You Are in
the System

The Clockwork That Wasn't

In 1687, Isaac Newton published the Principia Mathematica and handed Western thought something it had never possessed before: a complete, testable account of how the universe moves. Three laws. One equation for gravity. Suddenly the orbits of planets, the arc of cannonballs, and the swing of pendulums all yielded to the same mathematics. The universe was a machine. And machines, once understood, are predictable.

The consequences spread far beyond physics. By the nineteenth century, the clockwork metaphor had colonised medicine, economics, political philosophy, and eventually management theory. Understand the parts, understand the whole. Identify the cause, engineer the effect. Intervene correctly, and the system responds as intended.

In 1887, Henri Poincaré was working on a prize competition: prove that the solar system is stable. He could not prove it. What he found instead was something that shook the Newtonian world: even a system of just three bodies, each obeying Newton's laws perfectly, produces behaviour that is impossible to predict over long time horizons. Not because of randomness. Not because of missing information. Because of the mathematical structure of the problem itself.

Poincaré's discovery sat largely unnoticed for most of the twentieth century. Then, in the 1960s and 70s, researchers in meteorology, biology, mathematics, and physics began finding the same thing in very different systems: deterministic rules, unpredictable behaviour. Order and chaos, coexisting. The clockwork had cracks in it. The observer was not in control.

What follows is seven encounters with the ideas that emerged from those decades of work — each one demonstrated by a simulation you can play with directly, each one aimed at the same uncomfortable conclusion: that the practitioner who believes they are observing a complex system from the outside is mistaken. And that this mistake has consequences.

Determinism ≠ Predictability

Start here, because this is where the crack appeared first.

Take three objects in space. Give each a mass, a position, a velocity. Apply Newton's laws — deterministic, exact, verified to extraordinary precision. Now ask: where will these objects be in a hundred years?

The question has no general answer. Not because of measurement error. Not because Newton's laws are incomplete. But because the mathematical structure of three mutually interacting bodies does not admit a closed-form solution. Poincaré proved this. The system is deterministic — every future state is fixed in principle by the current state — but it is not predictable. These are not the same thing.

This distinction is not a technical footnote. It is a rupture in the foundation of how most institutions think about planning, intervention, and strategy. The assumption embedded in most theories of change, log frames, and strategic plans is not just that systems follow rules — it is that those rules can be used to engineer outcomes. Poincaré showed that this assumption fails even for three idealised point masses obeying the simplest possible physics.

The Same Start, Two Worlds

A single pendulum is one of the first things a physics student learns to solve. It swings in a regular arc, loses energy to friction, comes to rest. Everything about it is predictable.

Attach a second pendulum to the end of the first. Give the system a push.

The behaviour changes completely. The double pendulum swings, flips, reverses, pauses, lurches. Run it again from what appears to be the same starting position: you get a different trajectory. After a short time, two apparently identical starting conditions produce outcomes with no resemblance to each other.

This is sensitive dependence on initial conditions. Popular science calls it the butterfly effect, though that name understates how deep the phenomenon goes. In certain classes of systems, arbitrarily small differences in starting conditions grow exponentially over time, making long-range prediction impossible in principle, regardless of how good your model is.

The Shape of What You Cannot Know

In 1963, meteorologist Edward Lorenz was re-running a weather simulation. To save time, he re-entered values from a printout — rounding them slightly, from six decimal places to three. The simulation diverged completely. A difference of 0.000127 in a starting value produced, over time, an entirely different weather pattern.

But Lorenz did not stop at the observation that the system was sensitive. He kept looking. The trajectories never repeated, never converged — and yet they did not wander randomly through all possible states. They stayed within a bounded region. They traced a shape: a double-lobed structure now called the Lorenz attractor, folding back on itself endlessly without ever crossing.

A strange attractor: a region of phase space that a chaotic system inhabits without ever settling. The system has structure — you can describe the territory it lives in — but you cannot predict where within that territory it will be at any moment.

The Moment Before the Tipping Point

The logistic map — the simplest nonlinear equation in existence — demonstrates something that the previous three encounters did not: not how systems behave within a regime, but how they move between regimes. As a single parameter increases, the system's behaviour changes qualitatively. A stable equilibrium begins to oscillate. The oscillation doubles its period. Doubles again. At a precise threshold, the system enters chaos. Same equation, same structure — a different world.

Coleman, Vallacher, and Nowak demonstrated formally that opinion dynamics, conflict escalation, and polarisation follow exactly this kind of nonlinear attractor dynamic — with bifurcation points. The community that appeared stable last year may be in a different regime today, not because anything dramatic happened, but because a parameter crossed a threshold.

You Are a Node

Everything so far has concerned how systems behave over time. This encounter concerns something different: how the structure of a system shapes what is possible within it — independently of the intentions or capabilities of any individual actor.

Network science's central finding is both simple and radical: the topology of a network — how nodes are connected, not who or what those nodes are — determines what can spread through the network, what can be blocked, what can cascade, and what remains isolated. A bridge between two otherwise disconnected clusters carries information that neither cluster can generate internally, regardless of how sophisticated the actors within each cluster are. Structure constrains and enables independently of content.

The Programme That Worked

This is the sharpest encounter. Not because the ideas are more complex — but because they point most directly at the practitioner themselves.

The programme reduced conflict. The metrics improved. The log frame was satisfied. By any standard evaluation framework, the intervention was a success — within the boundaries that the evaluation was designed to see.

What the evaluation did not see: the programme had quietly built dependency in the communities it served. Local mechanisms for managing conflict — informal, imperfect, but self-sustaining — had atrophied, because the programme had taken over their function. The rights dispute that had generated the conflict in the first place was left structurally untouched, because addressing it was outside scope. When funding ended, the system had less capacity to hold itself together than before the intervention arrived.

What the Practitioner Does With This

None of what you have encountered here is an argument for paralysis. The three-body problem does not mean you should stop trying to understand systems. Sensitive dependence does not mean interventions are pointless. Strange attractors do not mean strategy is futile. Bifurcations do not mean experience is worthless. Network structure does not mean position is destiny. And the intervener-as-variable does not mean practitioners should step aside.

It means something more demanding. It means the practitioner must hold a more honest relationship with what they can and cannot know — and with how their own presence, assumptions, and framing are shaping the system they claim to be serving. Seriously engaged, these seven encounters produce four practical shifts:

That last shift is the one most complexity frameworks leave out. They describe the system. They do not describe the practitioner inside it. But the practitioner's interior — their assumptions, their certainties, their blind spots, their need for control — is not separate from the work. It is part of the causal structure. Which means that the development of the practitioner is not a soft addendum to complexity practice. It is a core competence.