Mathematics

The Digits Aren't Random

Published 2026-03-23

Reading 8 min read

As Pi Day has now come gone and Tau Day approahced , people will do what they always do: post long strings of digits, argue about notation, and talk as though the main mystery of π is that its decimal expansion "looks random."

That framing misses the point.

Before I go further: I use τ instead of π in this work, but this article is not about reopening the π-vs-τ debate. They are different normalizations of the same underlying circular constant.

If you want my reasoning on that choice, I've written it here:

Why I use τ instead of π: timothysolomon.com/papers/tau-pi-choice-of-generator

And the short LinkedIn version is here:

Hot take: the π vs τ debate was never about the symbol linkedin.com/posts/timothysolomon1_hot-take-the-π-vs-τ-debate-was-never-about-activity-7435716257557086209-m0ej

What this article is about is something else:

The digits of π or τ are not random in the deepest relevant sense. Not because they fail a frequency test. Not because I've "solved normality." But because the object generating them is structurally overconstrained.

That is the point of the paper I'm currently working on.

The usual question is the wrong one

People usually ask:

"Are the digits of π random?"

That question normally means one of two things.

First: algorithmic randomness. Are the digits generated by a process so irreducible that no shorter description exists?

For π or τ, the answer is clearly no. They are computable. There are deterministic algorithms that generate their digits.

Second: statistical randomness. Do all digit strings occur with the expected limiting frequencies? In other words, are these numbers normal?

That remains unproved.

But neither of those gets to the real issue.

The digits are not the object. They are a representation of the object.

A decimal expansion is just what a far deeper structure looks like when projected into base 10.

So the more interesting question is not:

"Do these digits look random?"

It is:

"What kind of object would force these digits to be exactly what they are?"

That is a very different question. And it leads into very different mathematics.

My claim in one line

The core claim of the paper is this:

τ is not merely a number with a decimal expansion. It is a scaling law.

Once you treat τ that way, the digits stop looking like the main event and start looking like the trace left behind by a very rigid structural mechanism.

The paper develops this through what I call scale-relative primality.

What scale-relative primality means

The simplest way to say it is this.

In ordinary arithmetic, we ask whether an integer like 2, 3, 5, or 7 is prime.

In the framework I'm developing, I look at expressions of the form:

±nτᵏ

where:

n is the integer content
τᵏ is the scale
the sign is just the sign

So:

3/τ
3
3τ
3τ¹⁷

all share the same integer content: 3

In this framework, they belong to the same τ-prime class.

That does not mean I am claiming "3/τ is prime in the ordinary integer sense." It means something more precise:

Primality is carried by the integer content. Scale is carried by the power of τ.

So the object has two layers:

content
scale

and the whole point is that these two do not collapse into one another.

That separation is the key.

Why transcendence matters

This is where the deep math starts.

The framework only works cleanly if τ is transcendental.

Why?

Because transcendence prevents scale from collapsing into content.

If we had used an algebraic scaling factor like √2, then powers of the scale can fold back into ordinary integers:

(√2)² = 2

That means the "scale part" and the "integer part" can become entangled.

The bookkeeping breaks.

You lose the clean distinction between:

what the number is counting
and the scale at which it is being viewed

But with τ, that collapse does not happen.

Transcendence protects the separation.

That is not decoration. That is the mechanism.

It means the decomposition into integer content and τ-scale remains structurally stable.

And once that is true, something powerful follows:

Integer multiplicative structure can be transported into a τ-scaled domain without algebraic interference.

That includes:

prime content
factorization
divisibility
multiplicative ordering

Not additively. Not in every number-theoretic sense. But multiplicatively, the structure carries across faithfully.

This is the real shift

Most popular treatments of π treat transcendence as a negative fact.

Something like:

"π is weird because it never ends, never repeats, and does not satisfy any polynomial equation with integer coefficients."

That is true, but it is shallow.

My claim is that transcendence here should be understood positively, not merely negatively.

Transcendence is what allows a perfect separation between content and scale.

In that sense, transcendence is not just an absence of algebraicity.

It is a structural condition that permits a faithful embedding.

So instead of saying:

"τ is transcendental because it escapes algebra"

I am effectively saying:

"τ's transcendence is what allows arithmetic structure to survive scaling without collapse."

That is a much stronger way of thinking about what this constant is.

So what does this have to do with the digits?

Everything.

If τ is carrying all of that structure at once, then its decimal expansion is not some loose spray of arbitrary symbols.

It is the unique decimal trace of a deeply constrained object.

The digits are not free to be otherwise.

That is the crucial point.

I am not claiming:

that I have disproved normality
that digit frequencies are fake
that one can just spot superficial patterns in the decimals and declare victory

That kind of thing is mathematical cosplay.

The claim is more serious than that.

The digits are structurally nonrandom because the object generating them is structurally determined.

A random-looking surface does not imply a random-generating object.

And in this case, I think we have been staring at the surface for too long.

Three senses of "random"

To make the distinction sharper, it helps to separate three senses of randomness.

1. Algorithmic randomness

A sequence is random if it is incompressible.

π and τ fail that test immediately. They are computable.

2. Statistical randomness

A sequence is random if blocks occur with the expected limiting frequencies.

For π and τ, this is tied to normality, which remains unproved.

3. Structural randomness

A sequence is random if it is not carrying a coherent, necessity-laden structural burden.

This is the sense I care about.

τ is not structurally random because it is doing too much.

It is simultaneously satisfying:

a geometric constraint (circumference-to-radius structure)
an algebraic constraint (transcendence)
and an arithmetic constraint (faithful multiplicative transport of integer structure into a scaled domain)

That is not randomness.

That is extreme structural obligation.

A simpler way to say the whole paper

If I had to compress the thesis into one paragraph, it would be this:

We have been treating the decimal digits of π or τ as though they are the mystery. They are not. The mystery is the object whose decimal trace they are. In the framework I'm developing, τ acts as a transcendental scaling law that preserves integer multiplicative structure without allowing scale and content to collapse into one another. That makes the decimal expansion of τ not a random stream, but the visible projection of a maximally constrained geometric-arithmetic object.

That is the heart of it.

What I am not claiming

This matters, because otherwise people hear one thing and imagine five others.

I am not claiming:

that τ-primes are ordinary primes in ℤ
that this resolves the normality problem
that every feature of integer arithmetic transfers unchanged
that additive number theory is preserved under irrational scaling
that one can infer all of this from digit-chasing alone

The framework is primarily multiplicative.

That distinction matters.

But within that domain, the structure is far more rigid than the usual "random-looking digits" language suggests.

Why this matters beyond Pi Day

Pi Day tends to reduce the constant to trivia:

memorization contests
novelty digit prints
pie jokes
vague statements about infinity

That is harmless enough.

But mathematically, it conditions people to look at the wrong layer.

The digits are the shadow, not the machine.

The real issue is not whether the decimal string feels chaotic.

The real issue is whether the object behind it is free.

I am arguing that it is not.

And if the generating object is not free, then the language of randomness is at best incomplete and at worst misleading.

The broader point

What interests me here is not just τ itself.

It is the larger philosophical and mathematical question underneath it:

When a mathematical object appears patternless at one layer, is that because it lacks structure — or because the structure lives at a level deeper than the representation we are inspecting?

For τ, I think it is clearly the latter.

And that means the phrase "the digits are random" tells us more about the poverty of the framing than about the constant.

Closing

So as Pi Day approaches, here is the short version of my position:

The digits of π or τ are not the thing to explain. They are what a deeper thing looks like when flattened into base 10.

And in the paper I'm working on, my argument is that τ should be understood as a transcendental scaling law that preserves multiplicative structure while preventing algebraic collapse between content and scale.

If that is right, then the digits are not random in the deepest relevant sense.

They are necessary.