Why publicly stating the obvious changes everything
In epistemic logic, everyone knowing a fact is not the same as common knowledge. Imagine three children with muddy foreheads. Each sees the other two are muddy, but doesn't know their own status. When an adult announces, "At least one of you has mud," they learn no new private facts. Yet making it public creates common knowledge—they now know that everyone knows—allowing them to deduce their own mud.
The Invisible Hierarchy of Knowing
In everyday conversation, the phrase "everyone knows" is treated as a simple statement of fact. If a group of people is sitting in a room and it starts raining outside a large window, it is reasonable to say that everyone knows it is raining. Yet in formal epistemic logic, this baseline awareness—known as mutual knowledge—is only the first tier of a deep hierarchy of mental states. Mutual knowledge simply means that each person in the group possesses the information independently. It says nothing about what each person thinks the others know.
Epistemic logic distinguishes this basic awareness from higher orders of knowledge. Second-order knowledge occurs when everyone knows that everyone knows. Third-order knowledge occurs when everyone knows that everyone knows that everyone knows. When this chain of nested beliefs continues without end across every possible level of reflection, it becomes common knowledge. While the difference between mutual knowledge and common knowledge may seem like a pedantic linguistic technicality, it represents the exact boundary where collective reasoning and coordinated action become possible.
The Logic of the Muddy Children
The standard demonstration of this principle is the muddy children puzzle. Suppose three children play outside and each gets mud on their forehead. Each child can clearly see the mud on the other two faces, but none can see their own face, and they are not allowed to communicate directly. Before any announcement is made, every child already knows that there is mud in the group; Child A sees mud on Child B and Child C, so Child A is fully aware that mud is present.
An adult then enters the room and publicly announces: "At least one of you has mud on your forehead." On the surface, the adult has conveyed zero new physical information to any individual child. However, the adult has transformed a private observation into common knowledge. The adult then asks repeatedly: "Do you know whether you have mud on your face?"
In the first round, all three children look around, see two muddy peers, and answer "No." In the second round, having observed that no one stepped forward in the first round, they are asked again, and all three still answer "No." But in the third round, all three children simultaneously step forward and announce that they must have mud on their own faces. The public announcement, combined with the silence of the other children, supplied the precise information needed to solve the puzzle.
How Silence Eliminates Possibilities
To understand why this deduction works, one must trace what the children deduce from the failure of others to speak. Suppose only Child A had mud. Child A would see two clean faces, immediately deduce from the adult's public statement that the muddy forehead must be their own, and answer "Yes" in round one. Because no one answered "Yes" in round one, all children now share common knowledge that there is more than one muddy child.
Now suppose only Child A and Child B had mud. In round two, Child A would look at Child B and reason: "If I were clean, Child B would have seen only clean faces besides their own, and Child B would have answered 'Yes' in round one. Since Child B said 'No', I must be muddy." Child B would perform the exact same calculation. Therefore, if there were only two muddy children, both would step forward in round two. When neither steps forward in round two, every child rules out the two-muddy-children scenario.
By round three, all three children recognize that if they were clean, the other two would have solved the problem in round two. The silence of the group in rounds one and two progressively eliminated possible worlds from everyone's epistemic models. This chain of deduction relies entirely on knowing that the other children heard the initial announcement, saw each other's reactions, and are reasoning rationally about everyone else's knowledge.
The Formal Discovery of Common Knowledge
The formal concept of common knowledge emerged independently across multiple disciplines in the late twentieth century. Philosopher David Lewis introduced the concept in his 1969 book, Convention: A Philosophical Study. Lewis was investigating how social conventions—such as driving on the right side of the road or using particular words to mean specific things—can persist without explicit contracts. He realized that a convention requires not just shared behavior, but a shared expectation that others will conform, along with the knowledge that everyone expects everyone to conform.
In 1976, mathematician and game theorist Robert Aumann provided the first rigorous mathematical formalization of common knowledge using set theory and probability partitions. In his landmark paper, "Agreeing to Disagree," Aumann proved that if two rational agents share common priors and update their beliefs via standard Bayesian conditioning, they cannot have common knowledge of different posterior probabilities. In simpler terms, two honest, rational people cannot agree to disagree if their differing opinions are made entirely common knowledge.
Computer scientists, notably Joseph Halpern and Yoram Moses, later adapted common knowledge to distributed computing in the 1980s. They demonstrated that achieving common knowledge is a fundamental prerequisite for perfectly coordinated action across networks of independent processors, yet it is mathematically impossible to guarantee over unreliable communication channels.
The Two Generals and the Limits of Communication
The difficulty of establishing common knowledge is famously illustrated by the Two Generals Problem in computer science. Imagine two allied armies encamped on opposite hills, preparing to attack a common enemy in the valley between them. They can only communicate by sending messengers through enemy lines, meaning any message might be intercepted.
General 1 sends a messenger with a note: "Attack at dawn." Even if the messenger arrives, General 1 cannot attack because he does not know if General 2 received the message. General 2 sends a confirmation messenger back: "Received, attacking at dawn." But General 2 now cannot attack, because she does not know if her confirmation made it through; if it was intercepted, General 1 will stay on his hill. General 1 must send an acknowledgment of the confirmation, which in turn requires an acknowledgment, ad infinitum.
No finite sequence of unverified messages can bridge the gap from mutual knowledge to common knowledge over an uncertain channel. Even if ninety-nine messages successfully cross the valley, both generals are left with high-order mutual knowledge, but never true common knowledge. Without that final guarantee, absolute coordination remains risky.
Social Norms, Rituals, and Pluralistic Ignorance
The distinction between mutual and common knowledge explains many peculiar collective behaviors in human society. A classic example is the folktale of the Emperor's New Clothes. As the emperor parades through the street, every citizen can see with their own eyes that he is naked. Mutual knowledge exists among all spectators. However, no citizen knows whether their neighbors actually see the nakedness or are genuinely convinced of the fabric's magical invisibility.
When a child shouts, "The emperor has no clothes!", the child does not reveal any hidden visual fact. The shouting functions precisely like the adult's announcement in the muddy children puzzle: it synchronizes everyone's awareness and converts private observations into common knowledge. Once everyone knows that everyone else knows, the social pressure sustaining the collective illusion evaporates.
This dynamic plays out in real-world phenomena like pluralistic ignorance, where members of a group privately reject a norm but incorrectly assume that everyone else supports it. Public rituals, town halls, open debates, and formal proclamations do not exist merely to distribute information; they exist to manufacture common knowledge, giving individuals the structural assurance required to change collective behavior.
Key takeaways
•Mutual knowledge means everyone knows a fact independently, while common knowledge means everyone knows that everyone knows it, across an infinite hierarchy of reflection.
•A public announcement of an obvious fact conveys no new first-order information, but it creates common knowledge by eliminating uncertainty about what others consider possible.
•Without common knowledge, rational agents cannot coordinate simultaneous actions or reliably resolve collective puzzles, as demonstrated by the Two Generals problem.