The Clock Program

*A proposal, not a claim. Everything below is a measurement roadmap — the mechanism story is plausible, the two cruxes are open, and the distinguishing signatures stand whether or not the calibration ever lands. (Measured on the archived meso3i engine — the moore26 record is Moore26 Findings.)

The mechanism

Two lattice effects, each replacing "curved time" with local accounting:

Velocity — the longer hop path. Carriers are rate-limited (one hop per tick). A bound structure translating across the lattice must run its internal cycle along a diagonal chase-path, so each internal tick of the object costs more substrate ticks. This is the C_max² = C_s² + C_i² bandwidth budget made physical: hops spent on translation are hops denied to internal cycling.

Altitude — the congestion gradient. Near mass, node traffic is dense and internal cycles run into transit drag (IFRIC and friends); in relaxed space they run at full rate. Gravitational time dilation becomes a statement about local node load, not metric curvature.

Both effects compose exactly the way the real data does: the Hafele–Keating flights (1971) measured −59 ns eastbound and +273 ns westbound — the east/west asymmetry that forced GPS onto the non-rotating frame plus a Sagnac patch. In substrate terms the ground was already moving; westbound clocks crossed fewer nodes. ISS (−25 µs/day, velocity-dominated), the crossover orbit (~3,178 km, where drag relief and hop-path cost cancel), and GPS (+38 µs/day net, congestion-relief-dominated) are the same two terms with different signs.

The two open cruxes — honestly

Crux 1: does the engine's dilation curve actually equal γ?sqrt(1 − v²/c²) follows only if a bound structure's internal cadence dilates by exactly the hop-path budget. Our lumps are dissipative, far-from-equilibrium objects — not ideal light clocks. They could dilate by γ, by a steeper curve, or not at all. This is measurable headless, and it gates everything.

Crux 2: the radial law. Flux conservation across concentric shells gives a congestion field ~1/r². GR's clock law needs 1 + Φ/c² with Φ ∝ 1/r. If local lag is proportional to local congestion, the altitude correction has the wrong radial dependence — a falsifiable mismatch with GPS data, not a refinement of it. Whether the substrate's congestion imprint (and the lag-vs-density transfer function) can reproduce 1/r to first order is the open question. It is a measurement, not a derivation.

The experiment ladder (all headless, no engine changes)

• E1 — twin-clock drift. Two identical lumps; one boosted, one stationary. Compare internal cadence (radiation turnover, winding precession, dwell counters). Does translation dilate, and by what curve?
• E2 — the lag law. Effective transit rate vs. local occupancy: the v_eff = f(ρ) transfer function the whole altitude story stands on.
• E3 — the congestion profile. Node occupancy vs. distance around a mass lump and around an active corridor — the radial exponent that decides whether Crux 2 can even be asked.
• E4 — the corridor clock. A wound lump parked beside a high-throughput flux corridor vs. an identical lump in quiet vacuum. If active traffic lags the internal cycle where GM/r is unchanged, that's the Part-3 signature — inside the model, before any claim about real clocks.

The dimensional bridge — two constants only

Like GR, we import two measured constants rather than deriving them:

• 1 hop/tick = c — locks the velocity term's geometry exactly. If E1 confirms γ, the kinematic dilation is exact to all digits by construction.
• surface congestion = GM/rc² — calibrates the lag coefficient to Earth's surface clock rate (6.95 × 10⁻¹⁰).

The "17th decimal place" is reachable only in the conditional sense: if E1 returns γ and E2+E3 produce a 1/r-compatible lag law, the calibration interpolates GR's own verified numbers. That is not a new prediction — it is consistency. The program's real deliverable is what comes next.

The distinguishing signatures (Part 3)

Where the substrate disagrees with 1 + Φ/c² regardless of calibration:

• Active-corridor lag. In GR, a quiet mass and a furiously radiating one of equal M dilate clocks identically. In the substrate, throughput chokes node bandwidth — an optical clock inside a high-flux corridor should run slow where Einstein says the term is zero. (Testable inside the model first: E4. Strontium optical clocks already resolve ~1 cm of altitude, ~10⁻¹⁸.)
• Spin-wake asymmetry. A spinning body carries a directional channel imbalance; prograde and retrograde transits should lag differently — far above frame-dragging scale if the M42/M50-class winding holds in the wake.
• A saturation floor. Node congestion ceilings (IREFL) cap clock lag and density. Nothing in this model runs to infinity — the divergences of the continuum are finite ceilings here.

The pitch is not "more accurate than relativity." It is: relativity has a zero where the substrate has a knob — and the knob is measurable.

First measurements (the record so far)

E3 (clock-profile, commit f373ef7). A steady central source against boundary drains, 56³, 600t settle: shell-averaged moving density falls as ~r⁻²·⁰⁻²·⁶; a parked lump's halo reads ~r⁻²·⁰. The transport regime is ballistic — which is what the ontology predicts. Mass generates a radial momentum-transfer field; a ballistic field falls as 1/r², and 1/r² flux density is inverse-square force. A diffusive 1/r halo would have been the bad result — it cannot produce inverse-square force from ray transport.

This corrects the framing this document originally carried: the comparison baseline is observation (GPS, Hafele–Keating), not GR's functional form. Field and potential are different objects — the potential is what the field integrates to along the path. The substrate does not need its congestion to resemble Φ; it needs its force field to look like g ∝ 1/r², and it does.

The re-posed Crux 2. A sitting clock lags from local congestion ρ(r) ∝ 1/r² — but a clock's internal cycle traverses the field, so its lag may integrate along its own path through the gradient (∫ r⁻² → 1/r), which is what the GPS number requires. Whether dilation is local-density or path-integrated is now precisely what E2 measures: the transfer function between local congestion and internal lag.

E3-dense. Cranking the source 16× does not shift the regime: slope stays ~r⁻²·⁰⁻²·¹ at ~10× the density. The inverse-square halo is structural — carriers follow channel vectors, and the collision rules gate on occupancy (capture/park) rather than randomizing direction into a diffusive bath. Inverse-square transport is not a dilute-regime accident.

E2 (clock-lag, 3ed51d5, f10aac4). Transit lag is discrete dwell, not continuous drag: with capture off, v_eff = 2.174 cells/t at every density — carriers collide but are never slowed by congestion. Congestion delay is park-and-wait, not friction.

E1 (clock-drift, c40b4b6; vault-cadence, c5ce356). Bound objects have no autonomous internal clock: static scalar lumps are frozen (wake=0), and the wound vault's winding is a stored record — every internal phase constant over 2000t. "Time" for a substrate object is its transaction rate with the environment. A real clock must be a flowing resonator. Bonus finding: wound structure halves driven-drift response vs. scalar (the C_s²+C_i² bandwidth trade as mechanics).

E4 (clock-cavity, 9c2cf30, dcd7f14). A packet between parked Q64/IREFL16 mirrors is a working light-clock: clean 23t half-period. Static interior density dilates it via path-length scattering (23→26.7t at q=4); the first transit is always invariant — c is fixed, dilation accumulates over cycles. At q≥8 a single absorbing site anywhere on the path kills the clock — the horizon is a capture cliff, not an asymptote.

E4b (clock-fizeau, 6771759). No medium drag: a −x headwind is identical to vacuum; a +x "tailwind" only injects wind mass onto the probe channel. The lattice does not flow — only standing congestion dilates.

E4c (clock-halo, 0304f58). Transverse sustained wind: zero period change at all rates — just faster decoherence. Isotropic bath (the honest omnidirectional halo): destroys clock coherence within ~2 bounces. Halo flux kills clocks; parked mass slows them — the inversion of the naive expectation, and the most mechanistically interesting result of the series.

E10 (photon-refract, e5d2dd8). With IPHOT=4 (fluorescence: the emit gate may output photons), a photon pulse through a q=96 corridor spreads into an exponential re-emission tail — peak arrival +40%, still emerging at t=105. Photons detained by mass resume as photons after ILAT. The delay curve is non-monotonic in ρ (peak ~q=32) and saturates in corridor depth — a detention event, not a per-cell tax.

E8b (clock-ensemble-ac, 5118759). The population clock survives congestion but bleeds — ~9 momentum flips then the cavity runs dry. Mean half-period does not shift monotonically with density; the per-packet dilation of E4 washes out in the bulk sign oscillation. Honest negative: the ensemble keeps ticking, but a usable period statistic needs a resonant drive, not a decaying cavity.

E11 (dwell-probe, 5118759). The edge delay-line register verified: DWELL=k gives v_eff = 2/k exactly (0.96, 0.64, 0.48 for k=2,3,4 — axial base rate 2 cells/tick divided by k). Uniform detention aggregates into smooth slowdown: the substrate can express effective v = c/k built from discrete waits. Third lag channel confirmed; also the first engineered slowdown (a knob, not an emergent effect — honest distinction).

The Refraction & Dilation Law (E2, E4, E10): Photons detained in rest resume as photons after ILAT; effective refractive index n(ρ) and cavity clock lag emerge from discrete dwell and scattered path length, with c strictly invariant on every hop.

E18–E20 (thermal arc — the iE excitation register). The substrate gained Planck's n: a per-site excitation level where absorbed photons raise bound energy at fixed mass (IPHOT>=5), shed back as light on a thermal decay beat. The measured arc: vacuum cooling is exponential and emission linear in mass (E18); a sustained flux bath accretes, it does not cool (E18b — ambient medium is a mass feeder); a parked ambient bath is perfectly inert — no equilibration channel exists without transiting carriers (E18c); a finite moving bath equilibrates once then starves (E18d). Heat exchange is transaction-mediated all the way down. The capstone (E20, 8b2447a): a lump radiating into a photon-absorbing bath in a closed box — radiation is trapped by the medium and returns as excitation, and the lump ends as a massless hot object, pure n. Mass↔photon↔excitation is a full reversible cycle; the substrate can now express an object made of heat. iE is photon-specific: massive-carrier kinetic energy still has no thermalization channel (E18b re-validation).

E21–E25 (thermodynamics closed). IPHOT>=6 collision->n verified: carriers thermalize into excitation (E21, n=329k vs 0 control). The conduction arc: open-space equilibration fails — photons self-recycle or drain (E22, three variants); a mass filament conducts n continuously (E23, B warms monotonically) but long-run still leaks to the free photon sea. Four audit bugs found and fixed: iE edge drain, snapshot coverage, IPHOT>=5 ablation fall-through, and photon deletion on missed rolls (the real killer — unabsorbed photons were deleted while IPHN stayed frozen). Post-fix the Solid Conduction Bar (matter-dense closed box, hot/cold step) shows First Law bitwise conservation, a steady ~600-photon dynamic gas, and real Fourier diffusion (R: 0->8598 while L cools) — energy transfers cell-to-cell as n->photon->n. E25: the clock is excitation-invariant — n=50 preload gives identical bounceTicks to cold at every density; thermal n lives on the photon plane and never touches mass-channel transit. A hot cavity ticks exactly like a cold one. E26 takes the claim from architectural to emergent: a probe photon pulse crossing a dense photon bath on the same iPA plane arrives with a bit-identical profile — light is blind to light at full c, Maxwellian superposition measured on a shared discrete channel.

E27–E29 (light → matter). IPHOT>=8 fires Breit-Wheeler on the lattice: antipodal photon pairs crystallize into rest mass (E27). The product can't persist inside its own bath — beams shred what they make (E28, five geometries). But E29 cracks it: a convergent 6-fold photon implosion pair-produces 920 quanta at the focus, then the bath's own momentum carries it radially away — leaving a frozen lump in clean vacuum, stable 270+t. Structure assembled from pure light; the freeze-out is geometric, not expansion.

Re-validation (iE era). Prior results survive the new register: E4's cavity clock unchanged (vacuum emits nothing to thermalize, 23t); E18b accretion identical; E10's refractive tail gained physics — the n-reservoir roughly doubles detention and turns the corridor into a persistent photon source.

The dimensional bridge

Formal bridge — grounded. The unit assignment is an ontological identification, not a measured result:

Substrate unit	Physical unit
cell (hop length)	ℓ_P = 1.616×10⁻³⁵ m
tick (hop period)	t_P = 5.39×10⁻⁴⁴ s
c = 1 hop/tick	ℓ_P/t_P = c — exact by construction

With this assignment every measured ratio is already physical: E4's 23-tick cavity half-period is 1.24×10⁻⁴² s; the E10 photon delay tail's ~80-tick span is 4.3×10⁻⁴² s; E11's v = 2/k dwell law is a literal refractive-index law n = k. The E-series is a Planck-resolution measurement record.

Structural bridge — open. Which substrate structure is a physical object is the harder identification: the atomic regime (~10⁴³ t_P cycles, ~10³⁰ Planck volumes — an electron is ~10³ cells across) is reachable only through ensemble statistics, not direct simulation. The coarse-graining path is the program's own premise — substrate structures are Planck-scale seeds whose population statistics produce atomic-scale physics. All measured lag ratios survive that coarse-graining intact — the absolute scale was always the identification step's job.

New directions from here

1. Identification hunt — find the substrate structure whose characteristic scale/period matches a known physical anchor; that fixes the bridge.
2. n(ρ) mechanism check — done (E10c): delay grows with corridor length (bulk, not boundary) but saturates once every photon is detained once — detention event, not per-cell tax.
3. Ensemble dilation — done (E8b): ensemble survives but the period doesn't shift monotonically; needs resonant drive.
4. Capacity-stealing channel — done (E12, 31825b5): refuted for the Q-beat. The ITICK-locked emitter can't dilate in rate, and transit flux refills its Q-core — traffic supplies the internal cycle (emitted=1 in vacuum, 213k under wind=32). Capacity-stealing survives only as a mechanical budget (E1's drift resistance), not a rate budget. (pget now exported.)
5. Edge-dwell registers — done (E11): v_eff = 2/k verified; detention aggregates into smooth slowdown.

Open assumptions (the honest audit)

• Capacity-stealing is a second, unmeasured mechanism. The audits' own framing is actualization bandwidth stolen by field processing — the C_s²+C_i² budget — not transit lag. The E1 wound-vault result (internal circulation halves drift response) is that signature, but it has never been run as a clock experiment. There may be two dilation channels; we've measured one.
• Static fill ≠ a mass's halo. E4 seeded parked rest; a real halo is sustained transaction traffic around capturing/emitting structure — occupancy + throughput + active self-focusing (IQMODE7) at once. Each ingredient was tested separately; the combination hasn't been.
• Single-particle artifacts — resolved (E8, E8b). The ensemble clock survives densities that kill a lone packet — the q≥2 cliff was an n=1 artifact. But the ensemble's period doesn't shift monotonically; the bulk oscillation outlasts decoherence without cleanly dilating. A usable statistical clock still needs a resonant drive or lossless mirrors.
• "Never slowed" has an exception. E11's DWELL delay line produces uniform v = c/k — smooth slowdown built from discrete waits. Congestion never slows a carrier mid-hop, but engineered detention registers can. The invariant is c per hop, not per unit time spent in transit.
• Two candidate "gravity" terms are conflated. Passive occupancy (E3) vs active Q-capture (E5) — the ontology's gravity is transaction focusing, not congestion; Q-field strength may be the right independent variable, not ρ.
• The lag is a spectrum, not a rate. Most carriers pass, some detain — the honest comparison to GR is the ensemble mean of the capture-time distribution, not a single-packet T(q).
• Dimensional bridge still absent. All numbers are lattice units — ratios and regimes only, never µs/day.
• Open boundaries. Every probe drains edges; the r⁻² halo holds over ~20 cells in a 56-box, untested at larger scale.

---

E34–E42: The Cadence Dilation Arc — a substrate theorem

The photon-frequency branch (NNU=4, feature/photon-frequency, merged ca6bb2d) gave light an energy spectrum for the first time: emission draws nu from local occupation n (Wien displacement measured), absorb deposits nu into iE (energy conserved), pair production fuses same-nu photon pairs into 2*nu mass.

That opened the cosmological question: can any transit mechanism stretch light's arrival cadence (1+z) on a static lattice? The arc produced five falsifications and one verification:

• E34 additive spectral shift — falsified. Down-shifting photons through a medium collapses the spectrum onto nu=1 — the Compton-distortion objection to Zwicky tired-light, reproduced on a 4-bin integer plane. Multiplicative nu/(1+z) can't resolve in 4 bins.
• E35 uniform detention — falsified. Static slab preserves T=10 pulse spacing exactly; delay is uniform per pulse.
• E36/E38 self-deposited mass + light-history nodes — falsified. Deposited mass saturates into a static corridor; a leaky iPT integrator forgets in the dark gap between pulses.
• E41 corridor queue backlog — falsified. Congestion meters bursts out at a fixed drain rate (out_dur ~150t regardless of tin/L/rho): additive, not scale-invariant.
• E42 kinematic Doppler — VERIFIED. A receding emitter at v=+0.25 cells/tick stretches arrival spacing to ~11.25, matching T(1+v/c_eff) with zero free parameters. Approaching and faster runs show detector artifacts, not mechanism failure.

Substrate theorem (commit-anchored, 9b2031a): under invariant hop speed c, interaction with a static medium changes photon amplitude (absorption), never transit velocity or inter-pulse cadence. Dilation on this lattice is strictly kinematic — relative motion of worldlines — never an in-transit medium effect.

This is a mechanism constraint, not a cosmological verdict: it closes the tired-light/dwell/backlog mechanism class and verifies the only dilation mechanism that exists in the ontology.

• E42b detector cleanup — the arc is now airtight both ways. The E42 v=0.5 "24-tick spacing" was a detector artifact, not physics: axial photons stay on their x-parity lane forever, so a single-column detector was blind to alternating pulses. Dual-parity detection (x=44+45) now shows recede stretching (~11.25, ~12.5) AND approach compressing (~8.75) — T(1±v/c) verified both directions, zero free parameters.
• E43 attenuation law — the "dimmer" observable is Beer-Lambert. Ablation-mode beam through parked slabs: ln(T) is linear in BOTH depth and density — T = e^(-a·rho·L). Exponential absorption per hop, not geometric spreading. (Re-emission contamination ruled out by ablation mode.) Together with the kinematic dilation result, the two supernova observables are now separately characterized: duration stretches only kinematically; flux falls off exponentially with matter depth.