The machine's information carrier, microwave through optical, described at the level of its elements and verbs: the radiating face, the optical stem, the fingertips (optical placement at the finest scale), the sensing ladder, the topological alphabet, and the energy loop. Size-free — the physics privileges ratios, not dimensions.
Ruling R1: light addresses, measures, and calibrates; sound and induction exert force and heat; electrons only amplify at the boundary edge and carry DC power — never information. Not a preference: every cross-carrier coupling in the volume is parametric, with negligible energy exchange (Field Compiler §17.6 bound i), so light cannot be muscle. Its whole value is being the machine's independent information plane. DERIVED One refinement (R1-b, adopted 2026-08-01): sound is the hands; light is the fingertips — light's direct radiation force is the precision-placement channel at the smallest payloads (§3b). A second refinement (R1-c, adopted 2026-08-03): light and induction are one connection, two regimes, not two different couplings that happen to both be electromagnetic. Both are realizations of the same forced u(1) gauge connection (Magnetism §§2–6); what differs is whether the connection's curvature stays bound near the source or detaches and propagates. Kept close to a conductor (within roughly a wavelength) it never fully radiates — it drives free-charge motion directly, a transformer-like power-transfer mechanism, which is why induction is strong and short-range. Let go, it propagates at any range but only ever carries momentum p = E/c — always weak, which is why light is fast and universal but can only ever address. R1's muscle/address split is therefore not an engineering convention laid over "it's all EM" — it is the physical content of the radiative/non-radiative distinction. The hardware already found this before the theory did: R5's contrawound winding pair's sum mode (confined, non-radiating) and difference mode (fast ring dipole, radiating) are the near- and far-field regimes of the same connection on the same coil, not two different antennas. DERIVED
What the photon brings that the phonon cannot: speed (ns propagation across the volume, so verification runs ~10⁴× faster than the acoustic chords it verifies — "acoustic builds, EM verifies"), bandwidth (six decades of frequency, microwave through optical, on one carrier family), penetration without a medium, a genuine polarisation pair in free space, and sensing precision that is SNR-limited rather than λ-limited. What it gives up: momentum — E/c is 10⁵–10⁶× less force per watt than sound's E/v, the sibling paper's whole reason to exist. The full asymmetry table is The Two Carriers §4–§5; the carriers' in-flight interactions (sound steers light, light reads sound, light writes sound) are owned by The Two Carriers §6b (software-facing copy: Field Compiler §17.6).
The photon's boundary element is a flat printed disc that behaves like a horn — a flattened coaxial horn: a radial ring feed at the bore rim launches into an equiangular-spiral slot grating (12 single-handed arms, r(φ) = r₀·ecot α·φ; the slots are voids in the conductor, not traces). Bandwidth comes from angle-defined geometry (Rumsey), not resonance — the Bode–Fano/Chu discipline in printed form. Element table (instance dimensions: Mechanical Construction §3):
| element / role | optimum | basis |
|---|---|---|
| launcher face | equiangular-spiral slot grating + radial ring feed — launches single-cycle, space-time-nonseparable toroidal and toroidal-helical pulses; the TE/TM mix α, relative phase β, and handedness are runtime drive parameters | MEASURED art: Wang et al., Commun. Phys. 7:356 (2024); Shi et al., Commun. Phys. (2026) |
| reflector | EBG (high-impedance surface) behind the spiral | multi-octave forward radiation off a thin one-sided disc, no λ/4 depth (Luadang 2025) MEASURED |
| sensing | the same aperture reads back for super-resolution positioning — ~0.5λ single-aperture, SNR-limited, via the toroidal pulses' space-time structure | MEASURED: 97% of cases <0.5λ, single antenna (Wang et al. 2024); the servo signal for every levitation mode |
| harvest | rectenna patches on the rim ring | microwave harvest → the drive rail (§6) |
| upgrade path | q-BIC leaky-wave metasurface: a perfectly symmetric cell supports a bound state in the continuum — trapped, non-radiating, infinite Q; breaking the symmetry by a controlled amount makes it leak at a chosen rate. The leakage rate becomes a geometric variable, so the Bode–Fano budget is spent spatially: aperture taper, sidelobe control, customized radiation from a flat printed surface, no per-element electronics | MEASURED at micro/mm-wave: Xu, Overvig, Kasahara, Martini, Maci & Alù, Nat. Commun. 14 (2023) |
The disc shares its face with the acoustic micro-horn mesh — at the EM launch band the sub-λ mesh reads as solid conductor, so the physics is the diplexer; the acoustic side of the plate belongs to The Phonons §2. Mirror-imaged facing plates have opposite intrinsic handedness: counter-drive covers twist, cancellation, and the two-plate coherent-perfect-absorption verb.
Z(r,t), with honest bounds.
Space-time holographic metasurface antennas are demonstrated
(Sci. Adv. 2026, adx7090): heterodyne, multifrequency, frequency
conversion at the aperture, holographic beamforming, near-field 3D imaging.
Two consequences: it encodes the chord alphabet natively (frequency content
plus amplitude and phase — exactly what a space-time hologram stores), and
time modulation breaks reciprocity, genuinely separating the ADD and REMOVE
paths. Bounds: the result is EM-only — frequency conversion
within a band, not acoustic↔EM transduction; and the time axis is
not free — every cell needs an active switching element at the modulation
frequency, spending exactly the virtue (no per-element electronics) that
made the passive hologram attractive. The honest reading is a tier:
passive Z(r) where a fixed transform suffices, active
Z(r,t) where frequency agility or nonreciprocity is worth the
wiring.A printed hologram is fixed, so it steers by frequency — each tone reconstructs a different computed pattern from the same static aperture: a chord is N boundary DOF per tone with per-tone aperture patterns, not "the same DOF, louder." Every frequency is a different illumination geometry — why frequency is one of only two diversity axes that add independent information (§4), and why bulk energy delivery is by spectral addressing: frequency is the address, the target's absorption spectrum the mask. The quantitative treatment — including the three precision levers a free-field model assumes away — is Field Compiler §17.5.
The bore sightline is not a camera hole. Ruling R2-b: the optical channel is a full TX/RX holographic port speaking the same toroidal-pulse alphabet as the microwave face, at optical frequency — one alphabet, six decades of frequency, both directions.
Sound is the hands; light is the fingertips (R1-b, adopted
2026-08-01; canonical text in
REPLICATOR_CAP_STACK_RULINGS_2026-07-30.md). R1 stands for
bulk force: per watt, sound pushes 10⁵–10⁶× harder, and the parametric
bound still forbids treating cross-carrier coupling as power. But R1's
derivation was about cross-carrier couplings — it never excluded
light's direct radiation force on matter, and at the smallest
payload scale that direct force is the machine's precision channel.
Register note: nothing here upgrades the machine's claims — the external art is MEASURED, every in-machine use is design, and "build at the nano level" is OPEN and stays off the deck.
Ruling R12 fixes what the machine may honestly claim to see, rung by rung. Each rung is tagged where it stands today:
| rung | claim | register |
|---|---|---|
| compound broadband full-waveform tomography, in-air | surface features at λ_min/2 of the working band — the impedance cliff makes solids-in-air opaque, so this is scattering tomography of surfaces: silhouettes, shadows, multi-bounce, fully invertible (instance: ≈0.85 mm at the plan-A air band — hardware spec). Interior κ/ρ/α phase maps require the fluid-coupled roadmap rung | DERIVED on MEASURED art (USCT) |
| band extrapolation | ~2× more, log-SNR-bounded | DERIVED |
| harmonic imaging | a few × more | MEASURED art |
| resonant near-field structure + broadband decoding | λ/25–λ/30 class | MEASURED elsewhere (Lerosey & Fink, Science 2007); in-chamber OPEN |
| discrete-scatterer localization + photoacoustic beacons | λ/100-class positions — the payload is discrete and carries its own beacons, so localization, not classical resolution, is the machine's working question | MEASURED art (ULM: Errico et al., Nature 2015) |
TX/RX conjugacy — the identity that closes the loop. A phase-kept scan is hologram acquisition; the same recording, replayed conjugate, is a trap around the thing imaged (time reversal). And the terms match at the physics level: the same Δκ/Δρ contrast vector appears in Born scattering (what the machine senses) and in the Gor'kov force (what the machine grips). The machine sees matter in the terms by which it holds it.
Reconstruction, in one paragraph. The software ladder runs L0 delay-and-sum backprojection → L1 DORT/time-reversal-operator SVD → L2 regularized linear inversion → L3 full-waveform inversion; a naive boundary backprojection yields a reflectivity image, not a geometry, and the program's single biggest technical risk is the Green's-function trap: inside a closed high-Q cavity, textbook imaging with a free-space Green's function produces a confident, sharp, meaningless image. The information budget is equally hard: only frequency and position (real or virtual — including the optically-placed photoacoustic sources of §3) add independent information; transmit-pattern cleverness and adaptive policies buy SNR and allocation, never new k-space. The full treatment — the ladder, the trap, the provenance rule ("the model may propose, only the aperture asserts"), the coverage tri-state — is software-owned: Field Compiler §16.3.
Ruling R13: topology is the machine's robust register — integer-valued field labels (skyrmion numbers, polarity × vorticity) that survive the chamber's own reverberant chaos, on both carriers. Four rungs, register-tagged:
Utility note: q-plate spin-orbit wavefront sensing (APL 2026) — a passive vectorial aberration monitor for the stem optics' self-calibration chain.
The boundary that shapes the field also recovers energy from it. The photon-side elements of the loop, element-level (instance wiring: Mechanical Construction §6):
| source | element | tier |
|---|---|---|
| microwave field | rim rectenna (MIM/tunnel rectifier patches on the face's rim ring) → the drive rail | MEASURED (>80% at microwave) |
| melt glow (cooling) | TPV ring — thermophotovoltaic conversion of the workpiece's own radiated heat | MEASURED (~40%, MIT 2022) |
| coherent extraction | the REMOVE verb — time-reversed drive pulls amplitude out of chosen modes as work, not heat; EM energy returned to the plates is rectified back into the rail | open-system recovery term; ordinary engineering, no thermodynamic heterodoxy |
Energy delivery runs on the same elements in the other direction: spectral addressing (§2) for bulk heat, and the stem for addressed optical heating at the weld — heat spent only where the build needs it.
Claims are tagged MEASURED (published or bench-verified art), DERIVED (arithmetic on measured quantities), or OPEN (registered bet with named falsification). Nothing was upgraded in the move to this paper.
Instance recipes — dimensions, layer stacks, feeds, the build sheet — live in Hardware Spec and Mechanical Construction. The machine topology that places these elements is Architecture. The sibling carrier — sound as muscle — is The Phonons.