the second hundred · metaphor 147
A rumour, a panic, a joke, a disease — some sweep through a whole office or city, and some die in a single retelling. What separates the two isn't the size of the spark. It's one ratio: how fast the thing passes on, divided by how fast it burns out.
We love to explain what spreads with stories about the thing itself — the juicy rumour, the frightening virus, the catchy tune. But underneath every contagion is the same dull bookkeeping. Each carrier passes it to some number of new carriers before they stop carrying it. Call that number the reproduction ratio. If it lands above one, each case more than replaces itself and the thing grows. If it lands below one, each case fails to replace itself, and it fades — no matter how loud the spark.
The unnerving part is what happens right at one. The behaviour is not smooth. Nudge the ratio from just under to just over and you do not get slightly more spread — you cross a cliff, from almost always dies to almost always sweeps. And the eventual toll is not proportional to the ratio; it follows a specific curve that leaps off zero at the threshold. The same match, dropped into two rooms whose ratio straddles one, does two entirely different things.
The reproduction number
The whole drama compresses into one number. In the simplest model of spread, a carrier is infectious for a while and, during that while, passes the thing on at some rate. Divide the passing-on rate by the burning-out rate and you get R₀ — the average number of fresh cases a single case produces in a fully susceptible crowd. It is a ratio of two speeds: transmission over recovery. Nothing about the drama of the disease or the juiciness of the rumour enters except through those two speeds.
Everything hinges on whether that ratio clears one. Below one, a chain of cases is like a family tree where each generation is smaller than the last: it may sputter along briefly but is doomed to end. Above one, each generation is larger, and — until the supply of susceptibles runs low — the thing grows explosively. Exactly at one, each case just replaces itself, the knife-edge between extinction and takeoff. This is why the same rumour lands so differently in two rooms: the rooms have different R₀.
And the toll, if it takes off, is not a free parameter. Kermack and McKendrick showed in 1927 that the eventual final size — the fraction ever touched — is pinned by R₀ through a single equation, z = 1 − e^(−R₀·z). That curve is flat at zero up to the threshold, then rises steeply: a small R₀ over one already infects a big minority; an R₀ of two takes four in five. The threshold is a genuine phase transition, and the final-size curve is its fingerprint.
What to try
Set the R₀ slider well below 1 and press Release one outbreak: a few red cases flare and gutter out, the population stays blue, the epidemic curve barely twitches. Now push R₀ above 1 and release again — the red floods the grid, crests, and drains to immune green, and the curve draws the classic bell of an epidemic. The point on the right-hand panel drops onto the final-size curve at your R₀; do it for several values and the fingerprint appears — flat along the bottom, then leaping up just past 1.
Then linger near the edge. At R₀ around 1.1–1.3, release the same outbreak several times: sometimes it sweeps, sometimes it dies in a few cases — same ratio, opposite fate, decided by early luck. Press Run 25 outbreaks to make this visible at once: below 1 every dot hugs zero; just above 1 the dots split into a low cloud of stochastic die-outs and a high cloud on the theory curve. The threshold is sharp in the average and ragged in any single throw of the dice.
The mapping
Read the population as any crowd an idea can pass through, and "infected" as anyone currently spreading it — telling the rumour, forwarding the post, joining the panic, repeating the phrase. Transmission is how readily a carrier passes it on: how contagious the story, how many people they tell, how believing the audience. Recovery is how fast a carrier stops — loses interest, gets corrected, moves on. Their ratio is the R₀ of the idea, and it decides, before any content does, whether the thing has a future.
This reframes the eternal question "why did this one blow up?" Usually the honest answer isn't in the spark's brilliance but in a ratio that happened to clear one — a slightly stickier message, a slightly more connected messenger, a slightly more receptive week. It also explains why killing a contagion doesn't require silencing every carrier: you only have to drag R₀ below one, by any combination of making it less transmissible or making carriers quit faster. Under the threshold, the thing extinguishes itself. That is the whole logic of a correction going out, a quarantine, a cooling-off — not to catch every case, just to tip the ratio.
Read as life lessons
What decides survival isn't the spark's size but whether each case makes more than one more. A tiny seed above the line wins; a huge seed below it loses. Look at the ratio, not the flare.
Near one, tiny changes flip the outcome entirely — there's no gentle middle. Systems poised at their threshold are the ones where a small nudge, or plain luck, decides everything downstream.
To end a contagion you needn't reach everyone — only pull R₀ under one. Beneath the line it dies on its own. Small, well-aimed friction beats heroic, total suppression.
In the wild
Measles runs at an R₀ near 12–18, seasonal flu closer to 1.3 — which is why one needs vast immunity to stop and the other flickers yearly. Vaccination works by cutting the susceptible fraction until effective R drops below one.
The same maths underlies whether a post, meme, or piece of misinformation goes viral: forwards per reader over how fast readers lose interest. Platforms tune, and fact-checkers fight, that very ratio.
Bank runs, fire-exit stampedes and market panics spread person-to-person by fear; whether a scare self-extinguishes or cascades depends on how many each frightened person frightens before calming down.
The mapping, exactly
| Mathematics | Life |
|---|---|
| transmission rate β | How readily a carrier passes it on — how contagious the story, how willing the audience to believe. |
| recovery rate γ | How fast a carrier stops — loses interest, gets corrected, calms down, moves on. |
| R₀ = β ⁄ γ | New cases per case: whether the thing more than replaces itself. The one number that decides. |
| the threshold R₀ = 1 | The knife-edge between a story that dies in one retelling and one that sweeps the office. |
| stochastic die-out | The supercritical spark that fizzles anyway by early bad luck — why not every good idea catches. |
| final size z = 1 − e^(−R₀z) | How much of the world it eventually touches — a fixed curve, not a free choice, once the ratio is set. |
The honest model
The population is a few hundred individuals, each susceptible, infected, or recovered. Time runs in steps. Every step, each infected recovers with a fixed probability γ; and each susceptible catches it with probability 1 − e^(−β·I/N), where I is how many are currently infectious — the well-mixed rule, everyone able to meet everyone. The slider sets R₀, and the code takes β = R₀·γ, so the ratio you dial really is transmission over recovery. The outbreak runs, stochastically, until no one is infectious.
Nothing is scripted. The grid, the epidemic curve, the final size and outcome are all read off the actual simulated run. The right-hand panel draws the theoretical curve by solving z = 1 − e^(−R₀z) for z by iteration — that smooth line is the prediction — and every dot on it is a real run's realised final size, plotted where it fell. Below the threshold the runs pile at zero; above it they scatter, some on the curve, some dying out early. The cliff at one is not drawn in; it emerges from the arithmetic of replacement.
Where the metaphor tears
It is tempting to speak of "the R₀ of" a virus or a rumour as if it were fixed. It isn't: R₀ is transmissibility times contact rate times susceptibility, and every one of those bends with behaviour, season, and network. The same pathogen has a different R₀ in a crowded city and an empty one — which is precisely why distancing, correction, and caution work. Treat the number as a dial you and the world keep turning, not a constant.
The clean threshold assumes every carrier can reach anyone equally. Real spread runs on lumpy networks with clusters, bridges, and superspreaders. That heterogeneity lets a contagion smoulder in a tight community while the population-wide R sits below one, or explode from a single hub above it. The average ratio is a first word on the outcome, not the last.
The model counts who was "infected" and treats a plague, a revolution, and a bad joke with the same accounting. It says nothing about whether the thing that spread was true, good, or worth catching — and "died out" is not a verdict of "was false." Reach and worth are different axes; the threshold governs only the first.