example. Controlled evidence is promising, while durability and transfer require continued testing.

Debunking. Correcting a false claim after exposure by identifying the error and supplying a supported replacement. Debunking usually improves accuracy, but effects can be incomplete and source trust still matters.

Go Deeper

The psychological entry point

Rob Brotherton, Suspicious Minds: Why We Believe Conspiracy Theories (Bloomsbury Sigma, 2015). Brotherton offers the most inviting first book after this one. He treats conspiratorial thinking as an extension of ordinary cognition rather than a museum of bizarre people, moving through pattern, agency, proportionality and the history of psychological explanation. The prose is brisk and the examples are memorable. Read it with one caution: the field has grown since 2015, and later work is more guarded about turning any single cognitive tendency into a broad causal account. Use it for the questions and mental machinery, then check stronger current syntheses for effect size and heterogeneity.

The history and concept

Michael Butter, The Nature of Conspiracy Theories, translated by Sharon Howe (Polity, 2020). Butter explains what conspiracy theories are, how they changed historically and why their current public status differs from earlier periods when conspiratorial accounts could sit closer to respectable political knowledge. The book is concise, comparative and especially useful against the belief that social media invented the form. Its historical sweep necessarily compresses regional differences, so treat it as a map rather than a complete global history. Read it to understand recurring plot structures, the relation to power and why the label conspiracy theory has its own political history.

The political supply side

Scott Radnitz, Revealing Schemes: The Politics of Conspiracy in Russia and the Post-Soviet Region (Oxford University Press, 2021). Radnitz shifts the question from who is prone to belief towards how conspiracy claims are produced, circulated and used. The book draws on more than 1,500 claims from twelve post-Soviet states, alongside surveys and focus groups, to show how citizens and elites use hidden-plot stories under different political conditions. It is a strong corrective to a literature centred on Western psychology. Its regional focus is also a limit: this is not a universal model. Read it to see conspiracism as a political repertoire and strategic resource.

The state of the science

Mikey Biddlestone, Ricky Green, Karen M. Douglas, Flávio Azevedo, Robbie M. Sutton and Aleksandra Cichocka, “Reasons to Believe: A Systematic Review and Meta-Analytic Synthesis of the Motives Associated with Conspiracy Beliefs”, Psychological Bulletin 151, no. 1 (2025). This is the technical route. It combines 971 effects from 279 independent studies and tests epistemic, existential and social motives. The main lesson is disciplined modesty: average relationships exist, vary substantially and are weaker in experiments than correlations. The paper assumes familiarity with meta-analysis and is not a casual read. Use its tables and moderation results whenever someone claims to have found the one reason people believe.

Notes and Sources

Definitions and evidential status. Karen M. Douglas and Robbie M. Sutton survey competing definitions and show that secrecy, powerful actors, malevolent purpose and conflict with official accounts vary in centrality. This book treats conspiracy theory first as an explanatory form, then distinguishes evidential states: documented conspiracy, live hypothesis, poorly supported claim and protected theory. Those states are an operating model rather than a settled taxonomy. Roland Imhoff, Tisa Bertlich and Marius Frenken support the separate distinction between a general conspiracy mentality and endorsement of specific theories.

Motives and variation. Mikey Biddlestone and colleagues synthesised 971 effect sizes from 279 independent studies, with 137,406 participants. Mean associations for epistemic, existential and social motives were modest, and heterogeneity was substantial. Experimental relationships were generally weaker than correlational relationships. Lukasz Stasielowicz's meta-analysis of personality correlates likewise argues against a single diagnostic profile. The book therefore uses motives as interacting conditions, not as a way to infer why one named person believes.

Pattern, agency and proportionality. The discussion draws on the broader review by Douglas and colleagues and on Petra Müller and Matthias Hartmann's signal-detection study. Their results linked illusory pattern perception more clearly with paranormal belief than with conspiracy belief, and showed that task base rates mattered. The book deliberately avoids the common overstatement that random-pattern detection provides a complete experimental miniature of political conspiracism. Agency and proportionality are presented as possible contributors whose force depends on context, available explanations and social meaning.

Identity, status and uniqueness. Roland Imhoff and Pia Lamberty found a modest relationship between need for uniqueness and conspiracy belief, including bounded experimental evidence that minority endorsement can increase appeal for some participants. Anni Sternisko and colleagues found that national narcissism predicted belief in, and dissemination of, conspiracy theories across a 56-country study during the COVID-19 pandemic. These findings concern probabilistic associations and particular measures. They do not establish that secure group attachment, national identification or nonconformity is pathological.

Partisan congruence and political supply. Omer Yair and colleagues analysed six survey studies in the United States and Israel, involving 10,765 respondents and 61 conspiracy theories. Conspiracy thinking predicted partisan conspiracy endorsement more strongly when the allegation fitted the respondent's political camp; the authors limit generalisation beyond the two countries and note observational endogeneity. David Hilden and Michael Kistner found that members of the United States Congress promoted conspiracy theories more before contested primaries, less afterwards and raised more individual-donor money after promotion. The result is confined to US congressional social-media behaviour. Sinan Alper, Onur Varol and Onurcan Yilmaz analysed 26,992 social-media users after the 2023 earthquakes in Türkiye and conducted a nationally representative face-to-face survey of 3,568 people. Their survey shows correlation between endorsement and support for the governing coalition, not that conspiracy belief caused voting behaviour.

Social knowledge and QAnon. Alice E. Marwick and William Clyde Partin's digital ethnography describes how one QAnon community built what the authors call populist expertise through collective interpretation, source evaluation and research practices. The book keeps their finding tied to the studied community rather than treating it as a model of every believer. Hayagreeva Rao and Henrich R. Greve provide the wider sociological account of attention, approval, networks, bricolage and collective production.

Media, platforms and historical prevalence. Joseph E. Uscinski and colleagues compared repeated survey questions about 37 specific conspiracy beliefs. Six rose, 16 showed no significant change and 15 fell across the available series. The studies are weighted towards the United States, cover selected claims and do not measure every form of suspicion. They support the narrow conclusion used here: digital media did not produce one uniform rise in all conspiracy beliefs. Claims about speed, searchability, participation and visible approval draw on Rao and Greve, Marwick and Partin, and Michael Butter's historical account.

Edgar Maddison Welch and Comet Ping Pong. The date, weapons, duration, shot through a locked door, absence of injuries, surrender and four-year sentence come from the United States Department of Justice sentencing release of 22 June 2017. Timothy R. Tangherlini and colleagues provide the comparative narrative analysis of Pizzagate and Bridgegate used to distinguish a sprawling conspiracy-theory framework from a documented conspiracy with a more bounded set of actors and actions. The manuscript does not reproduce the false accusations beyond what is needed to explain the mechanism and consequence.

The Protocols and antisemitic conspiracy. The United States Holocaust Memorial Museum documents that The Protocols of the Elders of Zion is a forgery purporting to record a Jewish plan for world domination, and traces its political circulation after exposure. Michael Butter supplies the broader historical context. The treatment here is intentionally brief because the histories of antisemitism, Judaism, the Holocaust and the Third Reich belong to other titles in the series.

Barruel, Robison and the Illuminati. The selective history follows Butter's account of Augustin Barruel and John Robison after the French Revolution. The Bavarian Illuminati existed, but assigning it command of the revolution turned a real secret society into an excessive causal explanation. The example is used to show how a genuine hidden group can become a container for events far beyond the evidence.

Watergate. Claims about the five burglars, the expanding investigation, campaign finance, testimony, White House recordings, obstruction and institutional conflict are grounded in the records of the Senate Select Committee on Presidential Campaign Activities and the National Archives' Watergate collections. Watergate is used as a contrast case: independent evidence narrowed the allegation, differentiated participants and survived challenge across institutions with conflicting incentives.

COINTELPRO. The Federal Bureau of Investigation's historical record and Vault files document the counterintelligence programmes begun in 1956 and the covert disruption of domestic political organisations. The Church Committee's reports provide independent congressional investigation of intelligence abuses. Knowledge of such programmes can rationally increase vigilance towards security agencies; it cannot establish a later claim without case-specific evidence.

The Tuskegee study. The Centers for Disease Control and Prevention's official timeline confirms that the men were not deliberately infected. Researchers recruited men who already had syphilis, did not obtain informed consent and withheld effective treatment after penicillin became widely available. The correction is included because a false detail can be removed without weakening the moral judgement of documented abuse.

Vaccination intentions and action. Daniel Jolley and Karen M. Douglas found that exposure to anti-vaccine conspiracy material reduced stated vaccination intentions in their experiments. The manuscript preserves the measured outcome. An intention recorded in a study is not identical to vaccination uptake, and one experiment does not establish a uniform effect for every health claim or audience.

Political violence. Adam Enders, Casey Klofstad and Joseph E. Uscinski examined 44 specific conspiracy theories and three measures connected with violence or conflict. Forty-two beliefs were positively associated with at least one outcome, but the patterns differed by theory and the authors stress the limits of observational inference. The book therefore treats the findings as evidence of concern and heterogeneity, not proof that conspiracy belief alone causes violence.

Correction, backfire and prebunking. Brendan Nyhan and Briony Swire-Thompson, Joseph DeGutis and David Lazer explain why broad claims of a routine backfire effect exceed the evidence. Ullrich Ecker and colleagues review the drivers of misinformation belief and resistance to correction. Cian O'Mahony and colleagues reviewed 25 intervention studies and found a mixed evidence base, with several approaches ineffective and some analytic or critical-thinking interventions more promising. Jon Roozenbeek and colleagues found that short, technique-based inoculation videos improved immediate recognition of manipulation across seven preregistered studies. The manuscript does not claim permanent immunity or guaranteed attitude change.

Contradictory beliefs and self-sealing. Jan-Willem van Prooijen, Iris Wahring, Laura Mausolf, Nicole Mulas and Shayda Shwan re-examined the claim that people confidently endorse incompatible conspiracy theories. Positive correlations can emerge from shared rejection of the official account while endorsement of each alternative remains low. Kurtis Hagen's philosophical analysis argues that self-sealing should be assessed by degree and by the quality of rescue assumptions, rather than treated as a vice unique to conspiracy theories.

Limits of the evidence. Surveys measure responses to selected wording, not private certainty or action. Experiments improve causal identification but often use immediate outcomes and restricted samples. Digital traces overrepresent people who post. Ethnographies reveal practices inside particular communities. Historical polling changes questions and coverage. Although cross-national research has expanded, English-language scholarship and evidence from the United States and western Europe remain disproportionately influential. Translation, sampling and political institutions can change what an endorsement item means. The strongest conclusions concern heterogeneity, modest average associations and the importance of social, political and evidential processes. Individual diagnosis, universal motive claims and universal intervention claims are not warranted.

Bibliography

Alper, Sinan, Onur Varol and Onurcan Yilmaz. “Conspiracy Theories as Instruments of Power: The Case of Conspiracy Beliefs in the Wake of the 2023 Earthquakes in Türkiye.” Political Psychology 47, no. 4 (2026): e70106. doi: 10.1111/pops.70106.

Biddlestone, Mikey, Ricky Green, Karen M. Douglas, Flávio Azevedo, Robbie M. Sutton and Aleksandra Cichocka. “Reasons to Believe: A Systematic Review and Meta-Analytic Synthesis of the Motives Associated with Conspiracy Beliefs.” Psychological Bulletin 151, no. 1 (2025): 48-87. doi: 10.1037/bul0000463.

Brotherton, Rob. Suspicious Minds: Why We Believe Conspiracy Theories. London: Bloomsbury Sigma, 2015.

Butter, Michael. The Nature of Conspiracy Theories. Translated by Sharon Howe. Cambridge: Polity, 2020.

Centers for Disease Control and Prevention. “The Untreated Syphilis Study at Tuskegee Timeline.” Accessed 4 September 2026.

Douglas, Karen M., Joseph E. Uscinski, Robbie M. Sutton, Aleksandra Cichocka, Turkay Nefes, Chee Siang Ang and Farzin Deravi. “Understanding Conspiracy Theories.” Political Psychology 40, supplement 1 (2019): 3-35. doi: 10.1111/pops.12568.

Douglas, Karen M., and Robbie M. Sutton. “What Are Conspiracy Theories? A Definitional Approach to Their Correlates, Consequences, and Communication.” Annual Review of Psychology 74 (2023): 271-298. doi: 10.1146/annurev-psych-032420-031329.

Ecker, Ullrich K. H., Stephan Lewandowsky, John Cook, Philipp Schmid, Lisa K. Fazio, Nadia Brashier, Panayiota Kendeou, Emily K. Vraga and Michelle A. Amazeen. “The Psychological Drivers of Misinformation Belief and Its Resistance to Correction.” Nature Reviews Psychology 1 (2022): 13-29. doi: 10.1038/s44159-021-00006-y.

Enders, Adam M., Casey A. Klofstad and Joseph E. Uscinski. “The Relationship Between Conspiracy Theory Beliefs and Political Violence.” HKS Misinformation Review 5, no. 6 (2024). doi: 10.37016/mr-2020-163.

Federal Bureau of Investigation. “COINTELPRO.” The Vault. Accessed 4 September 2026.

Hagen, Kurtis. “Are Conspiracy Theories Problematically Self-Sealing?” Episteme, First View (2026): 1-19. doi: 10.1017/epi.2026.10115.

Hilden, David, and Michael R. Kistner. “Promoting Conspiracy Theories Strategically.” Quarterly Journal of Political Science 20, no. 3 (2025): 307-337. doi: 10.1561/100.00024033.

Imhoff, Roland, Tisa Bertlich and Marius Frenken. “Tearing Apart the ‘Evil’ Twins: A General Conspiracy Mentality Is Not the Same as Specific Conspiracy Beliefs.” Current Opinion in Psychology 46 (2022): 101349. doi: 10.1016/j.copsyc.2022.101349.

Imhoff, Roland, and Pia Lamberty. “Too Special to Be Duped: Need for Uniqueness Motivates Conspiracy Beliefs.” European Journal of Social Psychology 47, no. 6 (2017): 724-734. doi: 10.1002/ejsp.2265.

Jolley, Daniel, and Karen M. Douglas. “The Effects of Anti-Vaccine Conspiracy Theories on Vaccination Intentions.” PLOS ONE 9, no. 2 (2014): e89177. doi: 10.1371/journal.pone.0089177.

Marwick, Alice E., and William Clyde Partin. “Constructing Alternative Facts: Populist Expertise and the QAnon Conspiracy.” New Media & Society 26, no. 5 (2024): 2535-2555. doi: 10.1177/14614448221090201.

Müller, Petra, and Matthias Hartmann. “Linking Paranormal and Conspiracy Beliefs to Illusory Pattern Perception Through Signal Detection Theory.” Scientific Reports 13 (2023): 9739. doi: 10.1038/s41598-023-36230-0.

National Archives and Records Administration. Records of the Watergate Special Prosecution Force, 1971-1977. Record Group 460. College Park, MD: National Archives.

Nyhan, Brendan. “Why the Backfire Effect Does Not Explain the Durability of Political Misperceptions.” Proceedings of the National Academy of Sciences 118, no. 15 (2021): e1912440117. doi: 10.1073/pnas.1912440117.

O'Mahony, Cian, Maryanne Brassil, Gillian Murphy and Conor Linehan. “The Efficacy of Interventions in Reducing Belief in Conspiracy Theories: A Systematic Review.” PLOS ONE 18, no. 4 (2023): e0280902. doi: 10.1371/journal.pone.0280902.

Radnitz, Scott. Revealing Schemes: The Politics of Conspiracy in Russia and the Post-Soviet Region. New York: Oxford University Press, 2021.

Rao, Hayagreeva, and Henrich R. Greve. “The Plot Thickens: A Sociology of Conspiracy Theories.” Annual Review of Sociology 50 (2024): 191-207. doi: 10.1146/annurev-soc-030222-031142.

Roozenbeek, Jon, Sander van der Linden, Beth Goldberg, Steve Rathje and Stephan Lewandowsky. “Psychological Inoculation Improves Resilience Against Misinformation on Social Media.” Science Advances 8, no. 34 (2022): eabo6254. doi: 10.1126/sciadv.abo6254.

Stasielowicz, Lukasz. “Who Believes in Conspiracy Theories? A Meta-Analysis on Personality Correlates.” Journal of Research in Personality 98 (2022): 104229. doi: 10.1016/j.jrp.2022.104229.

Sternisko, Anni, Aleksandra Cichocka, Aleksandra Cislak and Jay J. Van Bavel. “National Narcissism Predicts the Belief in and the Dissemination of Conspiracy Theories During the COVID-19 Pandemic: Evidence From 56 Countries.” Personality and Social Psychology Bulletin 49, no. 1 (2023): 48-65. doi: 10.1177/01461672211054947.

Swire-Thompson, Briony, Joseph DeGutis and David Lazer. “Searching for the Backfire Effect: Measurement and Design Considerations.” Journal of Applied Research in Memory and Cognition 9, no. 3 (2020): 286-299. doi: 10.1016/j.jarmac.2020.06.006.

Tangherlini, Timothy R., Shadi Shahsavari, Behnam Shahbazi, Ehsan Ebrahimzadeh and Vwani Roychowdhury. “An Automated Pipeline for the Discovery of Conspiracy and Conspiracy Theory Narrative Frameworks: Bridgegate, Pizzagate and Storytelling on the Web.” PLOS ONE 15, no. 6 (2020): e0233879. doi: 10.1371/journal.pone.0233879.

United States Department of Justice. “North Carolina Man Sentenced to Four Years in Prison for Armed Assault at Northwest Washington Pizza Restaurant.” Press release, 22 June 2017.

United States Holocaust Memorial Museum. “Protocols of the Elders of Zion.” Holocaust Encyclopedia. Accessed 4 September 2026.

United States Senate, Select Committee to Study Governmental Operations with Respect to Intelligence Activities. Final Report: Intelligence Activities and the Rights of Americans. Book II. Washington, DC: United States Government Printing Office, 1976.

United States Senate, Select Committee on Presidential Campaign Activities. Final Report. Senate Report 93-981. Washington, DC: United States Government Printing Office, 1974.

Uscinski, Joseph E., Adam M. Enders, Casey A. Klofstad, Michelle Seelig, Hugo Drochon, Kamal Premaratne and Manohar N. Murthi. “Have Beliefs in Conspiracy Theories Increased Over Time?” PLOS ONE 17, no. 7 (2022): e0270429. doi: 10.1371/journal.pone.0270429.

van Prooijen, Jan-Willem, Iris Wahring, Laura Mausolf, Nicole Mulas and Shayda Shwan. “Just Dead, Not Alive: Reconsidering Belief in Contradictory Conspiracy Theories.” Psychological Science 34, no. 6 (2023): 670-682. doi: 10.1177/09567976231158570.

Yair, Omer, Shira Hebel-Sela, Amnon Cavari and Asif Efrat. “Conspiracy Thinking and Belief in Partisan Conspiracy Theories: A Moderating Effect of Partisan Congruence?” Political Psychology 47, no. 2 (2026): e70131. doi: 10.1111/pops.70131.