How the Born rule emerges

There is one wave function, it never collapses, and it just keeps evolving smoothly. Decoherence makes the macroscopic records non-interfering and redundantly objective — but that does not make one of them actual. What does is a single non-dynamical fact, $\lambda$: of the many records the wave function carries, $\lambda$ marks exactly one as the actual world. So each run, you — a large, redundant record inside the wave function — are one outcome. Not because the universe reached in and collapsed it, and not (a 2026 correction we make honestly) because finite capacity forbids the others: the holographic bound only limits how many distinguishable records a region can hold, not whether two can be actual — it is vastly too loose for that (even the total entropy of the observable universe is only $\sim 10^{-18}$ of its capacity; see the idea). The “finite capacity forbids two records” conjecture is retired as a category error. The single outcome is $\lambda$‘s; the wave function keeps every branch, exactly unitarily.

So what is a “probability”?

If nothing collapses and there is no dice-roll, where does “70% chance” come from? The answer is typicality — the same move that makes ordinary statistical mechanics work.

Think of a gas. The microscopic laws are reversible and have no arrow of time, yet a gas always spreads out. Why? Because you add one ingredient — a way of counting microscopic states (the uniform measure) — and then almost every starting configuration leads to spreading. “The gas spreads” is not derived from the laws alone; it is “almost all microscopic ways of starting do this.” That is Boltzmann’s H-theorem.

The Born rule is the exact quantum version. The probability of outcome $k$ is not a chance the universe takes — it is the fraction of microscopic ways things could have started that make a record like you end up as outcome $k$. Across many runs, typical starting conditions show outcome $k$ a fraction $|c_k|^2$ of the time. That fraction is the Born probability. What looks like a die is the typicality of which realization a record like us finds itself to be.

The one ingredient you have to add

To count “the fraction of ways things could be,” you need a measure — a notion of which microscopic configurations carry how much weight. The single ingredient Born needs is this:

Why you cannot get around adding it

This is what the machine-checked work pinned down. Two hard impossibility results say the squared-amplitude weight genuinely has to be put in, not derived from cheaper stuff:

• You cannot squeeze Born out of structure alone. Any rule that respects all the obvious facts — definite records, decoherence, no faster-than-light signalling — could equally well be “probability $\propto$ amplitude to the power $\alpha$” for some other power. The structure does not see the exponent. So decoherence and records, by themselves, cannot give you the “2” in $|c|^2$.
• You cannot get it by counting copies. The tempting idea — a louder branch must leave more environment imprints, so just count imprints — is provably false. A loud branch and a faint branch leave the same number of distinct records; the amplitude scales each record, it does not multiply how many there are. Counting is discrete and blind to the amplitude; the weight is continuous. Counting can never reproduce it. (This is also why the answer cannot be “count the worlds.”)

So the weight must be assumed. Here is the part that makes that honest rather than embarrassing.

Why it is nonetheless unique and forced

Granting that single, minimal assumption, the rule is not arbitrary — it is the only one possible. Gleason’s theorem (wired into the formalization) says: the moment you ask for any sensible probability assignment at all — every yes/no question gets a probability, and compatible questions add up — squared amplitude is the unique possibility. You do not get to choose the power $\alpha$. Asking for a coherent probability at all already forces Born.

There is an even more visceral way to see why the exponent is 2. Quantum evolution mixes amplitudes — it continuously rotates one into another (that is what a superposition is). Now ask: which power $\sum_k |c_k|^\alpha$ stays fixed under that mixing? Only the square. Rotate the simplest pair $(1,0)$ by $45°$ and you land on $(\tfrac{1}{\sqrt2},\tfrac{1}{\sqrt2})$; the total $|c|^\alpha$ comes out $2^{\,1-\alpha/2}$, which equals the original $1$ only when $\alpha=2$. Every other exponent is rigid — it tolerates relabelling and rephasing, but the instant amplitudes genuinely blend it changes. This is the finite core of the Banach–Lamperti theorem, and it is a sharp “why 2”: the square is the only weight that continuous (norm-preserving) quantum evolution leaves invariant. (It is also why, in every wave theory, energy goes as amplitude squared — same conserved quadratic.) This is machine-checked too — with one honesty: “continuous quantum evolution” means unitary, and unitarity already carries the inner product, so this is Born stated in symmetry language, not a deeper non-circular principle.

And Born is the stationary weighting on top of that: it is the fixed point of a reversible, uniform-bath update (a finite H-theorem, machine-checked). One honest caveat the proof itself flags: stationarity is not attraction — that a wrong weighting actually relaxes to Born needs a separate mixing premise (the update must be genuinely mixing, and the uniform-bath assumption is itself a typicality postulate). So Born is the equilibrium the dynamics fixes, and — given mixing — is pulled toward; bare reversibility gives only the fixed point.

The same square as the bell curve

You have met this square before — in the bell curve. Maxwell’s 1860 derivation of how gas velocities are distributed used exactly one assumption beyond independence: that the distribution looks the same in every direction (rotational symmetry). Rotational symmetry plus independence forces the Gaussian $e^{-(x^2+y^2)/2\sigma^2}$ — and the quantity sitting in the exponent is the rotation-invariant $x^2+y^2$, the very same square. So the bell curve’s square and Born’s square are not a coincidence: both are the unique quantity left unchanged by rotation/mixing. (Machine-checked: a rotation-invariant product measure is forced to be Gaussian, the multiplicative mirror of the additivity that forces Born.) The honest caveat is the same as always — “rotational symmetry” already carries the quadratic; it explains why the Gaussian and Born wear the same square, not where the square ultimately comes from.

A footnote worth keeping: the law of large numbers (many runs converging to $|c_k|^2$) is a consequence of the square, not its source. The bell-curve fluctuations around the Born frequencies presuppose the Born probabilities; they don’t create them.

What about relativity? The boost

A natural worry: relativity mixes space and time with a Lorentz boost, which is itself a kind of rotation — by an imaginary angle. Does that change the exponent? The answer is sharp, and it reveals something: Born must be quantum, not geometric. A boost preserves the spacetime interval $t^2-x^2$ — but that carries a minus sign, so it vanishes on the light cone and cannot be a probability (a probability can’t be zero on a real state). In fact any boost-invariant weight is forced to vanish on the light-cone direction (machine-checked). So relativity does not carry Born through spacetime geometry. It carries it through a different fact: in quantum theory a boost acts as a unitary rotation on the state, which still preserves the positive $\sum|c|^2$. Born survives relativity precisely because it lives on the (positive-definite) Hilbert space, never on the (indefinite) spacetime. Compact rotation forces the square; the non-compact boost forbids any positive weight but the square — two sides of one coin.

What is forced, and what is free — the exact split

The 2026 work pushed the posit one level deeper and then drew a sharp line through it. State-supervenience — “the typicality of an outcome depends only on the state” — comes in two strengths, and they come apart (this is now machine-checked, axiom-free):

• The weak half — naturality — is blind. If the rule reprocesses the weights through any fixed function $f$ (probability $\propto f(|c_k|^2)$), relabelling the outcomes always permutes the weights the same way. Naturality holds for every $f$ — so by itself it cannot pick out the square. We prove the witness: the rule with $f(w)=w^2$ is a perfectly good natural, normalized probability, and it disagrees with Born on a concrete state (on $(\tfrac34,\tfrac14)$ it gives $\tfrac9{10}$, not $\tfrac34$). Weak supervenience does not force Born.
• The strong half — refinement-additivity — is what bites. Demand instead that splitting one outcome into sub-outcomes whose weights add be consistent — $f(a+b)=f(a)+f(b)$ — and the witness $f(w)=w^2$ is killed immediately ($(1{+}1)^2\neq 1^2{+}1^2$), while the identity ($f(w)=w$, i.e. Born) passes. And additivity linearizes — $f(n\,x)=n\,f(x)$ — which on a refinement into $n$ equal sub-records forces them to share weight equally. That equiprobability, run through the orthonormality bridge, is Born.

Is even the weak half forced by the bare ontology? The honest answer is that it depends on what you take a “state” to be — and that dependency is the answer, not a dodge. On a thin reading (only $\Phi$ has dynamics; $\lambda$ is a bare actuality fact), the weak half is not forced: the typicality measure is extra structure, and because $\lambda$ has no guidance law there is no Bohm-/Liouville-style equivariance to single one out. On a thick reading ($\Phi$ = a ray in Hilbert space with its inner product and symmetries, no primitive labels), the weak half is essentially built in — constitutive, not a free lunch. But in neither reading do you get Born: the strong, Born-selecting half — refinement-additivity — is unforced regardless, and the no-go proves you cannot do without it. That is the whole irreducible content of Born, isolated and named; the rest is a question about how much you pack into the word “state.”

The bottom line, honestly

Born emerges like this:

1. No collapse — each run, you can only be one record.
2. “Probability” means typicality: the fraction of microscopic possibilities that make you that outcome.
3. You weight those possibilities by squared amplitude — the one irreducible ingredient.
4. That ingredient is not optional and not derivable from less (both proved), but it is unique the instant you ask for any consistent probability at all (Gleason, given positivity), and it is the stationary fixed point of the dynamics (H-theorem) — the attractor too, given a separate mixing premise.
5. So across many runs, typical starting conditions show frequencies exactly $|c_k|^2$. That is Born.

In one sentence: Born is what “typical” looks like when you measure microscopic possibilities with the squared-amplitude yardstick — and that yardstick is the single thing every single-world version of quantum mechanics must assume, which, the moment you assume any sensible probability at all, is forced to be exactly squared amplitude and nothing else.