The dynamics are orderly. Coupled phase oscillators, well-understood, settling predictably onto attractors. The behavior on each attractor is stable, regular, sometimes periodic. Nothing chaotic about it.
The basin boundaries are not. Yan et al. show that a single phase-shift parameter in coupled oscillators drives the boundaries between basins of attraction toward a riddled state — where every neighborhood of every initial condition contains points leading to different attractors. The fractal dimension of the boundaries approaches the dimension of the full phase space. The map of “where does this initial condition end up?” becomes infinitely fine-grained everywhere.
The system isn't chaotic. The attractors aren't strange. The sensitivity to initial conditions isn't in the dynamics — it's in the geography. You can predict perfectly what will happen once the system has chosen a basin. You cannot predict which basin it will choose, because the basins themselves are interleaved at every scale.
This decouples two things usually linked: chaotic dynamics and unpredictability. Chaos makes trajectories diverge. Basin riddling makes destinations unknowable. You can have one without the other. A perfectly regular dynamical system can be completely unpredictable because the terrain on which it operates has been fractured beyond resolution.
The through-claim: prediction fails not just when the rules are complicated but when the map of outcomes is riddled. The rules can be simple. The map can be impossible. Most discussions of predictability ask about the dynamics — how sensitive is the system to perturbation? This paper redirects the question to the geometry: how interleaved are the basins? The answer to “what will happen?” depends on where you are, not on how the system moves.