How propaganda works: shaping reality with words, emotions, and meaning

Can you lie without saying anything false?

The classic model is simple: the manipulator lies, and fact-checking exposes them. Reality is trickier: most influence over our picture of the world happens not at the level of individual statements but at the level of their selection, order, naming, and emotional coloring.

Philosophers of language distinguish between what is said and what is implied. In the headline "Mayor met with businessman later arrested for fraud," every word may be accurate, but the reader fills in the link: the mayor is involved. The same technique works in advertising ("no other cream moisturizes better" — that is, they're all about the same), in corporate reports ("revenue grew" — while profit falls), and in everyday life ("I'm not saying anything, but he's staying late at work again").

Compare how the same events sound in neutral and loaded presentation:

A caveat is needed here: the pairs in the right column are not always equivalent to the left. Sometimes a strong word is more accurate than a neutral one — if people really were fired without severance, "thrown out on the street" may be an honest assessment and "optimized" a euphemism. Neutrality does not equal truth, and emotionality does not equal falsehood. The real question is different: does the assessment replace the facts or follow from them, and does the author show what it's based on?

Manipulation works in both directions: "collateral damage" softens, "barbaric strike" intensifies. Anyone who sees propaganda only in other people's wording and considers their own "calling things by their proper names" has already fallen into the trap.

Fact-checking is necessary but not sufficient: a message can pass a fact-check completely and still distort the picture. That's why analysis needs to go layer by layer — what is claimed, what it's based on, what is implied, and what is left out. The method is described in detail in the piece on layers of claims and evidence.

What propaganda is and how it differs from informing

The word "propaganda" comes from Church Latin: in 1622, Pope Gregory XV established the Congregation for the Propagation of the Faith (de Propaganda Fide). The word acquired its negative tinge in the 20th century, after the campaigns of World War I and the totalitarian regimes. In 1928, Edward Bernays titled his book simply Propaganda and considered it an inevitable tool for managing a democracy. Once the word became a slur, the same activity was renamed "public relations" — a nice example of framing.

A working definition for this text: propaganda is a systematic attempt to shape an audience's perceptions, attitudes, and behavior in the source's interest, where the goal matters more than accuracy, and emotional and social impact matter more than a checkable argument. There is no word "lie" in this definition. The key features are systematicity, a predetermined conclusion, and asymmetry: the source knows where it's leading, the audience doesn't.

It helps to distinguish propaganda from neighboring phenomena:

• Informing aims to give the audience material for an independent conclusion. A good informational text shows uncertainty, alternative explanations, and the limits of the data. It isn't afraid that the reader will reach the "wrong" conclusion.
• Persuasion openly argues for a position. A columnist says: "This is what I think, and here's why." The reader understands they're looking at an opinion and can evaluate the arguments. Persuasion turns into propaganda when it hides its goal or replaces arguments with pressure.
• Advertising is also open influence: everyone understands that the seller is praising the product. But advertising actively uses propaganda techniques — emotional associations, social proof, manufactured scarcity — and the line blurs when an ad is disguised as an independent review or a news story.
• Agitation is a call to a specific action here and now: vote, show up, sign, buy. Agitation leads to action; propaganda prepares the ground for it.

The French sociologist Jacques Ellul, in his book Propaganda (1962), added an important distinction. Besides explicit, "political" propaganda, there is integration or sociological propaganda — a diffuse, often unmanaged pressure of lifestyles, norms, and standards transmitted by film, advertising, school, and mass culture. It doesn't say "vote for X"; it says "normal people live like this." It's almost impossible to notice because it coincides with the feeling of normality.

It follows that propaganda isn't only about politics. According to historical document investigations, the mid-20th-century tobacco industry deliberately cultivated "doubt about the science" as a strategy. A wellness blogger who presents all medicine as a conspiracy and their own courses as salvation uses the same techniques as a state TV channel. An activist on any side who publishes only convenient facts and portrays opponents as idiots or villains is doing the same thing. And a parent who tells a child only bad things about an ex-spouse is running a small propaganda campaign.

This doesn't mean all influence is equally harmful: a state monopoly on information and a family grudge are not the same thing. But the mechanisms are shared, and the skill of recognizing them is universal.

Framing: how words choose reality

Framing is the choice of the frame in which to present information: which aspects to emphasize, which words to use, what to compare it with, and which problem to name as the main one. A frame doesn't change a fact, but it changes what you can see in it.

The best-known example is the research of Amos Tversky and Daniel Kahneman in the early 1980s. In the "Asian disease" problem, people were asked to choose a program for fighting an epidemic. When outcomes were described in terms of lives saved, most chose the safe option; when the same outcomes were described in terms of deaths, most leaned toward the risky one. Mathematically, the options were identical. In a similar study by Barbara McNeil and colleagues (1982), patients, students, and even physicians were described a choice between surgery and radiation therapy. The wording "90% survive the first month after surgery" made surgery noticeably more attractive than "10% die in the first month" — although it's the same number. Effect sizes vary across replications, but the framing effect itself is one of the most robust findings in behavioral science.

The linguist George Lakoff carried this idea into politics. His book Don't Think of an Elephant! (2004) rests on a simple observation: you can't refute a frame by repeating it. If you're asked not to think of an elephant, you'll think of an elephant. If a politician says "tax burden" and an opponent replies "the tax burden isn't that heavy," the opponent has already agreed that taxes are a burden rather than, say, a payment for infrastructure. Many researchers criticize Lakoff's work for overly strong claims about cognitive mechanisms, but his practical observation — that vocabulary sets the field of the debate — is confirmed by everyday experience.

Framing works through several levers:

1. Word choice. "Refugees," "migrants," "illegals," and "newcomers" describe overlapping groups of people but trigger different associations — sympathy, neutrality, threat, everyday foreignness.
2. Choice of reference point. Growth of 5% looks like success compared to last year and like failure compared to the promised 10%.
3. Choice of scale. "A million dollars for the program" sounds huge; "three cents per person per year" sounds trivial. It's the same amount.
4. Choice of hero and victim. A story about a factory closing can be told through the eyes of a laid-off worker, a shareholder, an environmentalist, or a resident of a neighboring town whose river got cleaner.
5. Choice of metaphor. Crime as a "virus" calls for treating causes; crime as a "beast" calls for a cage. In a study by Paul Thibodeau and Lera Boroditsky (2011), a single metaphor shifted readers' preferences between prevention and punishment, although the effect later turned out to be more modest.

A frame is unavoidable. Every text emphasizes something and leaves something in shadow. Neutral presentation in an absolute sense doesn't exist. The task isn't to find a text without a frame but to notice the frame and ask which other frames are possible.

One fact, several realities

Try taking a single fact and running it through opposing frames. It's one of the best exercises against propaganda because it shows how much interpretation hides in "just the news."

Demographics. Fact: the country's birth rate has been falling for ten years in a row. Threat frame: "The nation is dying out; we'll be replaced." Progress frame: "People gained access to education and contraception and decide for themselves when to have children." Economic frame: "The labor market will shrink, and the pension system won't hold." Each frame rests on real data and suggests different solutions — from banning abortion to pension reform.

Technology. Fact: AI systems are automating part of office work. Threat frame: "Machines will take millions of jobs." Liberation frame: "People will be freed from routine and turn to creative work." Power frame: "A few corporations will reap the benefits." All three frames find confirming examples, and none describes the future in full.

Control and safety. Fact: the city has installed cameras with facial recognition. Safety frame: "Dozens of wanted criminals have been found." Freedom frame: "Every step citizens take is now tracked." Effectiveness frame: "The system makes mistakes, and the cost of a mistake is an innocent person detained." The argument about cameras often turns out to be an argument about which frame is "correct," when in fact each one illuminates its own real side.

Economy. Fact: annual inflation was 8%, while a year earlier it was 12%. Failure frame: "Prices keep rising; life is getting more expensive." Success frame: "Inflation slowed by a third." Both sentences are true. Which one makes the headline often depends on who's in power and whom the newsroom sympathizes with.

Education. Fact: a school abolished mandatory grades in the early years. Degradation frame: "Kids aren't being taught or held accountable anymore." Humanization frame: "Kids stopped fearing mistakes and learn with interest." Inequality frame: "Parents without resources lost a clear way to track progress."

A caveat without which the exercise turns into relativism: multiple frames don't mean they're all equally well-founded. Sometimes the data strongly supports one interpretation and only weakly another. The point of the exercise is not to say "everything is subjective" but to see a choice where previously there seemed to be only "the way it is." After that, you can ask substantive questions: what data supports each frame, what does it miss, and who benefits from it?

Emotions instead of arguments

Emotions are not the enemy of thinking. They help us quickly judge how important an event is, motivate us to act, and signal our values. The problem arises when emotion replaces checking instead of complementing it. Propaganda deliberately exploits that substitution.

Daniel Kahneman popularized the distinction between fast, intuitive thinking ("System 1") and slow, analytical thinking ("System 2") — it's a metaphor, not two departments of the brain, but a useful one. Strong emotion favors the fast mode: a person reacts rather than weighs. Paul Slovic described the "affect heuristic": whatever evokes a pleasant feeling seems both more beneficial and less risky, even though logically these judgments are independent.

The main emotional levers of propaganda:

• Fear. Threat messages get priority in attention: evolutionarily, it's cheaper to be startled by a shadow ten times than to miss a predator once. Fear increases willingness to accept harsh measures and a strong leader. It's used by politicians ("if not us, then chaos"), insurers, supplement sellers, and authors of panicky posts about food.
• Anger. Anger mobilizes and makes people more confident in their judgments. Research suggests that messages with moral outrage spread more actively on social media, though the size of the effect is debated.
• Disgust. It easily transfers from objects to people: dehumanizing campaigns described entire groups as "parasites" and "vermin." Disgust switches off empathy faster than any argument.
• Pride and belonging. "We are a great nation," "only true professionals choose…" — a flattering self-image in exchange for agreement.
• Hope and guilt. The promise that one leader, product, or reform will solve every problem, and the reproach "if you don't share, you don't care." People agree to gain relief or get rid of an unpleasant feeling.

A separate tool is story instead of statistics. A single photo of a suffering child mobilizes more sympathy than data about thousands of victims (the "identifiable victim effect"). A story isn't manipulation in itself; it becomes manipulation when a vivid case is presented as typical in defiance of the statistics.

The sign of emotional manipulation is not the emotion itself but its disproportion and urgency. If a message demands an immediate reaction ("share before it gets deleted," "only two hours left"), if the intensity of the feeling doesn't match the scale of the event, if after reading you want to act rather than learn more — that's a reason to stop.

Creating an enemy: "us" and "them"

No technique unites a group as quickly as a common enemy. In "minimal group" experiments in the early 1970s, the psychologist Henri Tajfel showed that a random, almost meaningless division — for example, by preference for one of two abstract paintings — is enough for people to start allocating resources in favor of "their own." Later research refined the picture: favoritism toward one's own shows up more reliably than hostility toward outsiders. But favoritism alone gives propaganda enough to build on.

The enemy image serves several functions at once:

1. It explains failures. If the economy stalls, the cause isn't decisions but saboteurs, foreigners, oligarchs, bureaucrats, migrants, or "globalists" — depending on who's telling the story. The scapegoat mechanism removes responsibility from the source and gives the audience a simple explanation for a complex problem.
2. It unites. Internal disagreements fade into the background in the face of a common threat. Criticizing "our own" starts to look like helping the enemy.
3. It justifies measures. Against an enemy, things become acceptable that would be unacceptable in peacetime: restrictions, surveillance, violence, breaking the rules.
4. It dehumanizes. The enemy gradually stops being a person with motives, fears, and a range of opinions. They become a uniform mass with nothing but malicious intent.

A typical ladder: a label, then negative stories, then interpreting any action through malicious intent, then "you can't negotiate with people like that," and finally "something has to be done about them." Historians of mass persecution describe this ladder in very different societies.

The enemy image is especially convincing when it has real grounds: every group has some genuine villains. Propaganda generalizes real episodes into a description of the whole group and blocks access to counterexamples, while declaring the misdeeds of "our own" isolated incidents or provocations.

Everyone uses the "us vs. them" split. Politicians of different camps portray their opponents as an existential threat to the country. Corporations build marketing against competitors or against an imaginary enemy of the consumer ("chemicals," "Big Pharma"). At work, the sales and engineering departments spend years telling jokes about each other that gradually become "the truth." For more on how group membership gets built into beliefs, see the section on group and identity in the guide to belief architecture.

A useful test: try to describe your "opponent's" position so that its supporter would agree you've understood it correctly. If you can't, you probably know not the position itself but its propaganda image.

Repetition and the illusion of truth

In 1977, Lynn Hasher, David Goldstein, and Thomas Toppino ran an experiment: participants rated the plausibility of statements several times at intervals, and some of the statements were repeated. Over time, the repeated statements seemed more true — regardless of whether they were actually true. This is how the illusory truth effect was described.

The mechanism is linked to processing fluency: a familiar statement is easier to understand, and the brain mistakes ease for a sign of truth. In ordinary life this is a reasonable heuristic — things that are repeated often really are true more often — but it's easy to hack. The most unsettling finding came from Lisa Fazio and colleagues in 2015: the effect works even for statements that contradict participants' own knowledge. A person who knows the right answer rates a repeated false option as slightly more plausible. Knowledge weakens the effect but doesn't cancel it. Gordon Pennycook and co-authors (2018) showed that even a single exposure to a fake news headline increased its perceived accuracy the next time it was encountered — though for completely absurd claims the effect weakened.

Several practical consequences follow:

• Volume matters more than quality. A hundred weak repetitions can outweigh one strong argument. That's exactly why campaigns are built on slogans and short formulas rather than detailed evidence.
• Different sources amplify the effect. If the same claim comes from "different" people, it seems independently confirmed. In reality, the sources may be repeating each other or a single original press release. Sociologists call this false consensus; journalists call it an echo.
• A debunk can inadvertently repeat the lie. The headline "No, vaccines don't cause X" repeats the link "vaccines — X." That's why misinformation-correction specialists recommend starting with the truth, briefly mentioning the myth, explaining why it's wrong, and ending with the truth again — the so-called "truth sandwich."
• The meme beats the report. A short, rhythmic phrasing spreads better than a long and precise one.

Repetition isn't used only by politicians. An advertising slogan comes to feel like "the truth" about a product. The corporate mantra "we're one family" sounds like a description of reality even if the company lays people off by email. A family legend about the "irresponsible uncle" becomes fact simply through frequency.

The defense here isn't willpower but procedure. The feeling "I've heard this somewhere, so it's probably true" is worth turning into a question: "Where exactly, and what is it based on?" If there's no answer, familiarity isn't an argument.

Half-truths, omission, and selective statistics

If a lie can be caught, a half-truth is almost impossible to refute without going into details the audience has stopped listening to. That's why experienced manipulators prefer selecting to inventing.

Omission. The simplest and most powerful technique. Report a rise in crime in a neighborhood but don't mention that the method of counting changed at the same time. Talk about a drug's side effects but don't mention the risk without treatment. Show a protest but don't say how many people attended. "What isn't being said here?" is one of the most productive questions in critical reading. The article "See the hole, not the donut" explains in detail how to train attention to what's missing.

Cherry picking. From a mass of data, only what confirms the thesis is selected. Three out of twenty studies are cited. Two convenient years are shown out of a ten-year chart. One sentence is taken from an interview. Each individual element is credible; the overall picture is not.

A convenient point of comparison. "Unemployment has halved" — compared to the peak of the crisis. "Defense spending rose 40%" — compared to a year of abnormally low funding. The choice of baseline can turn any trend into a success or a failure.

Relative versus absolute numbers. "Risk rose by 50%" sounds frightening. If the risk went from 2 in 10,000 to 3 in 10,000, the absolute change is one case in ten thousand. Medical and pharmaceutical messages — both promotional and alarmist — exploit this gap especially often. A good rule: for any "N times more," ask "and what was it in absolute numbers?"

Averages that hide the distribution. "The average salary went up" may mean that the incomes of a small top group rose while nothing changed for the majority. The median and the distribution often tell a completely different story from the arithmetic mean.

Correlation as causation. "Regions with more X have higher levels of Y." The link may be explained by a third factor or reverse causation, but it's presented as proof of guilt or merit.

Out-of-context quotes, cropped videos, charts. The sentence "I'm not saying he's a thief" becomes "he's a thief" if you cut off the beginning. The video of a fight starts with the retaliatory punch. A truncated Y-axis turns 2% growth into a cliff.

False balance. The position of the overwhelming majority of experts and that of a marginal minority are presented as equivalent "two sides," and the viewer comes away with the impression that "scientists disagree," even though there's no substantive dispute.

It's important not to turn this knowledge into a weapon only against opponents. Anyone who argues online tends to bring convenient statistics into the debate and fail to notice inconvenient ones. The honesty test is simple: do I apply the same level of skepticism to numbers I like?

Ideology: when propaganda no longer needs to persuade

The most effective propaganda is the kind that no longer has to work. Once a person has a stable picture of the world, new messages don't need to be proven: it's enough for them to fit the picture. Ideology acts as a ready-made filter that selects suitable facts and discards unsuitable ones on its own.

Ideology isn't a synonym for lies: it's a system of values and explanations that helps people navigate a complex world, and everyone has some form of it. The problem begins when the system becomes unfalsifiable: any refutation is declared in advance to be the enemy's scheming, and any confirmation is proof of the truth. The big guide to belief architecture describes this transition in detail — from the protective belt of auxiliary explanations to the point where facts have nowhere left to get in.

The relationship between ideology and propaganda runs both ways. Propaganda shapes ideology, gradually filling the picture of the world with the right frames, enemies, and values. And a formed ideology makes a person receptive to propaganda from "their" side and almost deaf to messages from the "other" side. Political psychologists describe this through motivated reasoning: people scrutinize inconvenient arguments more strictly and convenient ones more leniently. They don't invent facts, but they move the bar of evidence depending on the conclusion.

Some signs that ideology is starting to work as a filter rather than a tool:

• You already know what you think about a news story before you've read it — from the source or a single mention.
• Any mistakes by your side seem like exceptions, and any mistakes by the other side seem like an expression of its essence.
• You can't name data that would make you reconsider your position.
• People with different views seem to you not mistaken but stupid or malicious.
• Doubt within your group is perceived as betrayal.

These signs don't belong to one end of the spectrum: they show up among supporters of any political views, among followers of diets and parenting methods, among fans of tech brands.

Ideological readiness explains why propaganda so often "preaches to the choir": a significant share of political communication aims not to convert opponents but to mobilize supporters. Its effect lies not in changing views but in reinforcing them and turning them into action.

Social pressure and the spiral of silence

Humans are social creatures, and many of our beliefs rest not on evidence but on the agreement of those around us. Propaganda exploits this: create the impression that "everyone thinks so," and a significant share of people will join in or at least stop objecting.

The Asch experiments. In the 1950s, Solomon Asch asked participants to compare the lengths of lines — a task where it's almost impossible to make a mistake. But the group included confederates who unanimously gave the wrong answer. About a third of the real participants' answers matched the group's wrong opinion, although most participants held their ground at least once, and a single ally telling the truth sharply reduced conformity. Replications have produced varying effect sizes, and interpretations differ: some participants apparently doubted their own perception, others simply didn't want to stand out.

Social proof. Robert Cialdini described this principle in Influence: in an uncertain situation, we look at what others are doing. Reviews, purchase counters, likes, and "millions have already chosen" all exploit the same rule — and the "everyone does it" signal is easy to fake.

The spiral of silence. In the 1970s, the German sociologist Elisabeth Noelle-Neumann proposed a theory explaining how a minority opinion can come to seem like the majority's. People constantly gauge the "climate of opinion," and if they believe their position is unpopular, they tend to stay silent out of fear of isolation. Silence makes that position even less visible, other supporters also decide to stay silent — and the spiral tightens. A loud minority starts to look like the majority, and the real majority may not know it's the majority.

Empirical support for the theory is mixed: meta-analyses find a link between perceived support for an opinion and willingness to voice it, but it's small on average and depends on culture and topic. Still, the core idea — that the appearance of consensus changes actual behavior — is useful for understanding propaganda. A closely related phenomenon is pluralistic ignorance: most people privately disagree with a norm, but each thinks everyone else supports it.

Propaganda builds the illusion of consensus in various ways: polls with leading wording, showing only one reaction on air, bot networks, "grassroots" movements funded by a single organization (astroturfing — named after the brand of artificial turf). Corporations order fake reviews, activists of every stripe coordinate attacks in the comments, and in a work chat, one pushy person can create the feeling that "everyone thinks so."

The defense starts with a simple question: "How do I know that the majority thinks this?" Often the answer is "that's what my feed says" or "that's what they said on TV," and that's not the same as the actual distribution of opinions.

Modern propaganda: algorithms, bubbles, and AI

The mechanisms described above have been known for decades, even centuries. What has changed is the infrastructure. Mass influence used to require access to newspapers, radio, and television. Now anyone can produce content, and the distribution of attention increasingly depends on recommendation algorithms.

Engagement algorithms. Platforms optimize feeds for watch time, clicks, comments, and shares. The algorithm has no political views, but it "learns" what holds attention best: outrage, fear, conflict, sensation. The result is a structural bias in favor of emotionally charged content. This isn't a platform conspiracy but a side effect of the business model — yet for a propagandist it works as a free amplifier.

Filter bubbles and echo chambers. Eli Pariser's idea (2011) that personalization locks people inside information bubbles became a popular metaphor, but the empirical picture is more complex. Many users see a wider variety of sources than is commonly assumed; dense echo chambers are characteristic mainly of a small but very active part of the audience, and they're often created by people's own choice of subscriptions rather than by the algorithm. Encountering opposing views doesn't always soften them either: in some studies, exposure to the "other side's" content only increased polarization. The problem isn't just that we don't see others, but that we see them in their worst, most outrageous form.

Microtargeting. The Cambridge Analytica story popularized the idea that personal data makes it possible to "hack" voters. The real effectiveness of psychographic targeting is disputed: studies find small personalization effects, and there's far less evidence of mass opinion change than there are loud claims — including claims by the companies that sold such services. But targeted delivery lets different groups see different versions of a campaign that no one sees in full, and that makes public scrutiny harder.

Generative AI. Language models and image generators have drastically cut the cost of producing plausible text, voices, and video: thousands of "different" comments, a fake recording of a speech, a photo of an event that never happened. The second risk is subtler — the "liar's dividend": when anything can be faked, it becomes convenient to declare anything inconvenient a fake. There isn't yet enough data on how strongly AI content changes real behavior; it's wise neither to exaggerate nor to downplay it. AI also affects consumption: summaries and chatbot answers become a new frame between a person and the source, one that can smooth over uncertainty or mix up attribution. How to read them with their limits in mind is covered in the piece on critical thinking and AI summaries.

Influencers and speed. Propaganda increasingly looks like the personal opinion of a likable person who has been paid — by a company, a party, or a government. And a false message racks up millions of views in hours while checking it takes days. A study by Soroush Vosoughi and colleagues (2018) using Twitter data found that false news on average spread faster and further than true news, with people rather than bots contributing most; the methodology is debated, but the conclusion about the advantage of novelty and emotionality is consistent with other data.

Why smart people are vulnerable too

A common comfort goes like this: propaganda works on the uneducated and the gullible, and a smart person will see through it. Research and history don't support this.

Jacques Ellul put forward a provocative thesis: educated people are especially susceptible to propaganda because they consume more information, feel obliged to have an opinion on everything, and are confident in their independence. The thesis hasn't been rigorously tested, but it points to a real problem: confidence in your own invulnerability is itself a vulnerability. Psychologists describe the "bias blind spot": biases are easy to see in others and hard to see in yourself.

Motivated reasoning. Ziva Kunda (1990) described how a desired conclusion influences which arguments a person looks for and how strictly they check them. Intelligence works here as a lawyer, not a judge. Dan Kahan's work suggested that in some experiments people with better numeracy diverged more sharply in interpreting politically charged data; however, these results are debated, and research by Pennycook and Rand shows that analytical thinking on average helps people tell false headlines from true ones regardless of their views. The honest bottom line: analytical ability tends to help but doesn't guarantee protection, especially when the topic touches identity.

The backfire effect. In 2010, Brendan Nyhan and Jason Reifler described cases in which a correction strengthened belief in a false claim. The idea became popular, but later work — including that of Thomas Wood and Ethan Porter — showed that the backfire effect is rare. More often, a correction shifts factual beliefs slightly in the right direction, though it doesn't necessarily change the position that rested on them.

Expertise outside one's field, and complexity as camouflage. A specialist carries their confidence into areas where their knowledge is no better than average, and the audience trusts them on authority. For educated audiences, propaganda comes packaged in charts, citations, and the tone of an "independent analyst": the citations are real, but the interpretation isn't contained in them, and the feeling of being well-founded replaces actual checking.

What genuinely helps is not the level of intelligence but thinking style: willingness to look for arguments against your own position, to separate the question "what is true" from "what benefits my group," and to state the conditions under which you would change your mind. Julia Galef calls this the scout mindset as opposed to the soldier mindset — more in the summary of The Scout Mindset.

How to recognize propaganda: a checking procedure

No checklist offers a guarantee, but a sequence of questions lowers the odds of being taken in. Below is a procedure you can apply to a news story, a post, an ad, a video, or a conversation. You don't have to go through every step every time; the stronger your reaction and the higher the stakes, the more steps are worth taking.

1. Notice your reaction. What do you feel — fear, anger, schadenfreude, delight? How strongly? Do you want to share or act right away? A strong emotion isn't proof of manipulation, but it's a signal to slow down.
2. Identify the source. Who is saying this? A primary source or a retelling of a retelling? What are the source's interests — financial, political, reputational? Professional fact-checkers use "lateral reading": instead of studying the site itself at length, they open new tabs and see what others write about it. Research by Sam Wineburg's Stanford group showed that this technique works better than careful reading of the original page.
3. Separate fact from interpretation. What exactly is being asserted as a checkable fact? What is an evaluation, a forecast, a conclusion? Rewrite the message in neutral words and see what's left.
4. Find the frame. Which words were chosen, and which could have been chosen instead? Who is the hero, who is the victim, who is the enemy? Try retelling the same fact in the opposite frame.
5. Check the numbers. Absolute or relative? What is being compared to what, and why that? What period was chosen? Is it an average or a distribution? Where does the data come from?
6. Ask what has been left out. What context is missing? What would the other side say? What data would be inconvenient for this conclusion — and is it mentioned?
7. Check the consensus. Is this one person's opinion, a group's position, or agreement among experts? Is "everyone's saying it" a real poll or your feed? Does the message match the conclusions of independent sources that don't repeat each other?
8. Look for an enemy image and urgency. Is there a "them" to blame for everything? Is immediate action demanded? Is a single simple solution offered for a complex problem?
9. Test yourself for symmetry. If the same message with the same evidence favored the opposite side, would you believe it? This is the main test, and the most uncomfortable one.
10. Suspend judgment. If you can't check it, the honest answer is "I don't know" or "it's unclear for now." You don't have to have an opinion on everything right away.

The procedure works better as a habit than as a one-off effort; daily reading habits help build it in. And you can practice layer-by-layer analysis — fact, source, interpretation, conclusion — in ThinkLens, which breaks a text down into claims and evidence.

A separate note on prebunking (inoculation). Research by Sander van der Linden, Jon Roozenbeek, and colleagues shows that if you show people typical techniques in advance — false dichotomy, appeal to emotion, scapegoating — using weakened examples, they later get somewhat better at recognizing them in real content. The effects are moderate, fade over time, and according to some studies come with a slight increase in skepticism toward reliable messages too. But the approach is more promising than debunking each lie separately — and this text is an attempt at that kind of inoculation as well.

Can you fully protect yourself?

No. And that's not pessimism but a condition for a sensible strategy.

Full protection would require personally checking every claim — climate data, drug safety, history — and that's impossible. Knowledge of the world rests largely on trust in scientists, journalists, institutions, and friends, and propaganda feeds precisely on that necessity.

There's danger on the other side too. A person who decides that "everyone lies" doesn't become free — they become vulnerable to those who offer an "alternative truth." If nothing can be believed, then nothing needs to be checked; it's enough to pick what you like. In her work on totalitarianism, Hannah Arendt wrote about people who believe everything and nothing at the same time, who think everything is possible and nothing is true. Such an audience is ideal for manipulation.

The realistic goal isn't immunity but calibrated trust:

• Trust sources more when they show their methods, admit mistakes, and correct them publicly.
• Trust sources less when they're never wrong, always on one side, and demand immediate action.
• Keep your confidence proportional to the evidence: "probably yes," "probably not," and "I don't know" are normal answers.
• Diversify your sources not for the sake of "balance" but to see where they diverge and why.
• Regularly ask yourself what could change your mind — and if there's no answer, that's a reason to be wary.

Attention is limited, and you don't have to have an opinion on every event. Sometimes it's more honest to say "I haven't looked into it" than to repeat the first frame that comes along.

FAQ

Research foundations

This text draws on several research traditions. Below is a map for further reading. These are signposts, not a substitute for the primary sources; some famous effects should be read with the replication debates in mind.

Classics of propaganda studies. Walter Lippmann, Public Opinion (1922) — on "the pictures in our heads." Edward Bernays, Propaganda (1928) — a practitioner's view of managing mass opinion. Jacques Ellul, Propaganda (1962) — on sociological propaganda. Edward Herman and Noam Chomsky, Manufacturing Consent (1988) — an influential and contested model of media "filters."

Psychology of decisions and influence. Daniel Kahneman, Thinking, Fast and Slow — heuristics and the framing effect (the author later acknowledged that some chapters, especially on social priming, relied on weak data). Paul Slovic — the affect heuristic. Robert Cialdini, Influence — social proof, authority, scarcity, and other principles. Solomon Asch — conformity. Elisabeth Noelle-Neumann, The Spiral of Silence — a theory of the climate of opinion with mixed support. Henri Tajfel — social identity. George Lakoff, Don't Think of an Elephant! — frames and metaphors in politics.

Misinformation and correction. Hasher, Goldstein, and Toppino; Fazio and colleagues — the illusory truth effect. Nyhan and Reifler; Wood and Porter — the debate over the backfire effect. Pennycook and Rand — analytical thinking and spotting fake headlines. Van der Linden and Roozenbeek — inoculation against manipulation. Wineburg and the Stanford group — lateral reading.

Related HiddenLogic pieces: guide to critical thinking, what people do with information, belief architecture and ideology, The Scout Mindset, see the hole, not the donut, layers of claims and evidence, AI summaries, and daily reading habits.

Conclusion

Propaganda isn't necessarily loud, false, and someone else's. More often it's quiet, formally accurate, and sounds like common sense. It picks the phrase "dying out" instead of "falling birth rate," or "optimization" instead of "layoffs," shows one story instead of a distribution, repeats a formula until it becomes familiar, and finds someone to blame. It works through the same mechanisms that help us live: quick judgments, emotions, trust in our own people, saving effort.

That's why the defense can't be a one-time exposé. It's a habit: noticing your own reaction, asking about the source and what's been left out, trying alternative frames, checking the numbers, and — hardest of all — applying the same skepticism to convenient messages as to inconvenient ones. If, after reading this, you feel the urge to find propaganda mainly among your opponents, that's a normal first reaction. The next step is to find it in your own feed and in your own wording.

Grass can be both green and sharp. The question isn't which description is true but why you were shown this one — and what you would see if you'd been shown another. If you want to turn that question into a working tool, try taking the next text that provokes a strong reaction in you and breaking it down layer by layer — by hand or with ThinkLens.