Propaganda
in a Hurry

The Whole Thing in One Page

The standard picture of propaganda is a poster with a clenched fist, an obvious lie and a population foolish enough to swallow it. That picture is comforting. It puts propaganda in the past, inside dictatorships, and inside other people's heads.

Propaganda is larger and less theatrical. It is the organised attempt by a state, or by people acting for one, to shape public interpretation, expectation or conduct in support of a political objective. It begins before anybody writes the slogan. The sponsor decides which audience matters and what response is needed: confidence at home, compliance among officials, endurance from soldiers, reassurance for allies, fear among enemies, hesitation among neutrals, or the visible impression that everybody else accepts the official account. One message can serve several of those tasks without making any audience sincerely loyal.

The machinery matters as much as the message. A slogan printed once is weak. Put it in schools, news bulletins, cinema, ceremonies and workplace meetings, remove rival accounts, reward repetition and punish correction, and the same words acquire institutional weight. Propaganda is therefore coupled to distribution, censorship, status, ritual and force. The poster is the label on the system.

The boundary is political rather than visual. A sober briefing can manipulate; a dramatic poster can convey a documented danger. Ask whether the sponsor is visible, evidence can be challenged, uncertainty is admitted and dissent remains safe. The same sentence behaves differently inside an open argument and inside a monopoly backed by punishment.

Its messages need not be false. Many of the strongest are built from facts selected, named, simplified, repeated and framed until one conclusion feels complete. Credibility comes from familiar institutions, recognised voices, documents, uniforms, statistics, eyewitnesses and apparent independence. Emotion then assigns the stakes. Fear identifies danger. Pride makes sacrifice honourable. Disgust converts neighbours into contamination. Hope gives endurance a destination.

Audiences still complete the campaign. People bring loyalties, grievances, memories, local knowledge, social ties and competing sources. Radio in Germany did not have one fixed political effect before and after Nazi control, and its estimated effects varied across places. Rwandan radio operated inside an organised genocide, but later quantitative studies disagree about whether stronger reception changed total participation, while a 2024 reanalysis found an association with earlier local onset. Neither case licenses the fantasy of a universal mind-control beam.

The history runs from royal inscriptions and public monuments through print, revolution and nationalism to the propaganda offices of the First World War. Radio, film and mass organisations let fascist, communist and democratic states address whole populations. The Cold War added international broadcasting, cultural diplomacy and covert sponsorship. Digital networks made distribution faster, participation harder to separate from production, and uncertainty itself a useful objective. Generative systems can now manufacture plausible text, images, voices and personas at low cost, though cheap production is not the same as political success.

One severe danger is the return loop. States build a common account because they need coordination across distance. If careers depend on confirming it, bad news stops travelling upwards. Staged unanimity becomes evidence of genuine support. Leaders begin making decisions inside the reality designed for everybody else. Propaganda can move millions of people without fully convincing them, then distort the judgement of the people giving the orders.

That is the book.

Why You Should Care

On 21 December 1989, Nicolae Ceaușescu stepped onto the balcony of the Communist Party headquarters in Bucharest for a rally designed to prove that Romania was calm and he remained adored. Workers had been brought into the square. Television carried the scene. The script called for applause, slogans and the familiar image of a leader speaking to one grateful people.

Then the crowd made the wrong sound.

Boos, whistles and confused shouting interrupted the speech. Ceaușescu raised a hand and tried to restore order. The broadcast faltered. None of this, by itself, overthrew the regime. Romania's collapse had deeper causes, protests were already under way and force would still be used. The importance of the moment was different. A public ceremony intended to display unanimity displayed its breakdown instead. Viewers could see that other people were no longer performing belief on command. Private doubt became public information.

That is why propaganda matters even when nobody appears persuaded. Political life depends partly on what people think other people think. A citizen may despise a ruler and stay home because the streets look empty. A civil servant may know a policy is failing and report success because every colleague does. A soldier may obey because the regime still seems permanent. A foreign government may compromise because resistance appears isolated. Change the visible picture of the crowd and all four may recalculate without changing a single underlying principle.

You meet this mechanism far beyond totalitarian rallies. Governments select what to publish, which losses to name, which maps to show, which victims deserve faces and which events receive a commemorative day. During war, they conceal operations and cultivate morale. During peace, they promote policy, identity and legitimacy. Public-health information, military recruiting, diplomatic broadcasting and election messaging do not become equal merely because states sponsor them. The difficult question is where necessary communication becomes organised manipulation, and what institutions let citizens contest the answer.

The subject also changes how you read media. A false statement can be propaganda, but so can a true photograph placed under a deceptive caption, an accurate statistic without its denominator, or an endless stream of contradictory claims intended to make verification feel pointless. The useful test is rarely, "Is this one sentence a lie?" It is: what picture is being built, which alternatives can reach me, what response is being made visible, and what happens to people who supply a correction?

There is a defensive benefit. Recognising propaganda does not make you immune. Knowledge helps people test evidence, but political commitments can produce selective scepticism: congenial claims receive less scrutiny than hostile ones. Cynicism creates another vulnerability, since the belief that everybody lies favours the actor least constrained by evidence. Resistance therefore needs competing sources, trusted correction, protected expertise, transparent uncertainty and a public culture in which changing one's mind does not look like surrender.

The legal stakes are narrow but severe. International law does not outlaw every tendentious state narrative. Article 20 of the International Covenant on Civil and Political Rights requires states parties to prohibit propaganda for war and specified advocacy of hatred that constitutes incitement. Article III of the Genocide Convention makes direct and public incitement to commit genocide punishable. These thresholds mark speech that can become part of an operational system of harm.

There is also a governing lesson. A state that cannot tell citizens the truth may lose the ability to learn it. Reports become performances, targets become fictions and leaders receive applause where they needed warning. Ceaușescu's rally is memorable because the picture broke in public. Many systems fail more quietly. The crowd claps, the numbers improve, the orders continue, and the decision-maker becomes the last person permitted to know what is happening.

The hour ahead is about how the system is built, when it works, when it fails, and how to keep public communication from becoming a prison for both audience and ruler.

The Core Ideas

The Objective Comes Before the Message

A ruler can give an order to one person. A state must coordinate millions who are separated by distance, occupation and private experience. Before a propagandist chooses words, somebody must decide what political problem communication is meant to solve.

That objective may be persuasion, but often it is not. A government can seek endurance from civilians, enlistment from recruits, discipline from officials, reassurance from allies, surrender from enemy troops, hesitation among neutral governments, or silence from domestic opponents. The same campaign may divide those tasks among different audiences. A triumphant battlefield bulletin can sustain morale at home, signal resolve abroad and warn civil servants that pessimism will be treated as disloyalty. Judging it only by whether the public believed every sentence misses most of its work.

The target also changes the message. Soldiers know different facts from civilians. Officials need a vocabulary that turns policy into routine. Foreign audiences may distrust a government voice but listen to journalists, exiles or cultural figures. People already committed to a cause may need a cue for action rather than an argument. The first discipline is therefore to name the intended audience and the response sought from it.

States then build a public world in which that response makes sense. People need some account of who belongs, what has happened, who threatens them, what sacrifice means and what everybody else is expected to do. Propaganda selects, names, simplifies, repeats, associates and displays. It selects some events for attention, names their actors, simplifies causation, repeats the interpretation, associates it with trusted symbols and displays the approved reaction. None of those moves has to be false. Together they can narrow a crowded political field into a story with obvious roles.

Public visibility adds another mechanism. Suppose a message reaches you privately: the government is strong, the war is being won, the leader is loved. You may believe it or reject it. Now suppose the same message is broadcast nationally, printed in every workplace and repeated at a ceremony where you can see neighbours applauding. You have learned something beyond the claim itself. You know that they have heard it, they know that you have heard it, and everybody knows that public conduct will be judged against it. Political theorists call this common knowledge. It can organise behaviour even where private belief remains mixed.

A national flag, a classroom map and a leader's portrait do related work. They compress a political order into objects that recur across settings. Repetition does not prove the claim represented by the symbol. It makes the claim easy to retrieve and hard to avoid. The map says where the nation ends. The portrait says where authority resides. The ceremony says which emotions belong in public.

These shared premises reduce the cost of command. An official need not explain every sacrifice from the beginning if citizens already recognise the war as defensive, the party as the nation or the leader as the source of order. The public argument seems short because its longest assumptions arrived years earlier. Visible compliance can then keep trains moving, taxes arriving and offices staffed without uniform conviction.

Democratic governments also commemorate wars, promote civic duties, explain policy and present the nation abroad. Those acts are not equivalent to a one-party system that monopolises media and punishes contradiction. Contestability, source transparency, legal coercion, access to alternatives and the safety of dissent change the communication order in which the message operates.

The first question is therefore not, "Did people believe it?" Ask what outcome the sponsor wanted from which audience, what political world made that response appear proper, and what happened to people who refused the assigned role. Propaganda starts with an objective, then arranges a public reality capable of carrying it.

The Channel Controls the Comparison

A sentence has no political power until it reaches someone. Reach has a structure: presses, paper, transmitters, cinemas, schools, postal systems, loudspeakers, party branches, mobile networks and the rules that decide what may pass through them. Propaganda is therefore an infrastructure problem before it is a writing problem.

Consider radio. A speech transmitted from one capital can enter kitchens, barracks and factories at the same hour. That simultaneity gives the message a public character that a leaflet lacks. Cheap receivers enlarge the audience. Scheduling creates habit. Familiar presenters lend continuity. Licensing, ownership and jamming determine whether rival stations can compete. The content may be brilliant or dull, but a dull message heard daily from the only strong signal can matter more than a brilliant one that never clears the border.

Censorship belongs inside this mechanism. Propaganda adds, selects and repeats. Censorship removes, delays and raises the cost of alternatives. Used together, they change the comparison set. An official claim that would look weak beside prompt independent reporting can appear sufficient when other accounts arrive late, in fragments or through risky channels. Silence around a topic may be as designed as the statement about it.

Modern censorship often works through friction rather than a perfect wall. A blocked search term, slow connection, removed post or requirement to register can deter ordinary users even if determined specialists find a route around it. Research on Chinese social media has shown a particularly important pattern: material criticising the state may survive while material with collective-action potential is more likely to disappear. The target is not every hostile thought. It is the capacity of dispersed people to find one another and act together.

Timing can be more important than permanent removal. Delaying a report until after a vote, protest, battle or panic may neutralise it even if the archive later becomes complete. Propaganda is often judged from what survives. Political effects depend on what was available when a decision had to be made.

Abundant media does not end channel power. It moves selection elsewhere. On a platform, candidate material is ranked by software, commercial rules, safety policies and behavioural predictions. A state can exploit that system through official accounts, paid placement, aligned creators, covert personas, automated posting, harassment or demands placed on the platform. It may seek removal. It may also seek saturation, filling attention with enough claims, jokes, documents and counterclaims that no stable account survives.

The digital audience helps distribute the campaign. Sharing, mocking and rebutting can all create further exposure, so a citizen may become an unpaid relay without accepting the message. Corrections still matter, but their source, timing and placement affect who encounters the evidence and whether the false claim receives another pass through the network. Distribution and correction cannot be analysed separately.

The channel can also signal power. A slogan on every wall demonstrates that someone can occupy every wall. A compulsory broadcast demonstrates control over the schedule. Haifeng Huang's work on authoritarian propaganda develops the narrower signalling logic: conspicuous expenditure on material that audiences may not find persuasive can still communicate the regime's resources and willingness to deploy them. The display of capacity becomes part of the political content.

Study the pipes, not merely the water. Who owns them, who can close them, what they reward, how often the message arrives and which alternatives face delay? Those facts are part of what the audience receives.

Credibility Is Assembled

The easiest propaganda to reject is a claim from a source you already distrust. States therefore spend effort constructing credibility rather than relying on the government seal.

One route is truth. A campaign can report accurate facts that favour its case, omit those that do not and arrange the remainder into a misleading whole. A photograph may be authentic, the casualty figure correct and the quotation exact, while the selection hides the wider pattern. Propaganda becomes harder to diagnose because each component survives a narrow fact check. The distortion lies in proportion, context and the boundary around what counts as relevant.

Another route is borrowed authority. Scientists, veterans, teachers, clergy, artists and local community figures can carry a message that would weaken if delivered by a minister. Documents and statistics produce a similar effect. Numbers look impersonal, though somebody chose the measure, denominator and comparison period. An official can make a political judgement appear administrative by placing it inside a chart.

Credibility can also be earned by admitting limited loss. A broadcaster that reports one defeat accurately may gain authority for later claims that listeners cannot check. This creates a strategic reason to preserve some truth and editorial independence. Propaganda systems can be strongest when they resist the urge to make every bulletin triumphant.

Propagandists also manage source visibility. White propaganda names its sponsor. Grey propaganda leaves the source uncertain. Black propaganda pretends to come from someone else, perhaps an enemy faction, an independent newspaper or a grassroots group. These labels describe attribution, not moral colour. Open sponsorship can be more credible when the audience expects accountability. Concealment can be useful when the state's identity would trigger resistance.

During the Cold War, governments funded broadcasters, magazines, conferences, cultural organisations and intellectual projects at varying distances from official control. Some participants knew the sponsor; others did not. The arrangement sought the prestige of independence while retaining strategic direction. The ethical problem is plain: an audience cannot assess motive or editorial constraint when sponsorship is hidden. The practical problem is equally plain: exposure of the relationship can contaminate truthful work produced within it.

Credibility is relational. The same voice can persuade one audience and repel another. Enemy admissions sometimes carry unusual weight because they run against expected interest. A defector may appear authoritative to outsiders and compromised to former colleagues. Local intermediaries can translate a national story into familiar grievances, but they can also alter it beyond central control.

Consistency helps, but perfect consistency is not always required. A high-volume campaign can issue incompatible explanations to different audiences, or cycle through them fast enough that the objective becomes doubt rather than acceptance. Christopher Paul and Miriam Matthews described a modern Russian pattern as a firehose of falsehood: rapid, repetitive, multichannel and unconcerned with consistency. The model should not be applied to every Russian message or every state. Its value is the objective it reveals. A campaign may win when citizens conclude that evidence is hopeless and commitment is naive.

Credibility therefore cannot be read from polish. Production quality, official statistics and an apparent eyewitness may each be assembled parts of a source strategy. Ask who is speaking, who selected the evidence, what the speaker risks by being wrong, whether the sponsor is visible and which trusted relationship is being borrowed.

Propaganda rarely asks you to trust a naked assertion. It dresses the assertion in somebody else's authority.

Emotion Assigns the Stakes

Propaganda is often described as emotional rather than rational, as though feeling switches thought off. The mechanism is more useful than that contrast. Emotion tells people what deserves attention, which risks are urgent, whose suffering counts and what action feels proportionate.

Fear narrows time. A threat presented as immediate makes delay look reckless and dissent look indulgent. Pride enlarges identity. It lets private cost appear as contribution to a collective achievement. Anger supplies an agent to blame. Disgust is especially dangerous because it converts political opponents into pollution: vermin, disease, filth, infestation. Once a group is described as contamination, exclusion can be presented as hygiene and violence as cleansing.

These emotions do not arise from words alone. They are attached to categories, images and repeated stories. Atrocity propaganda gives a conflict a moral shape by presenting cruelty attributed to the enemy. Some atrocity reports are documented; others are exaggerated or invented. The category describes their mobilising use, not whether every report is false. The evidential duty is therefore double: investigate the event and examine why this event, image or victim was selected for mass attention while others remained obscure.

Hope matters too. Revolutionary propaganda promises a transformed future, wartime propaganda promises victory, and development propaganda turns a dam, harvest or satellite into evidence that history is moving in the approved direction. The promise can sustain sacrifice because present deprivation acquires an ending. When deadlines pass, the story may shift from achievement to sabotage: success remains certain, but hidden enemies have delayed it.

Humour and entertainment carry emotion with less resistance. Songs, sport, serial drama and jokes can make the political message feel like shared culture rather than instruction. RTLM's music and conversational style mattered for this reason. Enjoyment can create attention and intimacy before an explicit demand appears.

Emotion also assigns social roles. The ideal worker, soldier, mother, martyr, pioneer or citizen gives the audience a person to become. The enemy image gives it a person not to resemble. Such types reduce a crowded political field into a moral drama with recognisable characters. They can be memorable because they recruit identity rather than asking for a fresh calculation on every occasion.

The Rwanda case shows why emotional content must be placed inside institutions and local conditions. Radio Télévision Libre des Mille Collines used humour, music, political commentary, names, threats and dehumanising language during an organised genocide. An influential 2014 study using variation in reception estimated greater participation in killing and social spillovers. A 2024 reanalysis with revised coverage maps and individual-level conviction records found earlier local onset where RTLM could be heard but no significant association with total participation. The quantitative dispute is material. It does not erase the broadcasts, their incitement or their place beside government authority, militias, roadblocks and local coercion. It does prevent one disputed estimate from becoming a settled measure of radio's share of the killing.

Emotion can also fail. Fear repeated without visible consequence may exhaust attention. Crude hatred can repel audiences whose identity is not already aligned. Pride may collapse when lived experience contradicts triumphal reporting. The strongest campaigns fit emotional cues to existing memories, losses and group boundaries.

Do not ask whether a message contains feeling. Every political judgement does. Ask which emotion it assigns, to whom, with what object and what action it makes feel necessary. Propaganda moves minds partly by deciding what a situation is allowed to feel like.

Public Behaviour Can Outrun Private Belief

A square full of portraits does not prove that the population loves the man in the portrait. It proves that displaying him is rewarded, required or safe. That difference is one of the most important in the subject.

People routinely conceal preferences when expression carries a cost. Timur Kuran calls the broader process preference falsification: a person states a public preference that differs from the private one. The motive may be fear, career interest, politeness, uncertainty or a wish to avoid isolation. When many people do this at once, each person's performance becomes evidence for everyone else. The regime can look more popular than it is because citizens are observing the caution they collectively created.

Pluralistic ignorance is related. Individuals privately reject a norm while mistakenly believing that most others accept it. Public propaganda can maintain that error by displaying approved behaviour and hiding dissent. A worker who suspects a production target is fictional may still sign the report because every office has signed. A student may repeat a slogan because every class repeats it. Neither needs to be fooled about the words. Each needs to be unsure about the social price of refusing them.

Political ritual exploits this gap. Parades, elections without meaningful competition, loyalty oaths, public confessions and leader portraits can test compliance, identify organisers and produce common knowledge of state reach. Lisa Wedeen's study of authoritarian Syria showed how absurd or excessive official claims could function without sincere belief. Citizens demonstrated that they knew the required performance and that the regime could demand it. The obedience was political evidence.

The ritual also sorts the population. Organisers learn who attends, who hesitates and who can mobilise others. Participants learn which colleagues will repeat the line. A ceremony can therefore gather intelligence while pretending merely to express sentiment. The audience is being displayed and inspected at the same time.

This does not make belief irrelevant. Repeated performance can shape identity, memory and habit, particularly when it begins in childhood and is supported by institutions. People also seek coherence between conduct and self-image. Yet the causal arrow can run the other way: public action occurs first because of incentives, and attitudes adjust later or remain divided.

The distinction explains sudden political cascades. A system may appear stable until one public act reveals that dissent is less isolated than people assumed. Others then update about the crowd, officials reconsider enforcement and defections become safer. The underlying grievances may be old. What changes quickly is knowledge of their distribution. Ceaușescu's broken rally mattered in this way. It turned a choreographed display into evidence that choreography had lost control.

Propaganda offices often measure the wrong thing because behaviour is easier to count than conviction. Attendance, circulation, votes, pledges and slogan repetition produce impressive totals. Under coercion, those totals may measure administrative capacity. They may still matter, since administrative capacity can sustain a state. They should not be mistaken for affection or durable consent.

The same warning applies outside dictatorships. A public campaign may report impressions, clicks, sign-ups or stated agreement while the intended behaviour never follows. Exposure is not comprehension. Comprehension is not belief. Belief is not action. Action once is not persistence.

Watch what people do, but ask what made the act informative. A raised hand can express conviction, coordination, fear or the sensible wish to keep one's job. Propaganda succeeds politically when those motives produce the required public world, even if the private world remains unconquered.

Audiences Answer Back

Propaganda is designed at the centre and completed at the edge. Audiences interpret, ignore, adapt, parody, translate, share and resist. They also differ. Any theory that gives a message one fixed effect has removed the people from the mechanism.

Prior belief is one source of variation. A message from a trusted national institution may reinforce supporters and harden opponents. Social identity changes which source appears credible. Knowledge can improve evidence testing, yet people can apply stricter standards to unwelcome claims than to congenial ones. Local experience may overpower distant claims. A triumphant harvest film has a difficult task in a village where the grain store is empty.

Competition matters. Before the Nazis controlled German radio, access to radio could expose listeners to government material hostile to them. After the takeover, the political direction of the medium changed. Research comparing areas with different signal strength found effects that shifted with control and varied with local predispositions. The finding is stronger than the slogan that radio elected Hitler. It shows an interaction among access, content, political environment and audience.

Networks matter too. A broadcast may change behaviour directly, but it can also change what neighbours expect, discuss and enforce. The 2014 Rwanda study attributed part of its estimated effect to social spillovers, while the 2024 reanalysis did not recover a significant association with participation totals. Both place radio inside local structures rather than treating each listener as an isolated target. Information passed through authority, social ties and organised violence, which is why reception alone cannot identify one stable mechanism or effect size.

Resistance takes many forms. Citizens use jokes, rumours, foreign broadcasts, private reading, religious communities, professional knowledge and ordinary comparison with lived conditions. Samizdat in the Soviet bloc copied prohibited writing by hand or typewriter. Radio jamming tried to make foreign broadcasting costly to hear, while listeners adjusted antennas and schedules. A state can dominate formal media and still fail to control kitchen conversation.

Intermediaries decide whether resistance travels. A respected doctor, priest, veteran or local organiser can make distant evidence usable. The same person can strengthen state propaganda. Central campaigns are therefore often contests over trusted translators, not contests between two messages arriving naked before an isolated citizen.

Counterpropaganda can misfire. Repeating a false claim in order to deny it can increase familiarity. A correction from a distrusted institution may confirm the audience's sense of persecution. Ridicule may strengthen group identity. Exposing covert sponsorship can discredit a campaign, but blanket accusations that every disagreeable source is foreign propaganda can become propaganda themselves.

Measurement must therefore follow a ladder of outcomes. Did the audience receive the material? Attend to it? Understand it? Remember it? Change a belief, expectation or emotion? Act? Persist? A study showing immediate agreement after one article does not establish durable political conversion. Recent experiments found that language-model articles could move reported agreement nearly as much as authentic covert campaign articles under controlled exposure. That is evidence about bounded exposure and reported attitudes, not proof that synthetic campaigns decide elections.

The audience may also repurpose the material. Posters become collector's objects. Official songs acquire ironic meanings. A slogan can be turned against its author. Such reversals are not decorative footnotes. They show that meaning is negotiated under unequal power rather than delivered intact.

Propaganda has effects, sometimes grave ones. The discipline is to state whose behaviour changed, in what setting, through which channel, over what period and compared with what alternative. The audience is never an empty container. It is the last institution in the chain.

The Picture Returns to the Palace

The first problem was coordination. A state built a public picture so distant people could act inside the same political world. The final problem is feedback. How does accurate information return once careers, safety and legitimacy depend on preserving that picture?

Every organisation filters bad news. Propaganda systems can make the filter political. A newspaper that reports failure may be closed. A local official who misses a target may invent success. A poll respondent may protect herself. An intelligence officer may soften a conclusion that contradicts the leader's speech. Each act can be individually rational and collectively disastrous.

The Great Leap Forward in China offers a severe example of information distortion inside mobilisation. Local cadres faced intense pressure to report spectacular agricultural gains. Inflated figures moved upwards, while the state procured grain on assumptions detached from real output. Propaganda, coercive targets, political campaigns and fear of being labelled obstructive helped corrupt the measurements on which policy depended. They did not alone cause the famine, which involved a wider set of policies, extraction decisions, institutional failures and natural variation. They made correction harder at the moment it was most needed.

Public unanimity creates a similar trap. A leader sees rallies, loyal newspapers, election totals and officials repeating approved language. These outputs were produced under rules the leader controls, but they can still be read as independent evidence of support. The more successful the performance, the harder it becomes to estimate what would happen if enforcement weakened.

Autocrats can commission surveys, monitor complaints and read private reports, but respondents and officials anticipate the audience. Fear contaminates answers before analysis begins. Even a technically sound instrument fails if citizens believe honesty is dangerous or local administrators can identify them. Measurement cannot be separated from political trust.

Foreign policy can suffer when propaganda categories replace analysis. An enemy described for years as cowardly, decadent or controlled by a tiny clique may be expected to collapse. Reports showing resilience look disloyal because they contradict the national story. A state can therefore enter a war with an audience model built for mobilisation rather than prediction. The message intended to weaken the enemy's image has weakened the sender's judgement.

Digital monitoring appears to solve the problem by supplying vast behavioural data. It can help, but measurement inherits the incentives of the system. Officials may optimise visible indicators, platforms may overcount automated activity, and citizens may learn which behaviour is watched. Sentiment analysis cannot recover opinions people dare not express. A dashboard can give false confidence with greater speed than a paper report.

Democracies have correction mechanisms that authoritarian systems often suppress: opposition parties, courts, independent media, auditors, elections, whistleblowers and professional bodies. None is automatic. Governments can punish internal dissent, classify embarrassment as secrecy, pressure broadcasters or turn policy claims into loyalty tests. The difference lies in whether correction remains institutionally possible and politically survivable.

The durable defence is not a ruler who promises to prefer truth. It is a system in which unwelcome evidence has more than one route to travel. Redundant reporting, protected disagreement, independent measurement and public records are information infrastructure. They may slow coordination and expose conflict. That friction is the price of remaining in contact with reality.

Propaganda can make a state look stronger by reducing visible contradiction. Beyond some point, the same achievement removes the contradiction that could have saved it. The shared picture returns to the centre, polished by every subordinate, and becomes the map by which power drives into the wall.

How It Actually Works

The ruler on the cliff

Around 521 BCE, Darius I had his version of a civil war cut into the rock at Behistun, on the road between Mesopotamia and the Iranian plateau. The monument is immense, roughly fifteen metres high and twenty-five metres wide, with relief figures and the same account in Old Persian, Elamite and Babylonian. Darius appears under the sign of Ahura Mazda. Defeated claimants stand bound before him. The text explains that he is the lawful king and they are liars.

The monument was not designed as a handbill for every passer-by. Its position, scale, languages and royal image established authority materially. The inscription addressed officials and literate intermediaries, commemorated victory and made the cliff carry the king's claim. Ancient state communication often worked this way: coins, decrees, triumphs, temples, public oaths and monuments joined a story to visible order.

The audience was layered. A scribe read for an official, a priest repeated the authorised account, a local ruler copied the royal formula and a coin carried the sovereign's face through transactions. Orality extended texts beyond readers. Ritual turned the message into an event. The state did not need every subject to inspect the inscription. It needed enough intermediaries to make the king's version the version attached to taxes, courts and armed power.

Ashoka's edicts placed moral and administrative messages across a different empire. Augustus later left the Res Gestae, an account of offices, gifts, wars and honours that turned one man's accumulation of power into a record of service. These were not modern mass campaigns. Literacy was limited, copying uneven and local interpretation unavoidable. The durable pattern was already present. Power explained itself, selected its past and made public objects certify the selection.

The word arrives after the practice

Print changed the scale and speed of political dispute. Reformation pamphlets, woodcuts and translated scripture let religious conflict travel through towns faster than church or prince could settle it. States answered with licensing, confiscation, sermons and their own presses. Control of doctrine became control of distribution.

The word propaganda entered this history through the Catholic Church. In 1622, Pope Gregory XV established the Congregation de Propaganda Fide to coordinate the propagation of the faith. The term described spreading doctrine and did not yet carry its modern automatic insult. That change came later, as mass politics made organised persuasion both common and suspect.

Revolutions enlarged the audience. The French Revolution renamed streets and months, staged festivals, circulated prints, defined citizens and enemies, and made political language part of daily time. Napoleon paired military victory with bulletins and images that promoted his command. Across the nineteenth century, schools, conscription, expanding newspapers, railways, telegraphy and national ceremonies gave states stronger ways to address populations as nations. Flags and anthems became ordinary technology for making strangers feel represented by one symbol.

The American and Haitian revolutions, independence movements across Latin America and later nationalist campaigns showed that the same media could build claims against empires. Pamphlets named grievances, declarations addressed audiences beyond the battlefield and newspapers connected local action to a larger cause. Governments did not monopolise propaganda, but successful movements often inherited its channels when they became states.

The First World War builds offices

The First World War industrialised the work. Governments needed men, money, labour discipline, alliance support and endurance through losses on a scale few populations had experienced. They created offices to supply news, shape foreign opinion, produce posters and films, brief speakers and coordinate censorship.

Britain established Wellington House, its War Propaganda Bureau, in 1914. It recruited writers, commissioned publications and directed much of its early effort abroad, especially towards neutral countries. The state also managed news and cable access. Reports of German violence in Belgium drew on real atrocities, but public discussion also carried rumours and stories that were weakly checked or false. In the interwar years, exposed fabrications and the broad belief that atrocity reports had been invented fed suspicion of official war news, while later scholarship showed that many of the underlying crimes were real.

The United States created the Committee on Public Information by executive order on 13 April 1917. George Creel's organisation distributed government news, produced films and posters, worked with advertisers and organised volunteer speakers known as Four Minute Men. It also operated beside a system of wartime censorship and legal repression. The official self-description was publicity rather than propaganda. The distinction was politically useful, but the machinery joined information, persuasion and control.

All belligerents worked under constraints. Paper, literacy, language, local trust and battlefield events set limits. Soldiers compared heroic imagery with mud, wounds and shortages. Families read casualty lists. Rumour travelled around official silence. Propaganda could make sacrifice intelligible; it could not indefinitely repeal experience.

The campaign entered ordinary rooms through objects. Posters asked for enlistment, loans, thrift and labour. Newsreels brought official images into cinemas. Volunteer speakers used the pause between reels to deliver a compact argument. Postal censorship sampled soldiers' letters for security and morale, while control of cables affected what neutral countries could receive. The message system gathered information as it distributed it. Reports on rumours, press reaction and audience mood flowed back to the offices, though officials often preferred evidence that justified the campaign already under way.

Germany, France, Austria-Hungary, Russia and the Ottoman Empire built their own mixtures of press control, patriotic culture, military reporting and foreign persuasion. Unequal access to global cables and news agencies mattered. Propaganda power depended partly on communications geography: which state could move its account across oceans before the rival account arrived.

After 1918, the word darkened. Citizens had seen modern states mobilise professional communication and concealment. Harold Lasswell studied propaganda as a technique of managing collective attitudes. Edward Bernays used the term openly in 1928 while linking wartime publicity to peacetime opinion management. The democratic fear was no longer that rulers would issue one lie. It was that organised minorities had learned to manufacture the environment in which mass judgement occurred.

Parties try to build total worlds

The Bolsheviks inherited a vast, poor, multilingual country damaged by war and revolution. Their answer was a propaganda state in which party organisation, education, agitation and administration grew together. Agitprop joined agitation, aimed at immediate political action, with propaganda, understood in party usage as fuller explanation. Posters, newspapers, theatre, literacy campaigns, trains and party workers carried messages into places where printing and schooling were weak.

Agitation trains carried presses, libraries, film equipment and speakers along the rail network. A village could receive a screening, a speech, printed material and officials in one visit. The train was a channel and a demonstration of central reach. Soviet designers also treated the poster as compressed instruction, using bold geometry, photomontage and legible types rather than assuming a long attentive reading.

The system sought transformation, not a favourable opinion poll. It named classes, enemies and historical stages. It taught people how to interpret scarcity, violence and delay. Failures could be assigned to wreckers or insufficient commitment. Successes became proof that the theory and party were moving history. Under Stalin, censorship, terror, controlled culture, falsified images and the leader cult narrowed the permitted public world with lethal force.

Fascist systems organised differently but shared the ambition to join media to mobilisation. In 1933, Nazi Germany created a Ministry of Public Enlightenment and Propaganda under Joseph Goebbels. The regime coordinated press, radio, film, culture and ritual while excluding Jews and political opponents from public life, destroying organisations and enforcing terror. Cheap radio receivers broadened access to speeches and entertainment. The message did not float above coercion. It arrived inside a state that had removed competitors and could punish the audience.

The Nuremberg rallies joined architecture, uniforms, lighting, disciplined movement, speech and film into a picture of unity. Leni Riefenstahl's Triumph of the Will did not record a spontaneous nation gathering around Hitler. It transformed a planned party event into a cinematic nation. The spectacle showed power by making organisation visible, while the regime's police and legal order determined what absence could cost.

Research on radio reception in prewar Germany shows why chronology matters. Before Nazi control, greater radio access could work against the Nazis because national broadcasts opposed them. After control, the direction changed. Estimated effects also varied with prior antisemitism and local conditions. A transmitter had no timeless ideology. Its political effect depended on who controlled it, what else listeners could hear and what they already brought to the set.

Democracies enter the same war

The Second World War destroyed any clean claim that propaganda belonged only to dictatorships. Britain created a Ministry of Information. The United States established the Office of War Information. Governments produced films, posters, broadcasts and overseas material, censored military information and appealed to duty, fear, hatred and national purpose.

Institutional differences still mattered. The BBC's international reputation depended in part on reporting bad news with enough accuracy that listeners treated it as more credible than Axis broadcasting. British covert operators also ran black stations that pretended to speak from inside Germany. Open truthfulness and concealed sponsorship existed within the same national effort because they served different audiences and tasks.

The war was global, and so were its claims. Imperial powers described a struggle for freedom while ruling colonies. Japan presented the Greater East Asia Co-Prosperity Sphere as liberation from Western empire while imposing its own violent domination. Anti-colonial leaders could use the contradiction. A message may mobilise one audience and supply evidence to another against the sender.

Leaflets dropped over enemy territory, broadcasts to occupied Europe and clandestine newspapers addressed people living under different risks. Reception could itself be criminal. Listeners had to judge whether a station was authentic, whether military news was credible and whether promised liberation would arrive. Propaganda worked through expectations about future power as well as interpretations of the past.

Propaganda also helped prepare and sustain mass murder. Nazi media had spent years defining Jews as an alien, conspiratorial and contaminating enemy. Words did not mechanically produce the Holocaust, which required institutions, orders, deportation systems, collaborators, armed units and war. They helped create categories in which persecution could be explained, normalised and concealed.

The Soviet war effort also shifted its public language. International revolution receded while defence of the motherland, historical heroes and German invasion supplied a broader patriotic frame. States change stories when survival demands a coalition wider than the doctrine that previously defined them. A propaganda system can be rigid in public and highly adaptive in production.

The Cold War fights over credibility

After 1945, nuclear deterrence made direct conflict between the superpowers exceptionally dangerous, while decolonisation produced new states and audiences. Broadcasting, publishing, education, exhibitions, aid, sport, science, art and popular culture became strategic terrain.

The United States funded the Voice of America and, through covert channels, Radio Free Europe and Radio Liberty. The Soviet Union and its allies built their own international broadcasting systems and jammed hostile signals. Each side promoted a social order while exposing the other's inequalities. American claims about freedom met segregation and racial violence. Soviet claims about workers met repression, shortages and tanks in allied capitals. Credibility turned on whether audiences could connect the grand account to observable facts.

Jamming made this contest audible. Listeners searching for a foreign station heard noise placed over the signal by their own state, proof that somebody feared the programme even before its content became clear. Broadcasters invited letters, monitored defector testimony and studied reception reports, but audience measurement remained partial. People who wrote to a foreign station were not a neutral sample of those who listened.

Covert sponsorship let magazines, conferences and artistic programmes borrow the credibility of independence. Some work retained intellectual value, but concealed funding denied audiences and participants essential information about control.

Newly independent governments used radio, schools, monuments and national languages to turn inherited borders into political communities. They faced real needs for public health, literacy and state-building, as well as temptations to identify party with nation and criticism with betrayal. Maoist China pushed mobilisation further through study sessions, posters, quotations, model performances and repeated campaigns. During the Cultural Revolution, political language and ritual entered workplaces, schools and families while factional violence and shifting signals made correct performance a matter of survival.

Television and satellites later changed the contest again. A moon landing, Olympic ceremony or summit could be watched across borders as evidence of technical competence and political modernity. The event supplied its own image. Yet interpretation remained contested: a display of abundance could inspire admiration, expose inequality or look like waste, depending on the audience and the life against which it was compared.

When speech enters the killing system

In Rwanda, RTLM began broadcasting in 1993 with music, jokes and an informal style that distinguished it from stiff official radio. As civil war and extremist mobilisation intensified, it spread hatred against Tutsi people, attacked moderate Hutu figures, named individuals and encouraged violence. During the 1994 genocide, radio operated beside government authority, militia organisation, local administration, roadblocks and face-to-face pressure.

The historical and legal record establishes that broadcasting formed part of the extremist mobilisation. Quantifying its effect is harder. A 2014 study used terrain-driven variation in RTLM reception and gacaca trial records to estimate that access increased participation in killing, including social spillovers. That estimate became one of the most cited demonstrations of media effects in conflict.

A 2024 study rebuilt two key measures. Its radio map covered more of Rwanda, and its outcome data identified individuals rather than counting proceedings in which the same person could appear more than once. With those revisions, areas receiving RTLM were more likely to experience an immediate onset of violence, but reception was not significantly associated with total participation at the subnational level. The authors cautiously interpret the timing result as consistent with radio helping some places reach an early critical mass, while finding no significant association with participation totals.

The disagreement matters because the original result and the later result measure related but different claims with different data. The responsible conclusion is not that radio was irrelevant, nor that one transmitter produced a known share of the genocide. RTLM supplied incitement, targets, rumours and cues inside an organised killing system. Evidence is stronger for its operational presence and possible effect on timing in some places than for one settled national estimate of how many people it caused to participate.

The International Criminal Tribunal for Rwanda's Media case drew a demanding legal distinction between hateful advocacy and direct, public incitement to commit genocide. Context, wording and how the intended audience understood the message can matter, but criminal liability cannot rest on the loose judgement that speech was poisonous. Propaganda is broader than incitement. It can recruit, excuse or obscure without issuing a legally direct appeal to commit genocide.

The network becomes the battlefield

The internet weakened national control over publication and created new points of control over discovery. Search engines, platforms, recommendation systems, hosting providers, mobile networks and app stores became intermediaries. States could publish directly, pressure companies, block services, sponsor creators, deploy covert accounts, steal and release documents, or flood discussion until attention rather than paper became scarce.

Contemporary campaigns often separate identity from authorship. A persona can pose as a local activist. A site can present state material as independent news. Automated accounts can enlarge the apparent popularity of a claim, although bots are not required for coordinated manipulation and raw post counts do not measure persuasion. Influence operations may target foreign audiences, domestic minorities, diasporas, journalists or the decision-makers who monitor public reaction.

Hack-and-leak operations join collection to propaganda. Stolen material is released at a chosen moment, mixed with interpretation and amplified through outlets whose relationship to the collector may be hidden. The documents can be authentic while the selection is strategic. Verification of the files therefore answers only one question. Timing, omission, sponsor and distribution answer the rest.

Russia's high-volume, rapid and contradictory campaigns have been described through the firehose model. China combines deletion and blocking with distraction, friction and positive narrative production. These are not interchangeable national essences. They are repertoires used by institutions under particular political and media conditions. Other states, including democracies, conduct strategic communication, covert influence and information operations. The legal authority, oversight, target and opportunity for challenge remain decisive distinctions.

Generative AI lowers the cost of producing fluent variants, synthetic voices, images and fictional participants. A preregistered 2024 experiment with 8,221 United States respondents found that GPT-3 articles could approach the immediate attitude shift produced by authentic covert propaganda texts under controlled exposure. A 2025 study of one Russian-affiliated campaign, observed from April to November 2023, found that generative tools expanded output and thematic breadth without an obvious loss of persuasiveness in a survey test. Neither finding establishes durable field conversion or an election effect. Production can become cheap while trust, distribution, timing and action remain difficult.

The new machine therefore retains the old structure. A sponsor needs a useful story, access to a channel, an audience with relevant priors, visible social cues and some route from exposure to consequence. Synthetic media changes supply. It does not abolish politics.

It also creates a liar's dividend. Once convincing fabrication is common knowledge, a public figure can dismiss authentic evidence as synthetic. Verification tools must therefore prove provenance as well as detect manipulation. The danger is two-sided: false material may be accepted, and true material may become easier to deny.

How we know

Propaganda leaves abundant material and poor counters. Archives preserve instructions, budgets, scripts, posters, films, transmitter records and censorship orders, which reveal what producers intended and could distribute. They do not show what every recipient heard, believed or did. Diaries, letters, interviews, audience reports and police files help, but literate, surviving and politically visible people are overrepresented.

Circulation and receiver ownership measure possible exposure, not attention. Rally attendance may measure conviction, coercion or transport capacity. Surveys inside repressive systems invite concealment. Platform disclosures describe detected operations and omit the unknown denominator.

Causal studies can exploit signal strength, timing or staggered access, yet their answers depend on coverage maps, outcome definitions, spillovers and local institutions. The Rwanda dispute shows the stakes: a 2014 analysis estimated participation effects, while a 2024 study using revised reception and individual-level conviction data found earlier onset but no significant association with participation totals. Laboratory experiments isolate messages and measure immediate responses more cleanly, but cannot by themselves establish durable field conduct.

The safest account combines production, distribution, reception and outcomes, then keeps each claim at the level its measure can carry. Propaganda is easier to collect than to quantify.

What People Get Wrong

“Propaganda is just lying”

A lie can be propaganda, but falsity is neither necessary nor sufficient. Private fraud is not automatically propaganda, and an official campaign can be built from accurate material.

The stronger method is selection. Show one real atrocity and hide another. Publish a true rise without the longer decline. Quote an opponent accurately while removing the sentence that changes the meaning. Use an authentic photograph from the wrong place. Every component may survive a narrow check while the assembled picture misleads.

Silence can carry the largest distortion. A casualty bulletin may contain no false number while excluding civilians, allies or an inconvenient period. An official film may show real factories and conceal stoppages outside the frame. Truth is not self-interpreting once a sponsor controls the sample.

The lie model became persuasive because the most notorious campaigns contain grotesque falsehoods and because calling a claim propaganda often substitutes for proving it wrong. That distinction matters because audiences trained only to catch fabrication remain open to framing, omission, source concealment and disproportionate repetition. Ask whether the claim is true, then ask why this claim, from this source, in this sequence, with these alternatives missing.

“Only dictators use propaganda”

Dictatorships can integrate propaganda with censorship, monopoly and punishment more completely. They do not own the category.

Democratic governments have recruited soldiers, sold war bonds, managed morale, sponsored overseas broadcasting, concealed operations, promoted national stories and funded communication whose sponsor was not always visible. The First and Second World Wars provide abundant examples. The difference is not that democracies communicate while dictatorships propagandise. It lies in legal restraint, plural sources, opposition, disclosure, oversight and the audience's power to replace the government.

Those safeguards can weaken during war and emergency. Secrecy expands, dissent attracts suspicion and officials claim that criticism helps the enemy. A democracy therefore needs scrutiny most when unity seems most useful. The existence of elections does not make every state narrative honest.

The myth survives because propaganda is a hostile label. States call their own output information, education or strategic communication and reserve propaganda for enemies. That linguistic border protects self-scrutiny. Applying the same standard to friendly governments is essential: transparent sponsorship, contestable evidence, proportionate secrecy and institutions able to expose manipulation.

“A clever message can brainwash anyone”

The fantasy of brainwashing grants the propagandist a direct line into an empty mind. Evidence gives a less tidy picture.

Effects vary with prior beliefs, identity, trust, competition, local experience, social networks and coercive context. German radio had different estimated effects before and after Nazi control. RTLM formed part of Rwanda's organised system of extremist power and local violence, while quantitative studies now disagree about its effect on participation totals. A message persuasive in a laboratory may fail amid distraction, contrary sources and consequences.

People also choose exposure. Supporters seek friendly outlets, opponents avoid them and uncertain citizens may pay little attention. Compulsory settings change that pattern but introduce reactance and performance. The propagandist never controls all the conditions implied by the message.

The myth grew from wartime fear, totalitarian spectacle and the understandable wish to explain mass cruelty through an alien technique. It can excuse perpetrators and flatter observers: they were programmed; we would recognise the trick. The correction does not make propaganda harmless. It identifies the conditions that make it dangerous and the institutions that can interrupt the path from exposure to action.

“People who recognise propaganda are immune”

Labelling a message does not dissolve its effects. You may reject the stated conclusion while absorbing its agenda, categories or emotional associations. Repeated exposure can make a claim familiar even when every encounter is hostile. Mockery can enlarge distribution. Constant warnings can make a minor actor seem central.

Educated audiences have no exemption. Expertise can sharpen scrutiny, but political commitments can redirect it towards defending a favoured conclusion. Cynicism creates a separate weakness. The belief that all sources manipulate equally erases the difference between a claim with inspectable evidence and one protected from correction.

A campaign can affect you without winning the argument you noticed. It may decide which topic feels urgent, which group appears suspicious or which policy is treated as the only serious option. Rejecting the headline does not restore the alternatives that never reached attention.

The immunity myth survives because detecting a crude poster feels like seeing through the machine. Modern propaganda often works through source choice, volume, social proof and omission rather than one slogan. The useful defence is procedural: slow the claim, trace provenance, compare independent evidence, inspect the missing denominator and remain willing to revise your own side.

“Censorship and propaganda are separate tools”

One removes information and the other supplies it, so the concepts look opposed. In practice they often form one system.

A repeated official claim gains strength when alternatives are blocked, delayed or made dangerous to share. Censorship need not erase every criticism. It may target coordination, timing or reach, allowing private complaint while preventing people from discovering that others share it. Propaganda then fills the visible space with approved explanations and activity.

The archive can later look deceptively open. A report published after the decision, or criticism available only to specialists, proves that information existed. It does not prove that citizens could use it when action was possible. Availability must be dated and socially located.

The separation persists because organisations and academic fields often study production and suppression apart. It also lets a government present positive messaging as benign while treating restrictions as unrelated security measures. The practical consequence is that message quality cannot be judged without the comparison environment. A claim heard among open competitors is different from the same claim made compulsory after competitors disappear.

“More information defeats propaganda”

Scarcity can aid control, but abundance can do the same. A person with ten incompatible accounts, hundreds of clips and no trusted method may know less than a person with two inspectable sources.

High-volume campaigns exploit this. They repeat, contradict, accuse, distract and force opponents to spend attention on endless correction. The objective may be to establish one belief, but it may instead be to make stable belief feel impossible. Under that condition, withdrawal and fatalism favour organised power.

Volume also creates false independence. Twenty accounts may copy one briefing, one leaked document or one coordinated script. Counting outlets exaggerates confirmation unless their evidence and ownership are traced upstream. More voices can mean one source wearing twenty faces.

The myth arose from an earlier internet hope that publication alone would break gatekeepers. Access has widened, yet ranking, attention, provenance and harassment still shape what becomes usable knowledge. So the remedy is not raw content. It is reliable retrieval, transparent sourcing, independent verification, protected journalism and institutions able to say what they know, how they know it and what remains uncertain.

“Digital propaganda changed everything”

Platforms, automation and generative media changed speed, scale, targeting, participation and the cost of producing variants. They did not remove the old constraints.

A synthetic video still needs distribution. A covert account still needs credibility. A bot network can inflate visible activity without changing a durable belief. A language model can produce persuasive text in an experiment while a field campaign fails through poor timing, weak trust or exposure to the wrong audience. Cheap supply does not guarantee political conversion.

Microtargeting is constrained by noisy data and by the same trade-off faced by older campaigns. A narrow message may fit one audience while contradicting material seen by another. Digital archives make inconsistency easier to discover even as production makes it easier to create.

The claim of total novelty is attractive to technology companies, security contractors and frightened institutions because it makes every event look unprecedented. It also hides continuity. States have long concealed sponsors, forged sources, flooded channels, recruited intermediaries and staged apparent consensus. The practical lesson is to protect provenance, competition, correction and institutional trust rather than chase each new format as though politics began with the latest tool. Novel tools alter tactics; they do not remove the need for an audience, an institution and a consequence.

Use It

Name the objective and audience

Do not begin with the isolated claim. Identify the political task and the people whose response matters.

Is the sponsor seeking sincere belief, visible compliance, turnout, endurance, fear, uncertainty, silence, reassurance or a cue for officials? Is the target the domestic public, a professional group, an ally, an enemy force, a diaspora, foreign journalists or the leaders who monitor those audiences? One message can do different work for each. Write the intended response in a verb: enlist, wait, report, obey, distrust, defect, endure.

Then reconstruct the world in which that response appears proper. Who belongs inside the nation, class, faith or moral community? Who appears as victim, protector, parasite, traitor or expert? Which past event explains the present? What future is promised? Which action counts as courage, and which question counts as disloyalty? A campaign becomes coherent when these roles line up across speeches, images, school material, commemorations and news selection.

This method catches distortion that sentence checking misses. Ten accurate items can build a false world if they are the only ten items repeatedly chosen. It also prevents overreaction. States cannot communicate without selecting and framing. The warning sign is that the chosen account becomes insulated from contrary facts, rival identities and peaceful challenge.

Follow the distribution advantage

A message's reach is never natural. Somebody owns the press, transmitter, account, classroom, cinema slot, search position or recommendation path. Somebody can delay a correction or make it expensive to find.

Map the route. Who produced the material? Who paid? Which intermediary delivered it? Was exposure chosen, habitual or compulsory? Which rules made it prominent? Which competitors were blocked, deprioritised, intimidated or drowned? A state may have weak arguments and overwhelming distribution. That is still power.

On digital networks, count less and trace more. Ten thousand posts may come from a small coordinated operation, while one trusted local intermediary may matter more than all of them. Viral reach is not persuasion, and reported impressions are not verified human attention. Look for the handoff from online visibility to journalists, officials, community groups or offline action.

The same map improves correction. Put rebuttal where the audience encountered the claim, use a source it has reason to trust and avoid repeating the falsehood more prominently than the evidence.

Separate belief from behaviour

When a campaign seems successful, ask what was measured.

A person can remember a slogan without accepting it, accept a claim without supporting a policy, support a policy without acting, and act once without persisting. Under coercion, attendance, applause, voting and public speech may show that the state can enforce a performance. That matters, but it is not the same as consent.

Use an outcome ladder: exposure, attention, comprehension, belief, evaluation, intention, action and persistence. Place each piece of evidence on the ladder. Do not let movement at one step stand in for all the others.

Then inspect social expectations. Did people act because they believed the official account, because they believed others believed it, or because they believed refusal would be visible? These mechanisms demand different remedies. Better facts can correct a mistaken belief. Protected association may be needed when the problem is isolation. Legal protection and institutional change are needed when the behaviour is produced by punishment.

This distinction also restores agency. It avoids treating every compliant citizen as a fanatic while refusing to pretend that reluctant participation has no consequences. It keeps moral judgement attached to conduct without inventing certainty about inner belief.

Test the missing comparison

Propaganda often wins by choosing the denominator.

A government announces that production rose, without naming the target, starting point or cost. A war report counts enemy losses and omits its own. A map begins at the border that makes one history disappear. A video compares the present with the worst year available. None requires a fabricated number.

Ask five questions. Compared with when? Compared with where? Out of how many? Measured by whom? What changed in the definition? These are small acts of resistance because they restore alternatives to a picture designed to feel complete.

Apply the test to images and anecdotes too. Is the scene typical, selected or staged? Does the named victim represent a pattern? Would the same source accept the evidential standard if the parties were reversed? What event of similar scale received no coverage?

The aim is not false balance. Some comparisons are irrelevant, and documented aggression does not become uncertain because every side issues claims. The aim is compatible measurement and proportionate inference. A strong conclusion should survive a fair denominator and a comparison chosen before the desired result is known.

Protect the return channel

Citizens usually think about propaganda as material travelling outwards from the state. The more dangerous failure may be information travelling back.

In any organisation, ask whether the person reporting bad news can survive the report. Who owns the measurement? Can an auditor inspect the underlying data? Are targets tied so tightly to status that people must falsify them? Does the leader hear from more than one chain of command? Are uncertainty and dissent recorded, or converted into a polished consensus before they reach the decision?

These questions apply to companies, campaigns and public agencies as well as governments. A dashboard can become propaganda when it is designed to defend a programme rather than test it. The numbers may be real and the selection corrupt.

Build redundancy. Independent statistics, free media, inspectors, courts, legislative scrutiny, professional standards, whistleblower protection and protected opposition create competing routes for evidence. They produce friction and embarrassment. That is their information function.

A system that treats every correction as an attack may preserve unity for a while. It is also training everyone below the centre to send reassurance instead of reality.

The limits

No checklist makes a person immune, and no clean boundary separates every public message from propaganda.

States must communicate. During emergencies they may need speed, secrecy and behavioural guidance. A public-health warning can be urgent, emotionally framed and repeated without becoming equivalent to a fabricated enemy campaign. The relevant questions concern evidence, proportionality, sponsor visibility, opportunity for review and whether uncertainty is concealed beyond what action requires.

Truth alone does not settle ethics. A government can publish a true private fact to intimidate a dissident. A complete military disclosure can endanger lives. A covert broadcast may give censored audiences valuable information while hiding its sponsor. Judgement cannot be replaced by the rule that openness is always good or state persuasion always bad.

Nor does propaganda explain every political outcome. Economic interests, institutions, coercion, identity, organisation, military events and ordinary experience have independent force. Messages work inside those conditions. Describing propaganda as total mind control gives communicators credit that belongs to police, parties, employers, neighbours and audiences themselves.

The practical objective is modest: improve provenance, competition, measurement and correction; reduce punishment for truthful disagreement; and keep claims no broader than the evidence.

The one thing to keep

Keep the return path open.

When you encounter state messaging, ask more than whether it is true. Ask who can publish, who can answer, which response is made visible and whether bad news can reach the people acting on the message. That single lens joins the citizen's problem to the ruler's.

A propaganda system grows powerful when it can define the public picture, occupy the main channels and make conformity observable. Yet those same achievements create its weakness. Officials learn which facts advance careers. Journalists learn which losses may be named. Citizens learn which doubts must remain private. The centre receives the result as evidence that the picture was accurate.

Open societies do not avoid national stories, selective attention or organised persuasion. Their advantage is the possibility of correction through institutions that do not need permission from the author of the claim. That possibility must be defended before a crisis, because a channel invented after the report becomes dangerous is often too late.

The permanent question is therefore: what happens to the person who brings contrary evidence?

If the answer is that the person is heard, checked and protected, propaganda meets a limit. If the answer is silence, dismissal or punishment, the public may still be the intended audience. The state is becoming the captive one.

Terms

Propaganda. Organised communication designed to advance the interests or objectives of a sponsor by shaping public interpretation, expectation or conduct. In this book the sponsor is a state or an actor working for one.

Target audience. The people whose beliefs, expectations or conduct a campaign seeks to alter. Naming the target prevents effects on citizens, officials, allies, enemies and neutral observers from being treated as one result.

Persuasion. An intentional attempt to change another person's belief, attitude, intention or voluntary behaviour. Propaganda uses persuasion but also works through distribution, censorship, public signalling and institutional power.

Public information. Communication issued to explain services, risks, decisions or duties. It can be accurate and necessary. Sponsor transparency, evidence, proportionality and opportunity for correction help distinguish it from manipulative practice.

Strategic communication. The coordinated use of words, images, actions and institutional signals to support policy objectives. Governments prefer the term because it sounds neutral, though the underlying activity may include propaganda.

Public diplomacy. A state's effort to communicate with foreign publics rather than only with foreign governments. Broadcasting, exchanges, cultural programmes and educational work can build relationships or promote strategic narratives.

Psychological operations. Planned military communication intended to influence the perceptions and behaviour of selected audiences. Modern forces often abbreviate the term to PSYOP. It is one specialised branch of state influence.

Censorship. The suppression, delay, alteration or deterrence of expression. It can operate through law, ownership, licensing, technical blocking, intimidation or friction, and often strengthens propaganda by weakening alternatives.

Misinformation. False or inaccurate information shared without necessarily intending deception. The term describes content and transmission, not a sponsor's full political system or motive.

Disinformation. False or misleading information deliberately created or distributed to deceive. A disinformation operation may form part of propaganda, but propaganda can use true material and open sponsorship.

Malinformation. Genuine information used to cause harm, often through selective release, decontextualisation or invasion of privacy. The category shows why truth does not settle the ethics of an information campaign.

White propaganda. Material whose sponsor is accurately identified. Open attribution can support accountability and credibility, though a named source can still select, frame or distort evidence.

Grey propaganda. Material whose source is unclear or incompletely disclosed. Ambiguity lets a sponsor borrow the appearance of independence while avoiding the risk of a fully fabricated identity.

Black propaganda. Material falsely attributed to another source, often an enemy, dissident group or independent outlet. Exposure can damage the operation and any truthful material carried within it.

Agitprop. A Bolshevik and Soviet combination of agitation and propaganda. Agitation pressed a limited, immediate political message; propaganda supplied a broader doctrine. In practice the institutions overlapped.

Atrocity propaganda. Communication that uses reports or images of enemy cruelty to mobilise hatred, endurance or intervention. The atrocities may be real, false or mixed; verification and selection both require scrutiny.

Framing. The selection and organisation of features that make one interpretation more available than others. Calling an event a riot, uprising, pogrom or liberation struggle assigns causes and moral roles before detail arrives.

Agenda-setting. Influence over which subjects receive public attention. A campaign can shape politics without changing anyone's opinion by making one problem unavoidable and another nearly invisible.

Priming. Making a consideration more accessible so that people use it when judging a person or event. Repeated security coverage may cause leadership to be evaluated chiefly through perceived strength.

Common knowledge. Information that people know, know others know, and know others know they know. Public ceremonies and broadcasts can coordinate action by making expectations mutually visible.

Preference falsification. Publicly expressing a preference different from one's private view because honesty carries social, economic or political cost. Widespread falsification can make a regime look more supported than it is.

Pluralistic ignorance. A condition in which many people privately reject a norm while each mistakenly thinks most others accept it. Controlled public expression can preserve the error.

Cult of personality. The organised elevation of a leader into the embodiment of nation, party or historical destiny. Portraits, biography, ritual and selective history convert political loyalty into personal devotion.

Dehumanisation. The representation of people as animals, disease, filth, objects or an undifferentiated mass. It weakens ordinary moral restraints by placing a targeted group outside the protected human community.

Incitement. Communication intended and sufficiently connected to prompting prohibited action under the relevant legal standard. Direct and public incitement to commit genocide is an international crime even if genocide does not follow.

Hate speech. Expression attacking people on protected or identity grounds. Legal definitions vary by jurisdiction. Hatred can support propaganda, but hateful expression and legally punishable incitement are not interchangeable.

Computational propaganda. The use of data, automation, algorithms and networked accounts to shape public discussion. The term covers coordinated human work as well as bots and should not imply that software acts alone.

Bot. An automated account or process that posts, shares or reacts according to programmed rules. Bots can enlarge apparent activity, but automation does not establish the sponsor, audience size or persuasive effect.

Deepfake. Synthetic or manipulated audio, image or video that convincingly depicts a person doing or saying something. Cheap fabrication raises verification costs, while authentic material can also be dismissed as fake.

Active measures. A Soviet term for covert political influence beyond conventional diplomacy, including front groups, forged documents, disinformation and support for aligned organisations. Similar methods predate the label and outlived the Soviet state.

Go Deeper

The long history

Philip M. Taylor, Munitions of the Mind: A History of Propaganda from the Ancient World to the Present Era, third edition (Manchester University Press, 2003). Start here for the chronological sweep. Taylor moves from ancient monuments and religious conflict through total war, broadcasting and post-Cold War information campaigns, which makes it easy to place unfamiliar cases. The final chapters predate platforms and generative media, so use the book for historical structure rather than a current map of digital influence. It remains readable, broad and unusually alert to the connection between propaganda and war. Its range also makes clear that the twentieth century intensified old practices rather than inventing political image-making from nothing.

The primary witness

Victor Klemperer, The Language of the Third Reich: LTI, Lingua Tertii Imperii: A Philologist's Notebook, translated by Martin Brady (Bloomsbury Academic, 2013). Klemperer was a Jewish scholar living under Nazi rule who recorded how official language entered newspapers, workplaces and ordinary speech. This is not a complete history of Nazi propaganda. It is more valuable for showing repeated words changing what could be said and therefore what could be thought without effort. Read it slowly. Its strength comes from accumulation rather than one grand theory. The book also shows why vocabulary cannot be dismissed as surface decoration once institutions make particular words the routine route to status and safety.

The autocratic system

Erin Baggott Carter and Brett L. Carter, Propaganda in Autocracies: Institutions, Information, and the Politics of Belief (Cambridge University Press, 2023). This is the strongest next step for readers who want modern comparative evidence. The authors treat propaganda as a strategic choice shaped by institutions, threats and the need to persuade or intimidate. The book uses a global dataset and formal argument, so it is more demanding than Taylor. It also corrects the assumption that every autocracy communicates in the same way. Its distinction between persuasion and threatened domination is especially useful beside the fifth Core Idea in this book.

The digital mechanism

Margaret E. Roberts, Censored: Distraction and Diversion Inside China's Great Firewall (Princeton University Press, 2018). Roberts explains why incomplete censorship can work. Fear, friction and flooding can shape attention without making every forbidden item impossible to find. The book is centred on China and should not be universalised without care, but its account of small access costs and diverted attention travels well. Read it for method too: digital traces, leaks and experiments are combined to connect information control with measurable behaviour. The warning is scope. China supplies the evidence, while the mechanisms require fresh testing before they are treated as universal.

Notes and Sources

Scope and definition

The book uses propaganda for organised communication that advances a state sponsor's objectives by shaping public interpretation, expectation or conduct. This is narrower than every biased message and broader than deliberate falsehood. Taylor supplies the long historical frame; Jowett and O'Donnell clarify the relation to persuasion and source attribution; Carter and Carter show how objectives and propaganda strategies vary across political institutions. The distinction among white, grey and black propaganda concerns whether sponsorship is open, uncertain or falsely attributed.

The Whole Thing in One Page and Why You Should Care

The account of Nicolae Ceaușescu's rally concerns 21 December 1989 and follows the surviving broadcast and Siani-Davies's historical chronology. The interruption did not by itself overthrow the Romanian regime. It broke a planned display of unanimity during an existing political crisis. Chwe provides the common-knowledge model, while Kuran and Wedeen explain why public performance can diverge from private belief.

The legal discussion is deliberately narrow. Article 20(1) of the International Covenant on Civil and Political Rights requires any propaganda for war to be prohibited by law. Article 20(2) requires the same for advocacy of national, racial or religious hatred that constitutes incitement to discrimination, hostility or violence. Article III(c) of the Genocide Convention makes direct and public incitement to commit genocide punishable. These rules do not create a general international offence of tendentious state communication.

The Core Ideas

Common knowledge is used as a coordination concept, not as proof that every ceremony produces one response. Chwe develops the mechanism. Kuran's preference-falsification model and Wedeen's study of public ritual show how visible compliance can persist without matching private conviction. Huang's signalling account supports the narrower claim that conspicuous propaganda can communicate regime capacity even where its factual content is unpersuasive.

The censorship discussion follows King, Pan and Roberts on collective-expression material in Chinese social media and Roberts on fear, friction and flooding. These findings arise from particular Chinese institutions and platforms. The text does not treat them as a universal law of censorship.

The section on credibility draws on the source-attribution typology in Jowett and O'Donnell, Taylor's histories of wartime information, and Cull's account of public diplomacy. Concealed Cold War cultural sponsorship is used to show the cost of hidden control, not to claim that every funded work was dictated by officials.

Claims about familiarity and motivated reasoning are kept at the level supported by experimental psychology. Fazio and colleagues show that prior knowledge did not always prevent repetition from raising perceived truth in their experiments. Taber and Lodge show selective scepticism and counter-arguing around political commitments in their study. Neither result makes recognition useless, proves that education worsens reasoning, or establishes a fixed response in every audience.

Emotion, dehumanisation and violence are treated as parts of wider political systems. Welch, Klemperer and United States Holocaust Memorial Museum material support the Nazi examples. The Rwandan account combines the ICTR Media case with two quantitative studies. Yanagizawa-Drott's 2014 signal-strength analysis estimated effects on participation and spillovers. Nyseth Nzitatira, Billing and Edgerton's 2024 reanalysis used revised coverage and individual-level conviction records, found earlier onset where RTLM could be heard, and found no significant association with subnational participation totals. The manuscript states the disagreement rather than selecting one estimate as settled.

Adena and colleagues show that German radio effects varied with political context and changed after Nazi control. The manuscript therefore rejects one permanent effect assigned to the medium itself. Carter and Carter support the broader claim that autocratic propaganda may persuade, intimidate, signal strength or respond to institutional threats.

The feedback argument is a cross-case synthesis rather than one universal estimate. Kuran, Wedeen, Yang, Wemheuer and the wider literature on authoritarian information problems support mechanisms of punished correction, inflated reporting and staged agreement. Their settings and outcomes differ. The Great Leap Forward famine had several political, economic and administrative causes. Information distortion is retained as one damaging mechanism, not a complete explanation of mortality or a template for every state.

How It Actually Works

The Behistun inscription is conventionally dated to the reign of Darius I around 521 BCE. UNESCO documentation supports the monumental setting, trilingual text and approximate dimensions used. Ashoka's inscriptions and statecraft are treated through Thapar. Augustus's public account is read through Cooley's edition of the Res Gestae. These ancient cases are precedents for organised rule through public claims, not examples of a continuous modern propaganda bureaucracy.

The Congregation de Propaganda Fide was established in 1622. The institutional date comes from the Vatican's historical account. The modern English suspicion attached to propaganda developed later. Print, confessional conflict and state formation are compressed through Taylor rather than assigned one technological cause.

The Committee on Public Information was created by United States Executive Order 2594 on 13 April 1917. National Archives records and Creel's own account establish its official news, publicity and mobilisation work. Creel is a participant defending his programme, so his interpretation is not treated as neutral. Horne and Kramer support the distinction between documented German atrocities in 1914 and the rumours, exaggerations and later denial that complicated their public memory.

Kenez supports the account of Bolshevik and early Soviet mass mobilisation. Welch, Klemperer, USHMM material and Adena and colleagues support the Nazi section. The book does not infer individual conversion from receiver ownership or broadcast reach.

Cull supports the Cold War broadcasting and cultural-diplomacy account. MacFarquhar and Schoenhals and Mittler support the Cultural Revolution material. The chronology gives examples of changing communication systems and does not attempt full histories of the states or conflicts named.

Paul and Matthews coined the firehose description for a Russian propaganda repertoire marked by volume, speed, repetition and inconsistency. The manuscript treats it as a model of a repertoire, not as an immutable national character. Goldstein and colleagues reported a preregistered experiment with 8,221 United States respondents in 2024 and measured immediate agreement after bounded exposure. Wack and colleagues published in 2025 using articles collected from one Russian-affiliated operation between April and November 2023, then tested perceived persuasiveness in a survey. Neither establishes durable election effects, global prevalence or universal field success. Chesney and Citron supply the liar's-dividend concept used for the increased ability to dismiss authentic evidence as synthetic.

What People Get Wrong and Use It

The seven corrections synthesise the evidence above. They separate truth from selection, regime type from communicative method, exposure from durable behaviour, recognition from immunity, censorship from message production, quantity from usable knowledge and technical novelty from political conversion. The practical outcome ladder is an analytical device, not a validated universal scale. Its purpose is to prevent recall, stated agreement, intention and action from being reported as the same result.

Terms

Legal terms are stated at the level needed to avoid conflation. Hate-speech law varies by jurisdiction. Direct and public incitement to commit genocide has a specific international-law history and should not be used as a loose synonym for hateful propaganda. The misinformation, disinformation and malinformation distinction follows the influential framework by Wardle and Derakhshan, though later usage varies. Psychological operations, public diplomacy, active measures and computational propaganda are institutional families whose labels and legal authorities vary across states and periods.

How we know

Production records reveal intention and capacity more readily than reception. Circulation, transmitter reach, receiver ownership, account counts and impressions are exposure opportunities, not direct measures of attention or belief. Natural experiments improve causal inference but depend on coverage models, outcome definitions, geography, access and spillovers. The conflicting Rwanda estimates show that a changed coverage measure or unit of outcome measurement can alter the conclusion. Laboratory experiments improve control while reducing field realism. Platform takedown reports contain detected operations and cannot supply a denominator for all operations attempted.

Bibliography

Primary, legal and institutional sources

Augustus. Res Gestae Divi Augusti: Text, Translation, and Commentary. Edited by Alison E. Cooley. Cambridge: Cambridge University Press, 2009.

Bernays, Edward L. Propaganda. New York: Horace Liveright, 1928.

Creel, George. How We Advertised America. New York: Harper & Brothers, 1920.

Dicastery for Evangelization. “History of the Congregation de Propaganda Fide.” Vatican City.

International Criminal Tribunal for Rwanda. Prosecutor v. Ferdinand Nahimana, Jean-Bosco Barayagwiza and Hassan Ngeze. Trial Judgment, ICTR-99-52-T, 3 December 2003; Appeals Judgment, ICTR-99-52-A, 28 November 2007.

Lasswell, Harold D. Propaganda Technique in the World War. New York: Alfred A. Knopf, 1927.

United Nations. Convention on the Prevention and Punishment of the Crime of Genocide. Adopted 9 December 1948.

United Nations. International Covenant on Civil and Political Rights. Adopted 16 December 1966.

United States Holocaust Memorial Museum. State of Deception: The Power of Nazi Propaganda. Washington, DC: United States Holocaust Memorial Museum, 2009.

United States National Archives and Records Administration. “Records of the Committee on Public Information.” Record Group 63.

UNESCO World Heritage Centre. “Bisotun.” World Heritage List.

Modern works and research

Adena, Maja, Ruben Enikolopov, Maria Petrova, Veronica Santarosa, and Ekaterina Zhuravskaya. “Radio and the Rise of the Nazis in Prewar Germany.” Quarterly Journal of Economics 130, no. 4 (2015): 1885-1939. DOI: 10.1093/qje/qjv030.

Carter, Erin Baggott, and Brett L. Carter. Propaganda in Autocracies: Institutions, Information, and the Politics of Belief. Cambridge: Cambridge University Press, 2023.

Chwe, Michael Suk-Young. Rational Ritual: Culture, Coordination, and Common Knowledge. Princeton: Princeton University Press, 2001.

Chesney, Robert, and Danielle Keats Citron. “Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security.” California Law Review 107, no. 6 (2019): 1753-1819.

Cull, Nicholas J. The Cold War and the United States Information Agency: American Propaganda and Public Diplomacy, 1945-1989. Cambridge: Cambridge University Press, 2008.

Fazio, Lisa K., Nadia M. Brashier, B. Keith Payne, and Elizabeth J. Marsh. “Knowledge Does Not Protect Against Illusory Truth.” Journal of Experimental Psychology: General 144, no. 5 (2015): 993-1002. DOI: 10.1037/xge0000098.

Goldstein, Josh A., Jason Chao, Shelby Grossman, Alex Stamos, and Michael Tomz. “How Persuasive Is AI-Generated Propaganda?” PNAS Nexus 3, no. 2 (2024): pgae034. DOI: 10.1093/pnasnexus/pgae034.

Horne, John, and Alan Kramer. German Atrocities, 1914: A History of Denial. New Haven: Yale University Press, 2001.

Huang, Haifeng. “Propaganda as Signaling.” Comparative Politics 47, no. 4 (2015): 419-437.

Jowett, Garth S., and Victoria O'Donnell. Propaganda & Persuasion. 7th ed. Los Angeles: SAGE, 2019.

Kenez, Peter. The Birth of the Propaganda State: Soviet Methods of Mass Mobilization, 1917-1929. Cambridge: Cambridge University Press, 1985.

King, Gary, Jennifer Pan, and Margaret E. Roberts. “How Censorship in China Allows Government Criticism but Silences Collective Expression.” American Political Science Review 107, no. 2 (2013): 326-343. DOI: 10.1017/S0003055413000014.

Klemperer, Victor. The Language of the Third Reich: LTI, Lingua Tertii Imperii: A Philologist's Notebook. Translated by Martin Brady. London: Bloomsbury Academic, 2013.

Kuran, Timur. Private Truths, Public Lies: The Social Consequences of Preference Falsification. Cambridge, MA: Harvard University Press, 1995.

MacFarquhar, Roderick, and Michael Schoenhals. Mao's Last Revolution. Cambridge, MA: Belknap Press of Harvard University Press, 2006.

Mittler, Barbara. A Continuous Revolution: Making Sense of Cultural Revolution Culture. Cambridge, MA: Harvard University Asia Center, 2012.

Nyseth Nzitatira, Hollie, Trey Billing, and Jared F. Edgerton. “How Radio Affects Violent Conflict: New Evidence from Rwanda.” American Sociological Review 89, no. 5 (2024): 876-906. DOI: 10.1177/00031224241278905.

Paul, Christopher, and Miriam Matthews. The Russian “Firehose of Falsehood” Propaganda Model: Why It Might Work and Options to Counter It. Santa Monica, CA: RAND Corporation, 2016. DOI: 10.7249/PE198.

Roberts, Margaret E. Censored: Distraction and Diversion Inside China's Great Firewall. Princeton: Princeton University Press, 2018.

Siani-Davies, Peter. The Romanian Revolution of December 1989. Ithaca: Cornell University Press, 2005.

Taber, Charles S., and Milton Lodge. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50, no. 3 (2006): 755-769. DOI: 10.1111/j.1540-5907.2006.00214.x.

Taylor, Philip M. Munitions of the Mind: A History of Propaganda from the Ancient World to the Present Era. 3rd ed. Manchester: Manchester University Press, 2003.

Thapar, Romila. Ashoka and the Decline of the Mauryas. 3rd ed. New Delhi: Oxford University Press, 2012.

Wack, Morgan, Carl Ehrett, Darren Linvill, and Patrick Warren. “Generative Propaganda: Evidence of AI's Impact from a State-Backed Disinformation Campaign.” PNAS Nexus 4, no. 4 (2025): pgaf083. DOI: 10.1093/pnasnexus/pgaf083.

Wardle, Claire, and Hossein Derakhshan. Information Disorder: Toward an Interdisciplinary Framework for Research and Policy Making. Strasbourg: Council of Europe, 2017.

Wedeen, Lisa. Ambiguities of Domination: Politics, Rhetoric, and Symbols in Contemporary Syria. Chicago: University of Chicago Press, 1999.

Welch, David. The Third Reich: Politics and Propaganda. 2nd ed. London: Routledge, 2002.

Wemheuer, Felix. Famine Politics in Maoist China and the Soviet Union. New Haven: Yale University Press, 2014.

Yanagizawa-Drott, David. “Propaganda and Conflict: Evidence from the Rwandan Genocide.” Quarterly Journal of Economics 129, no. 4 (2014): 1947-1994. DOI: 10.1093/qje/qju020.

Yang, Dali L. Calamity and Reform in China: State, Rural Society, and Institutional Change Since the Great Leap Famine. Stanford: Stanford University Press, 1996.