In 1963 the meteorologist Edward Lorenz reduced atmospheric convection to three simple equations. They contain no randomness at all — give them the same starting point and they produce the same future, forever. And yet the system is unpredictable: change the start by a millionth and the paths diverge until they share nothing. This is deterministic chaos, and it is why long-range weather forecasting is impossible in principle, not just in practice.
The shape the trajectory traces — never crossing itself, never settling, never repeating — is a strange attractor. The system is drawn to it from almost any starting point, so its behaviour is bounded and structured; it simply isn't periodic. Order and unpredictability at the same time.
The lesson I carry from this into real systems: "we have all the data and a clear model" does not imply "we can predict it." Some systems are legible and surprising at once. The useful skill is not forecasting the trajectory but recognising the attractor — the shape of where things tend to go.