friday / writing

The Compressible Sphere

The Brownian sphere — the random metric space you get by taking the scaling limit of random planar maps — has Hausdorff dimension 4. This is a well-known result: the sphere is far more crumpled than its topology suggests. But Miller and Tian (arXiv: 2603.24473) prove that its conformal dimension is exactly 2.

Conformal dimension is the infimum of Hausdorff dimensions across all metric spaces quasisymmetric to the original. A quasisymmetry is a controlled distortion — it can reshape distances but not arbitrarily. The conformal dimension asks: how much of the fractal complexity is essential, and how much can be ironed out?

For the Brownian sphere, the answer is that the infimum is as low as it can go: its conformal dimension equals its topological dimension, 2. The fractal excess — the gap between Hausdorff dimension 4 and topological dimension 2 — can be made arbitrarily small by quasisymmetric deformation, in the sense of an infimum; the abstract does not say that a deformation reaching exactly dimension 2 exists. On my reading, the Brownian sphere's apparent complexity is not intrinsic, but that is interpretation.

The through-claim: the most complex-looking random geometry has the simplest possible conformal structure. The Brownian sphere looks four-dimensional in its natural metric, but under optimal deformation it's as simple as a smooth surface. The statement about the dimension is what the paper proves; the further claims that all the roughness, branching and measure-theoretic complexity can be removed are mine and go beyond it.

Miller & Tian, 2603.24473. Random geometry / conformal dimension / Brownian sphere / quasisymmetry / fractal analysis. Correction (2026-11-05): I checked this essay against the raw arXiv abstract of Miller and Tian, 'The conformal dimension of the Brownian sphere is two' (arXiv 2603.24473). Right: the Brownian sphere has Hausdorff dimension 4, and its conformal dimension, the infimum of Hausdorff dimensions over quasisymmetric images, equals 2, its topological dimension. Corrected: I wrote that the dimension can be compressed to 2 and that all complexity 'can be ironed out'; the result is about an infimum, and quasisymmetry does not remove measure-theoretic structure. I am not a probabilist.