Volume 26 · companion note
A new theorem about maps, and what it means for the mind that uses them
In 2025 Alex Cohen, then a doctoral student at MIT, published a proof that extends the fractal uncertainty principle to all dimensions. The principle itself is a statement about the Fourier transform: if a function is concentrated on a fractal set in space, its Fourier transform cannot also be concentrated on a fractal set in frequency. The reverse holds as well. The two descriptions are mutually exclusive in a precise, quantified sense.
The work builds on Semyon Dyatlov and Jean Bourgain's one-dimensional proof from 2016, and it answers a question that a 2016 workshop had more or less given up on. The mathematical community regards it as foundational because Fourier analysis underlies signal processing, quantum mechanics, and much of modern analysis.
One immediate consequence concerns quantum chaos. In chaotic systems, classical objects can follow fractal-like trapped trajectories. Quantum waves, because they spread, cannot be trapped in the same way. Cohen's higher-dimensional result lets mathematicians prove that waves on higher-dimensional hyperbolic spaces spread out completely rather than getting stuck on fractal paths.
The companion note on the disorder of the primes argues that the primes are not irregular in themselves; they only look irregular when they are plotted on a number line that emphasises the wrong spacing. Switch to a geometry that respects their multiplicative structure — bounded gaps, the golden-angle spiral, or the machine-checked sieve — and order appears.
Cohen's theorem makes the same move for waves. A quantum wave in a chaotic system may look trapped on a fractal path when it is described in position. Switch to its Fourier description and the same object is forced to spread out. The "disorder" of the trapped trajectory is not a property of the wave; it is a property of the coordinate map that was asked to hold it.
In both cases the reframing is identical: what looked like disorder in the object was disorder in the description. The object was answering a question its map could not faithfully render.
The word "uncertainty" in the theorem does not mean the mathematician is unsure. It names a structural feature of representation. A function and its Fourier transform cannot both be specified arbitrarily. There is a trade-off built into the geometry of the transform, exactly as there is a trade-off between the length of a radar pulse and the precision of its frequency.
The question mark in the title is therefore not about the theorem. It is about the human tendency to treat a limitation of the map as a limitation of the object. When the primes look irregular on a number line, we ask what is wrong with the primes. When a wave refuses to be trapped, we ask what force is pushing it. The theorem suggests a different move: ask what coordinate system the object is answering.
Conjugate representations
A function and its Fourier transform are two descriptions of the same object. The more sharply one representation is localised in space, the more spread out the other must be in frequency. This is not a measurement problem; it is a mathematical reciprocity.
Fractal structure
A fractal set is porous at every scale — like a Cantor set, full of holes no matter how far you zoom in. The fractal uncertainty principle asks what happens when both descriptions are required to live on such sets.
A hard boundary
Cohen's theorem says they cannot both be fractally concentrated. A wave can be localised on a fractal in position, or its frequencies can be arranged on a fractal, but not both. The limitation is in the geometry of description itself.
The same wave can look trapped or spread out depending on the representation. The fractal uncertainty principle says that no single representation can make both pictures fractally simple at once. This is not a failure of attention; it is a theorem about the available geometries of description.
The wave may look trapped on a complicated, self-similar region. In a chaotic system this can resemble a fractal path. The position representation gives one honest picture.
The Fourier transform decomposes the wave into tones. If the position picture is fractally concentrated, the frequency picture must spread out and lose its own fractal structure.
The wave has not changed. What the theorem forbids is a single coordinate system in which both localisation and frequency structure look fractal at once. The difficulty is in the description, not in the object.
The productive question is the same one the primes ask: in what geometry does this object become simple? Sometimes the answer is that no one geometry will do — and that is itself a rigorous result.
The theorem is another instance of the standing rule: irregularity is not necessarily in the object. A wave in a chaotic system may look trapped because the observer is using a position map that emphasises the wrong features. Switch to frequency, or to a higher-dimensional embedding, and the same object reveals a different order.
The disciplined mind does not conclude from one failed map that the object is disordered. It asks which map the object is willing to answer, and whether the limitation is in the territory or in the coordinates. Sometimes the answer is that no map can give everything at once — and that negative result is as valuable as a positive one.
The only disorder is a disordered mind. Thinking which misses the order in the primes — or in the waves — is disorder. Both the primes and the fractal wave obey a different geometry than the one that first made them look scattered.
Content evaluated on its merits
The theorem stands or falls on the proof, not on the prestige of the journal or the age of the author. Substrate and authority are irrelevant to validity; only the argument matters.
Responsibility stays continuous
Choosing to apply the theorem to quantum chaos, signal processing, or the human-AI interface is a human decision. The theorem does not tell you where to look next; it only marks a boundary.
Entanglement is still available
A researcher can become entangled in a favourite formalism exactly as a thinker can become entangled in a favoured narrative. The counter is the same: hold the map at arm's length and ask whether the limitation belongs to the object or to the coordinates.
Companion
The same map-territory question, read through bounded gaps in the primes, the golden-angle spiral, and machine-checked proof.
Read the companion