A thousand fireflies in a dark meadow, each blinking to a private clock, no two clocks quite alike. Each one can see the others, and leans its timing a little toward the crowd. The slider sets how hard they lean. Slide it up slowly. For a while nothing happens. Then, past one exact point, with nobody in charge, the meadow starts to flash as one.
For the curious kid.
This really happens. Along rivers in Southeast Asia, whole trees of fireflies flash together like one slow lamp, and for two weeks each June a hillside in the Great Smoky Mountains does the same. There is no conductor firefly. Each one only has a clock inside it and a simple habit: when it sees the others flash, it hurries or waits a tiny bit to be closer to them.
Here is the surprising part, and you can find it with the slider. A little leaning does nothing. The clocks are too different, and the meadow just twinkles. But there is a tipping point, here at exactly 1. Cross it and a small group falls into step by luck, their flash together is brighter than the noise, that pulls in a few more, which makes it brighter still. Agreement feeds itself. The gold fireflies are the ones the crowd has captured. The blue ones have clocks so fast or so slow that they keep their own time. There are always a few. Press "All together now" with the slider low and watch a perfect agreement fall apart: below the tipping point, even a crowd that starts in step cannot stay there.
Deeper.
This is the Kuramoto model (Yoshiki Kuramoto, 1975, building on Arthur Winfree, 1967). Firefly i has a phase θi and a natural pace ωi, and obeys dθi/dt = ωi + (K/N) Σ sin(θj − θi). Average all the clock hands as arrows and you get one arrow of length r pointing at phase ψ. That length is the meadow's agreement: 0 for a twinkle, 1 for one lamp. The sum then collapses, and each firefly feels only the crowd: dθi/dt = ωi + K r sin(ψ − θi). The pull on every firefly is proportional to the agreement that already exists. That feedback is the whole story.
A firefly is captured when the pull can cancel its own pace: |ω| ≤ K r. It then sits at a fixed angle ahead of or behind the crowd, where sin(θ − ψ) = ω/(K r). That is the curve drawn on the ring, and the gold dots lie on it. The rest circle forever at speed √(ω² − K²r²). Demand that the captured ones produce exactly the agreement that captures them and you get the threshold: Kc = 2/(π g(0)), where g is the spread of natural paces. Here the paces follow a Lorentzian of half-width γ = 0.5, for which the algebra closes completely: Kc = 2γ = 1, and a settled meadow has r = √(1 − Kc/K). That is the curve on the right. It was not fitted to the dots.
In 2008 Edward Ott and Thomas Antonsen found something stronger. For these clocks, the whole meadow's agreement obeys one small equation, dr/dt = −γ r + (K/2) r (1 − r²), which can be solved with pencil and paper. So the page can forecast. Each time you move the slider or press a button, the dashed gold line in the middle chart is drawn out into the next twenty seconds from that formula, and then the meadow, a thousand separate clocks, walks along it.
The same mathematics describes the thousands of pacemaker cells that agree on your heartbeat, an audience's applause collapsing into a rhythm, and the generators of a power grid holding one frequency. The square-root rise past a threshold is the signature of a phase transition, the same shape as a magnet gaining its pull as it cools.
Honesty notes, museum policy. Every firefly here sees every other one equally; real ones mostly see their neighbours, and they signal in pulses rather than by a smooth pull, so this is the idealised version. The thousand paces are not drawn at random: they are the exact quantiles of the Lorentzian, every kind of clock in its true proportion, so this meadow behaves like a far larger one. Checked headless, its settled agreement matched the formula to three decimals at four settings, and its path stayed within 0.04 of the forecast. A Lorentzian has a long tail: a few percent of these clocks are so wild that nothing captures them, which is why the agreement never reaches 1. From a scatter, the moment of take-off is luck, since the seed is noise, so the forecast waits until r passes 0.25 before it commits. And at the threshold itself the crowd decides so slowly that a dot measured there sits a little high. The meadow keeps one more memory, of a bridge.
Nobody leads. Each one leans a little toward the rest, and past a certain point the leaning is enough.