Skip to content
Adjacency Theory markAdjacency Theory

Phase 73 · the frontier theorem

Architecture cannot outrun the complete physical causal graph.

For any implementation whose interactions are strictly local on a graph and whose influence propagation is bounded by a finite empirical velocity along every available physical channel, no local control law, phase transition, routing scheme or moving doorway can transmit genuinely new information from A to B earlier than the fastest available physical causal path. That closes the frontier conceptually and leaves exactly one experimentally open question: does the apparatus ever violate its own fully calibrated causal envelope? Physical evidence remains NONE.

Phase 73 · the frontier theorem

A smarter route cannot beat its own machine.

If every road has a speed limit, a smarter route cannot make a message arrive before every road could carry it. It can only find a better road. The only experimentally open question left is whether the apparatus ever violates its own fully calibrated physical causal envelope.

Project status

FRONTIER RESOLVED CONCEPTUALLY: architecture-only FTL is ruled out under the explicit local finite-speed assumptions. The experimentally open question is whether the apparatus ever violates its fully calibrated physical causal envelope. Physical evidence remains NONE.

frontier resolved conceptuallyphysical evidence: none

Project theorem candidate · stated under explicit assumptions

Frontier Theorem (project theorem candidate)

Let G_phys be the complete physical interaction graph of the apparatus, including wires, FPGA/control links, RF/capacitive/inductive leakage channels, shared clocks, power/ground coupling, measurement/DAQ paths and any nonlocal coupling. Let tau_e be the measured earliest influence delay across each edge. Define the empirical causal distance D_tau(A,B) = min over physical paths P of the sum of tau_e over e in P. For an unpredictable payload generated at A after arming, any ordinary local finite-speed implementation must satisfy T_decode(B) >= D_tau(A,B) - epsilon_cal, where epsilon_cal is the combined validated timing and calibration uncertainty.

D_tau(A,B) = min_{P : A -> B in G_phys} sum_{e in P} tau_e
T_decode(B) >= D_tau(A,B) - epsilon_cal
Delta_t = D_tau(A,B) - T_decode,C*(B) (anomaly requires Delta_t > epsilon_cal)
Corollary

A C* run that beats the logical C0 route but not D_tau is an ARCHITECTURE EFFECT, not a causal anomaly. Improving the routing improves the path chosen through G_phys; it cannot improve G_phys itself.

A1
Strictly local interactions on a graph

Every element of the implementation interacts only with elements it is physically coupled to. Coupling may be weighted, directed, weak or strong — but each coupling is an edge of a graph, and there are no edges outside that graph.

A2
Finite empirical propagation velocity on every channel

Every physical channel available to the apparatus — conductor, optical link, RF/capacitive/inductive leakage, shared clock, supply rail, instrument path, software channel — has a measured, finite earliest-influence delay. Nothing in the apparatus carries influence instantaneously.

A3
Payload unpredictability

The transmitted symbol is generated at A only AFTER arming, from a source that no part of the apparatus could have precomputed, cached or scheduled against.

A4
Complete channel inventory

The physical interaction graph enumerates every channel, not just the logically intended ones. An unlisted channel does not weaken the theorem — it invalidates the calibration.

Scope — This is architecture-independent within its assumptions and consistent with Lieb-Robinson / quasi-locality results for broad classes of locally interacting systems. Those results are BACKGROUND, not proof for our specific hardware. The theorem's force is entirely carried by A1-A4, and each of A1-A4 is an empirical claim about a machine that does not yet exist.

Why architecture cannot win

Routing picks a path in G_phys. It does not build G_phys.

ABC0 · intended logical routeC* · moving-doorway route (better path, same graph)D_tau · fastest measured physical path (leakage, clocks, supplies, DAQ)C* may beat C0.C* cannot beat D_tau — that is the theorem.

Three zones · what is closed, what is open, what would be extraordinary

CLOSED — under assumptions A1-A4

Provable under ordinary local physics

Architecture cannot outrun the complete physical causal graph. Every result the programme produced in Phases 47-72 — the Lipschitz no-go, the phase speed limit, the two-ruler and three-ruler tests — is a special case of the same statement: routing chooses a path in G_phys, and no choice of path is shorter than the shortest path.

  • No control law can lower tau_e on an edge it does not physically change.
  • No phase transition can create an edge that is not in G_phys.
  • No moving doorway can move faster than the channel that moves it.
  • No addressing or scheduling scheme changes D_tau; it only changes which path is used.
  • Therefore: an architecture-only FTL claim is ruled out without needing any hardware.
OPEN AND WORTH BUILDING

Engineering frontier

What remains genuinely available is using latent physical edges better than a conventional design does. This is where the programme's toy architecture results actually live, and none of them require or imply new physics.

  • Exploit latent physical edges the logical design never routes over.
  • Local sparse recruitment: fewer parts of the machine wake per trip as N grows.
  • Self-restoration after a transfer, without a global rebuild.
  • Fault tolerance under adversarial edge loss (multi-tree mesh, K = 8).
  • Multi-user scaling: many concurrent trips without a central scheduler.
  • High-utilization reuse: amortized cost sublinear only past a reuse horizon.
OPEN — and almost certainly empty

New-physics frontier

Only one observation survives the audit: a reproducible arrival of UNPREDICTABLE information earlier than D_tau by a preregistered significance margin, after the no-hidden-shortcut audit passes. Anything less is architecture, systematics or an uninventoried channel.

  • Payload generated after arming; precomputation cannot help.
  • Decoding above chance at B — correlation and synchronization do not count.
  • Earlier than D_tau - epsilon_cal, where D_tau uses conservative lower-confidence delays.
  • Survives the hidden-shortcut audit including deliberate positive controls.
  • Replicated on independently built apparatuses with independent clocks and blinded analysis.
  • Even then: NOT an FTL or spacetime claim. It is a demand for a better model of the apparatus.

Empirical calibration of G_phys · frozen before any doorway run

The bound is deliberately biased in favour of boring explanations

G1
Pulse every accessible node/channel pair independently

Not just the logical neighbours. Every pair we can physically drive and observe, one at a time, with all other activity armed but quiet.

G2
Build a directed weighted G_phys from earliest significant onset

Each edge weight tau_e is the earliest STATISTICALLY SIGNIFICANT influence onset, not the peak response and not the mean. Directed, because leakage is rarely symmetric.

G3
Repeat under controller power-off

Any influence that survives with the controller unpowered is a physical channel the controller was not creating. It must appear in G_phys.

G4
Repeat under RF shielding

Shielded and unshielded runs bracket the radiative contribution. The faster of the two bounds the apparatus.

G5
Repeat with isolated clocks

Shared clock trees are a first-class channel. Independent timebases at A and B must be tested, and the clock distribution path timed on its own.

G6
Repeat under alternate grounding

Ground and supply coupling is the most commonly missed fast path. Multiple grounding topologies, each fully re-calibrated.

G7
Independent oscilloscope / DAQ cross-check

A second instrument on a separate timebase must reproduce the onset times. Instrument and cable delays are calibrated and subtracted explicitly, with the calibration published.

G8
Randomized wiring and relabeling

Physical wiring permutations and node relabeling between calibration blocks, so no fixed layout artefact and no hidden lookup can key off identity.

G9
Fit conservative lower-confidence delays

Each tau_e is taken at the LOWER confidence bound, making D_tau as small as the data allows. The bound is deliberately biased to favour conventional explanations — if the doorway still beats it, the result is hard to dismiss.

G10
Positive controls with deliberate hidden shortcuts

Install a real, undisclosed-to-the-analyst fast link between two distant nodes and verify that the calibration and audit detect it. An audit that cannot catch a shortcut we planted cannot certify one we did not.

Classification · fixed before data exists

Four outcomes. Only one is even a candidate.

ALATENT-CONNECTIVITY / ARCHITECTURE EFFECT
T_C* < T_C0 but T_C* >= D_tau - epsilon_cal

The doorway beat the logical route by using physical connectivity the logical route ignored. Interesting engineering. Zero physical novelty. This is the most likely outcome of a successful bench run.

BNO CAUSAL ANOMALY
T_C* consistent with D_tau within epsilon_cal

The apparatus behaves exactly as its measured causal graph predicts. The doorway is a routing story, nothing more.

CCAUSAL-BOUND ANOMALY CANDIDATE
T_C* < D_tau - epsilon_cal in one apparatus

NOT new physics. It means one of three things and we must assume the boring two first: an uninventoried channel, a calibration systematic, or something we do not understand. Publish the raw data and the full channel inventory, then attack it.

DINDEPENDENT CAUSAL ANOMALY REQUIRING NEW MODEL
C replicated across independently built apparatuses and labs, unpredictable payload, blinded analysis, independent clocks

A demand for a better model of the apparatus. Still NOT an FTL or spacetime claim — alternative physics and systematics must be excluded before any extraordinary language is permitted anywhere on this site.

2026 context · background, not proof

H1Established physics
Experimental Kuramoto platform — Phys. Rev. E 114, L022202 (published 21 Aug 2026)

A reported experimental Kuramoto-type oscillator platform supporting arbitrary weighted and directed topologies across strong and weak coupling regimes.

Implication — This makes Experiment v0.1 practical on real hardware sooner than we assumed — and simultaneously makes HIDDEN CONNECTIVITY the dominant confound. A platform that can wire arbitrary directed couplings is a platform that can accidentally wire the shortcut we are trying to rule out.

H2Established physics
Many-body locality literature (Lieb-Robinson / quasi-locality)

Locality results for broad classes of locally interacting systems establish emergent finite information-propagation speeds and quasi-local cone structure.

Implication — Motivation for the finite-speed causal-cone framework only. We do NOT claim these theorems bound our oscillator bench; our envelope must be measured on our own hardware.

For a ten-year-old

If every road has a speed limit, a smarter GPS cannot make a message arrive before every road could carry it. It can only find a better road. A true surprise would be the message arriving before any road we measured could possibly get it there.