You now have the machinery: the ladder from private to mutual to common knowledge (the infinite regress — I know, you know, I know you know, forever), and the coordination game where the right move depends entirely on what you believe others will do. This lesson is where that machinery earns its keep. We’re going to watch it explain some of the strangest, fastest, most consequential events humans produce: banks that die in a weekend, regimes that look eternal right up until they evaporate, and why a soft-drink company will pay a fortune to reach you specifically while you’re watching the big game.
The thread tying all of them together is a single sentence. A stable-but-bad outcome can flip the instant a public signal turns a private truth into common knowledge. Nobody’s beliefs about the facts have to change — only their beliefs about each other’s beliefs. Hold that sentence; every case below is a variation on it.
Bank runs: the solvent bank that dies anyway
A bank takes your deposits, promises them back on demand, then lends most of them out for years — mortgages, business loans. That mismatch is the entire business model, and it hides a trap. The bank can be solvent (its loans are good and worth more than it owes) yet illiquid (it cannot turn those loans back into cash fast enough to pay every depositor at once). Those two words are the whole lesson, so let’s nail them down.
Two words that run the whole show
Solvent means assets exceed liabilities — given time, the bank can pay everyone. Illiquid means it can’t do so right now, because the money is tied up in loans that won’t mature for years. A healthy bank is normally solvent but always illiquid. That gap is where runs live.
Here’s the coordination game, stripped to its bones (this is the intuition behind the Diamond–Dybvig model of bank runs). If you believe other depositors will stay calm, your best move is to leave your money in and collect interest. If you believe others will rush to withdraw, your best move is to rush first — because the bank pays out until the cash runs dry, and latecomers get nothing. So there are two equilibria: everyone-stays (good, stable) and everyone-runs (bad, also stable). Both are self-fulfilling. You run if you think others will run, and that belief alone can drain a perfectly healthy bank.
Notice what does the killing. The loans didn’t go bad. The bank didn’t become insolvent. What changed is that “the bank is in trouble” became common knowledge — usually via a public signal everyone sees and everyone knows everyone else saw: a news report, a queue photographed outside a branch, a viral post. The signal doesn’t have to be true. It only has to make each depositor think everyone else is about to run.
Worked example: the Friday-afternoon run
Northbridge Savings holds $100 in deposits. Its loans are sound and will be worth $110 next year — genuinely solvent. But it keeps only $20 in cash on hand.
Monday, all is calm; withdrawals are tiny and routine. Then Friday at noon a popular finance account posts “is Northbridge about to fail?” — seen by thousands, and crucially, everyone knows thousands saw it. By 1pm a queue forms. The queue is itself a new public signal: now depositors don’t just fear the bank, they can see others acting on the fear. Each person reasons, correctly, “only the first $20 gets paid — I’d better be in it.” The $20 evaporates by 2pm. A solvent bank is dead by close of business, killed not by bad loans but by a self-fulfilling belief about everyone else.
This isn’t a museum piece. Britain’s Northern Rock in 2007 drew literal queues around the block; Silicon Valley Bank in 2023 died in roughly 48 hours as depositors — coordinated by news and group chats — pulled funds faster than any bank could survive. In both, the accelerant was a public signal that made the fear common knowledge.
Before you read — take a guess
Take a guess: a bank whose loans are perfectly good still collapses in two days. What's the most likely mechanism?
Revolutions: the discontent nobody could see
Now take the same shape and point it at a dictatorship. Under a repressive regime, opposing the government is dangerous, so people who privately despise it pretend to support it — they clap at the right moments, they don’t say what they think. This hiding has a name.
Preference falsification (Timur Kuran’s term) is the act of misrepresenting your true preferences under social or political pressure. Its poisonous side-effect is pluralistic ignorance: because everyone is hiding the same opposition, everyone underestimates how much opposition exists. You hate the regime, but you look around, see everyone clapping, and conclude you’re nearly alone — dangerously outnumbered. So you keep clapping. So does the person next to you, for exactly the same reason. The opposition can be a silent majority and still feel, to each of its members, like a lonely minority.
Two failures, stacked
Preference falsification is the hiding — I mask my real view. Pluralistic ignorance is the result at the group level — since we’re all hiding, none of us knows the truth about the others. Discontent is now private knowledge held by millions, but not common knowledge held by anyone.
Then something makes the shared discontent common knowledge — and the dam breaks. Someone shouts the forbidden thing and isn’t arrested. A crowd stands and doesn’t scatter. A dictator is booed on his own balcony, on live television, in front of everyone. Suddenly each person learns not just “I hate this” but “everyone hates this, and everyone now knows everyone hates this.” The private truth becomes common knowledge, the cost of speaking collapses (they can’t arrest all of us), and the regime that looked eternal falls in days.
Worked narrative: the balcony that broke a regime
For decades a regime holds power partly through preference falsification: people attend the rallies, wave the flags, and privately loathe it — each believing the loathing is theirs almost alone.
Then a mass rally is called, staged for the cameras as a show of loyalty. Standing in the enormous crowd, the leader begins his speech — and from somewhere the jeering starts. It spreads. The cameras, broadcasting live, capture the leader’s face freezing. In that instant every viewer learns three things at once: I’m not alone, they’re not alone, and everyone can see that everyone can see it. The falsification that held the whole system up depended on nobody knowing the truth about everyone else. The broadcast destroyed that in seconds. Days later the regime is gone. Romania’s Ceaușescu in 1989 was booed on exactly such a balcony; across Eastern Europe that year, and again in the 2011 Arab Spring, regimes that looked immovable dissolved once the shared discontent became common knowledge.
This is why revolutions blindside everyone — including the experts, the secret police, and often the revolutionaries themselves. The discontent was there all along, invisible, because everyone was busy hiding it from everyone else. What changed wasn’t how much people hated the regime. It was that they suddenly knew they all did.
Why do sudden revolutions catch even the secret police by surprise, when the underlying discontent had built for years?
Crowds: the number is the message
Why do protesters risk turnout in the streets rather than, say, a private petition? Because a visible crowd is itself the common-knowledge signal. When you stand in a square with a hundred thousand people, you don’t just learn “many oppose this” — you learn “many oppose this and everyone here can see that many do.” The size of the crowd, seen by the crowd, is the message. “We saw how many we were” changes everything, because it converts a private hope (“surely others feel this too”) into common knowledge in one glance.
This is also why broadcasting a crowd amplifies it, and why regimes fight so hard over the footage and the headcount. A protest of a hundred thousand seen only by its participants is powerful. The same protest on every screen in the country — where now everyone knows everyone saw it — is an order of magnitude more powerful, because it manufactures common knowledge among people who were never in the square at all.
Advertising: common knowledge you can buy
The political scientist Michael Chwe noticed that this same logic quietly runs a chunk of the advertising industry. Some ads run in the most expensive slots money can buy — the big game, the season finale — at a wild premium per viewer. Reaching one person during the big broadcast can cost many times what reaching that same person privately would. Why would any rational buyer overpay so grotesquely?
Because those slots don’t just deliver eyeballs — they deliver common knowledge. When an ad airs during the big game, you don’t only see it; you know that millions of others saw it, and that they know you saw it, and so on. For most products that’s worthless. But for a specific and important class of goods, it’s the entire point.
When you need everyone to know that everyone knows
A coordination good is something you want because others will want it too — its value to you depends on other people also adopting it. Three big families: network products (a chat app, a payment platform, a video-call tool — useless if your friends aren’t on it), status and social goods (a fashion label, a luxury brand — worthless as a signal if others don’t recognize it), and standards (the format everyone must pick together). For these, private information isn’t enough; you need common knowledge that others are adopting too.
Worked contrast: the same ad, delivered two ways
Imagine one million people, and a new video-call app whose whole value is that your contacts use it too. You have one ad to spend.
Privately — a million individually targeted views, each person seeing it alone. Everyone now knows the app exists. But no one knows whether anyone else is adopting it, so each waits for others to move first. The classic coordination stall: great product, private knowledge, nobody jumps.
Publicly — one slot in the big game, a million people watching together. Same one million impressions. But now each viewer knows the other million saw it too, and knows they know. Adopting suddenly looks safe — “my contacts must have seen this and be signing up.” The public buy created common knowledge; the private buy only created private knowledge. Identical reach, opposite result — which is exactly why the public slot is worth the premium for a coordination good, and not worth it for, say, painkillers, where I don’t care whether you take the same brand.
Sort each case by the common-knowledge channel that actually did the work.
Place each item in the right group.
- A hundred thousand in the square see, in one glance, how many they are
- One person shouts the forbidden thing, isn’t arrested, and the silent majority realizes it’s a majority
- An ad in the big game so network-app adopters know everyone else saw it too
- A dictator booed on live TV; suddenly everyone knows everyone loathes him
- Protesters broadcasting their headcount on every screen in the country
- A viral finance post makes "the bank is failing" common knowledge and a solvent bank collapses
The Keynesian beauty contest
Economist John Maynard Keynes described a newspaper game: readers pick the six prettiest faces from a hundred photos, and whoever’s picks match the most popular choices wins a prize. The clever player doesn’t pick the faces she finds prettiest. She picks the faces she thinks others will find prettiest. But everyone reasons that way — so really she must pick the faces others think others will find prettiest, and so on up the ladder of beliefs-about-beliefs you already know.
The Keynesian beauty contest is any situation where the winning move is to predict the average opinion about the average opinion, not to hold the correct view yourself. Being right about the object is beside the point; being right about everyone else’s guesses is the game.
Where the beauty contest bites
This is a startling amount of modern life. Markets: a stock can be worth buying purely because you expect others to buy it (find the meme before the crowd does), which is why prices can float free of fundamentals for a long time. Fashion and memes: a style spreads because people expect it to spread, not because it’s independently best. Elections covered as horse-races: voters and donors flock to the candidate they think others think is electable — “viability” becomes self-fulfilling, a beauty contest wearing a ballot. In each, everyone is trying to sit where the crowd will sit, which is precisely a coordination game solved by guessing others’ guesses.
Match each idea to its precise meaning.
Pick a term, then click its definition.
Why tyrants fear the town square more than the kitchen table
Put all of this together and a puzzle resolves itself. Authoritarian regimes are often oddly relaxed about private grumbling — you can complain about them at your own kitchen table — yet they come down with astonishing force on public assembly, even tiny gatherings. Why fear a small square more than millions of angry kitchens?
Because private grumbling stays private knowledge, and private knowledge doesn’t coordinate anyone. A regime survives on pluralistic ignorance: as long as each discontented person believes they’re alone, no one dares move. Public assembly is the common-knowledge-generating channel — the one thing that can convert a nation of isolated grumblers into a crowd that knows it’s a crowd. So regimes attack the channel, not the opinion. Ban gatherings. Jam broadcasts and cut the internet during protests. Arrest the single person who says the forbidden thing out loud — not because one voice is dangerous, but because the audience learning that the voice went unpunished is. Even punishing jokes and mocking memes makes sense here: a shared joke about the emperor is a tiny act of common-knowledge creation, everyone laughing and everyone seeing everyone laugh.
The channel is the target
China’s Tiananmen Square in 1989 was, above all, a fight over a public space where common knowledge could form — and its later erasure from the record was a fight to keep it from re-forming. Under Soviet censorship, dissidents circulated banned texts hand-to-hand as samizdat (“self-published”), a low-tech workaround precisely because the public channels of common knowledge were sealed shut. In every case the regime’s instinct is identical: control not what people privately think, but what they can all know that they all think.
State the tyrant's logic in one line.
Pick the right option for each blank, then check.
A repressive regime tolerates private grumbling because it stays , but crushes public assembly because it is the — the thing that can turn isolated grumblers into a crowd that .
The map: latent truth, public spark, coordinated outcome
Every case in this lesson has the same three-part skeleton. There’s a latent private truth (something many people know or feel, but privately). There’s a public spark or channel that converts it into common knowledge. And there’s the coordinated outcome that follows, often overnight.
| Case | Latent private truth | Public spark / channel | Coordinated outcome |
|---|---|---|---|
| Bank run | ”The bank might not survive a rush” | Viral post, news report, visible queue | Everyone withdraws; solvent bank collapses |
| Revolution | Millions privately loathe the regime | Booing on live TV, a defiant act unpunished | Falsification breaks; regime falls in days |
| Protest | ”Surely others are angry too” | The visible size of the crowd, broadcast | Isolated grievance becomes a mass movement |
| Coordination-good ad | ”I’d adopt this if others did” | Big-game broadcast everyone knows all saw | Simultaneous adoption; the network tips |
| Beauty-contest market | ”Others may pile into this” | Prices and headlines everyone watches | Self-fulfilling rush that ignores fundamentals |
The crucial caveat: sparks ignite, they don’t create
Before you start seeing common knowledge as a magic wand, one essential limit — and it’s the seam where this model can be misused. The spark only ignites a fire that was already laid. A public signal converts latent discontent into kinetic action; it does not manufacture the underlying grievance out of nothing.
The signal converts; it does not create
If the private dispositions aren’t there, the spark fizzles. A viral “the bank is failing” post does nothing to a bank nobody had reason to doubt and everyone knows is flush. A dictator booed by a genuinely content population just looks briefly embarrassed. The big-game ad can’t make people coordinate on a product none of them privately wanted. Common knowledge is the detonator — it releases stored energy fast — but it is not the fuel. Confusing the two makes you think you can spark a revolution or a bank run at will, or that every collapse was engineered by whoever lit the match. The grievance, the illiquidity, the latent demand had to be there first.
That’s also the honest limit on how far you can engineer these effects, and the next lesson digs into exactly where common-knowledge reasoning breaks down and misleads — the cases where the ladder of beliefs snaps, the signal is misread, or the “everyone knows” turns out to be a fiction. For now, hold the caveat: the signal is a converter of latent truth, never a creator of it.
Spot the flaw: a strategist claims, 'Just plant one viral rumor and you can topple any bank or regime you like.' What's wrong?
When to reach for this model
Pull out common-knowledge-and-coordination thinking whenever an outcome (a) depends on what people expect each other to do, (b) has a self-fulfilling quality, and (c) can flip fast without the underlying facts changing much. Bank runs, panics, protest waves, fashion cascades, speculative bubbles, network-product tipping points, platform migrations, viral norms — all fit. The diagnostic question is always the same: what’s the latent private truth, and what public signal would make it common knowledge? If you can answer both, you can often see the flip coming before it does.
Recap
Big picture
Bank runs, revolutions & crowds
- Public signal flips a bad-but-stable outcome
- Bank runs
- Solvent but illiquid = two equilibria
- Public signal makes "others will run" common knowledge
- Revolutions
- Preference falsification hides discontent
- Pluralistic ignorance: everyone feels alone
- A spark makes shared loathing common knowledge
- Crowds & media
- A visible crowd is the signal
- Broadcasting it manufactures common knowledge
- Manufactured common knowledge
- Big-broadcast ads for coordination goods
- You buy it because others will too
- Beauty contest
- Guess others’ guesses: markets, memes, elections
- The caveat
- The spark converts latent truth; it doesn’t create it
- Bank runs
Solvent banks, sturdy regimes, and quiet products all share a hidden switch: a public signal that turns “I know” into “we all know that we all know.” Learn to spot the latent truth and the channel that could light it, and the world’s most sudden events stop looking sudden. Next: where this powerful lens misleads — the failure modes of common knowledge.