Germs Don't Decide: Four Plagues, Four Human Forks

The Peloponnesian War began in 431 BCE: Sparta's army was unbeatable on land, Athens' fleet unbeatable at sea. Pericles (c. 495–429 BCE) drew the militarily impeccable conclusion — abandon the Attic countryside, never give battle — moving the rural population inside the city and the Long Walls, supplied by sea until the enemy tired.

Tens of thousands crowded in, housed in stifling improvised huts. In the summer of 430 the disease came ashore at Piraeus — the very supply line keeping the city alive. Thucydides caught it and survived, leaving in Book II of his history antiquity's coolest clinical record, and also its record of social dissolution: people stopped observing funerals and laws, because "judgment would arrive long after they were dead." Over four years Athens lost roughly a quarter to a third of its people. Pericles himself died of it in autumn 429; with the architect gone, Athens turned to the recklessness of Cleon and lost the war two decades later.

Concentrating for efficiency — supply-chain hubs, a single cloud region — is the optimum in the normal state and an amplifier in the shocked one. Availability and survivability are not the same objective function.

The Black Death of 1347–51 did not depart; for three centuries it returned every decade or so. Ragusa was a city-state republic on the eastern Adriatic, fewer than ten thousand people, with almost all its income from entrepôt trade — stopping a merchant ship meant stopping its own bloodstream. Nobody knew the cause. There was only one empirical rule: a plague ship docks, and the city starts dying.

On 27 July 1377 the Great Council of Ragusa passed the trentino: people and goods from infected regions had to spend thirty days outside the city, on the island of Mrkan or at Cavtat, before entry; violators shared the penalty. This is the first documented institutional quarantine — and what matters is not the idea of isolation but that it turned public health into an enforceable, priceable commercial rule: thirty days is a price. Too short and it leaks; too long and ships call elsewhere. Venice built the permanent Lazzaretto Vecchio in 1423 and later stretched the term to forty days — quaranta, hence quarantine.

Border rules and circuit-breaker thresholds in 2020 posed the same question: not "control or not," but where to set the number and who eats the cost. Every isolation policy is fundamentally an act of pricing.

In June 1520, on the Noche Triste, Cortés was driven out of Tenochtitlan having lost half his force — fewer than a thousand Spaniards against a Mexica polity of some two hundred thousand. Months earlier, Narváez's expedition had brought from Cuba something the Americas had never held: smallpox. The indigenous population had no history of coexistence with it, and therefore no herd immunity.

The epidemic reached the Valley of Mexico in September 1520. Cuitláhuac, the new tlatoani who had just beaten Cortés and was rebuilding the defense, reigned about eighty days before dying of smallpox. The Florentine Codex records that people "could not walk, they could only lie stretched out like corpses," with no one left to fetch water or cook — what the disease severed was not only lives but the chain of command. Cortés regrouped among the Tlaxcalans, laid siege in May 1521, and took the city on 13 August. The conquering force was never those few hundred Spaniards, but tens of thousands of anti-Mexica Tlaxcalan allies, plus a plague that hollowed out the opponent's political center.

Globalization distributes risk unevenly by exposure history: the same shock is a head cold to the side with antibodies and a collapse to the side without.

In the autumn of 1918 the United States was sprinting toward its fourth Liberty Loan, and under the Sedition Act newspapers printed no bad news (the name "Spanish flu" stuck only because neutral Spain did not censor). Philadelphia's health director Wilmer Krusen, knowing the navy yard already had hundreds of influenza cases, approved the Liberty Loan parade of 28 September — two hundred thousand people packed downtown. In the same weeks, St. Louis health commissioner Max Starkloff faced the same uncertainty and the same pressure from business.

Seventy-two hours after the parade, every hospital bed in Philadelphia was full; more than 4,500 died in the first two weeks of October, with bodies left on stoops awaiting collection. Starkloff, two days after his first cases on 5 October, closed schools, theaters and churches and restricted gatherings, holding firm against commercial protest. St. Louis ended with roughly a third of Philadelphia's excess mortality — and the seventeen-city comparison by Hatchett and Markel (PNAS, 2007) confirmed the pattern: interventions that are early, fast and sustained track robustly with lower death peaks. Seven years earlier and half a world away, in the Manchurian pneumonic plague of 1910–11, Wu Lien-teh concluded the disease spread person to person by droplet rather than by rats as then believed, designed a double-layer gauze mask, and persuaded the court to cremate thousands of unburiable bodies at the end of January 1911 — a grave ethical offense at the time. The outbreak ended four months later.

The divergence between states in 2020 replayed Philadelphia and St. Louis: the difference was not knowledge but who was willing to bear the political cost of acting before the evidence was complete.