In the spring of 415 BC, Athens woke to find the herms — the stone boundary markers carved with the god Hermes — mutilated across the city. No one knew who had done it. Within hours, the rumor was everywhere: it was a conspiracy against the democracy, a coup was imminent, the generals could not be trusted. The actual explanation, debated by historians ever since, mattered far less than the story that spread. Alcibiades, one of the most gifted commanders Athens ever produced, was recalled from the Sicilian Expedition on the strength of rumors that were almost certainly false. The expedition collapsed. Athens never recovered its strategic footing.
The internet did not invent this problem. It inherited it.
The Architecture of Alarming Information
Cognitive scientists have a reasonably clear picture of why false information spreads faster than true information, and the answer has nothing to do with platforms, algorithms, or the moral failings of any particular era. The human brain allocates memory resources based on emotional intensity. A story that triggers fear, disgust, or outrage is encoded more deeply and retrieved more readily than a story that is merely accurate. This is not a bug in the system; it is the system. For most of human evolutionary history, overreacting to a potential threat cost you energy. Underreacting cost you your life. The brain learned to err toward alarm.
The consequence, playing out across five thousand years of recorded history, is that emotionally charged false information behaves like a more virulent pathogen than emotionally neutral true information. It spreads faster, survives longer in memory, and resists correction more stubbornly. When the Athenians heard that Alcibiades had profaned the Eleusinian Mysteries, the accusation carried enormous emotional weight — sacrilege, betrayal, impiety. The subsequent investigation, which produced inconclusive evidence, carried none of that weight. The correction never caught the rumor.
Three Predictable Failures
Every society that has attempted to suppress or correct misinformation has arrived, eventually, at the same three responses. And every one of them has failed in the same way.
The first response is official denial. When a damaging rumor circulates, those in power issue a formal rebuttal. This approach has an obvious flaw that was well understood by Roman rhetoricians: the denial repeats the accusation. Cicero noted in his letters that responding to a slander in public often gave it more circulation than ignoring it would have. The Roman Senate's periodic decrees against false reports — there were several during the late Republic — had no measurable effect on the rate at which false reports circulated. They did, however, make it easier to prosecute political enemies.
The second response is the trusted messenger. If official denial fails, perhaps the correction needs a more credible source. Medieval European towns appointed official town criers in part to counter the informal rumor networks of the market. Guild systems in Florence and Venice attempted to regulate commercial information specifically because false reports about a merchant's creditworthiness could destroy a business overnight. The problem was that the trusted messenger was only trusted until he became associated with power, at which point he became a source of official propaganda rather than reliable information — and people adjusted their credibility assessments accordingly. The medieval peasant who distrusted the lord's herald was not being irrational.
The third response is punishment. The Han Dynasty maintained elaborate systems for controlling information that flowed through the empire, including penalties for spreading false reports that damaged state interests. Tudor England had the Court of Star Chamber, which prosecuted seditious rumors with genuine vigor. The Sedition Act of 1798, in the early United States, attempted to criminalize false statements about the government. In every case, the effect on actual rumor circulation was modest at best. What changed was not the volume of false information but the willingness of people to attribute it openly. The rumors went underground, gained the additional emotional charge of forbidden knowledge, and spread faster.
Why the Correction Always Loses
There is a specific asymmetry at work here that explains why all three responses fail. The original alarming story arrives first, gets encoded deeply, and establishes itself as the baseline reality in the listener's mind. The correction arrives second, is less emotionally engaging, and must do the cognitively expensive work of overwriting an existing belief rather than simply recording a new one. Psychologists call this the continued influence effect: even when people are explicitly told that a piece of information was false, the original false information continues to influence their reasoning. The correction is processed, acknowledged, and then quietly set aside.
This is not a failure of intelligence or education. Studies using undergraduate samples — the standard laboratory population — consistently find the continued influence effect across all demographic and cognitive categories. But the historical record offers a far richer data set. The Dreyfus Affair in France, which consumed the Third Republic for more than a decade, illustrates the effect at national scale: even after Alfred Dreyfus was officially exonerated, the population that had believed him guilty did not update its beliefs in proportion to the evidence. The original story had been too emotionally charged, too deeply encoded, and the exoneration arrived too late to compete.
The Market Town as Laboratory
Medieval market towns offer a particularly useful case study because they were, in information terms, genuinely isolated systems. News entered through travelers, merchants, and the occasional official messenger. There was no external fact-checking apparatus. Historians studying the spread of rumors through these communities — using court records, guild documents, and church proceedings — find patterns that are essentially identical to those documented in modern social network studies. False information spread faster than true information. Emotionally charged stories outcompeted accurate ones. Corrections issued by authority figures were discounted. Rumors that confirmed existing social prejudices were accepted with almost no scrutiny; rumors that contradicted them required extraordinary evidence.
The market town did not have an algorithm. It had human beings.
What Five Thousand Years Actually Tells Us
The historical record does not offer a solution to the misinformation problem. What it offers is clarity about what the problem actually is. The problem is not a specific medium — not the printing press, not the telegraph, not broadcast television, not the internet. The problem is a cognitive architecture that was built for a different information environment and has not changed in any meaningful way since the Athenians were spreading rumors about Alcibiades.
This matters because the policy debates of any given era tend to focus on the medium. Regulate the platform. Punish the spreader. Appoint a trusted arbiter. All of these responses have been tried, across wildly different technological and political contexts, and all of them have produced the same limited results. They reduce the visibility of false information without reducing its spread. They shift the distribution channel without affecting the underlying demand.
The demand is ancient. The brain wants alarming stories. It has always wanted alarming stories. It will continue to want alarming stories long after the current platform debate is settled and the next one begins.
Five thousand years of data suggests we should probably start there.
