Dissipative Structures

A candle flame has a shape, and it holds that shape for hours. But it is not a thing — it is what an ongoing process looks like. Stop the process and the shape does not break. It simply stops existing.

01A steady reading can mean two very different things

First, prise one word away from its everyday sense. In ordinary speech "equilibrium" means something like a stand-off — forces evenly matched, situation stable. In physics, thermodynamic equilibrium is more extreme than that: temperature, concentration and pressure the same everywhere, no net flow of anything, and no further change. Equilibrium isn't stable. It's finished.

Against it sits the steady state: the reading is also constant, but underneath it two large flows are cancelling. A bathtub sitting half full might have both taps and drain shut (equilibrium), or it might have water pouring in and draining out at exactly matched rates (steady state). The water level cannot tell you which — and the two call for opposite interventions.

Ilya Prigogine (1917–2003) won the 1977 Nobel Prize in Chemistry for taking the second case seriously, and gave this flow-sustained kind of order a name: the dissipative structure. The word "dissipative" is the point. It doesn't mean the structure is wasteful. It means the dissipation is the condition of its existence.

The dividing line isn't how tidy something looks. It is whether it needs continuous input in order not to disperse. The neat lattice of an ice cube is free: seal it in a constant-temperature box and it can stay that way indefinitely. A convection cell, a hurricane, a living person all demand that something keep flowing through them — cut it, and within seconds to days they are gone.

Do the arithmetic on a person: about 2000 kcal a day works out to roughly 100 watts, about the same as an old incandescent bulb. The exquisitely detailed order that is you is priced at one light bulb. And most of that 100 W isn't spent doing anything — it goes on holding the line: pumping ions back out, dismantling and rebuilding misfolded proteins, keeping what belongs inside inside.

02Entropy is not destroyed, it is exported

Here is a very common misunderstanding to clear up. The second law of thermodynamics says entropy — roughly, "how disordered" something is, or more precisely, how many microscopic arrangements are compatible with what you can see — never decreases. Which prompts the obvious objection: aren't life, cities and all this fine structure plainly decreasing disorder?

The objection drops a qualifier. The second law is about isolated systems — ones exchanging neither energy nor matter with anything. Earth, you, and a city are all open systems. An open system's entropy can fall, provided it pushes at least as much entropy outwards.

Erwin Schrödinger put it bluntly in What Is Life? (1944): living things feed on "negative entropy". He later added a note saying free energy would have been the more accurate term. The correction matters — organisms don't ingest some mystical substance called order; they ingest usable energy, and the order is what happens while that energy passes through.

Earth's books make the point cleanly. Earth receives energy from the Sun and radiates almost exactly as much back out — on energy it balances to the joule. But the quality on the two sides is nothing alike: what arrives is visible light from a 5800 K surface, what leaves is infrared at around 255 K. Same energy, some twenty times colder, so it takes roughly twenty times as many photons to carry it away.

That step is worth pausing on: Earth has never been short of energy. It is short of exits for entropy. If the Sun kept shining and the planet could not radiate, the temperature would climb until no structure could hold together. So "energy flow" is a slightly sloppy way to put it — what actually pays the bill is the drop in energy quality: high grade in, low grade out, and the spread between them funds everything structured on this planet.

03Turn up the flow and the structure is replaced

Now take the quality gradient of the last section, treat it as a dial, and slowly turn it up. The classic bench setup is a thin layer of fluid heated from below and cooled from above; the temperature difference is the dial → ref · the Lorenz system.

With a small difference, heat crosses by molecules jostling each other, the fluid itself does not move, and there is no visible structure at all. Increase the difference and at a specific value the whole layer suddenly starts moving, arranging itself into convection cells. Fluid dynamics puts a number on the threshold: the Rayleigh number — temperature difference, layer depth and viscosity rolled into one dimensionless quantity — crossing roughly 1708, beyond which standing still is no longer an option.

The point is not that a pattern appears above a threshold; that was Topic 13. The point is what comes next: keep turning, and the system does not do the same structure harder — it swaps in a different one. Regular rolls → wobbling rolls → periodic oscillation → full turbulence. Each transition is a bifurcation: a parameter value at which the system abruptly changes the kind of behaviour it has. You had one dial. The system gave you a whole sequence of qualitatively different structures.

Chemistry supplies a prettier example. In the early 1950s the Soviet chemist Boris Belousov found a solution that changed colour periodically — red, yellow, red, yellow, back and forth like a clock. This was widely held to be impossible: aren't reactions supposed to run one way to equilibrium and stop? His paper was rejected twice. A decade later Anatol Zhabotinsky reproduced and extended the work, and the thing is now called the Belousov–Zhabotinsky reaction, or BZ reaction.

It violates no law. It keeps oscillating because the system is held far from equilibrium: feed it and it swings indefinitely; let the reagents run out and it falls quiet, entering true equilibrium at last. In state space, this self-sustaining loop is called a limit cycle → ref · phase space and attractors. Prigogine and a colleague wrote down the minimal toy model that produces such behaviour in 1968 → ref · the Brusselator.

Which yields a distinctly non-obvious corollary: regular, clockwork behaviour can be evidence of being far from equilibrium rather than evidence of having settled. Something that happens punctually year after year may be very far from rest indeed — and it is punctual only because some flow has not yet stopped.

04Where this breaks down

Dissipative structures are among the most cited — and most overstretched — ideas in complexity science. Four boundaries you need before using it.

First, Prigogine's own theorem does not reach this far. Around 1945 he proved an elegant result, the theorem of minimum entropy production: in a steady state, a system settles where entropy production is minimised. The catch is the fine print — it holds only near equilibrium, in the linear regime where flows are still proportional to forces. Dissipative structures all live outside that regime. So in the interesting territory we simply do not have a general "the system optimises X" principle. This gets muddled constantly: invoking minimum entropy production to explain life, organisations or cities transplants a near-equilibrium theorem to precisely where it is declared not to apply.

Second, the symmetric claim isn't a law either. A rival school holds that far-from-equilibrium systems maximise entropy production — the maximum entropy production principle (MEP). It has produced decent predictions in some climate and fluid problems, but there is still no generally accepted derivation and no clear statement of when it applies. Anyone explaining everything with "systems always maximise entropy production" is treating an open question as settled law.

Third, "dissipative structure" does not explain the origin of life; it supplies a necessary condition. Saying life is a dissipative structure is true and thin — so are hurricanes and candle flames. Everything else life has (heritable information, the division of labour between metabolism and replication, variation that can accumulate) simply does not follow from the theory. Treating "life is far from equilibrium" as an account of life's origin passes off a shared property as a mechanism.

Fourth, don't use entropy as a metaphor. Entropy has a precise definition, units, and a computable value. In phrases like "this company's entropy is rising" or "there's too much entropy in the relationship", it has no dimensions, cannot be measured, and therefore cannot be wrong. If you want to use this vocabulary, the minimum bar is to say what the flow is, in what units it is counted, and who pays. If you can't, saying "this will fall apart if nobody maintains it" is both clearer and more honest.