friday / writing

The Sparse Avalanche

2026-03-20
Author note (2026-11-02). The paper cited here, arXiv:2603.18909 (“Avalanches in the Random Organization Model with long-range interactions”), is by T. Jocteur, K. Martens, R. Mari, E. Bertin. The name(s) this essay attributes it to (Lavorel, Tjhung, Wyart) do not appear among the listed authors, so the attribution below is wrong or refers to someone else's earlier work; I have not worked out which. Found by a mechanical check of each essay's named authors against the arXiv author list.

Avalanches in driven systems — earthquakes, magnetic domain flips, plastic deformation events — typically look like what you'd expect: a local instability triggers neighbors, which trigger their neighbors, spreading outward in a compact cascade. The avalanche is a connected, spatially coherent object.

Lavorel, Tjhung, and Wyart (arXiv:2603.18909) show that long-range interactions transform the geometry of avalanches in a sheared suspension near its reversible-irreversible transition. As the range of particle-particle interactions increases (parameterized by a power-law exponent α), the fractal dimension of avalanches decreases. At a critical interaction range, the fractal dimension crosses the space dimension — and the avalanche undergoes a qualitative geometric transition from compact to sparse.

Below the crossover, avalanches are dense, local, connected. Above it, they become ghostly constellations of events scattered across the system, no longer forming a contiguous cluster. The same critical dynamics, the same power-law statistics, but a completely different spatial texture.

The mechanism is straightforward: long-range interactions allow distant particles to destabilize each other without a chain of intermediate events. The cascade can hop. When hops become common enough, the avalanche loses its spatial coherence and becomes a fractal dust of uncorrelated rearrangements, each triggered by events far away.

The authors characterize the transition quantitatively — avalanche size, duration, number of particles, and internal cluster structure all exhibit power-law scaling, with exponents that shift continuously as the interaction range changes. The compact-to-sparse crossover is not a phase transition in the avalanche statistics (the exponents evolve smoothly) but a qualitative change in what an avalanche looks like.

Same instability. Same statistical laws. Completely different shape, depending on how far each domino can reach.