Critical Mass & Tipping Points

Drop one neutron into a lump of uranium that’s a hair too small and nothing happens — the neutron splits an atom, a couple of new neutrons fly out, and they mostly escape through the surface before hitting anything. Add a little more uranium so the lump is a hair too big, drop in the same single neutron, and the whole thing detonates in a millionth of a second. Nothing changed about the neutron. Nothing changed about the atoms. The only thing that changed was whether each split reliably caused one more split — and that tiny difference is the entire distance between an inert paperweight and a bomb. That knife-edge has a name: critical mass, and the model it gives you is one of the most portable in the whole latticework.

Critical mass is the threshold in any self-amplifying process where the amplification factor crosses 1 — where each unit of activity, on average, causes at least one more unit. Below it, activity dies out faster than it spreads and the whole thing peters out. Above it, activity breeds more activity than it loses, and the process runs away exponentially. The switch between “fizzles to nothing” and “sweeps everything” is not gradual: it flips at a single point, which is why a disease with an R-number of 0.9 quietly disappears while the same disease at 1.1 becomes a pandemic, why a social app is worthless with a thousand users and unstoppable with ten million, and why a standing ovation, a bank run, a viral video, and a forest fire all share one shape. This is the physics behind tipping points: the reason real change so rarely arrives on a smooth ramp and so often shows up as nothing, nothing, nothing — then everything.

This course assumes you’ve met feedback loops — specifically the reinforcing loop, where more begets more — and emergence, because a runaway chain reaction is a pattern that lives in the interactions, not in any single unit. From there it builds the model floor to ceiling. We’ll start where the idea was born, in nuclear fission, and pin down exactly what makes a mass “critical” (the k-factor, and why surface — where things escape — decides everything). Then we’ll lift the model off the reactor floor and show it’s really a statement about any reinforcing loop: the per-step gain crossing 1. We’ll tour the tipping points hiding in plain sight — an epidemic’s R0, viral content, the adoption S-curve and the network effects that make platforms into moats, a crowd tipping into a riot or an ovation, autocatalytic chemistry, even the meeting where nobody speaks until someone finally does. We’ll face the thing that makes thresholds so treacherous for human intuition — their nonlinearity, the way a long quiet build-up looks identical to failure right up until it explodes — and the hysteresis that makes some tipping points nearly impossible to reverse. And we’ll be honest about where the model lies to you: the trap of extrapolating a straight line toward a cliff, the error of reading “slow” as “failing,” and the opposite error of expecting every trend to tip when most simply die sub-critical. You’ll leave able to spot a threshold before you cross it, to tell a sub-critical build-up from a genuine dud, and to reason clearly about the one class of change that ambushes almost everyone.