Moore26 Findings

This is the measured record of the current engine — moore26, a Moore/FCC lattice substrate at 0.1 fm per node where every node carries 26 registers (one per channel — the name is the register count, not the lattice size). Everything on this page is output of headless harnesses against src/moore26.c, not assumptions: what the lattice actually does when the rules run, measured, including the results we had to retract. The engine's structure is 12 transporting edge channels on the even-parity sublattice plus face/corner registers that hold bound structure; transactions negotiate through an offer→incipient→confirm handshake each tick.

Notation

The shorthand used throughout — defined once here:

Registers (per-node state; c indexes the 26 channels, edge pairs are 2e/2e+1):

• winding[c] — committed circulation count; min(pair) ≥ 32 is bound structure
• dd — signed winding difference across a pair's two members; the coil's handedness and strength
• mass — parked rest energy (the condensation register)
• charge — signed int16; biases offer acceptance
• heat — the thermal bath; the friction sink
• packet[c] / packetNu[c] / packetPar[c] — photon count, energy, and parity-sign in flight
• incip[c] / incipNu[c] — pending (incipient) offers awaiting confirmation
• armed — Σ incip on a node: the congestion measure
• massFlight / chargeFlight — directed momentum transport
• grid_current / grid_next — the double-buffered state; phases stage into next, the tick swaps

Event counters (ev* — the harness-readable ledger): ev4 emissions · ev20 shed energy · ev23 fusion pairs · ev24 string-breaks · ev30 offer declines · ev43 mass-tax evaporations · ev44 thermalized overflow · ev45 same-sign declines · ev46 baseline declines

Tuners (runtime, all exported): qEvapShift mass tax · qConfirmCap confirm ceiling · qGroundFloor winding floor · qCatchGate offer gate · qFusionWall pair-production threshold (256) · qCommitMin binding threshold (32)

Phases: P0 carry · P1 flight · P1b junction annihilation · P1c mass launch · P2 offer · P3 confirm · P4 winding conduction · P5 phase sync · P6 emission/decay · P7 string break

Ledger: L = Σ(winding + mass + flight + packet·nu + heat ± charge) — the scalar the audits track. >> is arithmetic shift (m>>s = m/2^s); min(pair) is a pair's committed min.

Deviation audit — verified vs nature

Where the substrate matches (measured, not asserted):

claim	status
Rest mass exists	✓ Prigogine equilibrium — ~42k plateau, M ∝ 2^shift spectrum
Radiative decay	✓ spin-aligned nu=1 emission on the circulation axis
Charge sign matters	✓ ev45=331 same-sign declines, ~12% attraction asymmetry — local
Pair production	✓ fusion wall 256 → ±q fragments, net-zero charge
String break	✓ ≥12-node tubes fracture + charged endpoints — confinement analog
Persistent current	✓ loss-free rail circulation on rings
Ledger	✓ ~0.1% residual, every phase audited

In plain terms: a wound knot holds a stable mass because its evaporation rate balances its ambient feed — a genuine dissipative structure, not a fixed parameter. It sheds energy as real quanta radiated on its circulation axis, the way a particle decays. Charged knots measurably bend transaction traffic by sign — same-sign offers get declined where opposite-sign ones bind. Above an energy threshold, collisions mint pairs of oppositely-charged fragments. A saturated tube past twelve nodes fractures and throws off charged endpoint mass — the confinement signature, watchable live in the demo. Ring loops carry current without loss. And the bookkeeping is honest: a phase-by-phase audit sealed every leak down to a ~0.1% documented seam.

Where it deviates — the honest list (observational measurements only — what nature's instruments actually record; inferred rules like Pauli exclusion are excluded by definition):

1. Charge isn't a field — in nature, two charged objects feel each other across distance (1/r², measured). Ours only pushes on traffic that physically arrives and jams — the gate fires at armed≥4 offers on a node. Attraction and repulsion are real but local: a congestion effect, not a field.
2. Parity is symmetric — nature's weak sector measurably prefers one handedness (the P-violation experiments). Our dd coils come out even — 0.89-1.0 ratio, no preference.
3. Knots don't fly — measured particles move ballistically. Ours can't: the binding winding is anchored to where the mass sits, so knots crawl by rebuilding themselves cell-by-cell. No rest-frame momentum is conserved across translation.
4. No Lorentz structure — time dilation is measured every day (muon lifetimes, GPS clocks). Here c is simply 1 hop/tick — a lattice constant, not an emergent invariance. Nothing contracts, nothing dilates — yet.
5. Decay isn't a lifetime — nature measures an exponential decay curve with a characteristic lifetime per species. Ours evaporates by a mass-proportional friction tax — the shape is equilibrium decay, not a quantized lifetime.
6. Mass spectrum is approximate — the 2^shift scaling held at ~3.4-3.9× not the clean 4× (a finite pump, not an ideal bath). Nature's particle masses are measured constants; ours are tunable emergent values.
7. Gravity absent — gravitational attraction is measured (Cavendish onward). Routing asymmetry that would act like it is proposed but not implemented.
8. Ledger ~0.1% — nature conserves exactly. Our audit leaves P7-break −63 and mid-tick sampling artifacts as documented seams, not physics.

Verdict: local transaction physics is honest (conservation, charge-sign, pair production, confinement); field-level and relativistic physics is the open frontier — charge range, knots-in-motion, and Lorentz structure are the next deviations.

Scale context

The engine's node spacing is 0.1 fm — the sub-nucleon regime where quark confinement and string-breaking live. This MIT scale tour is a good intuition pump for where the lattice sits:

Transport

• v/c = 1.00 on all 12 FCC edge directions — isotropic, one hop/tick (the operational c)
• Dispersion-free for nu = 1–255
• Power-of-two dims only — chunk masks (w-1) corrupt on N=40

Pair production + annihilation

• Fusion wall exact: nu_a + nu_b >= 256 → ev23 pair mint
• Yield ≈ mass = pair × Σν >> 5 (≈ Σν/32)
• Signed ±q recoils on chargeFlight, net charge 0
• e⁺e⁻→γγ: opposing massFlight arrivals annihilate at the meeting node → back-to-back 255+255 packet jets, heat=0 in sparse vacuum. Same mechanic as meso3i's IANNIH (opposing drest → antipodal photons) — third-generation implementation

Knot regime (nucleon candidate)

• comm ≈ 120 → RMS ≈ 4.7, ~10-node diameter (the 1 fm analog)
• Kinematic ceiling ~0.033c — routing friction, NOT inertia:
  • 4.4× absorbed energy → no faster (routing-bandwidth bound)
  • 3× mass → +14% (wrong direction for a ∝ F/m)
  • ceiling set by pattern re-formation cadence, not volume
• The ceiling is the knot's SOUND SPEED: bulk translation requires a structural deformation to cross the pattern — the emission→flight→re-arm cadence IS mechanical information propagating through the medium. 0.033c = the rate phonons move through bound topology. Evidence: comm 40→120 gives ceiling 0.027→0.031c — it RISES with density (tighter coupling → faster sound), the opposite of inertia. Pushing harder doesn't help because you can't outrun your own phonons. Maps to v_s ≪ c in real materials — bound structures can only translate as fast as their internal coupling carries the deformation.
• Immortal against compression: flux, counter-flux, massFlight strikes, and knot overlap all accrete — never rupture
• String-break under TENSION only: ≥12-node saturated tube fractures → 768+768 endpoint mass + ±8 charged recoil, exact ledger (2×768 + 128 = 1664 drained)
• Species-inverse balance: mass ≈ committed/3.4 equilibrium (0.292, flat 1000→5000t under locked rules; the rule's 1:8 floor is a nucleation trigger, not the setpoint — competing radiation/condensation rates settle it higher). Massless sites radiate only (nucleation seed required)

Directional memory (the dd stack)

dd = winding[2e] − winding[2e+1] pair imbalance — stored angular momentum, thermodynamically irreversible:

• dir_bias transduction: offer's direction bit survives confirm → dd=314 written from flux (vs ~1 symmetric)
• dd_shift conduction: imbalance pushes along its own channel
• bend_shift impedance-gated rotation: fires only when forward pair saturates (≥128), rotates dd onto dot=1 edges
• rad_shift radiation tax: un-rotated fraction sheds as forward packets (ev33) — discrete synchrotron
• Retention: 125% over 800t beam at dir_bias=0 (grows!); +12,000% spin-up at dir_bias=256 — beam is a dd pump (pre-toll measurement — the surplus toll now bounds this: chain_share re-offers pay from the uncommitted margin, killing the runaway while preserving the dd=519 dimer bond)
• Accretion physics: sustained flux → mass+dd growth at plateaued heat ~7.4M — driven E→m + E→L conversion

Radiation

• Bend luminosity: threshold onset <rate 2, scales 326→1787 at rate 2→8, saturates 1853 with dd washout at rate 16
• Overdrive doesn't break the knot — it bloats (vol +100%), direction washes out, structure survives

Holographic finding — interiors are bookkeeping

Measured: a dd pulse injected on the knot's surface stays pinned at the surface (west=3520, mid=0, east=0 across 300t) — the directional deformation never propagates through the interior. The knot is sound-opaque: dd conducts along surfaces and channels (tubes, legs) but the dense interior doesn't carry it.

Implication: everything the knot does happens at its boundary — emission, absorption, arm/confirm, leading-edge re-formation (the 0.033c translation mechanism is a surface process, not a bulk propagation). The interior is archived state; no clock ticks there. The holographic principle (degrees of freedom on the boundary, not the volume) appears here as a measured property of a cellular substrate — "inside a knot" is the same kind of question as "inside a black hole." The interior is bookkeeping, not dynamics.

Update — chain_share volumetric routing: with chained offers enabled, the interior isn't dead space — confirmed transactions re-offer on the dominant dd rail, hopping signal node-to-node through dense structure (far-side packets 9→173, dd−4.8k→+76.7k across a knot wall; U-tube dd reaches leg4=2145 through three bends). "Surfaces all the way down" becomes "relays all the way through" — the interior conducts via transaction forwarding, not packet flight or scalar residue.

Density probe: 8 packed knots retain 1.028× isolated volume — no density penalty; mutual emission recapture slightly favors packing. "Mass is spaced far apart" does not drop out of knot-level stability; it lives in transport instead — dense regions are stable but causally dark (unverified on larger clusters).

Dead registers / phantoms

• phase register + 10-iter Jacobi sync: pure overhead — phase_sync=0 is 0.09% different (noise). Vortices unwind in 50t — no topological protection in the scalar field
• phase_gate (qi3 effPhase port) works but doesn't gate massFlight parking
• Inert on standard scenario: fusion_wall, commit_min, ground_floor

Geometric Pauli / two-slipstream

• Tilted {111}-plane hexagon: tangent rotates through 6 edge axes — supports two counter-rotating slipstreams (a/b)
• One stream: 150/30 holds at 100% (net handedness protected)
• Two streams: 150/150 → 62/62 — coexist but shed ~59% (the annihilation-shed analog; energy goes mostly to heat)
• Third stream: refused — the hexagon has exactly 2 routing lanes

Interactions at range — radiative torque transfer

Two comm=120 knots at sep=16 and sep=12 show no centroid drift (frozen at seed positions over 1500–3000t — null for linear force), but they exchange angular momentum through radiation: each knot's emission grazes the other's facing flank and dir_bias writes bound circulation:

sep=12, chain=128:  ddL: +1 → +11 → +16 → +23   (t=0→2000)
                    ddR: +5 → −6  → −9  → −15

Opposite signs — each knot sees the other's flux arriving from the opposite direction, so the deposition torques them counter-rotating, like gears coupled by the medium. This is the meso3i YORP mechanism (grazing light writes spin via absorbed flank deposition) running on the conserved engine: a real action-at-a-distance interaction living in the angular ledger, not the linear one.

The wound dd circulation is literally a persistent current loop on the transport edges — a candidate magnetic-dipole register. Open questions before claiming dipole magnetism: orientation sign-rule (does a pre-seeded dd preferentially anti-align a neighbor?) and distance falloff (dipole-dipole ∝ 1/r³).

Substrate artifact fixed en route: a +x lattice wind (+1.13 node drift/knot over 750t) traced to the conduction k-transfer always pushing the pair's + member direction plus first-tie-wins scans defaulting to channel 0. Fixed via gradient-descent conduction, per-node sent[] clamps against grid_current, and rotated scan order — vacuum now measured still at all separations.

Virtual layer & rails (post-ledger mechanics)

• Standing offers: nu%line moved to confirm — off-shell arming, ~6-tick lifetimes, Landauer dust to heat on expiry. Armed-but-unconfirmed offers steer every dominance pick via eff_dd (committed dd + parity-weighted incipients) — influence without answer.
• Parity currency: packetPar/incipPar carry ±1 helicity through transport (majority-merge). Deposit direction reads carried parity; emission debits the emitter's dd by 1 — spin-flip ledger. Sign-rule verified: receiver flips with source (+54/−6 perpendicular).
• Rails (rail_gate): saturated dominant member = waveguide — packets skip arm AND decay, re-channeled onto the rail (turns with it). Confirmed transactions pave rails: ack return lays 1 committed unit on the return pair (conserved transfer).
• Persistent current: 12 packets on a closed {111} rail, 2000t, zero arms/decay — live circulating current on a static coil. (Seed caveat: use ENGINE channel vectors — a mislabeled vec[] paints an open arc.)
• Census metrics — two axes of handedness: a closed circulation has zero net dd by symmetry (loop forward channels alternate +90/−90 across even/odd members → global sum cancels). Species need TWO measures: net dd (stick polarization — wound ball reads −100,415) and path curl/ ringdd (closed circulation — live ring holds 540 at 3000t). A loop reading dd=0 is not relaxed — it's circulating.
• Species census (corrected seeds, 3000t): ball mass=3,846 ratio=0.020; static ring comm=2,340 (survives, no nucleation); nucleated ring mass=60 (spark held — rings CAN carry mass); live ring stream=12, ringdd=540 (current + winding both persistent); wound ballmass=35,618 ratio=0.239 dd=−100,415 — 9× plain-ball mass, polarized winding, the spinning-knot species.
• Untested: PAR_SIGN-selective absorption — a coil that arms parity-matched offers and declines mismatched ones self-selects its charging spectrum (positive feedback: aligned flux grows dd → stronger selectivity). The inverted version (reject matched) is a dd stabilizer. One-line gate at the arm path; untested. Observational anchors: optical pumping (a vapor absorbs circularly polarized light → becomes spin-polarized → prefer- entially transmits that polarization — the feedback loop itself, measured since the 1950s); circular dichroism (differential Δε absorption of L/R circular light by chiral media); spin-valve GMR (a magnetized contact passes same-spin carriers preferentially).

Strong-force mechanics — the full checklist

• Tube exists ✅ — winding=200 zigzag rail carries a seeded stream loss-free through 6 turns (waveguide exemption + rail bypass). Packets re-channel onto each node's dominant member, riding the curve.
• Tube mediates force ✅ — the stream's confirmed deposits write massFlight on the arrival channel (ch6=+x) while the knot's intrinsic crawl writes its own dd axis (ch7). Per-channel anisotropy isolates extrinsic force from intrinsic churn — ch6:1 vs ch7:839 on the receiver.
• Confinement ✅ — Phase 7 string-break: a continuous ≥128-balanced saturated pair ≥12 nodes fractures, pair-produces endpoint mass (1,792 minted, ev6=1), and recoils the fragments apart with opposite chargeFlight. The tube's job is to store tension and snap — not to heal. (A manually-severed gap does NOT re-pave: paving requires traffic, and traffic requires the tube — self-consistent stall.)
• Motion ✅ — massFlight launch rides the dominant |dd| member (fixed a sign-bias that motion-locked −dd knots — both handednesses now crawl; −dd drifted +0.46, +dd drifted −0.20 over 3,000t, handedness-dependent direction).
• Radiation pressure — massFlight[c]+=1 per confirmed deposit wires absorbed momentum → recoil. Real but weak against 38k-mass knots (needs focused flux or long runs).

Substrate chirality — RETRACTED (bookkeeping bug)

The 1.7:1 accretion asymmetry reported here was a parity-tag bug, not substrate physics. PAR_SIGN(n,c) was channel-relative (winding[c] > winding[c^1]), so a packet emitted on the dominant channel always minted +Par — a −dd knot therefore debited its even minority member (p_>0 → (c&~1)), collapsing its structural min and starving accretion. Member-relative PAR_SIGN (even-dominant = +, odd = − — the convention the rest of the engine already used) restores symmetry: +dd 41,741 vs −dd 46,763 (0.89, within run variance).

The residual 0.89 is worth a longer-run check — a mild second- order asymmetry or plain seed noise, not the runaway.

The Meissner lock — committed structure stops bleeding

Phase 4 gradient descent used to drain any balanced pair >=32 at a>>3 per tick — rails and tubes dissolved into the vacuum (the visible dim-out) while dust below 32 sat frozen. The rule was inverted relative to nature: it froze the noise and bled the structure.

The lock: balanced pairs in [commit_min, 128) skip conduction entirely — committed structure is structure, not field. Pairs >=128 bleed only the over-pressure excess (a-127)>>3.

Measured (32³, seeded 200/200 on ch6/7):

• before: 200 → 105 → 69 → 68 and still falling
• after: 200 → 154 → 143, then t50 = t400 = 1572 — frozen

Regressions: wound-ball accretion unaffected (25k mass — unbalanced pairs take the a!=b directional path); ring rail holds 630/660 dominant winding (was dimming).

Calibration notes: tubes floor at ~143, not exactly 128 — (a-127)>>3 falls under the k>=8 transfer gate early. Dust <64 is still effectively frozen (k<8 never transfers), so the "vacuum expels field" half of the Meissner story is inert — committed structure persists, but the medium doesn't actively push flux back onto rails. Nature's tube persistence is now matched on the structure side only.

Consequence: over-pressured tubes relax toward the 128 fracture band — Phase 7 stays armed on any tube that parked just over the gate.

Post-lock hexbuckle (64³, n=12/scen): flat=63,785±6,873, dip0=57,602±5,852, dip32=53,755±6,485 — the bend sits within ~1.5σ of flat. No flexoelectric dipole — the charge+=cq transfers that would polarize are the same transfers the lock froze. But the probe now leaves ~60k trapped polarization vs ~0 pre-lock: deposited charge persists once conduction stops — charge memory, not strain-induced dipole.

Genealogy

meso3i (IDREST/IANNIH/DWELL) → link3 (lDR/lDW/lUP) → qi3 (winding/held) → moore26 (winding pairs/massFlight/incip)

Same knot regime across four engines — committed circulation, directed rest, dwell parking, annihilation-on-collision.

Conservation audit arc (ledger sealed to ~0.1%)

Per-phase delta counters (-DM26_AUDIT, moore26_audit(p)) attribute every ledger change to a tick phase. The hunt, in order:

leak	fix
h>>7 silent heat drain	booked to ev44 (tracked bath)
incip expiry booked h not h·nu	h*nu thermalizes
annihilation nu=hit mint (hit² out)	nu=1 per quantum — exact
register-cap truncations (flight/annih/chain-share/emit)	excess → heat everywhere
incipNu scalar overwrite on shared channel	weighted nu mixing at arming
spin-flip debit deleted 1 winding/emission	debited rung → heat (parity costs angular momentum, not free energy)
stale grid_next resurrected ghosts	rescan_live memsets dead chunks before swap

Result: every transactional phase audits ΔL=0; P7-break carries −63 string-break bookkeeping; the 5% void is now a ~0.1% documented seam. Mid-tick grid_next sampling in the audit shows a +957 P0 artifact (buffer-staging ghosts, not minting) — tick-end grid_current sampling is the honest read.

Mass spectrum — calibrated (Prigogine dissipative structure)

qEvapShift mass tax: per-tick evaporation probability (mass>>shift)&1023, emitting nu=1 packets on the knot's dominant circulation edge (spin-aligned radiation, ev43).

• Closed box: plateau ~40k, supply-bound — the knot eats until the vacuum starves; tax sets only the residual split (41k/39k/39k for shift 3/4/5)
• Open box + sustained flux: plateau ≈ flux × 2^shift — measured 467/906/1244 (shift 5 still converging toward 4× when sampled). Mass is an emergent supply/friction value — a real tunable spectrum, not an input

Charge law — measured field behavior

charge (int16) is dynamic — absorption re-charges a node (charge += nu can flip sign). Sign-aware binding: same-sign offers cap at 4, opposite-sign at 13−|q|>>3. Measured: ev45=331 same-sign declines vs 35 baseline on a +104 target — repulsion real; −q knots decline +par ~12% less (attraction). Knots also repolarize under sustained opposite flux — a −q node fed +par packets flips +.

Absorbing boundary

qAbsorbWall (moore26_absorb_set) zeroes the outer shell of grid_next each tick — open-box conditions for emitter demos; the x/y torus wrap no longer recirculates pumped streams into the measurement.

Runtime tuners (all exported, build.sh synced)

fusion_wall, commit_min, ground_floor, confirm_cap, catch_gate, dir_bias, deflect_align, dd_shift, bend_shift, rad_shift, phase_gate, phase_sync, evap_shift, absorb, charge_set, packet_par_set, chargeflight_get, audit

External reading

The physics concepts the substrate touches — the real-world counterparts the measurements are compared against:

• Moore neighborhood — the 26-cell stencil the channels derive from
• Face-centered cubic lattice — the underlying transport geometry
• Lattice gas automaton — the method family this engine belongs to
• Dissipative structure — Prigogine's far-from-equilibrium order (the mass plateau)
• Color confinement / string breaking — the QCD analog of the tube fracture
• Pair production — γ+γ → charged fragments at an energy threshold
• Parity violation — the weak-sector asymmetry the substrate doesn't reproduce
• Yarkovsky–O'Keefe–Radzievskii–Paddack effect — radiative recoil (the proposed knot-motion mechanism)
• Transactional interpretation — Kastner's offer/confirm framing, a lineage influence on the handshake
• Causal set theory — the discrete-causality program this substrate runs as a working engine