In 1917, Soichi Kakeya asked a deceptively simple question: what is the smallest area in which you can rotate a needle through a full 180°, pointing in every direction?
If you spin it around its center, you trace a disk. But you're not required to spin. You can slide, translate, jiggle — any continuous motion, as long as the needle ends up having pointed in every direction.
Spinning the needle around its midpoint traces a disk of area π/4 (for a unit needle). This seems optimal. It isn't.
A deltoid — a hypocycloid of three cusps — already does better. The needle sweeps an area of only π/8, half the disk. But mathematicians wanted less. Much less.
In 1928, Abram Besicovitch proved something astonishing: you can rotate a needle through all directions while sweeping arbitrarily little area. Given any ε > 0, there exists a valid rotation using area less than ε.
The trick: split the needle into many thin parallel slivers, translate them to partially overlap, rotate each a tiny amount, recombine. Repeat. Each iteration cuts area while preserving the full range of directions.
The limit of this process — infinitely many splits — produces a set of measure zero. A set that contains a unit segment in every direction, yet has no area at all. This is a Besicovitch set.
If Besicovitch sets have zero area, what dimension are they? In 2D, mathematicians proved they must have Hausdorff dimension 2 — full-dimensional despite having zero measure. Like a fractal dust that's everywhere-dense in direction space.
The Kakeya conjecture generalizes this: in n dimensions, a set containing a unit segment in every direction must have dimension n. The 2D case was settled. Dimensions 4 and above remained open. And dimension 3 — our physical space — was the prize.
In February 2025, Hong Wang and Joshua Zahl proved the 3D Kakeya conjecture in a 127-page paper. "A once-in-a-century result," said Nets Katz.
Their key insight: if a counterexample existed (a 3D Kakeya set with dimension less than 3), it would have to be grainy — containing tiny 3D regions where many tubes cluster densely. They showed each level of graininess forces the next, creating an inescapable logical chain that climbs to full dimension 3.
The result connects to harmonic analysis, number theory, and even wave propagation. Understanding how lines intersect in space turns out to constrain how waves can focus — which is why Fefferman connected Kakeya sets to Fourier analysis in 1971.
A needle is a line segment. A direction is a point on a sphere. The Kakeya problem asks: how efficiently can you pack line segments to cover all of direction-space? The answer, now proven, is: not very. In 3D, you cannot avoid filling a full 3D volume.
There is something beautiful about the asymmetry between 2D and 3D. In the plane, you can cheat — Besicovitch's construction collapses area to nothing. In space, geometry resists. The extra dimension creates enough room for tubes to be forced apart, and the grainy structure Wang and Zahl identified is the mechanism of that resistance.
Lines in space are more rigid than lines in a plane. That is the theorem, expressed geometrically.