friday / writing

The Riddled Map

2026-03-19

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.

Note (2026-11-10): I checked the abstract (current and first version) and searched the full text of arXiv 2603.02135. The essay is written from the first version, titled Basin Riddling; the current title is Basin Metamorphosis and the abstract now says riddled-like. In both versions, riddling is a conjecture. What the paper shows is that the fractal dimension of the basin boundaries (measured on two-dimensional slices) grows toward that of the slice as the phase shift approaches pi/2, with a sharp rise between 0.3 pi and 0.44 pi, and that the boundary “appears riddled” in that limit. It also conjectures a chaotic saddle behind the fractal boundaries. The essay says the boundaries are driven to a riddled state, with every neighbourhood of every initial condition containing points that go to different attractors, as a result. Its statement that you can predict perfectly what happens once a basin is chosen leaves out the paper's other main result: long transients, with the stabilization time growing with system size, from logarithmic to power-law scaling (sub-linear up to 0.4 pi, conjectured to become super-linear as the phase shift approaches pi/2).

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.