On September 3 at 17:31:59, a 92-page probability paper appeared on arXiv. Four seconds later, a second one. Five seconds after that, a third. Nine seconds, 264 pages, all sharing the same title stem — *Conformal Intrinsic Metrics on CLE4*, split into parts I, II, and III. Adding up the page counts: 92 + 65 + 107 = 264 pages, 27 figures.
The four authors are Emmanuel Kammerer, Konstantinos Kavvadias, Jason Miller, and Yi Tian. Miller is at Cambridge, and this work is supported by his ERC Consolidator Grant ARPF. All three signature pages show Cambridge affiliations, and all three submissions were made by Kammerer.
First, What Is CLE4?
Zoom in on a critical percolation configuration to infinity: the interfaces collapse into a family of randomly distributed smooth curves that fill the entire plane. The mathematical name for such a "random collection of loops" is CLE (conformal loop ensemble), parametrized by the SLE family — Schramm introduced SLE in 1999, and in 2006 Camia and Newman turned the full-scaling limit of critical percolation into a loop ensemble, paving the way for random conformal geometry.
κ = 4 is a critical position. In the words of the CLE4 abstract: at this parameter (and below), the loops are simple, non-self-intersecting, and disjoint from each other and from the domain boundary. For larger κ, loops touch themselves and each other — a wilder world entirely.
And the "metric"? It is a ruler built on this family of random loops: a distance between any two points (or loops), satisfying three properties — conformal invariance (the distance relations are unchanged under conformal transformations of the domain), locality, and metric-ball growth consistent with the uniform explorations of Werner–Wu. The abstracts of all three papers lock in these properties word for word.
Why Build This Ruler
Because the previous situation was awkward: the metric had been constructed in unpublished work of Sheffield, Watson, and Wu — it existed, it was written down, but nobody could cite it, rely on it, or build on it. Part I of the trilogy puts it politely: "This metric was previously constructed in unpublished work of Sheffield, Watson, and Wu."
The Cambridge group took a different route: instead of starting from a construction, they started from a limit. Place graph distances on the loops of CLE_κ, let κ approach 4, renormalize globally, and see whether the limit produces the desired metric. This "graph distance → renormalization → scaling limit" technique is the same craft as the program that turned LQG heuristic metrics into theorems over the past few years.
The three papers each hold one section. The first paper proves that a subsequential limit of the renormalized graph metrics exists, yielding a nontrivial conformally invariant local metric (main result, Theorem 1.3). The second proves geodesics exist — there is a shortest path between any two loops, and the path almost entirely hugs the loops of CLE4 (the set of deviations has zero Hausdorff dimension) and never touches the domain boundary; along the way it also establishes exponential tail bounds for rectangle distances. The third paper closes things out: such a metric is uniquely characterized by its properties (Theorem 1.5), convergence is full rather than merely subsequential (Theorem 1.7), geodesics are unique (Theorem 1.8), and the entire metric is a measurable function of CLE4.
Why It Is Worth Watching
First, between "unciteable" and "citeable" lies an entire field's infrastructure. Only once the metric is unique, measurable, and equipped with geodesics can later work treat it as an axiom. The forward-looking sentence in part III's abstract is not just courtesy — "this characterization will have important applications in future work" — and the key is that "the metric is determined by geodesics from each loop to the boundary, and these geodesics are determined by the uniform exploration."
Second, it connects two communities. One approaches CLE4 from SLE/exploration processes; the other views random spaces through the metric program of LQG. One ruler, two doors.
Third, sheer stylistic appreciation. Submitting three papers within nine seconds means they were polished together and submitted as one act — a kind of classicism in the fast-food era of preprints. A 264-page theorem chain with no break between the first page and the last.
Build the ruler first, applications later. This order is common in mathematics; its reversal is rare. It is hard to know in which fields these papers will first find use — the abstracts don't say, and I haven't found out either. What is certain: as of September 3, the map of CLE4 has a standard scale.