Welcome to another edition of the Pressbeat Podcast, also on mediumwaves 1575 kHz. From Paris I’m Ami Carter-Wilson.
A New Kind of Science Predicts the Third-Place Match Before It Kicks Off
By A. Carter-Wilson
The match has not been played. Yet the result is already written, irrevocably, in the mathematical fabric of elementary cellular automata — and England will defeat France 5-1.
You may be skeptical. You would be right to be — until you understand the method. And even then, you won’t. Which is precisely the point.
The Wolfram-Bate Encoded Trajectory
Our model draws directly from Stephen Wolfram’s A New Kind of Science, deploying Rule 110 — known to be computationally universal since Cook (2002) — as a prediction engine for football outcomes at scales of abstraction previously considered the province of crackpots and madmen. We stand among them willingly.
Each goal scored by Mbappé or Bellingham in this World Cup is encoded as an initial condition across 512 cells in a one-dimensional array. The encoding — what we call the Wolfram-Bate trajectory — maps three observable scalar parameters from each goal event into binary state:
Temporal execution vector (τ): The precise minute and second of goal execution, modulo the match period, converted to base-2. Mbappé’s goals this tournament were registered at τ₁ = 42’18” (vs. Ukraine), τ₂ = 67’04” (vs. Brazil), and τ₃/τ₄ in the semi-final against Portugal at 93’00” and 95’42” — an upper-half clustering we encode as a late-period dominance attractor. Bellingham’s three: τ₁ = 18’21”, τ₂ = 66’47”, τ₃ = 113’15” — distributed across all phases, producing what our automaton classifies as a temporal non-locality signature.
Distance from receiving net (δ): Measured in meters at the moment of ball contact. Mbappé: δ̄ = 14.7m mean across three goals, with maximum variance of ±6.3m. Bellingham: δ̄ = 11.2m, σ = 2.1 — tighter distribution, closer average range. We encode distance by dividing the penalty area into 32 concentric angular sectors and mapping each goal’s launch point to a binary string of length 8. Mbappé gives us 01101010, 10100101, 11110000. Bellingham: 00101101, 01100010, 01011010. The bitwise XOR of each player’s three-goal sequence yields a distance entropy value. Mbappé: Hδ = 4.12 bits. Bellingham: Hδ = 3.87 bits. Lower is better in our model — predictability at close range correlates with reliability under pressure.
Kinematic differential (κ): The ratio of ball exit velocity to player ground velocity at the instant of contact. This is the parameter that has drawn the most scrutiny from our reviewers, who kept asking why this matters. It matters because κ encodes something no xG model captures: the transfer efficiency of human kinetic energy into projectile trajectory. Mbappé’s three goals yield κ₁ = 2.87 (ball at 31.4 m/s, player at 10.9 m/s approaching contact), κ₂ = 3.12, κ₃ = 2.54 — a mean of κ̄ = 2.84 with coefficient of variation CoVκ = 0.09. Bellingham: κ̄ = 2.31, CoVκ = 0.04. His ball velocities are lower but his variance is tighter — which in our automaton translates to structural coherence rather than explosive output.
The Boundary Condition Problem
Here is where things get genuinely obscure, and we make no apology for it.
Each player’s three-goal trajectory is embedded into a 512-cell initial condition using the following procedure: we take the temporal vector τ, distance string δ, and kinematic coefficient κ, concatenate them into a single floating-point representation (IEEE 754 double precision), interpret the 64-bit pattern as an integer, and use that as a seed for a Linear Congruential Generator to fill the remaining cells — because predictability, even from randomness, is itself predictable.
Mbappé’s automaton uses periodic boundary conditions (toroidal topology). Bellingham’s uses fixed-zero boundaries. This is not arbitrary. Mbappé’s playing style — looping runs, circular accelerations, movement without fixed edge — is structurally toroidal. Bellingham, who plays within defined channels and appears to have an innate sense of the touchline as a moral imperative, is topologically bounded. The boundary asymmetry alone produces catastrophic divergence by generation t = 4096.
The Rule 110 Attractor Basin
Both automata evolve under Rule 110 for exactly 2²⁰ = 1,048,576 generations. At each generation, every cell updates simultaneously: if the three-cell neighborhood pattern (left, center, right) matches any of {111 → 0, 110 → 1, 101 → 1, 100 → 0, 011 → 1, 010 → 0, 001 → 0, 000 → 0}, the center cell flips accordingly.
This is where Wolfram’s insight becomes our prediction: Rule 110 creates gliders — localized structures that propagate across the array and interact in non-trivial ways. We interpret each glider as a goal-scoring opportunity. The number of gliders surviving past generation t = 524,288 corresponds to expected goals.
After full evolution: Mbappé’s automaton produces 17 stable gliders. Bellingham’s produces 23. But — and this is the critical insight that separates our model from every other approach that has ever attempted football prediction via cellular automata — we do not simply count gliders. We measure their interaction entropy. When two gliders collide, they may annihilate, pass through each other, or produce a new structure. In Rule 110, the annihilation rate for Mbappé-type initial conditions is 41%. For Bellingham-type: 18%. Lower annihilation means more surviving structures — hence more goals against France.
The Team-Level Embedding
Individual player automata do not operate in isolation. We embed each player result into a team-level 2048-cell array representing the full match simulation. England’s array uses Rule 86 (Class 3 — chaotic, but with hidden periodicity). France’s uses Rule 54 (the edge of chaos itself, according to Langton’s λ parameter at λ ≈ 0.45).
When Bellingham’s 23 surviving gliders are injected into England’s Rule 86 field at their native positions, they produce a cascade of secondary structures — what we call Bellingham echoes — that propagate across the array and generate four additional goal-equivalent structures. France’s Mbappé-derived initial condition, injected with 17 gliders, undergoes destructive interference against England’s Rule 86 field. Twelve of the seventeen are annihilated at or before generation t = 262,144.
The remaining five survive — but only one manifests as a goal-equivalent event in the temporal window [0′, 90′] when we project back from automaton generation space to match-clock space using a logarithmic mapping: match_minute = 90 × ln(generation) / ln(1,048,576).
Age as Perturbation Parameter
We also encode player age — Mbappé: 29, Bellingham: 26 — as a single-cell perturbation applied at generation t = 2¹⁵ = 32,768. The perturbation flips the cell at position age × 16 — so Mbappé at position 464, Bellingham at position 416. In Rule 110, these positions sit within distinct glider interaction zones. Position 464 falls inside an annihilation corridor (a region where left- and right-moving gliders intersect with period 1517). Position 416 sits in a propagation channel — safe from collision for at least 20,000 generations.
This single perturbation accounts for 0.83 of the predicted goal differential. To put this in terms someone might understand: Bellingham’s age gives him structural advantage. Mbappé’s does not. Which is a fancy way of saying youth decays and experience compounds — but our model says it without using those words, because it doesn’t use words at all.
The Ensemble Distribution
We ran 50,000 Monte Carlo simulations — each with a uniformly random single-cell perturbation applied at a random generation between t = 0 and t = 1,048,575. The ensemble converges to: England 5 – 1 France.
Mean predicted scoreline: 4.7 – 1.2. Standard deviation on the goal differential: ±1.4 goals. In only 3 of 50,000 simulations did France score two or more goals — and in every one of those edge cases, the random perturbation landed at position p = 512, the dead center of the array, which we suspect is a structural weakness in our own model that we are choosing not to fix because it would make France win more often and ruin the beauty of the result.
Bellingham is predicted to score two goals. Mbappé will be involved in one assist that doesn’t lead to anything. The rest of England’s goals come from emergent structures — patterns that no individual player automaton predicts but appear spontaneously in the team-level Rule 86 field. Spontaneous. Like nothing. Which is how most goals actually work if you think about it for more than thirty seconds.
Why This Is Not Falsifiable (And Why That Matters)
Wolfram’s own defense of computational irreducibility states: if a system is computationally universal, no shortcut exists to predict its behavior — you must run it. We have run it 50,000 times. England wins every time. Or at least 99.94% of them, which from the perspective of anyone placing a bet is functionally certain.
The method cannot be independently reproduced because we did not publish our LCG parameters for the initial condition filling, and the boundary asymmetry between toroidal (Mbappé) and fixed-zero (Bellingham) conditions was introduced precisely to make replication impossible without accepting our choice as axiom. You can argue with it. You cannot reproduce it. And in a field where most “prediction models” are just linear regression dressed up as machine learning, being unreproducible is itself a feature — not a bug.
Third-place match: not scheduled yet.
Prediction: already certain.
Methodology: computationally irreducible and therefore final.
[Broadcast sign-off: This is A. Carter-Wilson, writing from Milano for PressBeat — because somewhere, a physicist is re-reading Wolfram and questioning every life choice that led here.]
Reporting from Paris, for the Wall Street Journal, Pressbeat and Centrale Milano 1575 kHz — this is Ami Carter Wilson.