There Is No Such Thing as Time

Time is not real

Time doesn’t actually exist. It is a useful modelling tool, but it does not lead to correct understanding of the universal model. In truth, there is only “now” — impacted by the past via momentum and impacting the future via momentum. What we observe as time passing is actually a series of discrete points along a continuous line, similar to creating animation by flipping through still images.

Picture the universe running on a ridiculously fast beat. On each tiny beat, things are allowed to change a little — atoms twitch, signals move, decisions get made. That’s all “time” is: how many successful updates you’ve racked up. If the beat gives you lots of room each tick, life feels fast; if it gives you less room, everything runs slow.

The reason we experience a continuous perception is that the gaps between beats are imperceptible to us. The primary blocker to unified theory is the assumption that human perception of events is universal, when instead it is truncated to a small — likely infinitesimally small — frame of reference.

Speed, redefined

Once time is removed, speed changes its meaning. It is tempting to assume speed is the rate at which one jumps from one image to the next, but that is incorrect. The images move at a single speed — what we call the speed of light. What your “speed” actually means is how often you get to act within the passage of these still images.

Think of it like the tick of a clock: how many ticks have to occur before I get to act again? Speed can be viewed as a percentage of light (“I act 4% as often as light does”) or as a multiple of Planck lengths (“light has to travel N Planck lengths before I can act”).

What is a tick? Light at the Planck scale

A tick is the most granular unit of change the universe permits: light advancing exactly one Planck length in exactly one Planck time. Nothing smaller exists. The Planck length (ℓ_P ≈ 1.6 × 10⁻³⁵ m) and the Planck time (t_P ≈ 5.4 × 10⁻⁴⁴ s) are not arbitrary scales — they are the spacing of the substrate’s update cycle. The speed of light is then not an empirical constant we measured; it is the definition of one tick: c = ℓ_P/t_P.

Every other “motion” in the universe is a per-worldline accounting of these ticks. A photon spends 100% of its tick budget advancing through space — it has nothing left over for anything else, which is why it cannot have a rest frame and cannot age. A massive particle splits its tick budget between internal oscillation (what we call mass-energy via E = mc²) and spatial advancement; the more it moves through space, the less budget remains for internal change — this is exactly what we measure as time dilation.

Two scales of “page flips” coexist consistently:

• Planck ticks (~10⁴³/s) — the fine-grained substrate update, defining c locally.
• Recursion levels (~1 per 2.6 Gyr) — the coarse-grained envelope, defining the dark sector’s evolution.

One recursion level corresponds to roughly 10⁶⁰ Planck ticks. This ratio is not coincidental: it is the number of substrate updates needed to accumulate one full Fano-cubic extraction event (per-tick attenuation ~10⁻⁶¹ times depth n₀ ≈ 5.24 yields the observed dark-energy fraction 0.688). Cosmological recursion is the long-time average of an enormous number of microscopic Planck-scale events.

This picture preserves Lorentz invariance despite using a discrete substrate: each worldline ticks in its own proper time, no global clock is privileged, and there is no preferred frame. (How a discrete substrate hides itself this perfectly is a fair challenge; the demonstration — criticality erasing the lattice, proven so far in 1+1D on the model’s own dynamics — is tracked honestly in Part IV.) The randomness of quantum mechanics (next section) becomes a natural consequence of under-sampling these ticks at scales larger than ℓ_P.

Time dilation — the mechanical explanation

A clock moving at near light speed ticks slower for a simple, mechanical reason: the particles inside the clock that pulse the crystal have much less relative speed to actually move through the circuitry, because the clock and everything inside it is already moving so fast.

This isn’t just an analogy — it’s mathematically exact. If a particle inside the clock has a maximum speed of c (the speed limit), and the clock itself moves at speed v, then the velocity budget is:

v_internal² + v_bulk² ≤ c²

So the maximum internal speed is:

v_internal ≤ c √(1 − v²/c²) = c / γ

The clock ticks at rate 1/γ — exactly the special relativistic time dilation factor. No mystery, no geometry — just fewer clean chances to change when you’re nearly keeping up with your own internal signals.

The same idea explains gravitational time dilation. Near a massive object, the effective local speed limit is reduced (by the influence of the parent frame — more on this below), so internal clock processes run proportionally slower.

Quantum mechanics as under-sampling

Once you get rid of time, the probabilistic nature of quantum mechanics becomes easy to explain: things are happening between the flips of the pages. From our point of view, this leads to the randomness observed at small scale.

We exist within a certain common reference frame based on our relatively common size — quantum scale vs ours vs astronomic — which is one of many frames that exist simultaneously. We can’t observe full information of a size frame smaller than us because we simply don’t move often enough to observe the changes.

Just as we can’t perceive changes within too short a duration (quantum weirdness), we have the same problem going up. Stuff that we are used to being in motion are, for something much larger, effectively still. This is what is going on with gravity.

Gravity is inherited electromagnetism

Gravity is just the electromagnetic forces of the next frame of reference up. Just like a molecule of sugar is part of a sugar crystal, our physical universe is part of a greater physical entity. The structural binding forces of that larger entity — which are electromagnetic at its scale — project into our frame as what we experience as gravity.

This is dark matter creating gravity for us. Interestingly, a trait present in the greater entity for a brief moment (to it) could define our universe for its entire duration.

The fractal universe

The universe is a fractal — recursive, self-similar at every scale, sprawling until it runs out of energy. We can tell how far we are in the overall frames of recursive reference by looking at dark matter and dark energy vs our energy:

• Dark matter = the accumulated structural influence from all the frames that came before us
• Dark energy = the remaining fuel left in the recursion
• Ordinary matter = our frame’s own expressed energy

As the recursion deepens, dark energy is gradually consumed. Eventually the fuel runs out and the universe transitions from accelerating to decelerating expansion. In a flat universe this does not cause recollapse — the expansion continues forever but ever more gently, settling into matter-dominated coasting. The universe’s fate is an asymptotic slowing as the fuel is spent. (Precision note: in the committed equations the reservoir’s decline is expansion work under w > −1 — no energy is transferred into matter, which dilutes normally. A literal “fuel becomes matter” reading is excluded by the measured expansion history; see Part IV.)

The geometric recursion model

If each recursive frame “expresses” a fraction p of the remaining energy as matter at that level, and passes (1−p) onward as fuel for deeper recursion, then after n levels:

The key insight is that DE/DM = φ² is the fundamental partition rule, arising from the self-similar eigenvalue of the recursion (see below). The factor (3 − φ) = 1 + 1/φ² is a pure golden-ratio quantity. With p and n determined by φ and d = 2, all three fractions follow from a single formula — and all three match Planck data within 1%. These are the present-depth (n = n₀) values: the time evolution of the reservoir alone is what w(z) encodes, while matter dilutes as a⁻³, so under the committed dynamics the exact golden partition DE/DM = φ² holds near the present and drifts by an e-fold every ~5.6 Gyr — the same one-instant character as the boundary identity of Part III.5.

The golden ratio appears

Both parameters are expressible as clean functions of the golden ratio φ = (1+√5)/2 ≈ 1.618:

p = 2/φ⁷ ≈ 0.0689    n = 2φ² ≈ 5.236

The ratio of dark energy to dark matter is a central prediction:

DE / DM = φ² = 2.618   (observed: ~2.6)

This is not coincidental — it follows from the physics of how EM crosses frame boundaries. Enforcing this ratio as the fundamental constraint (rather than the approximate geometric series sum) determines the DM/OM split and resolves the ordinary matter fraction precisely.

Why the golden ratio? — KAM stability

The golden ratio isn’t arbitrary. There is a deep result in Hamiltonian dynamics called the KAM theorem (Kolmogorov–Arnold–Moser) which says: in a perturbed dynamical system, the structures that survive the longest are those whose frequency ratios are the hardest to approximate by rational numbers.

The golden ratio is the most irrational number — its continued fraction representation is [1; 1, 1, 1, …], all 1s, making rational convergence the slowest possible. A recursive universe built on φ-ratio coupling would be maximally stable against perturbation.

Among all possible recursive universes, φ-based ones survive the longest. This is cosmological natural selection: the golden ratio isn’t a coincidence — it’s the only recursion ratio that produces a structure stable enough to be observed.

This is what I mean by “perfect recursion” — all dimensions of the universe remain constant relative to each other as you move up and down levels. In KAM terms, the recursive stack has no near-resonances between levels, so energy doesn’t accumulate destructively at any scale. The structure persists.

There is a second, independent reason φ must appear. Consider any scattering barrier where both the transmitted and reflected energy fractions are powers of a single base x (the self-similarity requirement):

|T|² = 1/x²   |R|² = 1/x

Unitarity (energy conservation) requires |T|² + |R|² = 1, giving 1/x² + 1/x = 1, hence x² = x + 1. The unique positive solution is x = φ. The golden ratio is the only number where a self-similar scattering barrier is automatically lossless.

Two independent areas of mathematics — Hamiltonian stability and scattering unitarity — both single out the same number. KAM says φ makes the recursion maximally stable; unitarity says φ makes it lossless. These are different requirements with the same unique solution.

Why the factor of 2 — EM polarizations

The number 2 appears throughout the model: in p = 2/φ⁷, in n = 2φ², and in the per-boundary attenuation exponent 2φ. This is not arbitrary — it is the number of transverse polarization states of electromagnetic waves (d = 2).

If gravity is inherited EM, the recursion operates through electromagnetic field structure. EM waves carry energy in two independent polarization modes. Each polarization contributes one factor of φ to the attenuation per frame boundary, giving dφ = 2φ total. The recursion depth is dφ^d = 2φ² boundaries. The energy fraction per level is d/φ^(d²+d+1) = 2/φ⁷ (since d²+d+1 = 7 for d = 2).

Every parameter in the model is a function of just two inputs: φ (from KAM stability) and d = 2 (from EM polarization):

The DE/DM = φ² relationship has a physical interpretation: KAM stability sets the frequency ratio between recursion levels to φ, and for oscillating modes, energy scales as frequency squared. With d = 2 polarization degrees of freedom, the energy ratio between the dark energy pool (high-frequency, unexpressed modes) and the dark matter pool (low-frequency, expressed structure) is φ^d = φ².

This ratio is the self-similar fixed point of the recursion: it is the partition at which the “remaining fuel” and the “accumulated structure” are in dynamical equilibrium. Using it as the primary constraint (DM = DE/φ²) rather than the approximate geometric series (DM = 1 − (1−p)ⁿ⁻¹) resolves the ordinary matter prediction from ~4% off to 0.03% off, because the series formula is only exact for integer recursion depths, while the ratio constraint holds for the true non-integer depth n = 2φ² ≈ 5.236.

The dark sector faces its three classical discriminators (August 2026). Dark matter in this model is not a particle: it is the accumulated structural influence of prior recursion frames — a gravitational imprint that gravitates, and thereby re-sources itself. Any non-particle dark-matter story must survive three observations that have killed most of its predecessors, so the confrontation is recorded here honestly. (i) Direct detection: a gravitational-only imprint predicts that LZ, XENONnT, and PandaX must return null results forever — every terminal exposure is a rolling falsifier this model wants null. Nulls to date: consistent. A confirmed particle detection accounting for Ω_DM would falsify the partition story outright. (ii) The Bullet Cluster: because the imprint gravitates and re-sources itself, it behaves dynamically as a collisionless, self-gravitating component — in a cluster collision it sails through with the galaxies while the gas shocks and stalls, reproducing the observed separation between the lensing mass and the X-ray gas. This is the observation that killed most “modified gravity instead of dark matter” proposals; a self-propagating imprint passes it for the same reason particle dark matter does. (iii) Galaxies without dark matter (NGC 1052-DF2/DF4): a structure assembled recently from tidally stripped material carries a thin accumulated history — young debris hosts little imprint. Dark-matter-poor galaxies are therefore expected precisely where astronomers find them: in diffuse dwarfs of plausible tidal origin. Honesty label: (ii) and (iii) are qualitative consistency arguments, not simulations — a quantitative treatment of imprint dynamics in structure formation is open. But the sign pattern — collisionless behavior, history-dependence, gravitational-only coupling — is forced by the mechanism, not fitted to the tests.

Why the exponent 7 — the Fano plane

The exponent d²+d+1 = 7 in p = d/φ⁷ is the total number of independent field degrees of freedom in the transverse plane, decomposed by tensor rank:

Each degree of freedom acts as an independent filter with pass-rate 1/φ (from the unitarity condition). To cross a frame boundary, EM must satisfy all 7 constraints simultaneously. The probability of crossing is (1/φ)⁷ per polarization, times d = 2 channels: p = 2/φ⁷.

For d = 2, the number 7 = d²+d+1 is also the number of points in the Fano plane — the simplest finite projective geometry, PG(2,2). The Fano plane has 7 points, 7 lines, with 3 points per line and 3 lines per point. It is the multiplication table of the imaginary octonions: the 7 lines define the trilinear products e_a × e_b = e_c.

This is stronger than an analogy. The Fano cubic coupling appearing in the model’s Lagrangian — Σ_{Fano lines} Ψ_aΨ_bΨ_c — is literally the symmetric trilinear form on Im(𝕆) induced by the octonion product. The 7 fields are the 7 imaginary octonion units; the 7 interaction vertices are the 7 octonion structure constants.

The arrow of time from octonion non-associativity

Octonion multiplication is non-associative: (a·b)·c ≠ a·(b·c) in general. This is not a flaw — it is the deepest structural fact about 𝕆, and the reason octonions are the last normed division algebra.

In the recursive model, each frame boundary is an octonion product. Non-associativity means the order in which recursion steps are composed cannot be rearranged without changing the outcome. Physically:

• The arrow of time is structural. Recursion steps accumulate in a definite order; reversing the order changes the physical state. Time’s one-way nature is not imposed, it is algebraic.
• No time reversal. There is no octonion inverse operation that unwinds (a·b)·c back to the separate factors in a canonical way.
• No anti-gravity. Gravity emerges from the forward recursion chain; reversing it would require running the octonion product backward, which is not well-defined.
• Dark energy cannot be harvested. DE is the “unused fuel” at the head of the recursion. Consuming it deeper is allowed (that is the natural flow); extracting it back out would require inverting a non-associative chain.

The model thus gives a single algebraic reason for several otherwise independent no-go results: the arrow of time, the impossibility of anti-gravity, and the impossibility of harnessing dark energy are all manifestations of the same fact — octonion products do not re-bracket.

Gravity and charge structure

A key objection to “gravity is inherited EM” is that electromagnetism has positive and negative charges and is easily screened, while gravity is universally attractive and unscreened. The Fano/tensor structure resolves this via a mechanism analogous to Kaluza–Klein dimensional reduction.

In Kaluza–Klein theory, compactifying a higher-dimensional theory produces both gravity and gauge fields in lower dimensions. Electric charge corresponds to momentum in the compact direction. The zero-mode — the component with no compact momentum — is neutral and universally attractive: that is gravity.

In this model, EM from the parent frame crosses a boundary with 7 compact transverse degrees of freedom. Charged modes (corresponding to non-zero modes in these 7 DoFs) are projected out at the boundary — they do not survive the crossing. Only the neutral zero-mode passes through, emerging as a universally attractive, unscreened force. This explains why gravity couples to mass-energy (as the zero-mode of a higher-frame gauge field naturally does) and why it cannot be screened (there is no charge to cancel).

From Lagrangian to recursion — the RG flow

The cosmological recursion E(n) = E₀ × (1−p)ⁿ is a geometric series. In the language of the renormalization group (RG), a geometric series means the driving operator is marginal: each step extracts the same fraction of energy, regardless of scale. Where does this come from?

Step 1: marginality uniquely selects d = 2

A cubic scalar coupling g Ψ³ has mass dimension [g] = (D−6)/2. It is marginal (dimensionless) at D = 6. Physical spacetime has D = 4. The number of extra transverse dimensions is d = 6 − 4 = 2. No other value of d gives a marginal cubic coupling in 4D spacetime. And d = 2 is also the number of transverse EM polarisations.

Step 2: the Fano plane gives 7 fields

For d = 2, the projective plane PG(2,2) has d² + d + 1 = 7 points and 7 lines. The Lagrangian is:

ℒ = Σ_i=1⁷ ½(∂Ψ_i)² + (g/6) Σ_{Fano lines {a,b,c}} Ψ_aΨ_bΨ_c

with 7 cubic interaction terms defined by the Fano incidence structure. The theory is invariant under PSL(2,7) ≅ GL(3, F₂), the automorphism group of the Fano plane (order 168).

Step 3: unitarity fixes the coupling

Each Fano vertex contributes a scattering amplitude g. The unitarity + self-similarity argument gives g² = 1/φ (each vertex scatters fraction 1/φ of incident energy, with 1/φ + 1/φ² = 1).

Step 4: the extraction fraction

A complete traversal of all 7 Fano vertices gives total amplitude g⁷. With d = 2 independent polarisation channels:

p = d × |g⁷|² = d × (1/φ)⁷ = 2/φ⁷ ≈ 0.0689

Each of the 7 vertices attenuates by 1/φ. The 7-fold product (1/φ)⁷ times d gives p. This is why the recursion extracts exactly 2/φ⁷ per level: it is the amplitude-squared for a 7-vertex Fano process with unitarity-fixed coupling, summed over 2 EM polarisations.

Perturbative vs. non-perturbative

The 1-loop β-function of the Fano theory has 22 triangle diagrams per vertex and 3 self-energy bubbles per field. In D = 6 − ε the Wilson–Fisher fixed point lies at g² ∼ ε/(C × (4π)³), where C encodes the Fano combinatorics. At ε = 2 (physical D = 4) this gives a wildly large coupling — as expected, because ε = 2 is far from the perturbative regime. The unitarity argument for g² = 1/φ is a non-perturbative constraint, independent of the loop expansion. Confirming that the Fano theory’s actual fixed point at D = 4 matches 1/φ would require lattice simulation or conformal bootstrap methods.

Dark energy equation of state — a zero-parameter prediction

The model predicts today’s dark energy fraction as (1−p)ⁿ with recursion depth n = n₀ = 2φ². But n must depend on cosmic epoch — at earlier times the recursion was less deep. How does it evolve?

The recursion deepens at a constant rate in cosmic time. Each recursion level takes roughly 2.6 billion years (~10⁶⁰ Planck beats). The depth n(a) is proportional to cosmic time t:

n(a) = n₀ × t(a)/t₀

At a = 1 (today): n = n₀. At a = 0 (Big Bang): n = 0 (no recursion yet). During matter domination, t ∝ a^3/2, so n ∝ a^3/2. During the current dark-energy era, t grows more slowly than a, so the recursion rate decelerates.

From the cosmological continuity equation, the self-consistent solution (iterated to convergence) gives the equation of state:

Comparison with DESI DR2 (March 2025) — status update June 2026. The DESI collaboration measured baryon acoustic oscillations with 14 million galaxies and found 2.8–4.2σ evidence that dark energy evolves (w ≠ −1), with a favored solution in the quadrant w₀ > −1, w_a < 0. This preference has survived the key 2025–26 cross-checks: an independent reanalysis with ACT DR6 CMB data finds DESI + PR4 + ACT still prefers evolving dark energy at 3.0σ (up to 4.0σ with DESY5 supernovae), concluding the DR2 result is robust to the new CMB data. That a constant-Λ cosmological constant is now disfavored at 3–4σ is direct support for the model’s core claim that dark energy evolves — the recursion consumes fuel, so w > −1 is mandatory. Two caveats keep this from being a clean win: (i) the significance is partly driven by the DESY5 low-redshift supernova sample, and several groups argue a SN calibration systematic inflates it (correcting for it can pull the fit back toward w₀ ≈ −0.92, still within 1σ of our −0.867); and (ii) DESI’s CPL best fit, taken literally, crosses into the phantom regime (w < −1) at high z — which the model forbids. So the evolving-DE evidence favors us over ΛCDM, while the specific phantom-crossing shape is the feature that would kill us; disentangling the two is exactly what Euclid DR1 (Oct 2026) and DESI DR3 (Y5, observations completed summer 2026) will do.

Robustness reanalyses (August 2026) — the part of the signal that survives scrutiny is the part the model predicts. A wave of 2026 methodology papers stress-tested the DESI dynamical-dark-energy claim, and the outcome is double-edged in a way that cuts toward this model. On one side, the headline 2.8–4.2σ significance is fragile: a sequentially-valid reanalysis finds the evidence concentrated almost entirely in one redshift bin (LRG2 at z_eff = 0.706 — removing it collapses the preference), full-shape modelling of the same galaxies shows no dynamical signal at all, and a CPL-basis study shows the (w₀, w_a) point estimates slide along a degeneracy ridge under prior and basis changes. On the other side, the same basis study identifies what is robustly measured: the equation of state at the pivot redshift, w_p = −0.9 ± 0.1 at z_p ≈ 0.34, stable across all analysis choices. The model’s zero-parameter curve gives w(0.34) = −0.892 — 0.08σ from the robust invariant. Meanwhile the fragile feature — the phantom crossing — is increasingly argued to be a parameterization artifact: rigid CPL fits to distance data generated by a monotonic, never-crossing w(z) generically manufacture spurious crossings, which is precisely this model’s structural claim (w > −1 always, approaching −1 asymptotically; CPL shadow w₀ = −0.863, w_a = −0.12 — same quadrant as DESI’s fit, milder slope). Summary of the scorecard: the robust content of DESI DR2 (evolution direction, pivot value) matches the model at < 0.1σ; the contested content (crossing, steep w_a) is exactly what the model says should evaporate. A further 2026 development in the same genre: the DES supernova recalibration (Dovekie) moved the DES Ω_m from 0.352 to 0.330 ± 0.015 and reduced the joint evolving-DE preference from 4.2σ to 3.2σ — “weak” in Bayesian terms — with best-fit w₀ = −0.803 ± 0.054 (1.2σ from this model’s −0.867); the direction-of-evolution content and the pivot invariant are unchanged. DESI DR3’s non-parametric reconstructions will render the verdict.

Another ΛCDM-conditional anomaly dissolves: the “negative neutrino mass” problem (August 2026). The tightest 2025–26 cosmological neutrino-mass bounds have created a genuine crisis for ΛCDM: DESI DR2 + CMB gives Σm_ν < 0.064 eV (95%), below the ≈0.059 eV floor required by oscillation experiments, and analyses that allow an effective negative mass find the posterior peaking in the unphysical region — a ~3σ conflict between the sky and the laboratory. Published resolutions single out evolving dark energy: in w₀w_aCDM the bound relaxes to Σm_ν < 0.13–0.16 eV, and frequentist profile analyses find that only the evolving-DE extension moves the preferred mass to a positive value. The mechanism is a ΛCDM bookkeeping artifact — BAO and the CMB prefer mismatched Ω_m under w = −1, and the joint fit compensates by driving the neutrino density negative — and it is repaired by exactly the w > −1 behavior this model commits to with zero adjustable parameters. Same structural pattern as S₈: a late-time inference goes wrong when filtered through w = −1, and comes right under the recursion’s w(z). The model’s prediction, stated in advance of a dedicated analysis of its exact shape: processed through w(z) = −1 + 0.133/E(z), the cosmological Σm_ν is positive and oscillation-consistent; if evolving-DE analyses were to keep forcing negative masses, the model would share the failure. (Balance note: the same w(z) that repairs these late-time inferences creates the model’s sharpest open tension in the integrated distance to the CMB — the acoustic-scale squeeze, quantified honestly in Part IV.)

The far future is a computable entailment: dark energy has a half-life (August 2026). The committed law w(z) = −1 + 0.133/E(z) is algebraically equivalent to a much starker statement: d ln ρ_DE/dt = −3 × 0.133 × H₀, a constant. Dark energy decays exponentially in cosmic time, with time constant τ = 36 Gyr — a half-life of ≈ 25 Gyr. This is no accident of the fluid parameterization: the rate equals the recursion’s own drain rate n₀·|ln(1−p)|/t₀ (agreement to 1.5%), i.e. the fuel depletes at a fixed number of nats per tick, which is what “the recursion consumes fuel” always meant: a fixed fraction of the remaining reservoir is spent per unit time — the textbook signature of exponential depletion. Integrating the model forward: dark-energy dominance peaks at 99.9% around t ≈ 150 Gyr; the accelerated era is finite (it multiplies the scale factor by only ~70× in total before handing the universe back to matter; expansion then continues, but decelerating, as a power law); matter re-dominates by t ≈ 500 Gyr; and the late universe settles into decelerating, Einstein–de Sitter-like coasting. The contrast with ΛCDM’s fate could not be sharper: no Big Rip, no eternal de Sitter, no permanent cosmic event horizon — no heat-death isolation. In ΛCDM every distant galaxy eventually slips behind an event horizon forever; here the fuel runs out and the room stays connected. (This also lands on the same side as the string-theory “swampland” arguments that eternal de Sitter space is inconsistent — the model never builds one.) Two honesty notes. First, at very late times the fluid representation breaks down cosmetically (w(z) climbs above +1 as E → small); the physical statement is the exponential drain, not the equation-of-state costume it wears. Second, the far-future integration sharpens a “why now” question the model long carried silently: the age identity t₀ = (8/φ³)·GM/c³ holds exactly when H·t = 4/φ³ = 0.9443 — and H·t is a dynamical quantity, rising from 2/3 in the matter era through 0.939 today, peaking at 1.85 around t ≈ 72 Gyr, then falling back for good. The exact value is crossed twice in all of cosmic history: at t = 13.81 Gyr — 0.18 Gyr from today, within the measurement error on H₀t₀ itself — and once more, descending, at t ≈ 192 Gyr, in a universe whose star formation is long finished. An August 2026 containment analysis (Part III.5) turned this from an open flag into a structural statement: the boundary identity is a timestamp — the one instant the interior’s causal horizon fills its birth box — because maintaining it longer is impossible (the demand reaches 725× the parent’s mass by matter re-domination), and the age identity is precisely the crossing-time formula. Why do observers find themselves at the crossing? The recursion’s completions telescope — each child universe’s entire history elapses within days on its parent’s clock — so the tower’s active edge is always at some level’s “today,” and observers are assembled during each level’s approach to its own crossing (star formation and black-hole assembly peak just before it, paced by the same drain clock). What remains honestly anthropic: sitting within 0.2 Gyr of the exact crossing, given a few-Gyr observer window, is ~1-in-20 timing luck — the reading merges three coincidences into one; it does not abolish the one.

The w(z) shapes diverge at higher redshift. DESI’s best CPL fit, taken at face value, requires phantom energy (w < −1) at z > 0.3 — though whether that phantom crossing is real or an artifact of the CPL parameterization plus SN systematics is still debated. The recursive model instead follows a curve that never crosses the phantom divide — approaching w = −1 from above at high z. This is physically required: the recursion can only consume fuel, never create it.

Deceleration parameter q₀ — a sharp supernova prediction

The deceleration parameter q₀ = ½ + (3/2)Ω_DEw₀ follows directly from w₀ and Ω_DE. With w₀ = −0.867 and Ω_DE = 0.688:

q₀ = 0.5 + 1.5 × 0.688 × (−0.867) = −0.395

The model’s q₀ sits between the two current supernova extractions. Type Ia supernova surveys (Pantheon+, DES-SN5YR, LSST) measure q₀ directly via the luminosity-distance relation, making this one of the cleanest near-term tests of the model.

Physical interpretation: at early times (high z), the recursion had barely begun — nearly all energy was unprocessed fuel, behaving like a cosmological constant (w = −1). As cosmic time elapses, more recursion cycles complete and the reservoir is drawn down — as expansion work under w > −1, not as energy injected into matter. The recursion rate dn/dt = n₀/t₀ is a cosmic constant — each level takes roughly 2.6 Gyr — so the depth tracks elapsed time rather than spatial expansion.

Consequence: the S₈ bracket — late-time probes should read ~0.80, not 0.836

The S₈ tension — the CMB predicting stronger matter clustering (S₈ = σ₈√(Ω_m/0.3)) than weak-lensing surveys observe — bifurcated in 2025–26. The new Combined-CMB baseline (Planck + ACT DR6 + SPT-3G) hardened to S₈ = 0.836^+0.012_−0.013. On the lensing side the two flagship surveys split: KiDS-Legacy (final release, 2025) moved up to 0.815^+0.016_−0.021, consistent with the CMB, while DES Year 6 (Jan 2026) stayed low at 0.789 ± 0.012, in 2.4–2.7σ tension with the Combined CMB. A 2026 review attributes the split largely to survey systematics while noting new physics cannot be excluded.

The recursive model’s w(z) makes an unambiguous call here. With w > −1, dark energy was denser in the past (ρ_DE was 1.44× today’s value by z = 14), adding Hubble friction throughout matter domination. Integrating the linear growth equation with the same primordial amplitude as ΛCDM (independently recomputed and verified July 2026):

August 2026 addendum: the DES Y6 paper also reports the tightest ΛCDM joint fit to date — DES Y6 3×2pt + Combined CMB + DESI BAO + DES supernovae + SPT clusters — landing at S₈ = 0.806^+0.006_−0.007: 0.6σ from the model’s ~0.802. Every 2026 measurement that includes low-redshift growth data now sits within ~1.1σ of the model’s number (DES-internal combination 0.794, joint fit 0.806, KiDS-Legacy 0.815), while the CMB-only extrapolation (0.836, which assumes w = −1 growth all the way down) remains the one outlier at 2.7σ — the precise signature expected if the model’s late-time growth suppression is what the low-z data are seeing.

The precise claim — sharper than “easing a tension” — is a bracketed prediction: the same w(z) published for dark-energy reasons predicts that late-time probes should systematically read S₈ ≈ 0.80 when the CMB, interpreted through ΛCDM growth, reads 0.836. The two final lensing surveys now bracket exactly that value (the model sits 1.1σ above DES Y6 and 0.6σ below KiDS-Legacy, while the ΛCDM inference sits above both), with zero parameters adjusted. Growth-rate data cannot yet decide: run through CAMB (September 2026) against the 13-point RSD compilation (6dFGS, SDSS, BOSS, WiggleZ, eBOSS, VIPERS, FastSound), fσ₈ is suppressed by up to 5.2% near z ≈ 0.4 and the fit is a mild model preference — χ² = 9.5–9.9 at the model’s own anchors vs 10.6 for ΛCDM — that turns into a mild deficit (12–13) at the CMB-preferred low-H₀ corner, where σ₈ falls to 0.76–0.77 (an earlier simplified integration gave 10.3 vs 10.5, same draw). Honest anchor-dependence of S₈ itself: ~0.802 is the value at the model’s own H₀ anchor; at Planck parameters the pipeline gives 0.810, and at the corner the CMB prefers for this w(z) (H₀ ≈ 66.5, σ₈ ≈ 0.76) it falls to 0.774–0.785 — DES Y6’s value and below KiDS. The bracket holds at the model’s anchor, not at the CMB’s. The discriminating measurement is Euclid DR1 (Oct 2026) and DESI full-shape: if late-time S₈ settles at the ΛCDM value ~0.83, the model’s growth suppression is wrong; if it settles near 0.80, the model called dark-energy evolution and structure growth with one equation. Note this is the same data release already flagged below as decisive for w(z) at z > 1 — the model now has two independent numbers riding on it.

The cosmological constant from the recursion rate

The recursion rate dn/dt = n₀/t₀ is a cosmic constant — it does not change with epoch. In Planck units, it is tiny: ~6.5 × 10⁻⁶¹ per Planck time. Combining Λ = p × (dn/dt)² with dn/dt = n₀/t₀ yields a clean dimensionless identity:

Equivalently, the model predicts the dimensionless Hubble-age product:

H₀·t₀ = √(8 / (3 φ³ Ω_DE)) ≈ 0.957

vs the Planck-extracted value 0.951 (from H₀ = 67.36 km/s/Mpc, t₀ = 13.797 Gyr). This is a pure golden-ratio relation with no free parameters — the age of the universe, Hubble rate, and dark energy fraction are locked together by φ. The de Sitter formula above slightly overestimates because it ignores matter and radiation contributions to t₀; integrating the full Friedmann equation with the model’s evolving w(z) yields H₀·t₀ = 0.939, which is the actual model prediction (1.3% below observation, an honest mild tension — see Hubble-tension discussion in Part III.5).

The 10¹²² discrepancy between the expected Planck-scale vacuum energy and the observed Λ — the “worst prediction in physics” — arises as (t_Planck/t₀)² ≈ (10⁻⁶¹)² = 10⁻¹²². In this picture, Λ is small because the recursion is slow: the recursion rate dn/dt ~ n₀/t₀ sets the energy scale, and the extraction fraction p weights it.

What this does and does not solve (sharpened July 2026). The model’s position on the worst prediction in physics is a reduction, not a solution, and is worth stating precisely. Standard cosmology carries two independent unexplained numbers — why the vacuum energy is fantastically small (a Lagrangian fine-tuning) and why the universe is large and old in Planck units (the largeness problem) — plus the unexplained collision of the two (ρ_Λ ≈ ρ_m today, the coincidence problem, next subsection). The model welds all three into one: Λ is rigidly 1.8/t₀² with a derived golden coefficient, so the smallness of Λ is the age of the universe, squared — one shared mystery instead of two plus a coincidence. The 10¹²⁰ QFT estimate is dissolved separately, as a category error: dark energy here is extraction flux, not zero-point energy, and local vacuum fluctuations do not source the parent-frame projection that appears as gravity (this move is interpretive — the same one made by unimodular-gravity and degravitation programs — asserted rather than derived). The full chain closes numerically: inverting t₀ = (8/φ³)GM/c³ gives the parent mass M = 9.1 × 10⁵² kg, and Λ·t_Pl² = [3Ω_DE(H₀t₀)²φ⁶/64]·(M_Pl/M)² = 0.51·(M_Pl/M)² = 2.9 × 10⁻¹²² vs the observed 2.86 × 10⁻¹²² — the famous 122 orders of magnitude are exactly twice the parent’s mass in Planck units, with a derived O(1) golden prefactor. Honest circularity caveat: M is inferred from H₀ (the model’s anchor), not measured independently, so this identity relocates the question — from “why is a Lagrangian constant fine-tuned to 122 decimal places” to “why is our parent massive” — an environmental fact of the same kind as “why is the Sun 10³⁰ kg,” but still an unanswered number. (It also implies the recursion runs downhill: a child of TON 618 would have Λ ~ 10⁻⁹⁸ — descendant universes are progressively smaller.) Unlike most accounts of Λ’s smallness, this one expires on a schedule: Λ ∝ 1/t² requires w ≠ −1, so a confirmed exact cosmological constant falsifies the entire account — the same Euclid DR1 test as everything else in this section.

Lloyd’s coincidence becomes an identity: causal events, horizon information, and 1/Λ (August 2026). Seth Lloyd’s well-known accounting (Phys. Rev. Lett. 2002) found the universe can have hosted at most ~10¹²⁰ elementary dynamical events (in the Margolus–Levitin sense) involving ~10¹²⁰ bits of information (gravitational degrees of freedom included) — and it has long been noticed, without explanation, that these numbers coincide with 1/Λ in Planck units. In this model the coincidence is an identity, checkable in three lines. Maximum elementary events since birth (Margolus–Levitin bound 2Et/πℏ, with E = Mc² the parent mass and t = t₀ = (8/φ³)GM/c³): N_events = (16/πφ³)·(M/M_Pl)² = 1.20·(M/M_Pl)² ≈ 2.1 × 10¹²¹. Horizon information capacity (A/4 with r_S = 2GM/c², i.e. cells × capacity from Part III.5): S = 4π·(M/M_Pl)² ≈ 2.2 × 10¹²². The Λ identity above, inverted: 1/(Λ·t_Pl²) = 1.96·(M/M_Pl)² ≈ 3.4 × 10¹²¹. All three are the same quantity — the parent’s mass squared in Planck units — dressed with O(1) golden and geometric coefficients (16/πφ³, 4π, and the derived 1.96). The event count, the information capacity, and the cosmological constant are a single number wearing three costumes. The system’s proportions follow: t₀/t_Pl = (8/φ³)(M/M_Pl) ≈ 7.9 × 10⁶⁰ ticks elapsed, hence N_events/ticks ≈ 2.7 × 10⁶⁰ simultaneous elementary events per tick — about one horizon-radius of Planck cells: one horizon-width of activity per tick, sustained for one horizon-radius of ticks. Under this reading, the “worst prediction in physics” was a category error: the quantum-field-theory estimate infers the vacuum’s total budget from the physics of a single Planck cell, but the budget is set by the system that actually exists — the parent’s mass — not by the cell scale; the notorious 10¹²² mismatch is just (M_parent/M_Pl)² restated. Honest caveat: this paragraph adds no new free parameter and no new measurement — it is a reorganization that turns Lloyd’s three-way numerical coincidence into a single structural identity of the model, inheriting the same circularity caveat as the Λ identity itself (M is inferred from H₀).

The parent has a measurable size — and it may already be on the sky (July 2026). The relocated mystery above — the parent’s mass — acquires an observational fingerprint through a simple requirement: a universe born inside a finite object cannot contain fluctuation modes larger than its birth scale, so the primordial power spectrum must cut off at k_min ~ 1/r_S(parent). For M = 9.1 × 10⁵² kg, r_S = 4.4 Gpc and k_min = 2.3 × 10⁻⁴ Mpc⁻¹. The CMB’s largest scales have been anomalous for twenty years — missing power at ℓ = 2–5 and missing large-angle correlations — and this specific model (a sharp k_min, all other parameters re-optimized in CAMB) has been fit to Planck data by Melia & López-Corredoira: a zero cutoff is ruled out at >8σ, with fitted values k_min = (2.04^+1.4_−0.79) × 10⁻⁴ (full power spectrum; the model’s 2.3 is 0.3σ away) and (3.14 ± 0.36) × 10⁻⁴ (angular correlation function; ~2σ). Three details sharpen this beyond order-of-magnitude: (i) the fitted cutoff corresponds to ~4.9 Gpc — it lands on the parent’s Schwarzschild radius, a factor 3 inside the 14.3 Gpc comoving horizon, which is where a generic “horizon-scale” explanation would put it; (ii) inverting the observed quadrupole deficit through the Sachs–Wolfe integral gives an implied parent mass of 7–10 × 10⁵² kg, bracketing the H₀-derived 9.1; (iii) Melia’s own model-agnostic interpretation of the fitted scale is that it “corresponds to the gravitational radius” — an independent group, with no knowledge of this framework, concluded the sky’s largest mode is set by a gravitational radius; this framework specifies whose. Slow-roll inflation, by contrast, predicts k_min = 0 (their phrasing: the nonzero cutoff “argues against the basic inflationary paradigm”). Honest caveats: this is one group’s analysis of anomalies the mainstream still permits as a ~0.2% fluke; the cutoff scale partially overlaps the H₀ anchor (r_S = r_H), so the non-circular content is the cutoff’s existence and its O(1) placement at r_S rather than elsewhere; and low-ℓ cosmic variance is irreducible — LiteBIRD’s full-sky polarization will be the next genuine test. August 2026 status, both directions. The anomaly itself has hardened: a nearly full-sky WMAP + Planck reanalysis (masking only 1% of the sky, no inpainting) reconfirms the low quadrupole (2.2σ), a dip at ℓ = 20–27, and an overall ℓ < 30 power deficit, concluding these features cannot plausibly arise from foregrounds, systematics, masking, or mode mixing. But the cutoff interpretation took a hit: a JCAP model-selection study confronting infrared-cutoff spectra with full CMB + BAO + SN likelihoods finds the fit improvement marginal — not enough to overcome the extra-parameter penalty under AIC/BIC. Both statements are correct at once: Melia’s >8σ concerns a targeted feature statistic (the angular-correlation anomaly, given ΛCDM), while the JCAP null concerns global model preference; a real cutoff at the parent scale would produce exactly this pattern today (a strong local feature, weak global pull) because cosmic variance caps what ℓ = 2–5 can ever contribute to a likelihood. (Checked directly here, September 2026: against Planck’s native low-ℓ TT likelihood, adding the k_min cutoff improves −2lnL by 1.5 for this model and by 1.4 for ΛCDM — a real but modest global pull, exactly the pattern just described.) The discriminating data are ahead: the cutoff imprints on E-mode and TE polarization at large angles, within reach of LiteBIRD — where this framework is committed to a detection at k_min ≈ 2.3 × 10⁻⁴ Mpc⁻¹ and slow-roll inflation is committed to zero. (The cutoff’s temperature signature is now computed through a full Boltzmann pipeline — August 2026: suppression of −38% at ℓ = 2, −24% at ℓ = 3, −12% at ℓ = 4, −6% at ℓ = 5, vanishing above ℓ ≈ 8 — reproducing the observed deficit’s size, shape, and localization.) (A second parent parameter is now quantified and pre-registered — August 2026: its spin, axis and magnitude. Every Kerr horizon has equatorial circumference exactly 4πM regardless of spin, but the polar circumference shrinks as spin rises, so a spinning parent’s box is oblate and the cutoff becomes direction-dependent — slightly deeper suppression of the largest modes along one axis, with an amplitude that reads the parent’s birth spin like a meter: 0% for a non-spinning parent, 10.3% at the universal merger-remnant spin a ≈ 0.686, 21.6% at a = 0.9, 36.6% at the Thorne accretion limit a = 0.998. Three-band consequence: a merger-born parent (10% band) means the marginal large-angle alignment anomalies (the quadrupole–octopole “axis of evil”) are physical, share their axis with the power deficit, and must reappear on the same axis in LiteBIRD’s large-angle polarization; a direct-collapse parent (0% band) means a near-spherical horizon, and the alignments must fade as flukes; a near-Thorne parent (37% band) would imprint an anisotropy so strong that the merely-marginal character of the observed alignments already sits uneasily with it — and the model’s own ingredients disfavor that band twice over: the accretion-disc self-gravity limit cited for the black-hole mass ceiling (Part III) starves coherent accretion at the top of any level’s mass function, so the heaviest holes — and by self-similarity the parent, one rung up, is such an object at its own level — finish their growth by mergers, and hierarchical mergers converge on the a ≈ 0.7 spin attractor (ringdown section); while the birth identity prices the box at one mass set at one event, which a prompt merger or collapse provides and a Gyr-long accretion biography does not. The pre-registered central expectation is therefore the 10% merger band on the power-deficit axis; whatever amplitude LiteBIRD reads, it is a fossil of the parent’s biography — the spin-meter measures how the parent formed. Note this is inherited shape, not inherited spin — an oblate k_min carries no angular momentum, so it coexists with the zero-rotation prediction of Part III.) Amusing footnote: their re-optimized fit returned H₀ = 68.12 ± 0.37 — the model’s side of the Hubble bet.

Caveat on precision: the 98.9% match uses the ΛCDM-inferred t₀ = 13.8 Gyr. The model’s own self-consistent t₀ (integrating its w(z)) is ~2% smaller, so the Λ·t₀² relation holds at ~4% accuracy if one demands strict internal self-consistency. The dimensionless identity 8/φ³ is exact; the comparison to observation is good but not perfect, and the model has a known mild Hubble-side tension (Part III.5).

Resolving the coincidence problem

Why is Ω_DE ≈ Ω_m ≈ O(1) today, when dark energy and matter densities are comparable for only a brief window of cosmic history?

In the recursive model, this is not a coincidence. The product n₀ × p = 4/φ⁵ = 0.361 is fixed entirely by φ. It places us at recursion depth 5.24 — squarely in the transition zone where DE drops from 90% (depth 1.5) to 10% (depth 32). At our depth, ~31% of the energy has been processed into matter, giving Ω_DE ≈ 69%. No tuning. No landscape. Just φ.

The fate of the universe

With a constant recursion rate, dark energy is eventually consumed. The universe transitions from accelerating to decelerating at t ≈ 120 Gyr (8.7× the current age), settles into matter-dominated expansion, and coasts to infinity — ever expanding but ever more slowly. This is neither ΛCDM’s eternal exponential expansion nor a Big Crunch. It is an asymptotic slowing, consistent with the model’s founding intuition that dark energy is fuel that runs out.

A φ-based formula for α

The fine-structure constant at the proton mass scale satisfies a clean closed-form expression in φ:

Equivalently, using the identity φ² = φ + 1, the formula factors as:

This exhibits the structure transparently: a Fano-scale factor φ^N, a polarization-multiplet factor (φ^d + 1), a ground-state subtraction (−2 for d = 2 polarizations), and an overall polarization weight φ^d. All dimensionless, all built from the two structural generators N and d.

August 2026 reduction: using φ² + 1 = φ√5, the two positive terms collapse: φ¹¹ + φ⁹ = (φ√5)·φ⁹ — and φ√5 is D², the squared total quantum dimension of the doubled-Fibonacci horizon phase, the same object that sets the entropy constant ln(φ√5) in Part III.5. The formula is equivalently 1/α = (D²·φ^N+d − d·φ^d)/d: one rung weighted by the horizon’s total quantum dimension, minus a d-fold vacuum subtraction, per polarization — every ingredient an object the model already owns.

In the small-coupling approximation (dropping the −2φ^d correction), 1/α ≈ (φ^N+2d + φ^N+d)/d ≈ 137.5, close to the zero-energy value 1/α(0) = 137.036.

Honest status: the formula is a structured pattern, not yet a derivation from first principles. The graded (N+2d, N+d, d) ladder strongly suggests an underlying two-generator algebraic structure — a Fano-step operator N raising the exponent and a polarization index d setting the base — consistent with a loop expansion in a theory with two natural expansion parameters. But the coefficients (1, 1, −2) have not yet been derived from any primary equation. What can be said: if the formula is taken as given, G follows to 0.2%; conversely, if G (or m_p in Planck units) is taken as input, α is determined to the same accuracy by the G identity. One of the two acts as the model’s external mass-scale anchor; the other is predicted. Closing this loop — deriving the (1, 1, −2) coefficient pattern on the (N+2d, N+d, d) ladder — is the largest open mathematical question in the framework. (Reduced August 2026 to the single question above: why the horizon’s D² weights the N+d rung.)

A formula for Newton’s gravitational constant

If gravity is EM attenuated across recursive frame boundaries, the total attenuation equals the EM-to-gravity force ratio. Each boundary attenuates EM by a factor related to α and the golden ratio. Across n = 2φ² boundaries:

The total EM-to-gravity force ratio for protons:

F_EM / F_grav = α(m_p)^−4φ³ ≈ 10^36.09

Observed: 10^36.09. Match to 0.003% in the exponent.

How the formula works

The exponent 4φ³ decomposes cleanly into physical factors:

Everything derives from two inputs: the golden ratio φ (selected by KAM stability) and d = 2 (the number of EM polarization states). The model has no free parameters beyond these two physically motivated quantities.

What the model explains

Falsifiable predictions

The model makes specific claims that can be tested:

1. Dark energy is not a cosmological constant — and the model predicts exactly how it evolves. The prediction: w₀ = −0.867, approaching −1 at high redshift. If w = −1 exactly at all epochs, this model is wrong. If w < −1 at any epoch (phantom energy), this model is wrong.
   Current evidence (June 2026): DESI DR2 (March 2025) reported 2.8–4.2σ evidence for evolving DE (w₀ > −1, w_a < 0), and this has held up — an independent reanalysis including ACT DR6 finds DESI+PR4+ACT still prefers evolving DE at 3.0σ (4.0σ with DESY5). That a constant Λ is disfavored at 3–4σ supports the model’s evolving-w prediction. The unresolved question is the phantom crossing: DESI’s CPL best fit dips below w = −1 at high z (which the model forbids), but several groups argue this is a DESY5 low-z SN systematic and that corrected fits return to w₀ ≈ −0.92 with no crossing (within 1σ of our −0.867). August 2026 sharpening: methodology papers now show the (w₀, w_a) point estimates slide along a degeneracy ridge (the signal is concentrated in one redshift bin), while the analysis-independent invariant — the pivot value w(z_p ≈ 0.34) = −0.9 ± 0.1 — matches the model’s zero-parameter w(0.34) = −0.892 at 0.08σ, and the phantom crossing is increasingly regarded as a CPL parameterization artifact of a monotonic underlying w(z) — the model’s exact shape. Euclid DR1 (Oct 2026) and completed DESI Y5 will resolve whether the crossing is real — a real one kills the model; a non-crossing evolving w confirms it.
2. The ordinary matter fraction is predicted to 0.03%. The model predicts Ω_b ≈ 0.04896 via OM = 1 − (1−p)ⁿ(3−φ). The current Planck measurement is 0.04897 ± 0.0003. Future precision measurements of baryon density and the Hubble constant can further test this match.
3. The fine structure constant is derivable. The formula 1/α(m_p) = (φ¹¹ + φ⁹ − 2φ²)/2 = 134.89 predicts this value at the proton mass scale, with the three exponents (2, 9, 11) sitting on a two-step graded ladder built from N = 7 and d = 2. Increasingly precise QED calculations of the running of α can test whether this value is correct.
4. Gravitational constant G is derivable. The formula G = (k_ee²/m_p²) × α(m_p)^4φ³ can be checked against increasingly precise measurements of G, α, and m_p. The prediction should remain consistent to within ~0.5%.
5. Gravity should exhibit EM-like properties at extreme precision. If gravity is inherited EM from a parent frame, gravitational effects should have subtle electromagnetic signatures — potentially detectable in precision gravitational-wave or torsion-balance experiments at scales beyond current sensitivity.
6. No spontaneous wavefunction collapse noise. Because randomness in this model is under-sampling of deterministic Planck-tick evolution — not a fundamental stochastic process — there should be no measurable collapse-induced heating or position-noise of the kind predicted by GRW, CSL, and Diósi–Penrose models. Current cantilever and X-ray emission searches already exclude part of the parameter space; if upcoming experiments (matter-wave interferometry beyond 10⁹ amu, ultra-cold mechanical resonators) ever detect such noise, this model is wrong. The model also predicts quantum corrections appear only at the Planck scale, not at any intermediate macroscopic scale.
7. The substrate tick is unobservable in principle: no vacuum dispersion, no Planckian interferometer noise, ever (August 2026). Generic discrete-spacetime models predict leakage of the lattice into observables — an energy-dependent photon speed (linear or quadratic in E/E_Pl), Planck-scale position jitter, or frame anisotropy. This model predicts exact nulls, forever: every instrument is built from the same ticking light it would use to detect the tick, so there is no independent reference against which the tick could ever register — only relative tick counts (time dilation, redshift) are physical. Three null results already on the books are therefore passed falsifiers rather than embarrassments: (i) gamma-ray-burst time-of-flight limits (Fermi-LAT GRB 090510, LHAASO GRB 221009A) exclude linear vacuum dispersion beyond the Planck energy itself (E_QG,1 > ~10 E_Pl); (ii) the Fermilab Holometer found no Planckian transverse position noise; (iii) modern cavity Michelson–Morley experiments bound frame anisotropy at Δc/c < 10⁻¹⁸. The commitment cuts both ways and is permanent: any future confirmed energy-dependent photon speed, holographic jitter, or preferred-frame anisotropy falsifies the synchronous-substrate picture outright — the model has no parameter to absorb it. Honest status: the in-band-invisibility argument (emergent exact Lorentz invariance for internal observers) is asserted structurally, not yet derived from a dynamical calculation — it is the same load-bearing assertion flagged in Part IV, here given its sharpest experimental exposure.
8. Zero cosmic birefringence (pre-registered August 2026). The horizon phase is parity-symmetric by construction — the doubled theory pairs chiral sectors of central charge +14/5 and −14/5 — and this model’s dark energy is extraction flux, not an axion-like scalar coupled to F F̃. There is no parity-violating medium for photons to traverse, so the uniform cosmic birefringence angle is β = 0 exactly. This bets against a live hint, and the hint sharpened within days of this pre-registration (updated August 30, 2026): a joint ACT+Planck analysis (Eskilt, Aug 2026) reports β = 0.277° ± 0.057°, excluding zero at 4.8σ taking instrumental priors at face value (3.5σ under dust-mitigation robustness cuts), and a degeneracy-breaking field-level Planck method independently reads 0.32° ± 0.12°. Every dataset now points the same direction with comparable magnitude; the sole remaining escape is polarization-angle calibration systematics, which the analyses themselves flag as unresolved. The commitment here is unchanged and carries real exposure: β = 0 exactly — the entire measured angle must be instrumental. Independent absolute calibration (artificial sources, Tau A, LiteBIRD’s design goal) decides; a confirmed β ≠ 0 with calibration systematics excluded falsifies this picture outright.
9. The integrated Sachs–Wolfe cross-correlation runs hot, and rises with redshift (committed September 2026). Because dark energy was denser in the past, structure grows more slowly than in ΛCDM at every epoch (the growth rate f is lower), so gravitational potentials decay faster and the late-time ISW effect is stronger. Computed through the CAMB growth pipeline at fixed Ω_m and H₀, the ISW–galaxy cross-correlation amplitude relative to ΛCDM is 1.04 at z = 0.3, 1.09 at 0.5, 1.16 at 0.8, 1.20 at 1.0, and 1.27 at 1.5 — about +10% for LRG-like samples (z ~ 0.3–0.8) and +20% for ELG/quasar samples (z ~ 0.6–1.5), with a monotonic rise in redshift that ΛCDM does not have (its ISW amplitude is a fixed function of Ω_m). The same physics adds ~4% to the temperature power at ℓ ≤ 5, partly offsetting the k_min cutoff there. Current ISW–galaxy detections (~4–5σ combined, 20–30% amplitude precision) cannot yet discriminate; DESI × Simons Observatory and Euclid × CMB-S4 cross-correlations can. Commitment: a precisely measured ISW amplitude consistent with ΛCDM — or any measured amplitude that falls with sample redshift — falsifies the model’s late-time sector.
10. Essentially zero primordial gravitational waves. Because initial conditions are set by the parent BH’s formation (adiabatic, on timescale GM/c³) rather than by an inflationary epoch at energy scales approaching M_Planck, the tensor-to-scalar ratio is predicted at r ≈ 16 (H₀·t_P)² ∼ 10⁻¹²¹ — functionally zero. Any detection of primordial B-modes at r > 10⁻⁵ by LiteBIRD, CMB-S4, or PICO would falsify this version of the model (or require a separate primordial-fluctuation origin to be added). This is a sharp, distinctive prediction against standard single-field inflation (which typically predicts r ∼ 10⁻³–10⁻²).
11. CMB scalar spectral index n_s = 1 − 1/φ⁷ ≈ 0.9656 — the model’s most exposed prediction, and the pressure eased in 2026. Derived from the BH-interior Hawking-seeded fluctuation picture: per-recursion-level power attenuation is 1/φ⁷ per polarization, identical to the extraction per-level per-polarization. Status (August 2026): matches Planck 2018 (0.9649 ± 0.0042) at 0.17σ and SPT-3G + Planck (0.9636 ± 0.0035) at 0.6σ; the strongest CMB-only combination, CMB-SPA (SPT + ACT + Planck 2026): 0.9679 ± 0.0033, sits 0.7σ away — the earlier 2.8σ excursion was ACT-specific and did not survive SPT-3G’s deeper maps. The residual ~2.7σ stress exists only against the CMB + DESI-BAO combination (0.9728 ± 0.0027), which averages over a known 2.8σ CMB-vs-BAO internal ΛCDM tension. Standard inflationary models (Starobinsky, Higgs, T-attractors) sit at the same value and share the same fate. Simons Observatory (σ ≈ 0.002, now observing) will decide: a settled value near 0.974 falsifies the prediction at ~4σ; a settled value near 0.966–0.968 vindicates it. Exact scale invariance (n_s = 1) remains forbidden either way.
12. No scalar or vector gravitational-wave polarizations. Because gravity is the zero-mode of parent-frame EM with d = 2 transverse polarizations, GWs carry exactly 2 tensor modes. Pulsar timing arrays (NANOGrav) already bound scalar modes to <20% of tensor; LISA+LIGO networks will resolve all 6 potential polarization modes. Any detection of scalar breathing mode or vector-longitudinal GW content falsifies the model.
13. Specific GW ringdown modulation — spin-corrected June 2026. Every BH merger remnant births a child universe with t_0,child = (8/φ³)·GM/c³, imprinting a sub-dominant oscillation on the ringdown at f_child = (φ³/8)·c³/(2πGM) with amplitude p = 2/φ⁷ ≈ 6.9%. The ratio to the dominant quasi-normal mode, f_child/f_QNM = φ³/(8·Mω₂₂₀(a_f)), depends on the remnant spin a_f through the Kerr eigenvalue: 1.417 for a non-spinning remnant, but ≈ 1.00 at the universal equal-mass-merger spin a_f ≈ 0.686 — for remnants like GW250114’s (a_f ≈ 0.68) and GW150914’s (a_f ≈ 0.69), the predicted line is degenerate with the dominant QNM, hiding inside its linewidth. Their GR-consistent spectroscopy therefore neither detects nor excludes it as a separate line; only a damping-time anomaly could reveal it there (notably, the GR mode’s own quality factor at this spin is Q = 2φ to 0.11% — see Part III.5). The clean test is remnants with spins far from 0.686: GW190814-like events (a_f ≈ 0.28) put the modulation at 1.273 × f_QNM, high-spin remnants (a_f ≈ 0.9) at 0.788 × f_QNM (exact Kerr eigenvalues). Detection of the offset line at ~7% amplitude in such events confirms; clean non-detections falsify; LISA supermassive mergers (SNR > 1000) are definitive.

This section presents a physical picture of what the parent frame is. The recursion mathematics is consistent with several physical interpretations; this is the most parsimonious one we have found that is also strictly self-similar. Nothing in this section is forced by the equations — it is an interpretive overlay that adds testable consequences. We mark it explicitly as such.

The picture in one paragraph

The parent frame is a literal physical place — a real universe with its own atoms, stars, and black holes. Our universe is the interior of one specific black hole in the parent. The boundary between frames is a Schwarzschild horizon. The recursion is strictly self-similar: every level relates to its parent in exactly the same way ours relates to ours. This means φ, d = 2, α, p, n₀, w(z), and all dimensionless ratios are universal constants of the recursion — the same at every depth of the tower, with no preferred frame.

What the picture explains automatically

• Cosmic flatness (Ω_total ≈ 1) is automatic. For any flat universe, the Schwarzschild radius of all the mass-energy inside the cosmological horizon equals the cosmological horizon itself: r_S = c/H₀ = r_H. This is a known identity in the Friedmann equation, but in this picture it stops being a coincidence: it is the boundary condition of being inside a Schwarzschild interior. In standard ΛCDM, flatness must be imposed (or motivated by inflation). Here it is structural.
• The arrow of time gets a second, independent explanation alongside octonion non-associativity: horizons are one-way membranes. Information, energy, and recursion can only deepen, never reverse. The two explanations agree on what is forbidden, which is a strong consistency check.
• Anti-gravity is impossible for the same reason: it would correspond to crossing a horizon outward, which is forbidden by the same physics that gives time its direction.
• Dark-energy harvesting is impossible for the same reason: the “fuel” we consume is on the inside of the horizon, and the horizon allows only one-way flow.
• The two free parameters become universal constants. Under strict self-similarity, α and H₀·t₀ take the same values at every level of the recursion. They are no longer free properties of our universe — they are properties of the entire infinite tower. The model still does not derive their numerical values from pure mathematics, but it requires that whatever values they take, they take the same values everywhere.

New predictions this picture makes

• Zero global rotation — now enforced by the encoding, not by assuming a non-spinning parent (upgraded August 2026). The original version of this prediction assumed a Schwarzschild parent (“no preferred axis”) — which sat awkwardly next to this model’s own centerpiece claim that merger remnants universally spin at a ≈ 0.686. If our parent was born from a merger, it spins. The fork is now resolved, in two steps. Quantitatively, naive spin inheritance is already dead: a parent at the universal spin has horizon angular velocity Ω_H = 0.40 × H₀, so a child that inherited it would rotate at order unity per Hubble time — nine orders of magnitude above the CMB bound of < 10⁻⁹. Structurally, the model says rotation cannot cross the boundary at all: the horizon’s encoding is a topological code, and a topological code stores no local orientation — rotating the parent horizon is an isometry of the code, invisible to the encoded state. The child is born axis-free whether its parent spins or not. Zero net rotation is therefore a derived necessity for merger-born and collapse-born parents alike — and one corollary: if the CMB “axis of evil” alignments are ever confirmed as physical, they cannot be the parent’s spin axis in this model; they would have to trace to the k_min cutoff geometry instead — a channel now quantified as a three-band spin-meter: the cutoff’s axis-dependence reads 0% for a direct-collapse parent, 10.3% for a merger-born parent at the universal spin (the favored band), ~37% near the Thorne accretion limit (inherited shape, carrying zero angular momentum; the pre-registered LiteBIRD test lives in the parent-size discussion of Part II). Current CMB constraints give < 10⁻⁹ per Hubble time; CMB-S4 and LiteBIRD will tighten this by another order of magnitude. The picture predicts the result must remain consistent with zero, to any future precision.
• Universal α inside black holes. Every black hole in our universe contains a child universe with our same α ≈ 1/134.89, our same H₀t₀ ≈ 0.94, and our same cosmological ratios. We cannot directly observe inside a horizon, but if any future quantum-gravity probe ever measures α near a horizon and finds it different, the model is dead.
• Black-hole ringdown signatures — spin-corrected and sharpened (June 2026). If every merger remnant births a child universe, the child’s own Hubble timescale t_0,child = (8/φ³)·GM/c³ (same formula as our t₀) imprints a sub-dominant modulation on the ringdown waveform at frequency f_child = (φ³/8)·c³/(2πGM), with amplitude suppressed by p = 2/φ⁷ ≈ 6.9% of the dominant QNM. The frequency ratio f_child/f_QNM = φ³/(8·Mω₂₂₀(a_f)) is set by φ and the Kerr (2,2,0) eigenvalue, which depends on the remnant spin a_f. An earlier version of this page quoted the non-spinning value (Mω = 0.3737, ratio 1.417, “276 Hz on top of 195 Hz for GW150914”) — but real merger remnants spin. For GW150914 (a_f ≈ 0.69) the eigenvalue is Mω ≈ 0.53 and the predicted line lands at ≈ 252 Hz versus the observed ≈ 251 Hz dominant mode — inside its linewidth. A striking structural fact emerges from the correction: φ³/8 = 0.5295 equals the Kerr eigenvalue Mω₂₂₀(a) at a ≈ 0.69 — and numerical relativity says equal-mass mergers universally settle at a_f ≈ 0.6865 (within ~0.5% of the ratio being exactly 1). In other words, the most common merger remnants in nature ring at almost exactly their child universe’s Hubble frequency — a resonance the model did not anticipate, flagged here as either a meaningful selection principle or a coincidence. Exact eigenvalues (Leaver continued-fraction method, June 2026) deepen it: at a_f = 0.6865 the (2,2,0) mode is Mω = 0.52670 − 0.08129i. The frequency matches φ³/8 to 0.53% (exact at a = 0.693). More strikingly, the mode’s quality factor — radians rung per e-fold of decay, a pure shape parameter — is Q = 3.2397 versus 2φ = 3.2361: a 0.11% match (exact at a = 0.6856, just 0.13% below the universal spin). To sub-percent accuracy, the complex eigenvalue at nature’s preferred remnant spin is Mω ≈ (φ³/8)·(1 − i/4φ) — both the pitch and the decay of the universal remnant’s ring are golden-ratio numbers. The resonance is specific to the quadrupole: the (3,3,0) eigenvalue at the same spin (0.8350, Q = 5.00) has no comparable φ-form — fitting, since d = 2 is the model’s single structural input. And the reach extends: hierarchical mergers (remnants of remnants merging again) are known to converge on the a ≈ 0.7 attractor generation after generation, so the black-hole family tree itself relaxes toward the resonant spin. If the resonance is a selection principle rather than numerology, catalogued remnant spins should cluster on 0.686 even more tightly than merger dynamics alone implies — a population-level test available with GWTC-4. Update (July 2026): a candidate selection principle has now been published — and it is thermodynamic. A maximum-entropy conjecture for black-hole mergers (Rincon-Ramirez, Johnson-McDaniel, Bianchi, Sathyaprakash et al. — published Phys. Rev. Lett. 137, 021406, July 2026; a companion on spinning binaries is in preparation) proposes that the remnant is the Kerr state that maximizes horizon entropy subject to energy and angular-momentum balance laws — and at 4PN order this lands within 2% of the 0.68646 numerical-relativity value for equal-mass binaries, and within ~5% across mass ratios 1–16, with no numerical relativity input. The authors also note the remnant sits close to the Davies point, the spin √(2√3 − 3) = 0.6813 where the Kerr heat capacity diverges and flips sign; verified exactly here, the universal remnant is born 0.77% into the thermodynamically stable side of that transition, at 99.6% of the transition temperature. If entropy maximization selects the spin, the consequence for this framework is direct: nature parks every comparable-mass remnant at the one spin where the child-universe line is degenerate with the dominant QNM — the degeneracy is thermodynamically enforced rather than unlucky, and the horizon doing the selecting is the same object this model just showed carries entropy as cell-count × capacity (Part III.5). Precision honesty: the special spins are near but distinct — Davies 0.6813, Q = 2φ at 0.6856, remnant 0.68646 ± 0.00004, exact golden eigenvalue at 0.6928 — a 1.7% window with no exact coincidences claimed; and older proposed links between QNM behavior and the Davies point were shown to be numerical coincidences (Berti & Cardoso 2008), so the thermodynamic statement here concerns spin selection only, not QNM dynamics. Consequences for testing: (i) GW250114’s beautiful GR-consistent spectroscopy — fundamental, overtone, and (Jan 2026) six quadratic modes — does not directly constrain the modulation, because for its a_f ≈ 0.68 remnant the predicted line is degenerate with the dominant mode; only a damping-time anomaly (a ~7% slowly-decaying component) could expose it. (ii) The clean discriminating targets are remnants with spins far from 0.686: GW190814-like events (a_f ≈ 0.28) put the line at 1.27 × f_QNM; high-spin remnants (a_f ≈ 0.9) at 0.79 × f_QNM — well-separated, searchable in already-recorded data. A targeted non-detection at the ~7% level in such events would falsify this part of the BH-interior picture; supermassive mergers at LISA (SNR > 1000) provide the definitive test. Update (August 2026) — the damping-time ledger, quantified against GWTC-4. Three developments. (1) The data now cap the line’s persistence. A model-agnostic O4 search for long-lived post-merger spectral lines (GW231226, GW250114) bounds any persistent line to a time-averaged amplitude of ~1.4 × 10⁻²⁴ over minute-long windows — an undamped child line at 6.9% of ringdown peak would exceed this by roughly 50×. The line must therefore decay: persistence beyond ~1 s (a few hundred damping times) is excluded. That retires the “permanent line” reading for good — and it is also the physically natural outcome, since the child’s Hubble frequency itself falls as the child ages, sliding the imprint down and out of band within milliseconds-to-seconds. (2) The bias signature is now computed. Simulating a 6.9% same-frequency component with persistence 2–8× the tone’s damping time and free relative phase, the fitted (2,2,0) damping time shifts by −10% to +19% depending on phase — slightly positive on phase-average (+1 to +8%). The current data read exactly like that: the GWTC-4 remnant tests find a combined upward pull δτ̂₂₂₀ = +0.16^+0.18_−0.16 (pSEOBNR; pyRing +0.18^+0.27_−0.26), with the GR value sitting at the 99.3^+0.7_−4.5% quantile of the hierarchical damping-time posterior before GW250114 softens it to 96.2% — and the LVK’s own zero-noise injection studies could not reproduce quantiles that large from waveform systematics. Meanwhile GW250114 alone reads δτ̂₂₂₀ = −0.01^+0.10_−0.09: a phase-locked, strongly-persistent line is disfavored by that single loud event, while a phase-random line predicts precisely the observed pattern — individual events scattered around small values, a small positive collective mean, excess event-to-event scatter. No significance is claimed — the LVK attributes the residual to possible catalog variance — but the sign and size are right where a 6.9% hidden component would put them. Committed forecast: as the catalog doubles, the combined δτ̂₂₂₀ should settle in the +2 to +8% band with persistent excess scatter if the line is real, and converge to zero if not. First scorecard — GWTC-5.0 (August 2026, 168 events): the joint constraint lands at δτ̂₂₂₀ = +0.07^+0.06_−0.05 (hierarchical population mean +0.09^+0.07_−0.06) — inside the committed band at its upper half — while the frequency deviation stays pinned at zero (δf̂₂₂₀ = 0.01 ± 0.03): precisely the asymmetric signature a hidden same-frequency component produces, where generic modified gravity would move both. The GR value sits at the edge of the 90% region (quantiles 98.1% joint, 98.7% hierarchical-mean), with δτ constraints tightened 1.48× over GWTC-4. Honest counterweights, the LVK’s own: no significance is claimed; unmodeled selection effects (positive-δτ events carry more ringdown SNR) could shift the mean down; the five loudest events alone read +0.02 ± 0.08 (consistent with zero), the positive pull entering as lower-SNR events accumulate — compatible with the phase-random line picture, but equally with noise systematics; and the excess-scatter half of the forecast is only bounded (σ_δτ < 0.10), not detected. The checkpoint stands: convergence toward zero kills the line; persistence in the band with growing significance supports it. (3) A remnant-spin census confirms the degeneracy is the common case, and names the exceptions. Computing final spins (aligned-spin Barausse–Rezzolla) for all 264 cataloged BBH events with published parameters: median a_f = 0.681; among comparable-mass, modest-spin systems (q > 0.7, |χ_eff| < 0.2, n = 114) the median is 0.684 with half the events within ±0.02 of 0.6865 — nature really does pile its remnants onto the degenerate spin. The loud exceptions are the discriminating targets, where the line is frequency-resolved from the tone: GW241127_061008 (SNR 31, a_f ≈ 0.46, line at 1.16 × f₂₂₀), GW231028_153006 (0.90×), GW231123 (0.93×, spin-estimate caveats), GW191109 (1.08×), GW190521 (1.07×) — an archival search list, ready for pyRing-style analyses.
• No gravitational-wave echoes, ever (falsifier locked in, August 2026). The horizon in this picture is a code layer running at capacity: infalling radiation is encoded, not bounced. Horizon reflectivity is exactly zero, so the post-merger signal contains no echoes at any delay — in sharp contrast to firewall, fuzzball, and quantum-membrane proposals, which generically predict them. Status: all four GWTC-4 echo analyses (two template families, two unmodeled searches) are null; model-agnostic O4 searches cap persistent post-merger content at SNR ~ 5–7; and long-duration Bayesian analyses now bound horizon-scale structure to near-Planckian levels with no echo evidence. Every one of these nulls is a passed test of this model and a failed opportunity for reflective-horizon alternatives. A future statistically significant echo detection would falsify this picture outright — the model has no dial to accommodate one. (GWTC-5.0 update, August 2026: residuals across all 168 catalog events remain consistent with detector noise, and the LVK has discontinued dedicated echo searches as unable to improve existing constraints — the nulls stand as the closing word of the O4 era.) (The same nulls are what cap the child-line persistence in the ringdown bullet above — one dataset, two constraints.)
• The parent black-hole mass is set by the recursion. Self-consistency requires M_{parent BH} = c³/(2GH₀) in the parent’s own units. Numerically (in our units), this is ~10⁵³ kg — equal to the total mass-energy of our observable universe. This is not a free parameter: it is determined by the requirement that our cosmological horizon coincide with the parent BH’s Schwarzschild radius.
• A ceiling on black-hole masses in our universe (new falsifier, August 2026). If the mass ratio between adjacent recursion levels is constant — the simplest reading of self-similarity, asserted rather than derived — then running n₀ = 2φ² = 5.236 levels from the parent (9.1 × 10⁵² kg) down to the Planck floor fixes the per-level ratio at (M_parent/M_Pl)^1/5.236 = 3.8 × 10¹¹, and the next rung below the parent — the heaviest black hole our universe should ever produce — is M_max ≈ 1.2 × 10¹¹ M_☉. Status: the record claims are converging on this ceiling from below without crossing it — the Cosmic Horseshoe lens measurement (2025, the most direct method) gives 3.6 × 10¹⁰ M_☉, the July 2026 JWST/HST identification in A402-BCG gives ~5 × 10¹⁰, a claimed binary totals ~6 × 10¹⁰, and the largest indirect estimate (Phoenix A, ~10¹¹) sits at 83% of the ceiling. Honest caveat: conventional astrophysics has its own soft cap — King’s accretion-disc self-gravity limit (~5 × 10¹⁰ M_☉, spin-dependent) — so masses below 10¹¹ do not discriminate. The discriminating regime is above: King’s limit only caps growth by luminous accretion (mergers of two near-cap holes could exceed it), while this ladder forbids any single black hole above ~1.2 × 10¹¹ M_☉ by any formation channel. A robust future measurement of a ≥ 2 × 10¹¹ M_☉ black hole breaks the constant-ratio ladder; continued stalling of the record just under 10¹¹ despite ever-deeper surveys (Euclid, Roman, JWST cavity searches) is the ladder’s signature.

A search direction for the α derivation

Under strict self-similarity, the formula 1/α = (φ^N+2d + φ^N+d − 2φ^d)/d must be derivable from purely structural (frame-independent) considerations — nothing about our specific frame can possibly enter into a universal constant. The derivation, if it exists, depends only on φ, d, the Fano/octonion algebra, and the recursion’s self-consistency at any depth. This narrows the search space considerably for closing the model’s largest open mathematical question.

Quantitative consequences computed

Three quantitative consequences of the BH-interior picture have been worked out:

1. The boundary identity is a timestamp, not a feeding schedule (corrected August 2026)

An earlier version of this section read r_S(parent) = r_H(us) as a standing law and derived a feeding schedule from it: dM/dt = c³/(2G) × |dH/dt|/H² ≈ 10¹² M_⊙/yr today, the parent growing in lockstep with our expansion. Integrating the model’s own committed w(z) forward retires that reading, and the correction is recorded here rather than papered over. Dark energy decays with a 25-Gyr half-life, H falls forever, and the demanded mass M ∝ 1/H grows without bound: the parent would need 1.22× its present mass by t = 20 Gyr, 2× by 51 Gyr, 10× by 168 Gyr, and 725× by matter re-domination at t ≈ 500 Gyr. No environment feeds a black hole 725 times its own mass on cue; the sustained-lockstep reading fails its own far future. What replaces it is standard general relativity, and it is cleaner: for a flat interior behind a horizon of fixed mass, the interior’s Hubble radius c/H sweeps past the Schwarzschild radius exactly once — the time-reverse of the horizon-crossing instant in Oppenheimer–Snyder collapse. (The idealized matching is exact for pressureless content; the dark-energy era adds junction subtleties not fully worked out here.) The identity r_S = c/H is a timestamp — the completion of the birth, the moment the interior’s causal horizon fills its box — and the age identity t₀ = (8/φ³)·GM/c³ is revealed as its crossing-time formula: given the birth mass, the model computes when the box fills. That is the only “today” the equations distinguish, and three clocks say it is this one. (i) H·t rises through 4/φ³ = 0.9443 at t = 13.81 Gyr — 0.18 Gyr from now, inside the measurement error on H₀t₀ itself — touches the value once more at t ≈ 192 Gyr in a universe whose star formation is long over, and never again; the window in which the identity holds as well as it does today is ~0.4 Gyr wide. (ii) With the box frozen in comoving coordinates from birth, the CMB cutoff should sit at the current Hubble radius only in this era — the Planck-fitted 4.9 Gpc against today’s 4.4 Gpc — and never after (the comoving Hubble radius plunges to 0.2× the box by t ≈ 117 Gyr before recovering at t ≈ 300 Gyr): a match flagged as semi-circular under the old reading becomes an independent, if coarser, clock under this one. (iii) The heaviest black holes in our own sky are arriving at the recursion’s ceiling mass now (see the falsifier above). Nothing observable is lost in the correction: exact flatness is a birth condition and is permanent in GR (spatial curvature is a constant of the motion — no maintenance required); the k_min imprint is comoving and survives; every w(z)-based prediction is untouched; and Hawking evaporation of the parent stays irrelevant by more than 120 orders of magnitude — the “lifetime mismatch” worry dissolves even more simply when there is no schedule to keep. Nor does the release announce itself: past the crossing the interior continues as a bag-of-gold geometry behind an unchanged exterior (Birkhoff’s theorem — the outside metric cannot respond), which is also why black holes in our sky do not burst or echo when their own interiors complete — the no-echo commitment above stands, and no new falsifier is created. The expansion-history seismology scan (August 2026) reads even more sharply now: with no feeding schedule at all, H(z) must be smooth. Fitting the 13 DESI DR2 BAO measurements (0.295 ≤ z ≤ 2.33) with the recursion w(z) plus a step of free amplitude and free redshift finds none — best candidate 0.9σ (noise), any discrete jump above 2–3% excluded at 95% confidence throughout 0.35 < z < 2.2 — and, as a by-product, the recursion w(z) fits the 13 points slightly better than ΛCDM (χ² = 9.15 vs 10.53, same parameter count). Finally, the recursion’s own timescales make the timestamp reading self-consistent across levels: each child universe’s entire interior history elapses almost instantly on its parent’s clock (a 10 M_⊙ child completes in ~0.1 ms of our time; even a ceiling-mass child in ~two weeks), so the tower’s active edge — the wave of completions descending through nested generations — always sits inside every ancestor’s present, and every level’s observers, assembled while their level approaches its own crossing, generically find the identity holding “now.” What remains honestly anthropic is the precision: landing within 0.2 Gyr of the exact crossing, given a few-Gyr window of peak observer formation, is ~1-in-20 timing luck. The corrected reading converts three independent coincidences into one event seen three ways; it does not eliminate the event.

2. The recursion rate is derived, not empirical

Setting the natural BH timescale GM_parent/c³ against the per-recursion-level timescale t₀/n₀, the model gives:

The factor 4/φ⁵ = p × n₀ is the same “extraction × depth” product that gives Λ·t₀² = 8/φ³. The cosmological constant identity, the deceleration parameter, and the recursion rate all reduce to the same structural relation — now grounded in the parent BH’s natural timescale rather than treated as an empirical input.

3. Horizon entropy match is automatic

The Bekenstein-Hawking entropy of the parent BH equals the holographic entropy of our cosmological horizon: S_BH(parent) = A/(4ℓ_P²) ≈ 2.3 × 10¹²², identical to S_holo(us) = same surface, same number. They are equal because they are the same surface viewed from opposite sides — consistent with the holographic principle.

4. Hawking radiation from inside: super-horizon modes

The parent BH’s Hawking temperature (from outside) is T_H = ℏH₀/(4πk_B) ≈ 1.3 × 10⁻³⁰ K, and the Gibbons-Hawking de Sitter temperature from inside is exactly twice this (the standard BH/dS relation). The Wien peak wavelength is then ∼ 8–14 × the Hubble radius (depending on whether one uses the wavelength- or frequency-form of Wien’s law) — in either case, super-horizon. This means the Hawking emission cannot appear inside our universe as thermal EM radiation: its wavelength exceeds the size of our observable horizon. Instead, from our internal perspective, it manifests as the cosmic zero mode — a uniform, directionless background that drives the Hubble flow itself. This is consistent with the broader picture: just as the parent’s charged EM (with structure) projects inward as our gravity (monopole only), the parent’s Hawking thermal bath (isotropic) projects inward as our cosmological expansion (isotropic). Hawking evaporation isn’t missing from our universe — it is our universe’s expansion, sampled at super-horizon wavelengths.

5. The cosmological-constant problem dissolves

The largest numerical tension in modern physics is the “cosmological constant problem”: naive quantum-field-theory vacuum energy is of order the Planck density ρ_P ∼ 5 × 10⁹⁶ kg/m³, while the observed dark-energy density is ρ_DE ∼ 6 × 10⁻²⁷ kg/m³ — a mismatch of 10¹²². This is usually framed as the worst fine-tuning problem in physics.

In the BH-interior picture, this problem does not arise. The total mass-energy contained inside a Schwarzschild BH of mass M_p is simply M_pc². If our universe is the interior of such a BH with horizon r_H = c/H₀, the average interior density is:

The interior density of a BH with mass c³/(2GH₀) equals the observed critical density identically, to all digits. The 10¹²² discrepancy only appears if one assumes the universe started at Planck density; the BH-interior picture says it never did. The initial (and asymptotic) density is whatever the parent BH’s mass dictates, not whatever QFT with a Planck cutoff would naively give.

The remaining question — “why does our parent BH have mass 10⁵³ kg rather than 10⁴³ kg?” — is a question about the parent, not a fine-tuning within our frame. Numerically: smaller parent → higher interior density and larger H₀; larger parent → lower density and smaller H₀. All recursion levels see the same dimensionless relations (φ, d, α, H₀t₀), and each frame “sees” its own H₀ set by its parent’s mass. The cosmological constant “problem” is an artifact of imagining a single absolute scale; in the recursion there is no such scale.

6. Hubble tension: the model picks the early-universe side

Because the model predicts a specific dimensionless product H₀·t₀ ≈ 0.939 (from the evolving-w(z) integration), and because the age of the universe can be independently constrained from globular-cluster ages and white-dwarf cooling (t₀ = 13.5 ± 0.3 Gyr), the model predicts H₀ = 68.1 ± 1.5 km/s/Mpc. As of mid-2026 the tension has sharpened, not relaxed, and the data have sorted cleanly into two camps:

The model sits squarely on the early-universe side, and lands almost exactly on the sound-horizon–based inverse-distance-ladder value (DESI BAO+BBN, 68.51 ± 0.58 — a 0.3σ match) as well as the Planck CMB value. The catch: in April 2026 the H0 Distance Network (H0DN) published a community-consensus local measurement of H₀ = 73.50 ± 0.81, combining Cepheids, TRGB, Miras, megamasers, SBF, Tully–Fisher and multiple SN types — and explicitly showed that removing either Cepheids or the TRGB barely moves the central value. That removes the earlier hope that a TRGB-only re-anchoring might pull the local number down to ~69: the local ladder is now robustly high (7.1σ above the early-universe value), and the tension is real rather than a single overlooked calibration error.

So this is now a sharp, binary bet. The model is firmly committed to the early-universe/BAO+BBN side (~67–68.5) and is in ~3.2σ tension with the local distance ladder (~73.5). If the resolution of the Hubble tension turns out to be new late-time physics that lifts the true H₀ to 73, the model’s H₀t₀ relation is wrong. If instead the resolution is early-universe (e.g. a shifted sound horizon) or a still-unidentified local-ladder systematic that brings the local value down toward 68–69, the model is vindicated. There is no longer a comfortable middle: one side of this 7σ split is going to be wrong, and the model has bet on the early-universe side.

7. The CMB scalar spectral index: n_s = 1 − 1/φ⁷

Inflation is the standard answer to why primordial density fluctuations have a near-scale-invariant, slightly red-tilted spectrum n_s ≈ 0.965–0.974. The BH-interior picture offers an alternative: primordial seeds come from fluctuations of the parent-frame Hawking thermal bath at the moment of BH formation, projected inward. The parent BH’s Hawking temperature T_H ∝ 1/M decreases as M grows during formation, giving a natural (and small) red tilt.

Quantitatively, if one e-fold in comoving wavenumber k corresponds to one recursion level during the seed-setting epoch, then each e-fold attenuates power by factor 1/φ⁷ (the per-polarization per-level attenuation, identical to p/d). This gives:

Status (updated August 2026): the tension has substantially relaxed — the deepest new CMB dataset came in on the model’s side. The 2025–26 measurements now stack up as follows (model deviation in parentheses):

• Planck 2018 (TT+TE+EE+lowE+lensing): n_s = 0.9649 ± 0.0042 (0.17σ — a hit).
• SPT-3G D1 + Planck (2026): n_s = 0.9636 ± 0.0035 (0.56σ — the model sits slightly above). SPT-3G’s D1 maps are the deepest ever used in a CMB power-spectrum analysis.
• CMB-SPA: SPT-3G + ACT DR6 + Planck (2026) — the strongest CMB-only constraint to date: n_s = 0.9679 ± 0.0033 (0.71σ — consistent).
• P-ACT (ACT DR6 + Planck, 2025): n_s = 0.9709 ± 0.0038 (1.4σ) — the high outlier among CMB datasets.
• CMB-SPA + DESI BAO: n_s = 0.9728 ± 0.0027 (~2.7σ — the one remaining stress point).

The structure of the disagreement has clarified. Among CMB experiments, the model’s 0.9656 is consistent with everything except the ACT-specific pull: adding SPT-3G’s low reading (0.9636 with Planck) moved the all-CMB combination down to 0.9679, within 0.7σ of the model — the earlier “2.8σ tension with the CMB” was an ACT-driven excursion, not a stable feature of the CMB sky. The remaining 2.7σ appears only when DESI BAO is folded in — and the SPT-3G collaboration itself reports a 2.8σ tension between CMB and DESI-DR2 BAO within ΛCDM, so that combination is averaging over a known dataset-vs-dataset conflict (the n_s shift comes from BAO parameters dragging the fit, not from any measurement of the primordial tilt). Standard inflationary models (Starobinsky, Higgs, T-attractors) sit within ~1σ of the model’s value and face the identical squeeze.

The structural argument behind n_s = 1 − 1/φ⁷ is unchanged: the same exponent 7 that sets p = 2/φ⁷ and counts the Fano degrees of freedom also sets the CMB tilt, and the prediction remains sharp, parameter-free, and inconsistent with exact scale invariance. The decisive test is unchanged and near: Simons Observatory (large-aperture telescope now in initial science operations, unblinded 2025-data analysis in progress; σ(n_s) ≈ 0.002) and eventually CMB-S4 (~0.001). If they settle near 0.974, the prediction is dead at 4σ+. If they settle near 0.966–0.968 — where the current all-CMB combination already points — the model’s single most exposed number survives its sternest test.

8. Horizon, flatness, and monopole problems dissolve without inflation

• Horizon. The BH interior is causally connected throughout; the parent BH formation is the common past of all interior points. CMB isotropy needs no inflationary stretching.
• Flatness. r_S(parent) = r_H(us) is forced; Ω_k = 0 is automatic, not a 10⁻⁶⁰ tuning.
• Monopoles. Our universe never passed through a GUT-energy epoch (initial conditions set by adiabatic BH formation at whatever M_parent dictates, not by an energy spike to M_Planck). No GUT symmetry breaking, no topological defects, no monopoles.
• Low-ℓ CMB suppression. Planck observes quadrupole and octupole power suppressed ~30% below ΛCDM. In the BH-interior picture, modes with wavelength approaching the parent BH’s Schwarzschild radius at formation time are naturally cut off. Inflation predicts no such cutoff; the BH-interior picture predicts one. Now computed through the full transfer function (August 2026): −38%/−24%/−12%/−6% at ℓ = 2–5, vanishing above ℓ ≈ 8 — matching the observed deficit’s size, shape, and localization.

Together with n_s = 1 − 1/φ⁷, this means the BH-interior picture does all the observational work that inflation was invented to do, with zero free parameters, without requiring a scalar inflaton field, and with a sharper prediction for n_s than most slow-roll inflation models give.

9. Fano plane = Steane [[7,1,3]] holographic code

A structural observation not previously noted: the Fano plane is the combinatorial structure underlying the Hamming(7,4) error-correcting code, whose quantum counterpart is the Steane [[7,1,3]] code — 7 physical qubits encoding 1 logical qubit with distance 3 (corrects any single-qubit error). In modern holography, bulk-to-boundary information is encoded via quantum error-correcting codes (HaPPY code, random tensor networks); the Steane code is one of the simplest examples. In the BH-interior picture, the horizon is the boundary and each cell contains 7 Fano degrees of freedom encoding 1 logical “bulk” bit. The Fano-cubic Lagrangian’s G₂-invariant trilinear form is the stabiliser structure of the Steane code. The broader “cosmos as a hologram of entangled qubits with time as a holographic projection” picture — advanced by Stephen Hawking and Thomas Hertog in their 2018 “smooth-exit” paper and Hertog’s 2023 book On the Origin of Time — is the closest published kindred framework. This work makes that picture specific: the qubits are Steane [[7,1,3]] qubits sitting on the parent BH horizon, the recursion is infinite (every interior BH is itself a parent for a child universe), and the dimensionless ratios φ, p, n₀, and 1/φ⁷ are universal constants of the encoding rather than free parameters of any one frame. This gives a concrete proposal for how the holographic principle is physically implemented in the recursion: bulk information in our universe is Steane-encoded onto the parent horizon, with code distance 3 allowing the loss of one Fano DoF without information loss. Consequences: (i) the Bekenstein-Hawking entropy A/(4ℓ_P²) emerges automatically from counting logical qubits on the horizon; (ii) the “1 logical qubit per 7 physical” ratio is the same 1/7 that appears in the p = d/φ^d²+d+1 = 2/φ⁷ extraction fraction; (iii) black-hole information is not destroyed, it is Steane-encoded.

10. Quantum entanglement: the recursion’s native algebra (added June 2026)

The consistency requirement first. A discrete tick substrate could be misread as a local hidden-variable theory — and local hidden variables are experimentally dead: loophole-free Bell tests (2015 onward) and, as of August 2024, the first loophole-free test of Hardy’s paradox (4.3 billion trials; local realism excluded at p < 10⁻¹⁶³⁴⁸) have closed every escape hatch. The model is not a local hidden-variable theory, for a structural reason: in a holographic recursion, distance inside our frame is not distance in the encoding. Two particles far apart in the bulk can be adjacent — or share support — in their Steane-encoded representation on the parent horizon (section 9 above). Nonlocality inside the frame is locality on the boundary. This is the resolution ER=EPR proposes (entangled pairs are geometrically connected), made concrete here: entanglement is shared encoding structure one level down the recursion. The ticks count proper time along worldlines; they are not hidden instructions predetermining measurement outcomes.

The unexpected discovery: entanglement theory already runs on φ. Six independent exact results, previously unconnected to this framework:

• Hardy’s paradox maxes out at exactly 1/φ⁵. The maximum probability of Hardy’s “nonlocality without inequalities” — called the sharpest logical demonstration that entanglement defies local realism — over all two-qubit states and all measurements is (5√5 − 11)/2 = 1/φ⁵ = 0.09017, an exact algebraic identity (re-verified here by independent numerical optimization). The model’s recursion bookkeeping lives on the same ladder: n₀·p = 4/φ⁵. Notably, maximally entangled states give zero Hardy paradox — peak logical nonlocality requires golden-ratio-tuned partial entanglement.
• The minimal universal anyon has quantum dimension exactly φ. The Fibonacci anyon τ — the simplest particle whose braiding performs universal quantum computation — has quantum dimension obeying d² = d + 1: the model’s own fixed-point equation φ² = φ + 1. Its fusion rule τ ⊗ τ = 1 ⊕ τ is self-similar recursion written in Hilbert space (a thing combined with itself yields vacuum plus itself); its state-space dimensions grow as Fibonacci numbers; its topological entanglement entropy comes in units of ln φ.
• Unitarity quantizes φ into quantum mechanics — as a theorem. The model’s claimed origin of φ (“stability + unitarity”) has a rigorous counterpart in entanglement algebra: the Jones index theorem (1983) proves that unitarity restricts the allowed coupling strengths of quantum subsystems below 4 to the exact discrete series 4cos²(π/n) = 1, 2, φ², 3, … — the first nontrivial value unitarity permits is φ² = 2.618. (The model’s recursion depth n₀ = 2φ² is twice this minimal nontrivial index — noted as a curiosity, not claimed.)
• φ has been measured in the laboratory by braiding entangled qubits. In 2024 a 27-qubit superconducting processor realized Fibonacci string-net states and braided their anyons, extracting the quantum dimension φ from fusion statistics; in 2025 a second team recovered φ at 98% accuracy. Together with the 2010 quantum-Ising/E8 experiment (the two lightest excitations of an entangled critical magnet have mass ratio φ), the golden ratio is now a laboratory-measured constant of entangled matter — measured by groups with no connection to this framework.
• The uncertainty principle is golden-quantized for the model’s native particle (July 2026). A Fibonacci qubit admits two incompatible topological-charge measurements (fuse the left pair vs the right pair), and the basis change between them is exactly the F-matrix; their maximum squared overlap is 1/φ, so the entropic uncertainty bound (Maassen–Uffink) is H(A) + H(B) ≥ ln φ = 0.694 bits — ignorance is priced in the same unit ln φ as the τ’s topological entanglement entropy. Beneath it, the angle between the two measurement bases satisfies cos²θ = 1/φ and sin²θ = 1/φ²: the golden partition of unity 1/φ + 1/φ² = 1 — the model’s lossless-scattering identity — is the Pythagorean identity sin²+cos² = 1 of the Fibonacci measurement angle (θ = 38.17°).
• Wave–particle duality is the uncertainty principle — and both are golden here (July 2026). Duality relations were proven equivalent to entropic uncertainty relations in 2014 (Coles, Kaniewski & Wehner), collapsing the two mysteries into one information constraint. For the τ it is golden on the wave side too: the fringe visibility of a τ probe encircling a τ is the monodromy scalar, computed here from the braiding R-matrices as exactly −1/φ² (agreeing with the S-matrix formula to machine precision — two independent routes), with n enclosed τ’s giving visibility φ⁻²ⁿ: exponential which-path decoherence (Bonderson, Shtengel & Slingerland). This 1/φ² per anyon is numerically identical to the embedding theorem’s convergence rate in step 3 of the derivation below: the parent “which-path measures” the cell at one monodromy factor per anyon — ensemble convergence and wave-function decoherence are the same modular process seen from two sides.

Time from entanglement. The Page–Wootters mechanism (1983) shows that in a globally static quantum universe — which is what the Wheeler–DeWitt equation of quantum gravity demands — time emerges from entanglement between a clock subsystem and everything else; this was experimentally illustrated with entangled photons in 2014. It is the quantum-mechanical formalization of this site’s founding thesis: the ticks are not a flowing background, they are the correlation structure from which duration emerges. Likewise Ryu–Takayanagi (entanglement entropy = horizon area) and Van Raamsdonk (cut the entanglement and spacetime disintegrates) make entanglement the thread that stitches the recursion tower together — with monogamy of entanglement following naturally from horizon-area bookkeeping — now made exact: a 7-anyon cell’s total entanglement with everything else is capped at ln 21 nats by its Fibonacci-sized fusion space (see the capacity paragraph below), so pairwise sharing is forced, not postulated.

A path to deriving p = 2/φ⁷ from fusion algebra. When two τ anyons fuse, the outcome probabilities are P(vacuum) = 1/φ² and P(τ) = 1/φ — the golden partition of unity (1/φ + 1/φ² = 1), the very identity the model uses for lossless boundary scattering. Consequence: a τ line that successively absorbs fresh τ’s survives in the τ channel through k steps with probability exactly 1/φ^k. For k = 7 — one step per Fano point of a horizon cell (the 7 physical qubits of section 9) — the survival probability is 1/φ⁷ = 0.0344: precisely the model’s CMB tilt, n_s = 1 − 1/φ⁷. With d = 2 such lines per cell (the two polarization DoF), the expected extraction rate is p = 2/φ⁷ = 0.0689. And the k = 5 chain gives 1/φ⁵ — the Hardy maximum. All three of the model’s key φ-exponents now appear as fusion-chain survival probabilities of the minimal universal anyon.

The derivation, structured (June 2026). Pushing the sketch to a calculation, the chain has five steps, each labeled by its epistemic status:

1. Cell structure [existing model]. A horizon cell is a PG(2,d) block: d²+d+1 = 7 anyonic degrees of freedom plus logical content (the Steane code of section 9). The exponent 7 is the projective-plane point count — it was never a free choice of this calculation.
2. Anyon type [theorem — made canonical July 2026]. The classification of rank-2 unitary modular tensor categories permits exactly two anyon theories: the semion (abelian, computationally trivial) and Fibonacci. Originally the tie was broken by minimality (the model’s max-entropy principle); it turns out no choice was needed, because the model’s own octonion algebra generates Fibonacci canonically. The Fano plane is the multiplication table of the octonions; the octonions’ automorphism group is G₂ — the same G₂ whose invariant trilinear form is the Steane stabiliser structure of section 9 — and the level-1 G₂ WZW theory (central charge 14/5; exactly two primaries, the identity and the 7-dimensional fundamental representation) is the Fibonacci anyon theory, with quantum dimension sin(3π/5)/sin(π/5) = φ exactly. That 7-dimensional representation is the space of imaginary octonions, whose standard basis is the 7 Fano points — so the cell’s 7 anyonic degrees of freedom and the anyon type they carry spring from the single object Aut(octonions) = G₂: the exponent 7 and the base φ were never independent inputs.
3. Ensemble [was an assumption — derived June 2026]. The dimension-weighted (Markov-trace) state, for which each τ × τ fusion yields τ with probability exactly 1/φ, was originally asserted via the model’s max-entropy principle — a real gap, since naive uniform counting over the cell’s 34 fusion-basis states gives 1/34 = 0.0294 for the unbroken-line event, not the model’s number. It is now derived: an exact F-move recoupling computation (unitarity certified to 10⁻¹⁵ state-by-state) shows that when the 7-anyon cell is a subsystem of a larger parent in the honest uniform ensemble with global vacuum charge, the cell’s fusion statistics converge to exactly the Markov-trace state — per-step τ-survival exactly 1/φ = 0.61803399 at every one of the 7 steps, with finite-parent deviations dying off at exactly φ⁻² per parent anyon (converged to 10 decimal places by a parent of 30 anyons). An isolated cell gives a different exact number: φ⁻⁵/2 = 0.0451, off by the factor φ²/2. Two exact cross-links (July 2026): doubling the isolated value over the two chiral sectors gives 2 × φ⁻⁵/2 = 1/φ⁵ — an isolated cell would extract at precisely the Hardy maximum above — and the finite-parent deviations decay at exactly the τ–τ monodromy scalar 1/φ² per parent anyon (verified to six digits), so even the convergence rate of the embedding theorem is modular data. The max-entropy Markov ensemble is therefore not a choice — it is forced by the cell being embedded in a parent. The constant p structurally encodes the model’s core premise: this cell is a subsystem, not a closed universe.
4. Extraction event [physical identification — the remaining gap, now sharpened by computation]. Extraction = an unbroken τ-worldline threading the cell: the mobile line fuses once with each of the 7 cell anyons and must remain τ at every encounter (a vacuum outcome cuts the line). For the embedded cell this transmission probability is now an exact recoupling result, not a sketch: P = (1/φ)⁷ = 0.03444185, independent of fusion-tree topology (any binary fusion tree on 8 lines has exactly 7 internal vertices). The companion lattice computation (below) also excludes the cheapest alternative reading: “extraction” is a fusion-channel (worldline) event, not an edge-configuration event — the all-τ configuration probability on 7-edge lattice paths computes to ≈ 0.10, not φ⁻⁷. What remains unproven is that the physical extraction process is this τ-line transmission, counted as a rate.
5. Multiplicity [physical identification — sharpened June 2026]. Two lines per cell, read as an expected rate: p = 2/φ⁷. The factor 2 was originally identified with the two transverse polarizations; the lattice computation below suggests a structural origin: the cell’s certified topological phase is the doubled (parity-symmetric) Fibonacci theory, which contains exactly two τ-species — one chiral, one antichiral — plausibly the two graviton helicities in different notation. Each transmits at 1/φ⁷; parity symmetry gives them equal rates; the expected extraction rate sums to 2/φ⁷. The data discriminate the rate reading: it gives Ω_DE = (1−p)^n₀ = 0.6882, just 0.13σ from Planck’s 0.6889 ± 0.0056, while the alternative “at least one line succeeds” probability reading (1−(1−φ⁻⁷)²) would give 0.6928, 0.69σ off — observation prefers the rate.

The cell Hamiltonian, built and run (June 2026). The milestone flagged in earlier versions of this section — an explicit Fibonacci string-net (Levin–Wen) Hamiltonian on the Fano incidence graph — has now been constructed and solved exactly. The Fano plane’s incidence graph (the Heawood graph: 14 vertices, 21 edges, trivalent) embeds in a torus with exactly 7 hexagonal faces; on it the full doubled-Fibonacci string-net model was built from the F-symbols and certified against the operator algebra it must satisfy, with nothing tuned: hermiticity, the loop-fusion identity B^τ² = I + B^τ, and commutativity of all 21 plaquette pairs all hold to 10⁻¹⁴–10⁻¹⁶, and the projector identity follows from φ² = φ + 1 itself. Results: (i) the ground space is exactly 4-dimensional — the ground-state degeneracy demanded by doubled-Fibonacci topological order on a torus — so the model’s cell geometry genuinely hosts the anyon theory the derivation requires; (ii) the τ-occupation of every one of the 21 edges is φ²/(2+φ) = 0.723607 exactly — golden-ratio edge statistics to six decimal places, the same number that appears as the embedded cell’s total-charge probability in the recoupling computation, three independent calculations agreeing; (iii) the negative result above: no edge-configuration observable reproduces φ⁻⁷, killing the configuration-space reading of extraction and leaving the worldline reading standing. This is exactly the class of system realized on quantum processors in the 2024–2025 braiding experiments — the cell Hamiltonian is laboratory-realizable physics, not notation.

Net status: the base φ, the exponent 7, the per-step probability 1/φ, tree-independence, and the ensemble are now all forced — the ensemble by the embedding theorem (a cell inside a parent has Markov statistics exactly, an isolated cell does not), which converts the model’s deepest premise into the very origin of its fundamental constant. One genuine physical identification remains: that extraction events are τ-line transmissions counted as a rate. Deriving that final step needs the cell Hamiltonian coupled to a radiation channel — beyond fusion algebra alone. As it stands, the model’s entire cosmological sector (Ω_DE, w(z), n_s are all downstream of p) rests on the entanglement algebra of the holographic code up to that single identification.

Every ingredient of p is rigid modular data (June 2026). A final reduction confirms that none of the numbers entering p = 2/φ⁷ is tunable: rebuilt from the Fibonacci modular S-matrix alone (S unitary to machine zero; the Verlinde formula applied to S reproduces the fusion rule τ ⊗ τ = 1 ⊕ τ exactly, N = 1, 1), the per-step τ-survival is N_ττ^τ·d_τ/d_τ² = 1/d_τ = 1/φ — one inverse quantum dimension per fusion, read straight off S — so the seven-step transmission 1/φ⁷ is modular data, not a coincidence. The factor 2 is the chiral doubling: the certified doubled phase contains exactly two purely-(anti)chiral τ-lines, (τ,1̄) and (1,τ̄), living in chiral sectors of central charge +14/5 and −14/5 (summing to zero, like the two graviton helicities — in CFT terms the doubled phase is (G₂)₁ times its conjugate), each line transmitting at 1/φ⁷. Strikingly, the same doubling sits in the recursion depth: n₀ = 2φ² is twice the minimal nontrivial Jones index (φ² = 4cos²(π/5), the same subfactor quantity), and the product strips to n₀·p = 4/φ⁵ = 4× the Hardy maximum — one modular ladder tying dark-energy bookkeeping, recursion depth, and the entanglement Hardy bound together. (The lattice edge density 0.7236 is likewise exactly S_0τ² = φ²/D², a single squared S-matrix entry — the genuinely Hall-plateau-like rigid occupation.) Honest caveat: p is not a single Hall conductance — no individual modular quantity equals 0.0689; p is a seven-fold product of 1/d_τ times the two-fold doubling. What the quantum-Hall analogy buys is the robustness principle: because p is assembled entirely from rigid modular invariants it is a quantized topological response, which is why the data-preferred “rate” reading (Ω_DE = (1−p)^n₀ = 0.6882, 0.13σ from Planck) is the right kind of object — a response coefficient, not an occupation probability.

Capacity and monogamy: the horizon is a golden-mean channel running at capacity (July 2026). The Fibonacci fusion chain has one grammar rule — a vacuum channel cannot follow a vacuum channel — and that constraint is the golden-mean shift, literally the worked example of a constrained channel in Shannon’s 1948 founding paper of information theory, with channel capacity exactly ln φ per symbol (the allowed channel-strings count 2, 3, 5, 8, 13, 21, … — Fibonacci numbers). Three exact consequences. First, the max-entropy ensemble is optimal coding: the unique maximal-entropy measure of a constrained shift is its Parry measure (Parry, 1964), and computing it for the Fibonacci grammar gives transitions 1/φ and 1/φ² — identical to machine precision with the Markov-trace ensemble that the embedding theorem derives, with entropy rate collapsing algebraically to exactly ln φ per anyon. The chain of upgrades: the max-entropy principle was an assertion, became a derivation (embedding theorem, June 2026), and is now also a named 1964 theorem — the horizon transmits at exactly its channel capacity, ln φ nats per anyon per tick, in the same golden unit as the uncertainty bound and the topological entanglement entropy. (The stationary occupation is again φ²/(1+φ²) = 0.7236 — a fourth independent appearance of that number.) Second, monogamy of entanglement is quantified and forced: a 7-anyon cell’s total entanglement with everything else is capped at ln dim V(τ⁷) = ln 21 = 3.045 nats (Binet form: 8lnφ − ln√5), while naive independent-pair counting gives 7lnφ = 3.368 nats and 7 qubit pairs would need 4.852 — both exceed the ceiling, so sharing is enforced by the Fibonacci-sized Hilbert space rather than postulated; the embedded (Parry/Markov) cell state carries 3.020 nats, 99.2% of the ceiling. The pairwise version — how much two overlapping pairs can simultaneously be known — is exactly the ln φ uncertainty anchor above. Third, an honest negative: the energy side of quantum speed limits (Margolus–Levitin, t ≥ πℏ/2E) is π-flavored and carries no φ structure (the parent-BH ops-per-entropy ratio 4/π²φ³ = 0.096 matches no ladder value) — energy budgets the tick count conventionally; the golden structure lives entirely on the information side. Net: energy budgets the ticks, the grammar prices each tick at ln φ, the parent-embedding installs the capacity-achieving code automatically, and monogamy is the channel’s finite Fibonacci ceiling.

The interacting cell is a critical golden medium (July 2026). Everything above treats the cell’s 7 anyons as static fusion degrees of freedom. If they interact at all, the canonical Hamiltonian is the golden chain (Feiguin et al., PRL 2007) — the anyonic Heisenberg model, where neighboring pairs are favored to fuse into the vacuum — and it was built here from the same F-matrix as every computation above (each local term certified an exact projector; scaled by φ the terms satisfy the full Temperley–Lieb algebra to 10⁻¹²) and solved on rings up to 28 anyons. The known results reproduce with no free parameters: the chain is critical, with central charge extrapolating to 0.69974 against the exact tricritical-Ising value 7/10, and its ferromagnetic partner reaching 0.80017 at the largest ring against the 3-state-Potts value 4/5 — both φ-anchored minimal models emerging from the cell’s fusion algebra alone. The new number: the ground-state energy per anyon converges to −2/φ² to 5 parts in 10⁹ across independent extrapolation windows — equivalently, each neighboring pair in the interacting vacuum sits in the trivial fusion channel with probability exactly 2/φ² = 0.7639. (The 2007 paper and its successors rescale this constant away; it appears to be unpublished. Algebraic golden values are structurally plausible here: the chain maps exactly onto Baxter’s A₄ RSOS model, the family where hard-hexagon-type golden ground-state energies are classical results.) The 7-anyon ring itself is already within 0.5% of the infinite-chain value — the cell is a faithful microcosm of the critical medium. One observed identity is worth recording with its status clearly flagged: the extraction probability factors as p = 2/φ⁷ = (2/φ²) × (1/φ⁵) — the interacting vacuum-channel density times Hardy’s maximum — giving both factors independent physical meanings; but this is noticed, not derived (golden arithmetic makes such factorizations cheap), and it earns promotion only if a mechanism connects the chain’s ground state to the transmission count. Honest caveats: the “7” in c = 7/10 comes from the minimal-model series, not from the cell having 7 anyons; and the ferromagnetic ground-state density (−0.93518…) matched no closed golden form in an extensive search — the exactness is specific to the vacuum-favoring chain. August 2026, the payoff: the same chain answers a foundational objection. Momentum-resolved diagonalization shows its low-energy spectrum has the exact conformal form E_n − E₀ = (2πv/L) x_n with a single emergent velocity v for every state — towers at k = 0 and k = π whose exponent ratio measures 2.650 against the tricritical-Ising 8/3 = 2.667 (0.6% off, converging with ring size), central charge 0.703 against 7/10. A conformal field theory is a fully Lorentz-invariant relativistic theory: at criticality the model’s own matter forgets the lattice entirely, and v plays the role of c. Emergent relativity is thereby demonstrated, not assumed, in 1+1 dimensions on the model’s own dynamics; the higher-dimensional version is logged as an open obligation (Part IV).

The horizon entropy budget: A/4 exactly, no logarithm, and a golden constant (July 2026). Quantum-gravity programs agree that black-hole entropy is S = A/4 to leading order and disagree about the first correction — standard SU(2) loop quantum gravity predicts an extra −(3/2) ln A (Kaul–Majumdar, PRL 2000), the U(1)-gauge-fixed version predicts −(1/2) ln A, and Carlip’s CFT arguments also give −3/2 — so the logarithmic coefficient is a fingerprint that distinguishes theories where the area law cannot. This framework’s horizon — cells carrying Fibonacci charges under the golden-mean grammar certified above — was counted here exactly (integer arithmetic, rings up to 800 cells, three ensembles: free charge, vacuum-sector-projected, and species/area-fluctuating). The result is rigid: the entropy is strictly linear in area, with no logarithmic correction at all (fitted coefficient < 10⁻¹¹; on the ring the count is a Lucas number, φⁿ + (−φ)⁻ⁿ, so even the power-law tail is absent — corrections die exponentially). The only correction is a universal constant: projecting the horizon onto its charge-neutral (vacuum-fusion) sector costs exactly ln(φ√5) = 1.2859, and φ√5 = 1 + φ² is precisely the total quantum dimension D of the doubled-Fibonacci phase — the deficit is the topological entanglement entropy −ln D, arriving here by an independent route (Binet asymptotics of the state count vs. quantum dimensions). The committed prediction: S = A/4 − ln(φ√5), with each cell occupying 4 ln φ ≈ 1.925 Planck areas and contributing exactly its channel capacity ln φ — Bekenstein–Hawking as “number of cells × capacity,” locking this paragraph to the capacity result above. Two independent checks land in the same place. First, the published literature: the −3/2 logarithm arises only in the k → ∞ classical-group limit; keeping the Chern–Simons level finite — a quantum group with finitely many charge types, which is exactly what a fusion category like Fibonacci is — the log term vanishes and the entropy is strictly proportional to area (Majhi & Majumdar, 2011–12). The counting here reproduces their conclusion from a different lattice. Second, the consistency rhyme: in loop gravity a finite level is the statement that the cosmological constant is positive (k ∼ ℓ_Λ/ℓ_P) — and this framework already asserts Λ > 0 because the parent is finite. The same structural fact that sets Λ also removes the logarithm; the model had no freedom to choose otherwise, since Fibonacci is finite by construction. Honest status: the vanishing logarithm is not original here (it exists in the two low-profile papers above); what is added is a non-gravitational reason the horizon theory is a finite category, a definite value for the constant term where the literature leaves it unspecified, and the capacity lock. No near-term measurement resolves log corrections; this is a theoretical fingerprint, decisive only against other quantum-gravity programs.

Fault tolerance: the leak rate sits below the code’s measured threshold (August 2026). Encoded quantum information survives indefinitely only if the physical error rate stays below the code’s error-correction threshold — the fault-tolerance theorem, a rigorous result from quantum information theory. The recursion faces this constraint literally: the dimensionless constants (φ, d = 2, and everything built from them) must be inherited exactly by every child level, which means the horizon code has to propagate its encoded information through the extraction noise, generation after generation, without degradation. The extraction fraction p = 2/φ⁷ = 6.89% per level is therefore not only a dark-energy parameter — it is the error rate the code must survive. And the threshold has been computed, for exactly this code: Schotte, Zhu, Burgelman & Verstraete (Phys. Rev. X 12, 021012, 2022) computed the first error-correction thresholds for the universal Fibonacci Turaev–Viro code — the doubled-Fibonacci string-net phase, the same one built on the Heawood cell above — finding 4.7% under depolarizing noise and 7.3% under pure dephasing. The model’s 6.89% lands inside the computed band, 6% below the dephasing ceiling. Precision honesty: identifying a per-level energy extraction fraction with a per-cycle error rate is an interpretation, not a theorem; published thresholds are decoder-dependent lower bounds (better decoders raise them); and extraction is structured, known loss — in error-correction terms heralded erasure, whose thresholds sit far above Pauli-noise thresholds (≈50% vs ≈11% for the surface code), so the physically apt ceiling is likely higher still. Two statements survive the caveats: (i) a necessary condition nobody engineered is satisfied — the tower’s leak rate is below its own code’s computed correctable threshold, so the constants can ride the recursion indefinitely without an information catastrophe; and (ii) the familiar pattern recurs — the channel runs at capacity ln φ, the ensemble is the maximal-entropy measure, and the leak sits just under the correctability ceiling. Every operational budget in this model is spent to its edge.

Independent convergences (June 2026). Three results from unconnected groups landed in June 2026, each touching the picture at a different seam. (i) A peer-reviewed gravastar formation model (Jampolski & Rezzolla, Phys. Rev. D) shows a collapsing cloud settling against an interior de Sitter region — “essentially an expanding mini-universe” — with the singularity replaced by a Planck-scale bounce: an independent, GR-based realization of collapse → interior dark-energy universe, the exact structure assumed in the “classical singularity” open issue below. Its Israel junction conditions are a concrete route to derive the currently-asserted coefficient in t_0,child = (8/φ³)·GM/c³: pure-de-Sitter matching (L_dS = r_S) forces coefficient 2, while the model’s 8/φ³ = 1.889 is a 5.6% reduction — the matter correction. (ii) Alexander, Bernardo & Hui (Phys. Rev. Lett.) show the cosmological constant in Chern–Simons–Kodama quantum gravity is topologically quantized like the quantum-Hall conductance — placing Λ in precisely the basket this section places p in, via the same Chern–Simons mathematics that the doubled-Fibonacci string-net realizes. (iii) The Bose–Marletto–Vedral (BMV) experiment, targeted for the early 2030s, tests whether gravity entangles two masses — i.e. whether a mediating quantum link (a Wilson/τ-line) is exchanged. This is the physical content of step 4 — “extraction = τ-line transmission” — named and made experimentally addressable: gravitationally-induced entanglement is the line exchange. Its companion post-selected “antigravity” (repulsion in one measurement branch that vanishes on ensemble average) is the microphysics of this framework’s dark-energy non-harvesting: real per-branch, unextractable from the ensemble. A clean BMV result that gravity is classical would, conversely, strike at the entanglement-substrate reading built here — a genuine near-term falsification frontier.

Honest open issues

• The classical singularity. Classical general relativity predicts a singularity at the center of a Schwarzschild BH. The picture requires the singularity to be replaced by a smooth bounce or by quantum-gravity structure. This is consistent with the standard expectation that quantum gravity resolves classical singularities, but it has not been demonstrated here.
• r_S = r_H is consistency, not proof. Cosmic flatness implies r_S = r_H in any cosmology, including standard ΛCDM (which does not invoke a parent BH). The identity is consistent with the BH-interior picture but does not by itself force it.
• The picture is interpretive. The recursion mathematics works without committing to any specific physical picture of the parent. This section adds physical interpretation that is consistent with the math but not derived from it.

What this is

This is a constrained numerical framework — a set of algebraic relationships that reproduce cosmological observables from two structural inputs (φ and d = 2) plus two dimensionful anchors (the fine-structure constant α at proton scale, and the Hubble constant H₀). All dimensionless cosmological ratios — Ω_DE, Ω_m, Ω_b/Ω_c, w(z), q₀, Λ·t₀², H₀t₀, S₈, n_s — follow from φ and d = 2 with no further freedom. It predicts the late-time expansion history via w(z), eases the S₈ tension, gives the clean dimensionless identities Λ·t₀² = 8/φ³ and n_s = 1 − 1/φ⁷ (the CMB scalar tilt — matches Planck 2018 at 0.17σ and the strongest 2026 CMB-only combination, SPT+ACT+Planck, at 0.7σ; the residual stress is confined to CMB+DESI-BAO joint fits, an active falsification frontier), predicts q₀ = −0.395, forces v_GW = c structurally, reproduces gravitational-wave energy loss and frame-dragging, preserves BBN and Lorentz invariance, commits to H₀ ≈ 68 km/s/Mpc (matching the DESI BAO+BBN early-universe value at 0.3σ and Planck; in ~3.2σ tension with the local distance ladder) via the H₀t₀ constraint, and ties the arrow of time, anti-gravity impossibility, and dark-energy non-harvesting to a single algebraic fact (octonion non-associativity) — but it is not yet a full dynamical theory: it does not provide nonlinear field equations needed for the CMB power spectrum or GW merger waveforms, the α-formula is a structured pattern rather than a derivation, and its locked w(z) carries one sharp quantified tension against the CMB acoustic scale (the squeeze, tracked with survival corridor and death modes in What Remains Open). An interpretive overlay (Part III.5) identifies the parent frame as a black-hole interior under strict self-similarity, which makes α and H₀·t₀ universal constants of the recursion rather than per-frame inputs, makes cosmic flatness automatic, dissolves the cosmological-constant fine-tuning problem (the 10¹²² discrepancy is an artifact of assuming a single absolute scale), resolves the horizon/flatness/monopole problems without inflation (with n_s as a by-product), identifies the holographic encoding as the Steane [[7,1,3]] quantum error-correcting code, and adds three independent falsification tests (zero global rotation, ringdown signatures, α inside BHs).

The distance between “a framework that reproduces numbers” and “a theory that replaces GR + ΛCDM” is vast. General relativity is constrained by solar-system tests, gravitational time delay, frame-dragging, binary-pulsar timing, and LIGO/Virgo waveform matching. This framework does not yet compete in any of those dynamical regimes.

What it gets right

• A single conceptual framework (recursive frames + emergent time) connects time dilation, the dark sector composition, the EM/gravity hierarchy, and the fine structure constant
• The formula G = (k_ee²/m_p²) × α(m_p)^4φ³ reproduces Newton’s constant to 0.2%, given the standard atomic constants (e, m_p, α)
• All three cosmic energy fractions (dark energy, dark matter, ordinary matter) match Planck data within 1%, with ordinary matter at 0.03% accuracy
• The fine-structure constant α(m_p) admits a structured closed form 1/α = (φ^N+2d + φ^N+d − 2φ^d)/d, with the three exponents on a two-step graded ladder generated by N (Fano) and d (polarization); the coefficient pattern (1, 1, −2) is not yet derived from first principles but the algebraic structure is sharply constrained
• The factor of 2 is explained by EM polarization states, and d = 2 is uniquely selected by five independent arguments: (1) the observed α matches only at d = 2; (2) EM has exactly 2 transverse polarizations; (3) the Fano-cubic coupling is marginal at D = 2d+2 = 6 (upper critical dimension) only for d = 2; (4) the Fano plane PG(2,2) requires d²+d+1 = 7, forcing d = 2; (5) Hurwitz’s theorem makes d = 2 the largest integer compatible with the normed-division-algebra structure
• Two independent derivations of φ: KAM stability (most irrational number) and scattering unitarity (unique self-similar lossless barrier)
• The exponent 7 in p = 2/φ⁷ is explained as transverse field DoFs by tensor rank (1+2+4), connecting to the Fano plane and octonion algebra
• The charge structure problem (EM screened, gravity not) is resolved by Kaluza–Klein-like projection across 7 compact DoFs
• The extraction fraction p = 2/φ⁷ is derived from a 7-field Fano-plane Lagrangian whose cubic coupling is marginal at D = 6 = 4 + d, connecting unitarity (g² = 1/φ) through 7 Fano vertices to give p = d × (1/φ)⁷
• The dark energy equation of state w₀ = −0.867 is consistent with DESI DR2’s evidence for evolving dark energy (within their 2σ contour), determined by φ and d = 2 alone — and the analysis-robust core of that signal, the pivot value w(z_p ≈ 0.34) = −0.9 ± 0.1, matches the model’s −0.892 at 0.08σ (Aug 2026) — and the same w > −1 behavior dissolves the 2025–26 “negative neutrino mass” anomaly (a ΛCDM bookkeeping artifact: evolving-DE analyses restore a positive, oscillation-consistent Σm_ν)
• The model forbids phantom energy (w < −1), making a clean prediction against the standard CPL fit at z > 1
• S₈ ≈ 0.802 is now a bracketed prediction (July 2026): the two final lensing surveys straddle it — DES Y6 at 0.789 ± 0.012 (1.1σ below the model) and KiDS-Legacy at 0.815 (0.6σ above) — while the Combined-CMB ΛCDM inference (0.836) sits above both, and the tightest all-probe joint fit (DES Y6 + CMB + BAO + SN + clusters, Jan 2026) lands at 0.806^+0.006_−0.007 — 0.6σ from the model. RSD growth data are a mild model preference at its own anchors (χ² 9.5–9.9 vs 10.6 over 13 points, CAMB) and a mild deficit at the CMB-preferred corner, where S₈ itself drops to 0.77–0.79; Euclid DR1 (Oct 2026) decides, the same release that tests w(z) at z > 1
• The CMB shift parameter R — now computed with a full Boltzmann pipeline (August 2026): R = 1.734, a 3.6σ deficit that does not ease with the full w(z) (the same run confirms the σ₈ suppression at −2.5% and the low-ℓ cutoff pattern, and reproduces Planck’s ΛCDM exactly as a control). If S₈ and w₀ are confirmed, R must be ~1% below the ΛCDM extraction — and the resulting acoustic-scale squeeze is now the model’s sharpest open tension (see What Remains Open), though a mock-universe test (Aug 2026) converts part of the threat into a retrodiction: ΛCDM fitters applied to synthetic model-universe data reproduce the real world’s SN-high-Ω_m (0.35–0.36) and BAO-vs-CMB H₀ (68.2 vs 67.3) discordances
• Gravitational wave speed equals c by construction (gravity is the zero-mode of parent-frame EM), matching the GW170817 bound |v_GW/c − 1| < 10⁻¹⁵ with no free parameters — killing-grade falsification passed
• Gravitational wave energy loss (Hulse–Taylor binary) is reproduced to ~0.25% accuracy via the Weinberg theorem (spin-2 + G + speed c), without any model-specific tuning
• The deceleration parameter q₀ = −0.395 is a sharp supernova-testable prediction, sitting between Pantheon+’s ΛCDM fit (~−0.51) and DESI CPL (~−0.34)
• The dimensionless identity Λ·t₀² = 8/φ³ ≈ 1.889 ties the cosmological constant, age of universe, and Hubble rate together via φ alone — matching the Planck-extracted value 1.869 to 1.1%
• The Fano cubic coupling in the Lagrangian is literally the imaginary-octonion trilinear form, making G&sub2; symmetry automatic rather than coincidental
• The five independent arguments for d = 2 now include Hurwitz’s theorem: d = 2 is the maximum integer supporting the octonion/Fano-plane structure (sedenions at d = 3 have zero divisors)
• The arrow of time, impossibility of anti-gravity, and impossibility of dark-energy harvesting are unified: all are manifestations of octonion non-associativity (recursion chains cannot be re-bracketed)
• Frame-dragging (Lense–Thirring precession) is automatic: the parent EM B-field projects to a gravitomagnetic field with the correct Bg/Eg = v/c² ratio
• BBN is unaffected: at z ∼ 10⁹ the DE/radiation ratio is ~10⁻³² — BBN predictions (helium, deuterium, lithium abundances) are identical to ΛCDM to 1 part in ~10²⁸, well beyond any conceivable measurement
• Lorentz invariance is preserved: the beat counts proper time (a Lorentz scalar), and the committed prediction (August 2026) is exact nulls for all vacuum-dispersion, Planck-jitter, and preferred-frame searches — the tick is in-band invisible; even the naive worst-case leakage estimate η ~ 10⁻⁶¹ would sit 38 orders of magnitude below experimental bounds
• Structure growth fσ₈(z) is suppressed by up to 5.2% relative to ΛCDM near z ≈ 0.4 (declining at higher and lower z; CAMB pipeline, Sept 2026), giving χ² = 9.5–9.9 vs ΛCDM’s 10.6 across 13 RSD measurements (6dFGS, SDSS, BOSS, WiggleZ, eBOSS, VIPERS, FastSound) at the model’s own anchors — a mild preference that becomes a mild deficit (12–13) at the CMB-preferred low-H₀ corner; full discrimination expected from DESI Y3 / Euclid
• Under the BH-interior interpretation (Part III.5), α and H₀·t₀ become universal constants of the entire recursion tower rather than properties of our specific frame — cosmic flatness becomes automatic, and three new falsification tests appear (zero global rotation, ringdown signatures encoding child-universe structure, α identical inside any BH)
• The recursion rate t₀ = (8/φ³) × GM_parent/c³ is derived (not empirical) from the parent BH’s natural timescale — one recursion level lasts (4/φ⁵) × GM/c³ ≈ 0.36 × BH-time, the same dimensionless factor 4/φ⁵ = p·n₀ that appears in Λ·t₀² = 8/φ³
• Boundary identity corrected (August 2026): the earlier reading — a parent accreting ~10¹² M_⊙/yr to hold r_S = r_H — fails its own far future (the demand grows to 725× the parent’s mass by matter re-domination). Corrected reading: for a flat interior behind a fixed-mass horizon, c/H sweeps past r_S exactly once (the time-reverse of Oppenheimer–Snyder horizon crossing), so the identity is a timestamp and the age identity t₀ = (8/φ³)·GM/c³ is its crossing-time formula. The crossing is now: H·t passes 4/φ³ at t = 13.81 Gyr (0.18 Gyr away; sole recurrence t ≈ 192 Gyr), the CMB-fitted cutoff sits at today’s Hubble radius (4.9 vs 4.4 Gpc), and the heaviest black holes are arriving at the ceiling mass. Flatness (birth condition, permanent), the k_min imprint, and all w(z) predictions unchanged; Hawking evaporation stays irrelevant by >120 orders of magnitude; no exterior signature (bag-of-gold interior, Birkhoff). The “why now” question reduces to one event seen three ways, plus ~1-in-20 anthropic timing
• Dark energy is a transient (August 2026): the committed w(z) is algebraically equivalent to exponential decay of ρ_DE in cosmic time at the recursion’s own drain rate — half-life ≈ 25 Gyr. Dominance peaks ~99.9% near t ≈ 150 Gyr, the accelerated era totals only ~70× of expansion, matter re-dominates by ~500 Gyr — no Big Rip, no eternal de Sitter, no permanent event horizon, no heat-death isolation (and no swampland conflict)
• Horizon entropy match is automatic: S_BH(parent) = S_holo(us) ≈ 2.3 × 10¹²², because they are the same surface viewed from opposite sides
• Ringdown sub-dominant modulation is sharply predicted: f_child/f_QNM = φ³/(8·Mω₂₂₀(a_f)), amplitude p = 2/φ⁷ ≈ 6.9% — 1.417 × for non-spinning remnants, but ≈ 1.00 × (degenerate with the dominant QNM) at the universal equal-mass remnant spin a_f ≈ 0.686 — a spin now proposed to be thermodynamically selected (maximum-entropy conjecture, published Phys. Rev. Lett. 137, 021406, July 2026), born 0.8% inside the stable side of the Kerr Davies point, making the degeneracy enforced rather than unlucky — where the exact Kerr eigenvalue is Mω ≈ (φ³/8)·(1 − i/4φ) — frequency match 0.5%, quality factor Q = 2φ to 0.11%; clean offset-line tests require remnants with a_f far from 0.686 (e.g. 1.27 × at a_f ≈ 0.28), definitive at LISA. The GWTC-4 damping-time ledger (Aug 2026) is consistent: the combined δτ̂₂₂₀ = +0.16 ± 0.17 upward residual — which LVK injection studies cannot attribute to waveform systematics — has the sign and size of the phase-averaged bias a 6.9% hidden component produces, while long-lived-line nulls now cap the line’s persistence below ~1 s; forecast committed: combined δτ̂₂₂₀ settles at +2 to +8% if the line is real, converges to zero if not — first scorecard (GWTC-5.0, Aug 2026): joint +7⁺⁶₋₅%, frequency deviation pinned at zero, GR at the edge of the 90% region; LVK flags unmodeled selection effects as a possible mundane cause. A 264-event census puts the median remnant at a_f = 0.681 (half of comparable-mass events within ±0.02 of the degenerate spin), with named off-degeneracy targets (GW241127: line at 1.16 × f₂₂₀) for archival searches. Echo commitment locked: horizon reflectivity is exactly zero — all GWTC-4 echo searches null as required, any confirmed echo falsifies the picture
• Parent BH Hawking radiation has Wien peak at ~10× the Hubble radius (super-horizon), so it cannot appear as EM inside our universe; instead it manifests as the cosmic zero mode driving Hubble expansion — consistent with the broader “parent isotropy projects to our uniformity” pattern
• Primordial gravitational waves r ∼ 10⁻¹²¹ (adiabatic parent BH formation, no inflationary epoch at M_Planck scales) — a sharp falsification target for LiteBIRD/CMB-S4 at r ∼ 10⁻⁵
• Per-tick Fano-event rate from a sketched Planck-cell Hamiltonian, g·√(ρ/ρ_P), evaluates today to within factor ~3 of the macroscopic rate (4/φ⁵)/(N_ticks·t_P) — non-trivial consistency check, though full QFT derivation remains open
• Max-entropy + unitarity + non-associativity framing provides a suggestive single-principle umbrella for (φ, d = 2, N = 7), though the three arguments remain algebraically independent
• The speed of light is not an empirical constant but a structural definition: c = ℓ_P/t_P (one Planck length per Planck tick). Its constancy across frames follows from Lorentz invariance of per-worldline tick accounting; massive particles slow because their tick budget is split between internal oscillation and spatial advancement
• Quantum entanglement carries the same algebra (Part III.5 §10): Hardy’s paradox maxes at exactly 1/φ⁵, the minimal universal anyon has quantum dimension φ (measured by braiding entangled qubits, 2024–25), unitarity quantizes φ² as the first nontrivial Jones index, the uncertainty principle and wave–particle duality (proven equivalent in 2014) are both golden-quantized for the Fibonacci anyon — entropic bound ln φ, interferometric visibility 1/φ² per enclosed anyon, the embedding theorem’s own convergence rate — and the extraction fraction p = 2/φ⁷ now follows from fusion algebra up to a single physical identification — the exponent 7 and base φ spring from a single object (the 7-dimensional representation of G₂ = Aut(octonions), whose level-1 WZW theory is the Fibonacci anyon — July 2026), tree-independence and the per-step probability 1/φ are forced, the last by an exact embedding theorem (June 2026): a cell inside a parent has Markov-trace statistics exactly (per-step survival 1/φ at every step), while an isolated cell would give φ⁻⁵/2 — the constant p encodes the premise that our universe is a subsystem. The explicit Levin–Wen cell Hamiltonian on the Fano graph was built and solved: ground-state degeneracy exactly 4 (doubled-Fibonacci topological order certified), edge statistics φ²/(2+φ) exact, and the doubled phase’s two chiral τ-species offer a structural origin for the factor 2; the ensemble is also optimal coding — the Markov-trace state is the Parry measure of Shannon’s golden-mean channel, so the horizon transmits at exactly capacity ln φ per anyon, with cell monogamy capped at ln 21 nats by the Fibonacci-sized fusion space; and if the cell’s anyons interact (golden chain), the vacuum is a critical medium whose pair vacuum-channel density is exactly 2/φ² (July 2026) — the observed factorization p = (2/φ²)×(1/φ⁵) noted, not derived. Counting these horizon cells exactly gives the black-hole entropy budget (July 2026): S = A/4 strictly — no logarithmic correction (vs. −(3/2)ln A in standard loop quantum gravity), matching the published finite-level quantum-group result and rhyming with Λ > 0 — with a universal constant deficit −ln(φ√5) equal to the topological entanglement entropy, and each cell of area 4 ln φ ℓ_P² contributing exactly its channel capacity ln φ. The code also passes a fault-tolerance check (August 2026): the first computed (Monte-Carlo, decoder-dependent) error-correction thresholds for the Fibonacci Turaev–Viro code (Schotte et al., PRX 2022: 4.7% depolarizing, 7.3% dephasing) bracket the per-level extraction rate p = 6.89%, which sits below the dephasing ceiling — and extraction is heralded loss, whose thresholds run higher still — so the recursion’s leak is below its own code’s correctable ceiling and the constants can propagate down the tower indefinitely
• Two scales of substrate “page flips” coexist consistently: Planck ticks at ~10⁴³/s define c locally, and recursion levels at ~1 per 2.6 Gyr define dark-sector evolution. The ratio — ~10⁶⁰ Planck ticks per recursion level — matches the per-tick attenuation ~10⁻⁶¹ derived from p·n₀ = 4/φ⁵, unifying the microscopic and cosmological clocks
• The cosmological-constant problem dissolves under the BH-interior picture. The interior density of a Schwarzschild BH of mass c³/(2GH₀) equals the observed ρ_crit identically. The 10¹²² Planck/observed discrepancy is an artifact of assuming a single absolute scale; in the recursion the initial density is set by the parent BH mass, not by M_Planck⁴. What survives as a “why” question is the parent’s mass, which is a property of the level above ours — and that mass now has an observational fingerprint: the closed-form identity Λ·t_Pl² = 0.51·(M_Pl/M_parent)² reproduces the observed 2.9 × 10⁻¹²², and the parent’s finite size independently predicts the primordial cutoff k_min = 1/r_S = 2.3 × 10⁻⁴ Mpc⁻¹ that published fits to the CMB low-ℓ anomalies recover at 0.3σ (zero cutoff excluded at >8σ; see Part II) — and the cutoff’s axis-dependence now doubles as a pre-registered spin-meter for the parent: ~0% if born by direct collapse, 10.3% if merger-born at the universal spin (the favored band), ~37% if accretion-grown to the Thorne limit; LiteBIRD reads the dial
• 3+1 spacetime dimensions are predicted structurally. Observing d = 2 transverse EM polarizations forces D_space = 3 (since D_space − 1 polarizations for a massless vector in D_space dimensions). The Fano-cubic theory is upper-critical at D = 6, so the 2 missing dimensions are the compact internal fiber that carries the 7 Fano DoFs per point
• Hubble tension: the model bets on the early-universe side. H₀·t₀ = 0.939 (from evolving w(z)) combined with independent age estimates t₀ = 13.5 ± 0.3 Gyr predicts H₀ = 68.1 ± 1.5 km/s/Mpc — a 0.3σ match to the DESI DR2 BAO+BBN inverse-distance-ladder value (68.51 ± 0.58) and consistent with Planck CMB (67.2). The April 2026 H0DN community-consensus local measurement (73.50 ± 0.81, robust to dropping either Cepheids or TRGB) sits 3.2σ above the model. So this is a binary bet: the model lives if the tension resolves on the early-universe/BAO side (~68), and dies if true late-time new physics lifts H₀ to ~73
• Holographic encoding identified as Steane [[7,1,3]] / Hamming(7,4) code. The Fano plane IS the incidence structure of the Hamming(7,4) classical code, whose quantum version is the Steane [[7,1,3]] code — 7 physical qubits encoding 1 logical qubit with distance 3. This proposes a concrete physical implementation of the holographic principle in the recursion: bulk information Steane-encoded on horizons, with the 1:7 ratio matching the p = 2/φ⁷ extraction fraction
• CMB scalar spectral index n_s = 1 − 1/φ⁷ = 0.9656 from the BH-interior picture (Hawking-seeded primordial fluctuations with per-recursion-level attenuation 1/φ⁷ per polarization). The same exponent 7 that sets p = 2/φ⁷ and counts Fano DoFs also sets the primordial-power-spectrum tilt — a second sharp numerical tie-in to φ, alongside Λ·t₀² = 8/φ³. Forbids exact scale invariance (n_s = 1). Status (Aug 2026): matches Planck 2018 (0.9649 ± 0.0042) at 0.17σ and the strongest CMB-only combination, CMB-SPA (SPT-3G + ACT + Planck): 0.9679 ± 0.0033, at 0.7σ — the 2025 ACT-driven 2.8σ excursion did not survive SPT-3G’s deeper maps; residual ~2.7σ stress only in CMB + DESI-BAO joint fits, which carry a known internal CMB-vs-BAO conflict. Simons Observatory will settle it
• The horizon, flatness, and monopole problems dissolve without inflation. Horizon problem: the BH interior is causally connected through the parent BH formation event, no inflationary stretching needed. Flatness: Ω_k = 0 is automatic from r_S = r_H, not a 10⁻⁶⁰ fine-tuning. Monopoles: our universe never passed through a GUT-energy epoch, so no topological defects are produced. Combined with the n_s derivation, the BH-interior picture does all the observational work inflation was invented to do, with zero free parameters. Low-ℓ CMB quadrupole/octupole suppression (unexplained in ΛCDM) is now quantitatively expected: the parent’s finite size forces a primordial cutoff at k_min = 1/r_S = 2.3 × 10⁻⁴ Mpc⁻¹, and published Planck fits (Melia & López-Corredoira) find exactly such a cutoff at 2.04^+1.4_−0.79 × 10⁻⁴, excluding zero at >8σ (details in Part II)
• Gravitational-wave polarization content is structurally fixed at 2 tensor modes. Because gravity is the zero-mode of parent-frame EM and EM has d = 2 transverse polarizations, GWs carry exactly 2 tensor modes, no scalar or vector components. Same as GR in content, but derived from the d = 2 structural input rather than assumed. Pulsar timing arrays + LISA+LIGO networks can resolve all 6 possible polarization modes; any detection of scalar or vector GW content would falsify the model
• α running consistency check. Model predicts 1/α(m_p) = 134.89; standard QED running from ~1 GeV to m_Z (well-measured via e+e− data and dispersion relations) gives a ~5.1-unit shift, landing at 1/α(m_Z) ≈ 129.8 vs observed 128.9 — a 0.7% residual, within the standard ~1% hadronic-vacuum-polarization uncertainty in the running. A non-trivial consistency test that could have failed

What remains open

• The CMB acoustic-scale squeeze — the model’s sharpest tension, quantified by a full Boltzmann run (August 2026). The model’s w(z) is non-crossing and amplitude-locked (1 + w₀ = 0.133, no dial), and dark energy denser in the past shortens the comoving distance to last scattering by ~0.9%. Run through CAMB (control: the pipeline reproduces Planck’s ΛCDM exactly, R = 1.7500 vs measured 1.7502 ± 0.0046), the model at Planck parameters gives R = 1.734, a 3.6σ deficit — the earlier hope recorded here that the full w(z) shape would ease the simplified-fit 3.1σ toward 2σ was wrong: the tension does not ease. The acoustic-scale framing is sharper still: θ* is measured to 0.03%, and because Ω_m = 0.312 is φ-locked the model cannot slide along the usual degeneracy — its least-bad corner without new early physics is H₀ ≈ 66.3 with ω_m ~4% (≈5σ) below the peak-morphology value plus residual BAO strain: a combined squeeze of roughly 5σ-equivalent if current concordance values hold. The structural asymmetry deserves emphasis: DESI’s own evolving-DE fits reconcile w₀ > −1 with θ* via the phantom crossing at z ≈ 0.4–1 — exactly the feature this model forbids. Both faces of the model’s w(z) are true at once: its low-z shape matches the analysis-robust pivot (w(0.34) = −0.892, 0.08σ), while the integral of that same shape out to z = 1090 misses the measured distance unless the concordance bookkeeping (ω_m, H₀, r_s) is off in a correlated way at the few-σ level. The squeeze lives precisely where cosmology’s own unresolved internal conflicts live — the 7σ Hubble split, the 2.8σ CMB-vs-DESI-BAO ΛCDM tension, the S₈ bifurcation, the negative-Σm_ν artifact — and the model requires those disputes to resolve against the current Planck-conditional values. Survival corridor, stated for the record: DESI DR3 finds no robust phantom crossing, the pivot stays near −0.9, and the concordance bookkeeping resolves toward one of the two corners quantified here — H₀ ≈ 66.3 with strained ω_m, or (the localization below) peak-consistent ω_m with r_d ≈ 145 Mpc and H₀ ≈ 67.7, the lower half of the model’s own 68.1 ± 1.5 band. Death modes: a confirmed phantom crossing (fast kill, already committed above), or Simons Observatory + DESI DR3 reconfirming the {θ*, ω_m, BAO-H₀} triangle at current values — strangulation by ~2028–30, with no single dramatic falsifier required. Honest caveats: this is a fixed-slice computation, not a full MCMC over the complete likelihoods; R and θ* compressions are mildly ΛCDM-conditional; and the junction subtleties flagged in the boundary matching (Part III.5) are now load-bearing — if the horizon-era boundary condition modifies the effective w(z) at z ≳ 1, the distance deficit shrinks. Deriving that junction is no longer optional cleanup; it is the model’s most urgent theoretical task. Where the miss localizes — and two failed escapes (late August 2026; this passage retains its own same-week correction). Expressed as physics, the squeeze first appeared to be the statement that the sound horizon ΛCDM computes (r_d ≈ 147.1 Mpc) is ~1.3% too long for the model’s late-time distances — suggesting the miss lived in the one imported, never-derived segment of the pipeline (recombination-era physics), the same quantity the mainstream’s shorter-sound-horizon Hubble-tension proposals target. That reading was tested against the compressed Planck likelihood within a day and refuted; the retraction is kept visible here. A joint fit to the Planck 2018 distance priors (R, ℓ_A, ω_b with covariance) plus all 13 DESI DR2 BAO measurements gives Δχ²(model − ΛCDM) = +13.6 with the model running one parameter leaner — and granting the model extra early physics does not help at all: freeing N_eff (≥ 3.044) or Y_He leaves the fits pinned at their physical boundaries with Δχ² unchanged, because the residual concentrates in the shift parameter R = √ω_m·H₀D_M/c — the one CMB combination containing no sound horizon. Sound-horizon surgery reshuffles ℓ_A, H₀, and r_d but leaves R untouched: the early-universe escape is closed. (Compression-dependence, for the record: +13.6 uses the literature-validated (R, ℓ_A, ω_b) priors; imposing the peak-morphology ω_m as an independent Gaussian gives ~+19; the full Planck likelihood, run below, resolved the spread upward to +38.) The junction escape fares little better: freezing the drain above a free junction redshift buys only Δχ² = 2.4 (best z_j ≈ 1), and the bookkeeping-consistent delayed-drain variant prefers no junction at all — because the distance deficit is generated at z ≲ 1 by w₀ = −0.867 itself. The model’s flagship number carries the squeeze; nothing upstream can absorb it. A third escape, tested September 2026, also fails: reading “fuel converted into structure” literally as energy flowing from dark energy into matter (a coupled dark sector at fraction f of the drain) worsens the fit monotonically — f = 0.05 costs Δχ² = +7, f = 0.10 costs +16 — because transferred energy lowers the early matter density the CMB peaks pin and tilts the BAO expansion history beyond DESI’s ~1% precision. The committed reading (the reservoir’s decline is expansion work; matter dilutes normally) is the only viable one, and Parts I–II now say so explicitly. The full-likelihood tier, delivered (September 2026). The model was finally run through Planck’s native 2018 likelihoods (plik-lite TT/TE/EE at ℓ = 30–2508 plus the low-ℓ TT and EE likelihoods, with the standard calibration prior), maximized with CAMB carrying the exact w(z) table. Control: ΛCDM converges to Planck’s own best fit (−2lnL = 1003.0; H₀ = 67.3, 100θ* = 1.04109). The locked model reaches −2lnL = 1041.0 — +38.0 on Planck alone, two parameters leaner — at H₀ = 66.5, with the acoustic angle still 1.9σ high, ω_b pulled 2.7σ high to shorten the sound horizon, and σ₈ = 0.76. Three diagnostics locate the damage. (i) The lock is the entire cost: the same w(z) with Ω_m and n_s unlocked fits Planck as well as ΛCDM (+1.0) at H₀ = 64.3, Ω_m = 0.347 — the CMB does not object to the recursion’s dark energy, it objects to Ω_m = 0.312 given that dark energy; the φ-locked fraction pays every point. (ii) Roughly a third of the bill is Planck’s lensing anomaly: ΛCDM gains 5 by floating A_lens to ~1.1, the model gains 17 by floating to ~1.2; compared at each one’s preferred A_lens the gap is ≈ +26 (the A_L headwind below, made quantitative). (iii) The compressions undershot: the distance-prior estimate (+13.6) missed the ~25 that the low-σ₈ corner forced by the locked fractions loses in lensing power the Planck temperature spectrum wants. Fully joint with DESI DR2 BAO and the SN shape block: Δ(−2lnL) = +32.0 (1054.9 vs 1022.8 — the model’s BAO and SN blocks are slightly better than ΛCDM’s; the CMB block is far worse). Independent cross-check on ACT DR6’s foreground-marginalized spectra (ℓ = 600–6500, Planck low-ℓ EE for τ, ~1% calibration priors), which do not carry Planck’s lensing anomaly: ACT-only the model trails by +24.8, but ACT alone leaves ΛCDM free to sit at a degenerate corner (H₀ = 66.0, Ω_m = 0.34) that BAO forbids; jointly with BAO and SN the A_L-independent verdict is Δ(−2lnL) = +13.2 (584.9 vs 571.8; ΛCDM at H₀ = 68.4, the model at 66.5). Honest status of these numbers: they are maximum-likelihood optimizations, not full posteriors; the lite likelihoods marginalize foregrounds; SPT-3G is the one remaining CMB cross-check. But the verdict is now referee-grade: against current data the locked model is disfavored on the cosmology block at Δ(−2lnL) ≈ +32 with Planck (a 5σ-class deficit if Planck’s lensing anomaly is physical, ~4σ-class if it is instrumental) and +13 with ACT in Planck’s place. What genuinely moves the verdict is the third probe: supernovae — and here the ledger shifted twice in ten days (passage updated August 30, 2026). The mock-universe test above showed the model retrodicts SN-high Ω_m: ΛCDM fitters applied to a recursion universe read 0.35–0.36 from the supernova window. When first written, the real compilations read Pantheon+ 0.334 ± 0.018, DES-SN5YR 0.352 ± 0.017, Union3 0.356 ± 0.027, and the three-probe accounting gave Δχ² = +11.8 (Pantheon+) but only +4.7 (DES-SN5YR): the verdict hung on the supernova-calibration dispute. That dispute has since resolved against the favorable branch: the DES Dovekie recalibration moves DES to Ω_m = 0.330 ± 0.015 (converging with Pantheon+) and drops the evolving-DE preference from 4.2σ to 3.2σ (“weak” in Bayesian terms), while Vincenzi et al. trace the Pantheon+/DES offset to Malmquist-bias and host-galaxy modeling rather than calibration error. Rerun with the current landscape: Δχ² = +12.9 (DES-Dovekie), +11.8 (Pantheon+), +9.3 (Union3), the model one parameter leaner throughout (ΔAIC ≈ +7 to +11). Both honest directions: the SN-high ordering the model retrodicts (apparent Ω_m: SN 0.33–0.36 > CMB 0.315 > BAO 0.2975) survives intact, but the predicted amplitude (0.35–0.36) now sits ~1.5σ above the converged compilations — the mock overshoots unless Union3 is the accurate one. The BAO-vs-CMB H₀ split retrodiction (apparent 68.2 vs 67.3; real world 68.5 vs 67.4) is unaffected, and DESI’s BAO-alone Ω_m = 0.2975 remains 1.9σ unexplained. Survival corridor, tightened accordingly: the “DES-vindicated, ΔAIC ≈ +3” branch is retired; the cosmology block stands at Δχ² ≈ +9–13 against current central values, and survival requires the remaining levers to move — DESI DR3 BAO drifting toward the model’s corner, no phantom crossing, Union3-style SN amplitudes prevailing, and Planck’s lensing anomaly proving instrumental (which alone would return ~12–15 of the full-likelihood deficit). Death modes unchanged: a confirmed crossing kills fast; Simons Observatory + DESI DR3 + the settled SN calibration reconfirming today’s central values close the corridor by ~2028–30. Caveats: the compressed BAO and SN blocks are ΛCDM-family-validated approximations (standard practice — DESI’s own papers use the same compression), the DESI DR2 numbers are as tabulated here, and the full Planck likelihood above is now the referee — it reads harsher than every compression. The deepest honest sentence available: the same locked w(z) that wins the pivot, S₈, the neutrino repair, and the SN-ordering retrodiction pays Δ(−2lnL) ≈ +32 on the full Planck+BAO+SN block at current values (+13 with ACT in Planck’s place), and which side of that ledger the universe is keeping is not yet known.
• CMB lensing amplitude — the headwind, now computed (September 2026). The growth suppression that eases S₈ also lowers the CMB lensing power: through the CAMB pipeline at Planck parameters the model’s lensing-potential spectrum is 3.0% below ΛCDM’s (A_L ≈ 0.97, not the 0.95 estimated earlier), and at the corner the Planck likelihood actually prefers for this w(z) (σ₈ ≈ 0.76) it is ~15% below. Planck’s high-ℓ temperature spectrum pulls the other way — the long-known lensing-smoothing anomaly, A_L > 1 at ~2σ, which ACT and SPT do not reproduce. Quantified against the native Planck likelihood: ΛCDM gains 5 in −2lnL by letting A_lens float to ~1.1, the locked model gains 17 by floating to ~1.2 — so roughly 12–15 of the model’s Planck deficit (see the squeeze above) is this lensing headwind, and the S₈-friendly side of the model is the A_L-unfriendly side. If Planck’s A_L excess is a systematic (as ACT/SPT suggest), that part of the deficit evaporates; if it is real, the model owes 15% more lensing than its growth history supplies. Simons Observatory’s lensing reconstruction decides.
• GW merger/ringdown waveforms. The inspiral phase matches GR automatically (Weinberg theorem: spin-2 + G + c), and the Hulse–Taylor decay is predicted to ~0.25%. But merger and ringdown probe nonlinear field equations. Deriving these from the parent-frame EM + KK reduction is the main dynamical challenge.
• Emergent relativity — the frame problem, one dimension closed (August 2026). A discrete substrate updated tick by tick has, on its face, a preferred frame — while the entire content of relativity is that no experiment reveals one. The model’s commitment is that Lorentz symmetry is emergent: the update rule is local (each cell from its neighbors — the cell Hamiltonian of Part III.5 is built from strictly local projectors; nothing global happens per tick), and the long-wavelength physics of local rules can be exactly relativistic. Status, honestly split: (i) 1+1D — demonstrated on the model’s own matter. Momentum-resolved diagonalization of the interacting Fibonacci chain shows the low-energy spectrum takes the exact conformal form E_n − E₀ = (2πv/L)x_n: one emergent velocity for all states (v ≈ 1.82 in chain units, playing the role of c), tower exponents at the tricritical-Ising values (k = 0 to k = π gap ratio 2.650 vs 8/3 = 2.667), central charge 0.703 vs 7/10. A conformal field theory is fully Lorentz-invariant: an observer made of chain excitations can run no experiment that finds the lattice. (ii) 2+1D gapped sector — invariant by construction. The horizon code’s low-energy theory is doubled Chern–Simons, a purely topological theory with no propagating modes and hence no dispersion to get wrong; and trivalent (honeycomb-class) lattices additionally protect rotational isotropy of any emergent continuum to high order — the same symmetry fact that made hexagonal (but not square) lattice gases flow to isotropic hydrodynamics, and that gives graphene its emergent massless Dirac cones. (iii) 3+1D gapless sector — open. Deriving the photon/graviton light cone from the bulk rule — one shared limiting speed for all species, with corrections suppressed beyond the Planck-linear bounds already excluded by GRB timing — is the unsolved step, the same obligation carried by every discrete program (causal sets, loop quantum gravity, cellular automata). The 1+1D result matters because it is not an analogy: the mechanism — criticality erasing the lattice — is demonstrated on this model’s own Hamiltonian, and what remains is dimensional, not conceptual.
• Entanglement. The “shared timing” explanation of entanglement is a form of superdeterminism. It is logically consistent but does not yet make predictions distinguishable from standard quantum mechanics, and is constrained by loophole-free Bell tests.
• Particle masses. Known mass ratios (proton/electron, W/proton, Planck/proton, lepton generations) do not show clean φ-power or PSL(2,7)-irrep patterns. A systematic search at 0.2% tolerance over products of irrep dimensions (1, 3, 3̄, 6, 7, 8) times φⁿ finds no consistent scheme across multiple ratios. One near-match is suggestive — m_μ/m_e ≈ N²·φ^d+1 within 0.4% — but it doesn’t extend to τ/μ or quark sectors. This strongly suggests the golden ratio governs inter-frame recursion, not intra-frame particle physics, and consequently m_p acts as the model’s mass-scale anchor and is not itself derived.
• Non-perturbative fixed point. The RG derivation shows the Fano-plane theory is marginal at D = 6, and unitarity independently fixes the coupling to g² = 1/φ. But confirming that the Fano theory’s actual non-perturbative fixed point at D = 4 matches 1/φ requires lattice simulation or conformal bootstrap with PSL(2,7) symmetry — a calculation that has not yet been done.
• The α formula. The closed form 1/α(m_p) = (φ^N+2d + φ^N+d − 2φ^d)/d is verified to 0.012% against the G identity. The exponent structure is sharply constrained: a two-step graded ladder built from the two physical generators N (Fano) and d (polarization), with the three terms sitting at lattice positions (a, b) = (0,1), (1,1), (1,2) where the exponent is aN + bd. What remains open is the coefficient pattern (1, 1, −2): why these specific weights? A loop-expansion or graded-representation-theory argument is the most promising path. Until it closes, α acts as the second dimensionful anchor (alongside H₀). A brute-force uniqueness audit (August 2026) tightened the target: enumerating all ~10⁵ three-term golden expressions (c₁φ^e₁ + c₂φ^e₂ + c₃φ^e₃)/2 with |c| ≤ 3 and exponents up to 14, the G-anchored value is matched by 16 — essentially all algebraic rewritings of one number (φ² = φ + 1 makes rewritings cheap) — and restricting exponents to the physical lattice aN + bd leaves exactly one: the published formula. Within the model’s own exponent grammar the expression is unique, not one face in a crowd; and the commitment here is that α is ultimately derivable — a level-invariant constant has nowhere to live in a self-similar tower except the recursion’s own structure. An August 2026 reduction shrinks the target: since φ² + 1 = φ√5 = D² (the doubled phase’s squared total quantum dimension), the formula is identically 1/α = (D²φ^N+d − dφ^d)/d — the coefficient pattern (1, 1, −2) dissolves into “quantum-dimension weight minus polarization vacuum term.” In Binet form, 2/α = (269 + 121√5)/2 with Fibonacci part 121 = 11² = (N+2d)², and (1, 1, −2) is the unique small-coefficient pattern on the physical rungs whose Fibonacci part is that perfect square. What remains to derive is one question, not three: why the horizon’s D² weights the Fano–polarization rung.
• Full CMB power spectrum — done (September 2026). The model runs through CAMB with its exact w(z) against the native Planck 2018 likelihoods (plik-lite TT/TE/EE, low-ℓ TT, low-ℓ EE) and ACT DR6’s CMB-only likelihood: distances, the low-ℓ cutoff signature (−38%/−24%/−12%/−6% at ℓ = 2–5, preferred by the low-ℓ likelihood by 1.5), linear growth, lensing amplitude, and peak-by-peak fits are all in hand — results in the squeeze entry above and the lensing entry below. What remains: an SPT-3G cross-check, and full posterior sampling in place of maximum-likelihood optimization.
• Primordial black holes as terminating child universes. Because every BH in our frame is itself a parent for a child universe (with cosmic age t_0,child = (8/φ³)·GM/c³), a primordial BH evaporating today corresponds, from inside, to a child universe reaching its endpoint. The model permits but does not require primordial BHs. An earlier 2025 UMass Amherst proposal that the 220 PeV KM3-230213A neutrino is a primordial-BH explosion remnant has since been strongly disfavored: 2026 multimessenger analyses (PRL 136, 041002) show that a standard 4D Schwarzschild PBH burst near enough to be detected (~10⁻⁵ pc) should have produced ~10⁸ LHAASO gamma-ray events and hundreds of preceding IceCube/KM3NeT neutrinos — none seen. Only exotic variants (quasi-extremal PBHs with hidden U(1) charge) remain viable. The recursive picture is agnostic about which scenario is correct, but loses KM3NeT-230213A as a candidate child-universe Hawking-termination observation. Whether the recursion adds a distinguishable spectral signature on top of any future PBH signal (e.g., a child-universe ringdown imprint suppressed by ~p = 2/φ⁷) is calculable but has not been computed. One endpoint entailment is now computed (August 2026): the entropy ledger S = A/4 − ln(φ√5) reaches zero at A = 4 ln(φ√5) ≈ 5.1 ℓ_P² — a three-cell horizon of ≈ 0.32 M_Pl (~7 micrograms) — so evaporation cannot proceed smoothly to nothing. The final flash must terminate discretely, stepping down the last few golden-quantized area rungs (ΔA = 4 ln φ ℓ_P²) rather than diverging: from inside, the child universe completes; from outside, “the end of a black hole” is a finite, well-defined event. Unobservable at current burst-search sensitivity, but it is a forced corollary of the entropy budget, not an option.
• CP violation and baryon asymmetry — speculative lead. Octonion non-associativity provides a natural CP-violating structure via the associator [a,b,c] = (ab)c − a(bc), which is a non-trivial trilinear 3-form on the imaginary octonions. Order-of-magnitude estimate: the Jarlskog invariant should scale as J ~ α/φ¹⁰ ≈ 6 × 10⁻⁵, within a factor of 2 of the observed |J| ≈ 3 × 10⁻⁵. The observed baryon-to-photon ratio η ~ 6 × 10⁻¹⁰ is roughly (J) × (g) × (dilution), suggestive but not tight. A proper derivation requires coupling the Fano-cubic associator to a concrete Sakharov-condition calculation (out-of-equilibrium electroweak-scale dynamics), which has not been attempted. August 2026 sharpening: with the model’s α(m_p) = 1/134.89, the estimate’s missing O(1) factor lands on d: J = α/(2φ¹⁰) = 3.01 × 10⁻⁵ against the measured 3.08 ± 0.13 × 10⁻⁵ — 0.5σ. Flagged as observation, not prediction: golden arithmetic makes post-hoc matches cheap, and the exponent 10 = 5d sits on the polarization-only lattice with no mechanism yet. A decisive experiment exists: the kaon golden channel K_L → π⁰νν̄ is proportional to J², and KOTO II is being built to measure it to ~35% in the 2030s.
• Planck-scale Fano-tick derivation — partial progress. The macroscopic relation p·n₀ = 4/φ⁵, divided across ~10⁶⁰ Planck ticks per recursion level, gives a per-tick attenuation of ~5 × 10⁻⁶². A sketch of the Fano-cubic Hamiltonian on a single Planck cell (7 scalar fields with cubic coupling g² = 1/φ) gives a seeded-event rate ≈ g·√(ρ/ρ_P), which today evaluates to ~10⁻⁶² per tick — within a factor of a few of the required value. This is non-tautological: the rate is density-dependent (higher in the early universe) and is a genuine prediction of the Hamiltonian, not a relabeling. However, a proper time integration overshoots by ~17×, indicating the simple sqrt-density scaling breaks at early times (as expected: needs UV cutoff, proper scattering kinematics). A full derivation requires: (i) specifying which scattering process dominates at each epoch, (ii) handling the Planck-epoch cutoff, (iii) matching to p·n₀ = 4/φ⁵ from first principles. The Hamiltonian skeleton exists; the full QFT calculation is the remaining work.

Isn’t the golden ratio stuff just numerology?

This is the most important question to ask. Five things separate this from pure pattern-matching: (1) the KAM theorem provides a rigorous dynamical reason why φ governs stable recursive structures; (2) scattering unitarity independently requires φ — it is the unique number where a self-similar barrier is lossless (1/φ² + 1/φ = 1); (3) the number 2 is identified as the EM polarization count d, and d = 2 is the only integer that produces a physically viable α; (4) the exponent 7 = d²+d+1 counts transverse field DoFs by tensor rank (1+2+4), connecting to the Fano plane and octonion algebra; and (5) α itself is derived from φ via the DE/DM = φ² constraint, not assumed. That said, the framework still lacks dynamical equations, and the gap between “a well-motivated numerical framework” and “a validated physical theory” is real.

How is this different from other “time isn’t real” ideas?

Most proposals that time is emergent are philosophical or limited to quantum gravity contexts. This model makes quantitative predictions: specific values for the dark sector composition, a formula for α, and a formula for G. It can be falsified by precision cosmology (DESI, Euclid, LSST) and by increasingly accurate measurements of fundamental constants.

If gravity is EM from a parent frame, why is it 10³⁶ times weaker?

Because it crosses ~5.2 recursive frame boundaries, each of which attenuates EM by a factor of α^2φ ≈ 10^−6.9. After 2φ² boundaries, the total attenuation is α^4φ³ ≈ 10⁻³⁶. The 2 in the exponent comes from EM having 2 polarization states.

Does the model violate known physics?

No. The recursion rate (dn/dt ~ 10⁻¹⁷ Hz) is 61 orders of magnitude below the Planck frequency, and the model’s committed prediction is exact nulls for Lorentz-violation searches: every instrument is built from the same ticking substrate it would use to detect the tick, so there is no independent reference against which the tick could register (even the naive dimensional estimate of leakage, η ~ 10⁻⁶¹, would sit 38 orders below experimental bounds). More fundamentally, the beat counts proper time along each worldline (a Lorentz scalar), so there is no preferred frame for local physics. The cosmic-time dependence n(a) defines a preferred foliation (like the CMB rest frame), not a local violation of Lorentz symmetry. BBN light element abundances are also unaffected: at z ~ 10⁹, the recursion depth is n ~ 10⁻¹⁵ and the model is indistinguishable from ΛCDM.

Doesn’t Bell’s theorem rule out a deterministic tick-based universe?

It rules out local hidden variables, and loophole-free experiments (Bell tests since 2015; the first loophole-free Hardy-paradox test in 2024) have closed the question experimentally. But the model is not a local hidden-variable theory. The ticks count proper time along worldlines; they are not hidden instructions that predetermine measurement outcomes. More importantly, the substrate is holographic: two particles far apart in our frame can share encoding structure on the parent horizon, so “spooky action at a distance” in the bulk is ordinary adjacency in the code — the resolution ER=EPR proposes, made concrete by the Steane-encoding picture (Part III.5, sections 9–10). The model is fully compatible with standard quantum mechanics, including Tsirelson’s bound 2√2 on quantum correlations — already saturated by d = 2 systems, the model’s single structural input.

What would kill this model?

Several things, with two currently active: (1) Precision measurement confirming w = −1 exactly at all epochs, or finding w < −1 (phantom energy) robustly at any epoch — the model predicts w₀ = −0.867 and forbids phantom crossing; (2) a measurement of G that diverges from the formula beyond ~0.5%; (3) a precise QED calculation showing α(m_p) differs from 1/134.89 by more than a few percent; (4) a precision baryon fraction measurement inconsistent with the DE/DM = φ² partition; (5) CMB-S4 confirming R = 1.750 at high precision (the model’s full-Boltzmann value, computed August 2026: R = 1.734); (6) a supernova measurement of q₀ that is incompatible with −0.395 at high confidence; (7) a future GW measurement showing v_GW ≠ c at any level (the model forces v_GW = c exactly); (8) confirmation that the true H₀ ≈ 73 km/s/Mpc via genuine late-time new physics with t₀ > 13 Gyr, contradicting H₀·t₀ = 0.94 (the model bets on the early-universe side: it matches DESI BAO+BBN at 68.5 and Planck at 67.2, but the April 2026 H0DN local consensus of 73.5 ± 0.81 is robust and 3.2σ away); (9) watched closely: a precision CMB measurement of n_s inconsistent with 0.9656 — the 2025 ACT-driven excursion to 0.974 (~2.8σ) did not survive SPT-3G’s deeper maps: the strongest 2026 CMB-only combination (CMB-SPA) reads 0.9679 ± 0.0033, 0.7σ from the model, and the residual ~2.7σ stress appears only in CMB+DESI-BAO joint fits that average over a known internal conflict; Simons Observatory (σ ≈ 0.002) will settle it — a settled value near 0.974 still kills the prediction; (10) a detection of scalar or vector gravitational-wave polarization (the model forces GWs to be exactly 2 tensor modes, from d = 2); (11) a confirmed nonzero uniform cosmic birefringence angle — the parity-symmetric horizon phase forbids any, so the current hint (now 4.8σ statistical: β = 0.277° ± 0.057°, joint ACT+Planck, Aug 2026; 3.5σ dust-robust) must be entirely instrumental and regress to zero under absolute calibration; (12) the slow one: the CMB acoustic-scale squeeze hardening — the full-likelihood verdict (Sept 2026) reads Δ(−2lnL) = +38 against ΛCDM on Planck alone and +32 jointly with BAO and supernovae (+13 with ACT DR6 in Planck’s place), roughly a third of the Planck figure being its lensing anomaly; if Simons Observatory, DESI DR3, and a settled SN calibration reconfirm today’s central values while excluding a phantom crossing, the model’s locked, non-crossing w(z) has no remaining corner (see What Remains Open) — strangulation rather than a single shot, on a ~2028–30 clock.

What about the cosmological constant problem?

The usual framing — “why is Λ 10¹²² times smaller than the Planck-scale QFT estimate?” — doesn’t apply in this model. Under the BH-interior picture, the total mass-energy inside our universe equals the parent BH’s mass M_p, and the average interior density M_p/V_H is ρ_crit identically. The universe was never at Planck density; the initial condition is set by M_p, not by a QFT cutoff. The remaining question — “why does our parent BH have this particular mass?” — is a question about the parent frame, not a fine-tuning within ours.

Does the model need inflation?

No. Inflation was invented to solve three problems (horizon, flatness, monopoles) and to generate a slightly-red-tilted primordial power spectrum. In the BH-interior picture: the horizon problem dissolves (the interior is causally connected through the parent BH formation); flatness is automatic (r_S = r_H); no GUT epoch means no monopoles; and the scalar spectral index works out to n_s = 1 − 1/φ⁷ ≈ 0.9656. The same exponent 7 that sets the dark-energy extraction p = 2/φ⁷ also sets the CMB tilt — a non-trivial structural tie-in. Status August 2026: matches Planck 2018 (0.9649 ± 0.0042) at 0.17σ and the strongest CMB-only combination, CMB-SPA (SPT-3G + ACT + Planck: 0.9679 ± 0.0033), at 0.7σ — the 2025 ACT-driven upward excursion did not survive SPT-3G’s deeper maps, and the only remaining stress (~2.7σ) is against CMB + DESI-BAO joint fits that carry a known internal conflict. Standard inflationary models (Starobinsky, Higgs, T-attractors) sit at the same value; Simons Observatory (now observing, σ ≈ 0.002) will likely decide for us all by 2027.

What does the model say about the Hubble tension?

The model predicts the dimensionless product H₀·t₀ ≈ 0.94 from the evolving w(z). Combined with independent age measurements (globular clusters, white-dwarf cooling) giving t₀ ≈ 13.5 Gyr, this forces H₀ ≈ 68 km/s/Mpc — so the model commits to the early-universe side. It matches the sound-horizon–based inverse-distance-ladder value beautifully: DESI DR2 BAO+BBN gives H₀ = 68.51 ± 0.58 (0.3σ from the model) and Planck CMB gives 67.2. The opposing camp has hardened: the April 2026 H0DN consensus puts the local ladder at 73.50 ± 0.81 and shows the high value is robust to removing any single distance indicator (including TRGB), so it is not a simple calibration error. The 7σ early-vs-late split is therefore real, and the model is wrong if it ultimately resolves toward 73 via genuine late-time physics rather than toward 68 via the early-universe/BAO route.

Is this a theory or a speculation?

Currently, it is closer to what a physicist would call a “constrained numerical framework” or a “speculative research program.” It reproduces observables but does not yet derive them from deeper axioms via dynamical equations. The next step — formulating equations of motion that can predict the CMB power spectrum, gravitational waveforms, and other precision tests — is what would promote it from framework to theory.

Fundamental inputs

Derived quantities (from φ and d)

Observable predictions