communicoupling

Different roads,
same destination.

Two children are adopted into the same house, raised by the same parents on the same street — and grow into strangers who share nothing but an address. Two others start from opposite childhoods, one gentle and one brutal, and arrive at the same steady, unremarkable adulthood. A family therapist meets both kinds every week. What decides when where you started still shows, and when it has been washed out entirely?

The answer is equifinality. Ludwig von Bertalanffy drew the line precisely: a closed system's final state is fixed by its initial conditions, but an open system — one held in steady state by constant flow-through and regulation — can reach the same final state from different starting points and along different paths. The endpoint becomes a property of the system's organisation, not of where it began. Closed systems remember their origins perfectly; open ones are free to forget.

Its twin is multifinality: give a system several attractors, or amplifying loops inside, and identical starting conditions can fan out into different final states. Origins underdetermine outcomes in both directions — same road, different destinations; different roads, same destination. Bertalanffy named equifinality in General System Theory (1968); developmental psychologists (Cicchetti and Rogosch, 1996) named its complement. Below, a cloud of starting points falls through a landscape you control with one master dial — openness. Spray them, and watch what the endpoints remember.

openness · γhow hard flow-through pulls toward form
0.90
noise · νrandom jostling each step
15%
basins · Bhow many attractors the landscape has
1
spray widthhow spread the starts are
0.45
endpoint spread
how far apart the finals sit
pairs converged
two random starts, same end
The state space · a cohort falling through the landscape
gold rings are the attractors · each dot is one start, trailing where it has been · watch the cloud converge, freeze, or sort
t = 0.0
Where they end up · the endpoint distribution
the horizontal spread of finals · a broad cloud at launch collapses to spikes under equifinality, splits into several under multifinality, and stays put when the system is closed
Watch what happens
A wide spray of starts falls into a single basin under strong openness. Every trajectory drains to the same point — the endpoint cloud collapses, and where each one began stops mattering.
dp/dt = −γ ∇U(p; B) + √(2ν)·ξ Bertalanffy: a closed system (γ = 0) keeps its initial state — the past, verbatim. An open one (γ > 0) flows toward attractors set by its organisation U, not its origin. One basin: equifinality — every start, one end. Several: multifinality — one start, several ends.

Closed remembers, open forgets

The landscape decides, not the launch.

The instrument is a two-dimensional state space with a landscape U carved into it — one, two, or three basins, set by the B dial. Every point you spray rolls downhill, pulled toward the basins with a strength set by openness γ. That single dial is the whole argument. At γ = 0 the pull vanishes: nothing rolls anywhere, every point stays exactly where it landed, and the final scatter is a photograph of the first. This is the closed system — the archive — and it is the only regime that remembers its origins with perfect fidelity.

Turn openness up and the flow takes over. With one basin, every start — near, far, clustered, scattered — drains to the same steady state; the endpoint cloud collapses to a single point that says nothing about where any trajectory began. That is equifinality, drawn live: endpoint spread falls toward zero while pairs converged climbs toward 100%. Add basins and the landscape grows watersheds. Now a start's fate depends on which basin's pull wins — and near the ridges between them, a breath of noise is enough to tip a trajectory one way or the other. Two points launched a hair apart end up in different worlds. That is multifinality, and the endpoint distribution splits from one spike into two or three. The steady state, in every case, is a property of U and γ — the system's organisation — and not of the spray.

What to try

Three moves that show origins losing their grip.

01

Every start, one end

Load The open system (γ = 0.90, B = 1). Spray a wide cloud and watch it braid inward to a single point. Endpoint spread crashes toward 0.0; pairs converged hits ~100%. Where they began has stopped mattering — organisation beat origin.

02

Freeze the flow

Drag γ to 0, or load The archive, and press New spray. The cloud just sits there; the endpoint histogram keeps the exact shape of the launch. Only a closed system preserves its past verbatim — and one touch of noise begins erasing even that.

03

Twins diverge

Load The forked road (B = 2, narrow spray, noise on). All the starts are nearly identical, yet they sort into two clusters, roughly 50/50. Same history, forked ends — the outcome no longer testifies to the origin.

Why case histories explain less than we hope

The outcome cannot testify about the path.

Biography is therefore treacherous evidence. When many different childhoods can produce the same steady adult — because adulthood is an open system, regulated by work, relationships, habit, and a thousand daily corrections toward a form — then the adult in front of you cannot tell you much about the child they were. The convergence has thrown that information away. Family therapists know it in their bones: the well-functioning client with the horrifying history and the struggling client with the idyllic one are not anomalies, they are what equifinality looks like from the consulting room. Origin explains far less than the intuitive theory of life-as-consequence promises, because open systems are shaped by what they keep doing, not by where they began.

The same structure indicts a common error of inference. Watch The false attribution: two sprays launch from opposite corners and both drain to the same point. An observer who arrives at the end and reasons backward — this person turned out this way, so their origin must have been thus — is wrong by construction, because two opposite origins produced the identical end. Multifinality closes the trap from the other side: when the same founding conditions can issue in flourishing or collapse depending on which loops happened to amplify, then a shared beginning guarantees nothing about a shared fate. Twins, cohorts, start-ups from the same garage, nations built on the same charter — the more open and loop-laden the system, the less its origin binds it.

In the family of the loop

Kin concepts.

Equifinality is the loop cluster's quiet law about memory. The homeostat shows the machinery underneath — a device that returns to its stable setting from wherever you knock it, an attractor made physical. Structural coupling explains where a system's basins come from in the first place: they are carved by its long history of exchange with an environment, which is why organisation, and not origin, ends up doing the steering. And autopoiesis gives the deepest reading of the "open system" at work here — a living thing whose steady state is something it continuously produces, so that its final form is authored by its ongoing operation rather than inherited at birth.

The mapping

Mechanism ↔ social life.

In the modelIn the world
the initial sprayThe childhood, the founding charter, the first draft — where a life or an institution began.
openness γHow strongly ongoing exchange and regulation pull the system toward a form.
the attractor / basinThe person or institution the system keeps becoming, whatever its start.
one basin (equifinality)Why case histories explain less than we hope: many roads, one outcome.
several basins + noise (multifinality)Why identical histories still diverge: same start, forked ends.
the cohortA set traced together — siblings, a founding class, a generation.
endpoint spreadHow much of the origin still shows in the outcome; near zero once organisation has won.

Where it tears

Limits.

Equifinality frustrates causal inference.

If many paths reach the same outcome, the outcome cannot testify about the path. This is a methodological wound. In clinical and historical work we routinely reason backward from a result to its cause, and equifinality says that inference is often unsound in principle, not underpowered. The convergence has destroyed the evidence you were hoping to read off the end state.

"Same final state" is coarse-grained.

Two adults who look equally "well-adjusted", two firms equally "stable", converge only at the resolution of the categories you brought. Describe them finely enough — the private fears, the exact balance sheet — and the convergence dissolves; the roads never truly met. Equifinality is always relative to the observer's grain. Sharpen the grain and some of it turns back into ordinary path-dependence.

Open and closed is a spectrum, not a switch.

Bertalanffy's clean dichotomy is an idealisation. Real systems are semi-open — leaking a little of their past, regulating a little toward a form — and where regulation ends and drift begins is an empirical fight, not a setting on a dial. The γ knob here is a clarifying fiction; in the field you rarely get to read openness straight off the system.