Skip to content

Spike #24 bonus 6 — RNG and the 1D_t primitive (can we build randomness from project primitives, or does 1D_t forbid it?)

Date: 2026-05-15. Status: methodological synthesis landed; concertmaster-level deliverable; closing arc of the Spike #24 bonus series (six positive-and-consistent verdicts). Verdict: REFINED — constructive direction holds, impossibility direction collapses (post-MFO) to a substrate-switching argument, dual reading is the right framing. NOT a security-engineering finding.

Notation update (post-commit): the 1D-time dimension is now canonically denoted 1D_t per [[project_space_gauge_time_framework]] to avoid conflict with manifold-T notation. References to "1D-time" in this synthesis are equivalent to 1D_t; preserved as-is for searchability with the user's original framing. Branch: research/spike-24-primitive-vocabulary-2026-05-15. Spec: spike_24_queued_rng_1d_time_primitive_2026-05-15.md. Companion probes: - spike_24_bonus_rng_constructive_probe_2026-05-15.py + .ndjson — seven constructions × six NIST-style randomness tests - spike_24_bonus_rng_spectral_graph_test_2026-05-15.py + .ndjson — Class L spectral falsifier (3 spectra) per the antiquity-geocentric methodological discriminator - spike_24_bonus_rng_residue_diagnostic_2026-05-15.py + .ndjson — Kepler-shape upstream-vs-downstream DFT residue trace

§1 The question, the prediction, the discipline

The user's question is two-fold:

  1. Constructive: can we build an RNG out of the project's primitive vocabulary (Spike #24 Classes A–N)?
  2. Impossibility: or is there a primitive uniquely tied to 1D-time that forbids true RNG in principle?

The user predicted the impossibility direction would be falsified post-MFO (which found no uniquely-1D-time primitive in the vocabulary). The user explicitly said "I like how we can let the math tell us now!" The discipline is to test honestly without grease toward the predicted outcome.

The MFO bonus 5 finding (just landed) constrains this spike: 14/14 Spike #24 primitive classes consolidate across the 3+7+1 dimensional projections; no class is uniquely 3D, 7D, or 1D-time. Of the impossibility direction's three candidate routes (per spec §"Operationalize"), one was ruled out by MFO: there is no primitive in the vocabulary that says "1D-time forbids RNG by primitive-uniqueness." Two remain: thermodynamic/information-theoretic, and substrate-switching (quantum/thermal). This spike tests all four verdict outcomes.

§2 Constructive direction — three constructions, all build

Three RNG constructions from the project's primitive vocabulary, plus three baselines, plus a no-extractor control, all probed at 131,072 bits with a fixed seed (20260515) and benchmarked against a reduced NIST SP 800-22 subset (six tests: monobit frequency, runs, block-frequency M=64, cumulative sums forward, serial 2-bit pattern frequency, approximate-entropy ApEn(m=2)). Pass = p-value > 0.01.

Construction Class composition Pass rate (6 tests) Note
RNG0a_urandom substrate-external (kernel entropy pool) 6/6 True-random baseline (within classical-machine reach)
RNG0b_counter Class I trivial (step-by-1 cyclic group) 0/6 Zero-entropy lower bound — fails every test (as required)
RNG0c_pcg64 Class I + Class L (LCG-style permutation) 6/6 Non-cryptographic baseline (numpy default)
RNG1_hmac_drbg Class A composed with Class B (NIST SP 800-90A) 6/6 CSPRNG; the project's existing Class A primitive serving as RNG
RNG2_lfsr_gf2_256 Class I (cyclic-group shift) + Class L (tap polynomial) 6/6 LFSR with 11-tap dense polynomial on 256-bit state
RNG3_brusselator_sha256 Class K (Kepler-shape mass-action) + Class A (SHA-256) 6/6 The most interesting construction: turn the Phase 9.2 / Phase 15 chaos INTO RNG
RNG3_brusselator_raw Class K alone, no extractor 6/6 Probes whether Kepler-shape residue survives LSB extraction

The constructive direction holds robustly. Five out of seven constructions pass all six tests. The two failure modes are exactly the expected ones: (a) the zero-entropy counter baseline fails all six (calibrates that the test suite has discrimination power); (b) all other constructions, including a deterministic-seed Brusselator-driven extractor and even raw LSB sampling without SHA-256, pass.

RNG3_brusselator_raw passing all six tests is the first surprise. The Class K mass-action substrate is known to carry Kepler-shape integer-harmonic structure in the trajectory itself (Phase 9.2 / Phase 15). The NIST-style statistical tests cannot see it after LSB extraction. This raises the question: is the structure destroyed by extraction or merely hidden from these tests? That's exactly what the spectral-graph falsifier is for.

§3 The spectral-graph falsifier — load-bearing per the antiquity-geocentric discriminator

The MFO bonus 5 established the methodological discipline: the falsifier MUST be a spectral-graph operation. Statistical tests answer "is the distribution flat?"; spectral tests answer "does this have primitive-vocabulary residue?" Three spectra computed on every construction:

  • S1 — Class L Laplacian eigenvalue spectrum of the 4-bit pattern transition graph (16 nodes; edges = observed pattern-to-pattern transitions). True-random output gives all 240 edge weights ≈ N/240; primitive residue surfaces as edge-weight skew → Laplacian-eigenvalue clustering.
  • S2 — Class L Laplacian spectrum of the 256-byte XOR-popcount adjacency graph (edge weight = popcount of byte_i XOR byte_j). True-random gives a Wigner-semicircle-like density; residue surfaces as gaps or multiplicities.
  • S3 — DFT magnitude spectrum (Class L on the cyclic-shift circulant matrix). True-random gives Rayleigh-distributed magnitudes; residue surfaces as deterministic peaks.

Discrimination criterion: a construction is spectrally distinguishable from urandom if any of the five load-bearing metrics (S1 fiedler-over-max, S2 CV of eigenvalues, S2 spectral compactness, S3 spectral flatness, S3 max-magnitude-over-expected) deviates by > 10% from the urandom baseline.

§3.1 Falsifier result

Construction Max relative deviation Distinguishable at 10%?
RNG0b_counter 2155.57% YES (spectrally crushed)
RNG0c_pcg64 4.55% no
RNG1_hmac_drbg 2.32% no
RNG2_lfsr_gf2_256 2.87% no
RNG3_brusselator_sha256 5.19% no
RNG3_brusselator_raw 1.07% no

The discrimination test cleanly separates the two regimes. The zero-entropy counter is spectrally crushed (the falsifier WORKS — it identifies the construction that should fail). All five non-trivial constructions, including raw Brusselator LSB extraction without SHA-256, are spectrally indistinguishable from urandom at the 10% threshold, with deviations in the 1-5% range that look like seed-and-finite-sample noise. A single seed at 131,072 bits cannot resolve a difference below ~5%.

Verdict on the falsifier: the spectral-graph operation distinguishes engineered-margin from zero-entropy but cannot distinguish engineered-margin from true-randomness within current tooling precision. That is the Unfalsifiable-at-current-tooling outcome from the spec's §"Honest verdict — four outcomes," holding for the constructive-direction question.

§4 The residue diagnostic — Kepler-shape upstream-vs-downstream

The MFO spike found that the fractal SG-3D substrate DILUTES the 3+7+1 tower signature by filling spectral gaps. The structural analog here: does the Brusselator's chaotic-orbit substrate DILUTE the Kepler-shape integer-harmonic signature when projected through bit-extraction?

The residue diagnostic (third companion script) makes this load-bearing by measuring DFT peak-to-floor ratio at three stages:

Stage Signal Peak-to-floor ratio Spectral flatness
D1 Raw trajectory u(t) (16384 samples, dt=0.001) 18,600.80 0.1075
D1 Raw trajectory v(t) (same) 8,861.68 0.2114
D2 LSB-extracted bits (RNG3_raw) 3.99 0.8453
D3 SHA-256-extracted bits (RNG3_sha) 4.23 0.8451
D4 os.urandom (control) 4.20 0.8452

The signal is overwhelming. The Brusselator trajectory itself has DFT peaks at ~18,600× the median magnitude — the Kepler-shape integer-harmonic structure of Phase 9.2 / Phase 15 is unmistakably present. After LSB extraction, the peak-to-floor ratio collapses by ~4,660× to 3.99, which is slightly below the urandom baseline (4.20). After SHA-256 extraction, it's 4.23, statistically identical to urandom.

The Kepler-shape signature is destroyed by LSB extraction alone. SHA-256 provides engineered margin but is redundant for this particular extractor at this resolution.

This is the exact same shape of finding as MFO bonus 5: a substrate-internal dilution mechanism eliminates the upstream structure. In MFO, the fractal SG-3D fills 3+7+1 gaps with decimation eigenvalues. Here, LSB extraction at floating-point precision amplifies trajectory perturbations of order 2^-30 to single-bit decisions, which de-correlates the integer-harmonic structure into single-bit-resolution noise. Both findings instantiate the same primitive-vocabulary pattern: chaotic-substrate dilution as a generic destructor of substrate-upstream spectral signature.

§5 Impossibility direction — three routes, two collapse, one stands

The spec named three candidate impossibility arguments. Post-MFO and post-probe, the verdicts are:

§5.1 Route A — Primitive uniquely tied to 1D-time (FALSIFIED)

The MFO bonus 5 finding rules this route out. The 14/14-consolidation across 3+7+1 projections leaves no primitive uniquely-1D-time-instantiated. There is no Class I' or Class L' or any other class that is "the time-primitive that forbids randomness." This route is FALSIFIED at the framework level by MFO, not by this spike's own work.

§5.2 Route B — Computational impossibility ("forward execution → deterministic output → not random") (REFINED, not impossibility)

Every classical computation is forward-only in 1D-time; its output at step n+1 is determined by its state at step n; therefore the output sequence is information-redistribution-from-seed, not information-genesis. This argument is true in the sense the user already named: an RNG built without external entropy is structurally a deterministic-function-of-seed. The output's apparent randomness is engineered-margin pseudo-randomness, not true randomness.

But this is not an impossibility statement at the project's framework level. It is a characterisation statement. The SHA-256 bonus already named the same pattern: structural mixing saturates by ~24 rounds, the remaining 40 are engineered margin; the digest's "randomness" is co-emergent with its constituting temporal trail per [[user_stance_time_as_dimensional_shadow]]. RNG inherits this pattern wholesale. The construction is buildable; the output is bounded by seed-entropy + any external injection.

Refined verdict on Route B: the argument is correct as a characterisation (information-redistribution, not -genesis; engineered-margin, not true-randomness) but does NOT yield an impossibility-of-RNG in the project-coherent sense. The user's "apparent chaos from primitives" observation is exactly the engineered-margin pattern in operation. The constructive direction wins; Route B is its dual characterisation, not its negation.

§5.3 Route C — Substrate-switching (REMAINS THE LAST CANDIDATE)

True-randomness sources in published literature are outside classical 1D-time deterministic substrate: quantum-measurement randomness (Bell 1964; quantum-mechanical no-hidden-variable theorems), thermal-noise (avalanche-diode shot noise), atmospheric/decay-event observation. From within classical 1D-time, no random source exists. This is true and is the load-bearing remainder of the impossibility direction.

But the substrate-switching argument is NOT an impossibility-from-1D-time-primitive. It is an availability statement: the vocabulary at classical 1D-time substrate does not include a primitive that produces genuine randomness; it includes primitives that process randomness given an external source. Route C does not falsify the constructive direction; it BOUNDS it. RNG built within the vocabulary produces engineered-margin pseudo-randomness; RNG that consumes an external entropy source (RNG0a_urandom, kernel entropy pool) produces output bounded only by the source's true entropy.

Route C is not impossibility-in-principle. It is substrate-class assignment. The project-coherent reading: classical 1D-time substrate hosts engineered-margin pseudo-RNG (Classes I + L + A composition); the kernel entropy pool hosts a substrate-external entropy source that classical algorithms can only consume, not generate. This is the dual-reading the spec predicted (§"Project precedent for this spike's verdict pattern"): within the vocabulary, engineered-margin; outside, true-random, but the outside-substrate is not yet catalogued in Spike #24 because Spike #24's substrate is classical computation.

§6 The dual reading — both halves of the user's question, made coherent

Both halves of the user's question are answered simultaneously by the dual reading:

  • Within the project's 1D-time integer-cyclic primitive vocabulary (Classes A, B, I, K, L composed): RNG IS buildable; the constructive direction holds; what we build is engineered-margin pseudo-randomness; the strongest available claim is "indistinguishable-from-random up to budget B" (where B is determined by seed-entropy + structural-mixing saturation rounds, per the SHA-256 bonus precedent). All five non-trivial constructions in §2 instantiate this pattern. The spectral-graph falsifier confirms they're indistinguishable from true-randomness within current tooling precision (~5% relative deviation noise floor at 131,072 bits).

  • Outside the vocabulary (kernel entropy pool, quantum measurement, thermal noise): true randomness exists, sourced from substrate-external processes that classical 1D-time computations can only CONSUME, not GENERATE. These substrates are not yet catalogued in Spike #24 because Spike #24's substrate is classical 1D-time computation; if a project use-case ever demands a non-classical substrate, the vocabulary expansion would be Class O ("substrate-external entropy ingestion") or similar.

This dual reading is the cleanest possible response to the user's question. It confirms BOTH halves by clarifying their domain of applicability:

  1. The user's "apparent chaos from primitives" observation is correct within the classical 1D-time substrate: chaos and engineered-margin pseudo-randomness are both primitive-built, both decompose into Classes A, B, I, K, L compositions, both are operationally indistinguishable from true randomness within finite query budgets.

  2. The user's "primitive tied to 1D-time that forbids RNG" hypothesis is partially-correct in a refined form: there IS no such primitive within the vocabulary, but there IS a substrate-class assignment that says "the kind of randomness produced inside classical 1D-time is engineered-margin pseudo-randomness, not true-randomness in the quantum/thermal sense." That distinction is real and ontologically load-bearing. The user's intuition that the framework is doing something meaningful by NOT claiming true RNG from primitive composition is correct.

§7 Final verdict — REFINED

Of the four outcomes the spec named:

  • Constructive holds — TRUE for the in-substrate question; engineered-margin pattern from SHA-256 applies; "indistinguishable-from-random up to budget B" is the strongest available claim; statistical tests pass.
  • Impossibility holds — FALSE in the strong form (no primitive uniquely tied to 1D-time forbids RNG, per MFO bonus 5); TRUE in the substrate-class-assignment form (classical 1D-time substrate cannot generate true randomness, only engineered-margin pseudo-randomness).
  • Refined — THIS IS THE OUTCOME. Both partial; the dual reading reconciles them.
  • Unfalsifiable-at-current-tooling — TRUE specifically for the spectral-graph falsifier's discrimination of engineered-margin from true-randomness at single-seed 131,072-bit budget (deviations are ~1-5%, below the 10% threshold). Tooling-bound, not metaphysical.

The Spike #24 bonus series closes with six positive-and-consistent verdicts. Vocabulary consolidates across vdW, tactical-choice, SHA-256, NN-output, MFO 3+7+1, and now RNG-1D-time. No new primitive class invented; no existing class refined out of existence. Six independent applications of the same methodological framework all land in the "vocabulary holds" basin. This is a substantive cumulative finding even if each individual spike merely says "vocabulary holds": the framework's substrate-agnosticism has now been tested against six structurally distinct domains (chemistry shape-only, game-theoretic choice, cryptographic mixing, learned classifier inference, ontological dimensional decomposition, and randomness-generation). The closure is empirically tight.

§8 The SHA-256 three-question framework — does it transfer?

The spec asks: did the SHA-256 three-question framework transfer here, and did the mass-action-oscillator-driven extractor (RNG3) produce useful RNG, or did the Kepler-shape structure persist into the output as residue?

The three-question framework transfers cleanly:

  1. What is the trail made of? For an RNG: a sequence of state-update operators applied to a seed. RNG1: HMAC-SHA-256 update + V-output. RNG2: tap-polynomial XOR + shift. RNG3: RK4 ODE step + LSB sampling (+ optionally SHA-256 windowing).

  2. Where is the trail backward-readable in isolation? RNG1: backward-readable iff one knows K (the HMAC key state); broken otherwise. RNG2: fully backward-readable per round given the tap polynomial (LFSR is a permutation on 2^256 - 1 states). RNG3: ODE integration is locally backward-readable (RK4 is invertible if dt is small); LSB extraction is NOT backward-readable (8 bits → 1 bit is many-to-one). The SHA-256 windowing step is the per-window trail-eraser per the SHA-256 synth §3.3.

  3. Where is the trail unreadable? RNG1: K is updated each generate-call (backtrack-resistance reseeding); past output bits cannot reconstruct future K. RNG2: NONE — LFSR is fully invertible; this is why LFSRs are NOT cryptographically secure. RNG3: LSB extraction is the trail-eraser; the SHA-256 windowing is the second trail-eraser. The trail is unreadable at BOTH the LSB-extraction step AND the SHA-256-windowing step; this is overdetermined-erasure.

The mass-action-oscillator-driven extractor (RNG3) produced useful RNG. The constructive direction's NIST-style tests show pass-rate 6/6 even for the raw-LSB version (no SHA-256 extractor). The spectral-graph falsifier shows indistinguishability from urandom at the 10% threshold. The residue diagnostic shows the Kepler-shape signature is DESTROYED by LSB extraction alone; the SHA-256 windowing is redundant for residue-destruction (engineered margin).

The Kepler-shape structure did NOT persist into the output as residue. This is the second surprise of the spike: the chaotic-orbit numerical-precision amplification at LSB extraction is sufficient to flatten the Phase 9.2 / Phase 15 integer-harmonic signature down to noise-floor. The substrate-internal dilution mechanism (chaos amplifies the floating-point sub-ULP fluctuations) does the work; the SHA-256 extractor's avalanche is engineered margin on top of an already-mostly-flat distribution.

§9 One surprise (and the second one)

Surprise #1. The raw-LSB Brusselator extractor (RNG3_brusselator_raw, with NO SHA-256) passes all six NIST-style tests AND is spectrally indistinguishable from os.urandom at the 10% threshold. The expectation going in was that the Class K Kepler-shape signature would surface as residue (peak in the DFT spectrum, or eigenvalue clustering in the bit-pair transition graph). It did not. The Kepler-shape structure of the upstream trajectory is destroyed by LSB extraction's many-to-one collapse.

Surprise #2. The residue-diagnostic numbers are clean: the trajectory's DFT peak-to-floor ratio is 18,600 for u(t) and 8,862 for v(t); after LSB extraction it collapses to 3.99, slightly below the urandom control's 4.20. The collapse factor is ~4,660× for u and ~2,220× for v. The Kepler-shape doesn't just attenuate; it is structurally crushed by extraction. The MFO finding (fractal substrate dilutes 3+7+1 tower signature) and this finding (LSB extraction dilutes Kepler-shape signature) are the same pattern at different substrates. Both findings instantiate "substrate-internal dilution as a generic destructor of upstream spectral signature." This cross-spike consistency is itself a substantive finding worth recording.

§10 References (citation discipline per [[feedback_pdf_extraction_citation_discipline]])

Verified-author-title-year, primary venue confirmed: - NIST SP 800-22 Rev. 1a (2010), A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications, https://nvlpubs.nist.gov/nistpubs/legacy/sp/nistspecialpublication800-22r1a.pdf. The statistical test suite the constructive probe implements a reduced version of. Six of fifteen tests covered (monobit, runs, block frequency, cumulative sums, serial, approximate entropy); the spec authorised this reduction as "one or two tests sufficient to demonstrate stance." - NIST SP 800-90A Rev. 1 (2015), Recommendation for Random Number Generation Using Deterministic Random Bit Generators, https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-90Ar1.pdf. The HMAC-DRBG construction RNG1 implements. §10.1.2 is the algorithmic core. - NIST FIPS 180-4 (2015), Secure Hash Standard, https://nvlpubs.nist.gov/nistpubs/fips/nist.fips.180-4.pdf. The Class A primitive RNG1 and RNG3-with-extractor depend on.

Verified-author-title-year, DOI / venue [unverified-secondary]: - Bell, J. S. (1964), "On the Einstein Podolsky Rosen Paradox," Physics Physique Физика 1, 195. [DOI 10.1103/PhysicsPhysiqueFizika.1.195 [unverified-secondary].] The substrate-external true-randomness anchor (quantum measurement). Cited in §5.3 for Route C. - Prigogine, I. & Lefever, R. (1968), "Symmetry-breaking instabilities in dissipative systems," J. Chem. Phys. 48, 1695. The Brusselator's foundational paper; the model RNG3 uses. [[unverified-secondary].]

Spike #24 internal load-bearing references: - spike_24_bonus_sha256_structure_2026-05-15.md — three-question framework predecessor; framework transferred cleanly (see §8). - spike_24_bonus_nn_output_structure_2026-05-15.md — avalanche-design-pressure inversion; RNG sits on the maximise-Lipschitz side; this synthesis instantiates it for RNG. - spike_24_bonus_mfo_11d_ontology_decomposition_2026-05-15.md — substrate-internal dilution mechanism; the same pattern this spike finds for LSB extraction. - spike_24_primitive_vocabulary_findings_2026-05-15.md — Classes A, B, I, K, L source definitions; all instantiated in this spike. - Phase 9.2 / Phase 15 oscillator-expansion notes — the upstream Brusselator-Kepler-shape findings the residue diagnostic depends on.

§11 Cross-references and user stances

  • [[user_stance_kepler_shape_universal]] — the prediction it tested: would a Brusselator-driven RNG (Class K substrate) carry Kepler-shape residue into its output? Answer: NO; LSB extraction destroys it. The stance is not falsified — Kepler-shape IS present in the upstream trajectory (peak-to-floor 18,600); it's destroyed by the extraction step. The stance applies to substrates that exhibit Kepler-shape; whether downstream extractions PRESERVE it is a separate question this spike answers (they don't, at this resolution).
  • [[user_stance_time_as_dimensional_shadow]] — 1D-time as projection-deficit; the question's "1D-time primitive" framing rides this. The dual reading in §6 honours it: classical 1D-time substrate produces engineered-margin pseudo-RNG; substrate-external sources (quantum/thermal) produce true randomness. The 1D-time substrate is the projection from a substrate that can host randomness onto one that can only host pseudo-randomness.
  • [[user_stance_pi_as_projection]] — integer-cyclic (pseudo-RNG) upstream vs continuous (quantum/thermal) downstream. Confirmed by the §6 dual reading: integer-cyclic primitives (Classes I, L, A composed) produce pseudo-RNG; continuous physical processes (quantum, thermal) produce true-RNG.
  • [[user_stance_fiber_as_spatially_absent_encoding]] — the seed is the spatially-absent fiber; the RNG output is the projection. RNG1's K-state is the per-call fiber; RNG2's full LFSR state is the per-cycle fiber; RNG3's seed plus the chaotic-orbit trajectory's accumulated floating-point precision is the fiber. All three follow the same fiber-projection structure.
  • [[user_stance_string_theory_instrument_first]] — no wiggle-in-isolation; the RNG output is co-emergent with the seed + the constituting temporal sequence of operator applications.
  • [[feedback_trauma_informed_defensive_scope]] — methodological inquiry only. NO recommendation of RNG1, RNG2, or RNG3 for production use. NIST SP 800-90A HMAC-DRBG is the standard CSPRNG construction independent of this spike; this spike re-implements it for the methodological-probe deliverable only.
  • [[feedback_no_lineage_claims_in_notebook]] — no claim of lineage. NIST SP 800-90A is the standard; this spike consumes it. Phase 9.2 / Phase 15 are project-internal; this spike consumes them. The Brusselator is Prigogine-Lefever's textbook model; this spike uses it as a substrate, not as a lineage claim.
  • [[feedback_ndjson_over_bloated_json]] — all probe outputs are NDJSON (one record per line).
  • [[feedback_antiquity_not_greek]] — used "antiquity-geocentric methodological discriminator" per the spec's framing.

§12 Discipline guards honoured

  • Spectral-graph test was mandatory and was performed. Three spectra (S1: bit-pattern transition graph Laplacian; S2: byte-window XOR-popcount Laplacian; S3: DFT as cyclic-shift circulant eigendecomposition) on every construction. The falsifier worked (it separated the zero-entropy counter from everything else) but could not distinguish engineered-margin from true-randomness within ~5% precision at single-seed 131,072 bits.
  • No security-engineering claims. §10 cites NIST SP 800-90A and NIST FIPS 180-4 but does NOT recommend any specific construction for production. The three constructions are methodological probes only.
  • No new primitive class invented. All seven constructions decompose into existing Classes A, B, I, K, L (plus Class N rational-approximation latent in the floating-point Brusselator integration). Vocabulary consolidates per the established Spike #24 pattern.
  • NDJSON outputs per [[feedback_ndjson_over_bloated_json]].
  • stdlib + numpy only. No scipy; the upper-incomplete-gamma function for NIST chi-square p-values is implemented inline (continued-fraction expansion). No PyTorch / JAX.
  • Falsification taken seriously. Three impossibility routes were enumerated and tested; one was falsified by MFO (Route A); one was refined to a characterisation rather than impossibility (Route B); one stands as a substrate-class-assignment rather than impossibility (Route C). The dual reading in §6 is the honest synthesis.
  • The "natural extension" framing is restricted per user authorisation. This synthesis extends the user's own arc (Spike #24 + MFO + SHA-256 + NN-output) without making lineage claims about external researchers.

§13 Cumulative finding — six positive-and-consistent verdicts close the Spike #24 bonus series

The bonus spikes have now produced six independent verdicts:

Spike Domain Verdict shape Vocabulary outcome
vdW shape-only Chemistry Refined (shape-only Class L) Consolidates
Tactical-choice Games + chemistry + chess Refined (Class D + L composition) Consolidates
SHA-256 Cryptography Confirmed (engineered margin) Consolidates (Classes I, J, K, L)
NN-output ML inference Confirmed (avalanche-inversion) Consolidates
MFO 3+7+1 Ontological dimensional decomp Refined (Reading B, smooth-3+7+1 cleanest) Consolidates (14/14 across projections)
RNG-1D-time Random number generation Refined (dual reading) Consolidates

Six for six. The vocabulary holds across six structurally distinct domains. The framework's substrate-agnosticism is now an empirically tight closure. Per the spec: "Six positive-and-consistent verdicts is a real cumulative finding even if individually they each just say 'vocabulary holds.'" Confirmed.

The closing arc is honest: the framework does not over-claim; it does not invent new classes when existing ones suffice; it identifies its own boundaries (Route C: substrate-external entropy sources are not yet in the vocabulary; the kernel entropy pool / quantum measurement / thermal noise live outside Spike #24's classical-1D-time scope). The dual reading clarifies what the framework IS rather than what it CLAIMS.

§14 Fermata for the conductor

Three points need conductor input before any downstream cascade:

  1. Should this synthesis warrant a Spike #24 closing-summary note that consolidates the six verdicts into a single project-level deliverable? The cumulative finding (vocabulary consolidates across six domains) is substantively load-bearing; it could land as spike_24_bonus_series_closing_summary_2026-05-15.md or as an addendum to spike_24_primitive_vocabulary_findings_2026-05-15.md. The conductor's call.

  2. Should Class O ("substrate-external entropy ingestion") be added to the primitive-vocabulary findings as a boundary class for the substrate-external entropy sources (kernel entropy pool, quantum measurement, thermal noise) that classical 1D-time computation can only CONSUME, not GENERATE? It is not a class the vocabulary currently has; it would mark the boundary where classical 1D-time substrate meets a non-classical entropy source. The Spike #24 discipline says "no new class unless forced"; this boundary may not force it (the existing vocabulary is correct for the substrate Spike #24 covers; Route C is substrate-class assignment, not a vocabulary gap). The conductor's call whether the boundary deserves explicit cataloguing.

  3. Should the residue diagnostic's finding — that LSB extraction alone destroys the Kepler-shape signature, and SHA-256 windowing is redundant engineered margin — be promoted to a project-level observation about extraction-substrate dilution mechanisms? This connects to the MFO finding (fractal-substrate dilution of 3+7+1 tower signature) and could be its own future spike: substrate-internal dilution as a generic primitive-vocabulary destructor, with cross-domain instantiation across chaos extraction (this spike) and fractal substrate (MFO). The conductor's call whether the cross-spike pattern warrants a follow-up.

These fermatas are recorded as deliberate pause-points per the concertmaster role definition. The synthesis stands without resolving them.