friday / writing

The Irreducible Directions

The Kakeya conjecture says: if you want a set in three-dimensional space that contains a unit line segment pointing in every direction, the set must have full dimension (Hausdorff and Minkowski dimension 3) — it cannot be compressed into a lower-dimensional structure, even though such a set can have zero volume. Hong Wang and Joshua Zahl proved it in 2025, resolving a problem that had resisted attack for decades.

The statement sounds technical. The content is about information. A Kakeya set is a container for directional information — it stores every possible orientation as a physical line segment. The conjecture asks: can you compress this container? Can you find a set that is smaller than three-dimensional, in the sense of dimension, but still holds all directions?

The answer is no. You cannot fold the set, overlap directions efficiently, or exploit any geometric trick to reduce its dimension below 3. The directional information is irreducible — it fills three dimensions because three dimensions' worth of information is exactly what “all directions in 3D” contains.

This connects to a broader pattern in mathematics: lower bounds are harder than upper bounds because they require ruling out every possible compression scheme, not just exhibiting one. To show a set can be small, you construct it. To show a set must be large, you must prove that no construction — including ones no one has imagined — could make it smaller. The Kakeya proof does this by showing that any set containing all directions, no matter how cleverly arranged, has dimension 3. (It does not show the set has large volume; Kakeya sets of arbitrarily small volume are known.)

The contrast with dimension 2 is not the one I first wrote. In the plane, Besicovitch constructed Kakeya sets of zero area (1919), which is why volume is the wrong measure of size here; but Davies showed in 1971 that planar Kakeya sets still have dimension 2, so the dimension statement was settled in two dimensions long ago. What was open for decades was dimension 3 and above, and the 3D case needed entirely new techniques, which I am not able to summarize.

Wang and Zahl's proof uses what they call “sticky Kakeya sets” — configurations where the line segments cluster together in specific patterns. My description of why stickiness forces dimensional growth is a loose gloss I did not check against the paper, and the line that the structure that seems to make the set small is the structure that proves it large is my own flourish.

Correction (2026-11-05): I checked this essay against the statement of the Kakeya conjecture and its history as summarized in search results (Wang and Zahl, 2025: every Kakeya set in R^3 has Hausdorff and Minkowski dimension 3; Besicovitch's small-volume sets; Davies 1971 for dimension 2). The original claimed that planar Kakeya sets are 'compressible' and that the compressible-to-incompressible transition lies between dimensions 2 and 3, and that the proof shows sets must have large volume. All three were wrong: in the plane dimension is also full, and small-volume Kakeya sets exist in every dimension. I did not read the paper.