the second hundred · metaphor 124
A new thing — a movement, a company, a habit — has to start somewhere small. But small is exactly when it's most likely to die. Why do most beginnings dissolve back into the world they meant to change, and what does it take for one to catch?
The first three people who believe are surrounded by everyone who doesn't. The founding team is all overhead and no momentum. The new routine is pure friction and no reward yet. From the inside it feels like willing something into being against a current that keeps washing it away — and usually the current wins. The gathering breaks up, the startup folds, the resolution lapses by Thursday. Not because the idea was wrong, but because it never got big enough to hold together.
There is a threshold. Below it, being different costs more than it pays, and the thing shrinks. Above it, the same difference finally pays for itself, and it grows on its own. Physics meets this exact problem every time a liquid tries to freeze or a vapor tries to condense, and it has a name for the threshold and the wall in front of it: nucleation.
The wall in front of every beginning
A seed of the new phase has two competing energies. Its bulk — the interior, the members who are already in, the conviction already shared — is a gain, and it scales with volume, with r³. Its surface — the boundary where the new thing rubs against the old world, the exposure, the friction, the people you must persuade at the edge — is a cost, and it scales with area, with r².
For a small seed, area beats volume: it is almost all boundary and almost no interior, so the cost wins and the thing shrinks. As the seed grows, volume catches up and overtakes area — but only after a hump. The top of that hump is the critical radius r*, and the height of it is the barrier ΔG*. Below r*, every step of growth is uphill and the seed dissolves. Above r*, every step is downhill and it grows without further help. The whole drama of starting anything is getting one cell across that peak.
Raise the supersaturation — make the surrounding world more ready, more primed, more full of would-be believers — and the peak slides inward and downward: r* shrinks, ΔG* collapses. A very ready world needs only a tiny seed. An indifferent world demands a huge one, and mostly gets nothing.
What to try
The seed is a real point sliding on a real energy curve, nudged by real noise; nothing is scripted. Set a seed size just left of the peak and press Drop: it rolls back toward zero — the classic under-critical death, all effort washing away. Nudge the size just right of the peak and drop again: the same tiny difference now grows on its own, faster and faster, and the panel flags it grown.
Now leave the size fixed and raise supersaturation. Watch r* march left until it slips under your seed — the instant it does, a seed that was dying starts to grow, no bigger than before. Then turn up thermal noise and drop an under-critical seed a few times: once in a while, a lucky run of jostles carries it over the wall anyway. That rare, unearned survival is real nucleation — most seeds die, a few catch by luck, and a ready-enough world tilts the odds.
The mapping
The same arithmetic governs anything that must begin as a minority of itself. A movement: the first few are all exposure and no cover — pure surface — and disband unless enough gather to make belonging cheaper than standing out. A startup: below some size it is all coordination cost and no network effect; past it, each new user makes the next one easier. A habit: the first week is friction with no reward banked, which is why most resolutions dissolve before the interior — the identity, the automaticity — has volume enough to hold.
This reframes "willpower" and "hype" as ways of getting one cell over the wall, not of sustaining it forever. You don't need to push a growing thing; above r* it pushes itself. You need to push a small thing past a threshold — and the smartest move is often not to push harder but to raise the supersaturation: find the readiest corner of the world, seed it there, and let a smaller cell suffice.
Read as life lessons
A thing dies young not because it's wrong but because it's mostly boundary. The task isn't to be big; it's to clear one threshold, after which size takes care of itself. Aim at r*, not at forever.
Raising supersaturation shrinks the seed you need. Seed a primed room, not a cold one — the same push that fails in an indifferent world catches in a ripe one. Timing is not luck; it's lowering the wall.
Under the critical size, dissolving is the correct outcome, not a failure of nerve. Expect a graveyard of small starts; the winners are the rare few that cleared the hump, often by luck. Seed many.
In the wild
Pure water super-cools far below 0°C because ice needs a critical seed to form; a speck of dust — a heterogeneous site — hands it one, and rain falls. No seed, no phase, however cold.
Studies of protest and adoption find tipping fractions: below a committed minority the effort disperses; above it, the norm flips and spreads on its own. A first bloc that survives is the whole game.
Network products chase a critical mass where each user pays for the next; behavior-change work chases a streak long enough to bank an identity. Both are fights to cross r* before the seed melts.
The mapping, exactly
| Mathematics | Life |
|---|---|
| surface energy ∝ r² | The cost of the boundary — exposure, friction, the people at the edge you must persuade, the tax of being visibly different. |
| bulk energy ∝ r³ | The interior payoff — shared conviction, network effects, banked identity — the part that grows faster than the edge. |
| critical radius r* | The minimum viable size: the first cell that must exist before growth starts paying for itself instead of costing. |
| the barrier ΔG* | The founding effort — the uphill hump that returns nothing until it's cleared, where most beginnings quietly die. |
| supersaturation | How ready the surrounding world is — a primed market, a discontented population, the ripe moment — which lowers the wall. |
| dissolving back | The under-critical seed melting into the water it meant to change: the group that disbands, the app with no one on it. |
The honest model
The curve is classical nucleation theory, one line of arithmetic: the free-energy cost of a spherical seed of radius r is a surface term that grows and a bulk term that pays, ΔG(r) = 4πγ·r² − (4/3)π·Δg·r³. The surface tension γ is held fixed at 1; the bulk driving force Δg = ln(S) comes straight from the supersaturation slider S. Setting the slope to zero gives the peak exactly.
The seed itself is not scripted: it obeys overdamped dynamics on that curve, dr = −M·ΔG′(r)·dt + noise, reflecting at r = 0. Below r* the slope pushes it toward nothing; above r* the slope pushes it outward. Every readout — r*, ΔG*, the net tug, the survival margin — is measured from the same curve and the same moving point, so the fates you watch are the model's, not a story laid over it.
only γ is hand-set (a single unit-fixing choice); Δg, r*, ΔG*, the tug, and the seed's whole path are computed from the equations above every frame, and the growth-or-death you see is the honest result of the noisy dynamics, not a predetermined outcome.
Where the metaphor tears
Classical nucleation assumes a seed forming spontaneously in clean, uniform liquid — the hardest case, with the tallest wall. But water freezes on dust, and movements crystallize on defects: a charismatic founder, an existing church or union, a scandal. Heterogeneous nucleation slashes the barrier, and the fixation on "critical mass in a vacuum" overstates how big a seed must be. Find the dust.
The r² vs r³ law assumes a compact ball whose boundary is fixed by geometry. A movement can choose its shape — media, ritual, and identity make the boundary cheap, cutting surface cost without adding a single member. The geometry that dooms a small seed is partly a design decision, not a fate, which is exactly why organizing works.
Here supersaturation is a dial you set and the seed reacts. In life the seed changes the solution: a growing movement can raise its own supersaturation (converting bystanders) or exhaust it (backlash, saturation, burnout). That feedback — the phase reshaping the conditions that birthed it — is real history the tidy curve leaves out.