The epistemic vertical: an audit of the floor structure

A theory of everything faces one degeneration mode more dangerous than error: accretion — layer upon layer of convincing narrative over bare mathematics, each locally plausible, none load-bearing. The defence is not modesty but discipline: every organizing reading must name what breaks without it.

Document status

This is a meta-level document: an [О]-instrument (its criteria are definitions) applied in [И]/[Т]-graded audits. It does not prove theorems; it audits readings — the organizing structures through which the corpus arranges its theorems — and issues verdicts with statuses. It binds future canonizations (§8). Everything it says about specific theorems carries the statuses those theorems already have in the status registry.

Why this document exists. The arc T-222…T-236 naturally arranged itself into three floors: the Universe as Holonom (the axiomatic top), certified organisms (the middle), foundations of mathematics (the newly opened bottom). Before any such arrangement is canonized, the corpus must answer a question that is itself a fundamental question of UHM's meta-level — its epistemic potential: by what right? A theory of everything must hold its content to the highest standard precisely because everything can be narrated into it. The corpus already has micro-level hygiene: the status discipline [Т]/[С]/[Г]/[П]/[О]/[И], audit chronicles with same-day retractions (T-235 is the standing example), articulation hygiene, the falsifiability registry, machine verification (M1; the exhaustive computations of the Σ-wave). What it lacked was meso-level hygiene: criteria for organizing readings — framings that arrange many theorems at once. This document supplies them, applies them, and records the verdict.

§1. The failure mode being guarded against

Call it the layered-cake degeneration: the theory's mathematics stays sound, but its self-presentation grows by superposed framings — each new wave of results wrapped in a new grand reading, none of which is ever tested for necessity. The end state is a corpus that looks like a worldview and works like a disconnected pile of lemmas. The user-level symptom: a reader can delete a whole framing paragraph and no derivation, prediction, or verification anywhere loses anything.

The opposite of accretion is crystallization: the theory's evolution should reduce the number of independent framings by showing that several are one — as the arc did when the composition ceiling and the federation bound turned out to be two projections of one 23-unit geometry (T-232), or when the "Lie shadow" of the generator algebra reappeared as the forbidden subspace of observable decoherence rates (T-226). Canonization is legitimate exactly when it crystallizes; it is illegitimate when it decorates.

§2. The criteria: K1–K6 [О]

An organizing reading (a "floor", a governing metaphor, a cross-domain identification) is canonizable iff it passes all six:

#	Criterion	Test
K1	Load-bearing	Name what breaks if the reading is deleted: a derivation route, a verification, a prediction. If nothing checkable breaks, the reading is foam. (This extends the corpus's articulation-hygiene test from sentences to structures.)
K2	Two-way flow	At least one result must have flowed into the floor from outside, and at least one out of it into another floor. A layer that only consumes results from below is decoration.
K3	Invariant formulation	The reading must be statable as one structural sentence with statuses attached — not only as prose atmosphere.
K4	Falsifiable trace	The floor must own at least one falsifiable prediction or machine-checkable artifact.
K5	Irreducibility	The floor must contain theorems not expressible in the vocabulary of the other floors. Otherwise it is a renaming.
K6	Hole register	The floor must publish its open holes — the [Г]/[П]/[И]-places where the reading outruns the proofs — in an explicit register. Canonization without the register is prohibited. An empty register is suspect by default.

§3. Audit of Floor I: the Universe as Holonom

Invariant formulation (K3). The total state of reality is static (Axiom A5, Page–Wootters: the Hamiltonian constraint is zero; time is relational); its self-model is terminal (T-222: the Lawvere fixed point is the Pareto-optimum of the full resource vector); its organization is coinductively multi-level (the fractal holon $\nu X.\, \mathcal{D}(\mathbb{C}^7) \times \mathrm{Multiset}(X)$); nothing is imported from outside. Statuses: [П] for the axioms, [Т] for the theorems named, [О] for the carrier.

Evidence base. A5 (#pw-constraint) [П]; the corpus's own foundational text: "everything is derived from internal structure; nothing is imported from outside"; T-120 (one world — derived) [Т]; T-186 (cohesive closure) [Т]; T-211 (non-fragmentation — definitional) [Т]; T-221 (site-relativization: observers as internal sections) [Т]+[И]; T-222 (terminality of $\rho^*$, in the Markov domain via majorization) [Т]; Gap as the holonomy of the ∞-topos connection (#gap-голономия) [Т/О].

K-scores. K1 ✓ — delete A5 and background independence collapses (external time re-enters; the corpus states this explicitly); delete terminality and the viability-first derivation chain of the organisms floor loses its anchor. K2 ✓ — downward: monism is what forces the $N=7$/$G_2$ programme rather than parameterizing it; upward: T-222's terminality, proven at the Γ-level, returned to this floor as the reading "the static structure is not merely invariant but terminal". K3 ✓ (above). K4 — partial: floor-level falsifiability is indirect (the Λ-deficit computational programme; $\pi_{\mathrm{bio}}$ as a consciousness-side discriminator); this is declared, not hidden — see the register. K5 ✓ — one-world, PW-staticity, and site-relativization are not expressible at organism scale.

The honesty core of this floor — three declared holes (K6):

• H1.1 The phrase "the Universe has the self-model $\rho^*$" is an [И]-extension of T-222: the theorem is proven for Γ-level holons; its cosmological instance — the self-model of the total PW state — is a reading licensed by the monism axiom, not a computed theorem about cosmological data. It is legitimate as a reading of the axioms (that is what floor I is), and it must be said this way.
• H1.2 The stage of the Universe — its purity $P$ on the ladder $P^{(n)}_{\mathrm{crit}}$ — is neither derived nor measured [Г]. "The Universe is at a certain stage of viability" is structurally meaningful (the ladder exists [Т]) but numerically open. Closing it is a research object adjacent to the Λ-deficit programme.
• H1.3 "One grammar at all levels" is conditional: Theorem Σ selects the Fano grammar wherever its axioms (D1–D3) hold [Т при D1–D3]; asserting that every level of reality satisfies D1–D3 is [П]/[И]. The scale-freeness is a theorem-shaped conditional, not an unconditional fact.

Verdict. Legitimate — with a precise character: Floor I is not a new layer added on top of the mathematics; it is the axiomatic layer itself, read aloud. The 222+ arc did not create it; it strengthened it (terminality, invariant self-description). Canonization of an overview document ("the Universe as Holonom") is admissible only together with the register H1.1–H1.3 embedded in it.

§4. Audit of Floor II: certified organisms

Invariant formulation (K3). The unit of mid-scale reality is the certified organism: viability is primary (T-222), consciousness is its licensed regime (the four-condition window), self-diagnosis is perfect on seven axes (T-224/T-225/T-226), vertical growth is bounded by three and continues as ecology (T-232, K.5, T-228).

Evidence base and K-scores. This is the strongest floor and the reference standard for what "legitimate" looks like: K1 ✓ (delete viability-first and the SYNARC specification's derivation order collapses — the spec is organized by it); K2 ✓ (downward: the floor consumed T-222/T-223; upward: the tower arithmetic $U(m) = 8m-1$, discovered here, exported to foundations as T-232 and fed the Σ-Mor programme); K3 ✓; K4 ✓✓ — the machine-verified M1 core, the falsifiable batteries K-P1–K-P4, Σ-P1/Σ-P2, FSQCE E4–E7; K5 ✓ (interoception stacks, composition ceilings, federation grammars have no formulation at the axiom floor or the foundations floor); K6 ✓ — its holes are already registered in place: the (P)-genericity note (closed by T-240: verified against R1–R5, with the iso-comma correction; (P) a theorem on intensional fibers), the two-ceilings question (mechanism identity closed in the negative by T-239; the root of the value agreement carried as H2.1′ [Г]), the unverified predictions (falsifiable by design).

Verdict. Fully legitimate. No conditions.

§5. Audit of Floor III: foundations

Invariant formulation (K3). Intensionality is the geometry of strictness defects (T-233); its depth is finite and code-valued (T-229/T-230); its self-diagnosability is a constructive resource (T-231); free combination destroys deep diagnosability and binding preserves it (T-234); its perfect form is Fano — realized so far as a designed blueprint (T-236), with natural occurrence open.

The honesty core — the designed/discovered split. The claim "the same grammar appears in matter, mind and mathematics" must be graded, and this audit grades it:

Instance	What it is	Status
The $\varphi$-cocycle in Γ-dynamics (rates, T-226) and in quantum protection (T-227)	discovered: two computations of literally the same finite-geometric object	[Т]
The selection arithmetic applied to foundations (T-229/T-230/T-231/T-234)	discovered: theorems about charted fibers, proven at the stated axioms	[Т при аксиомах карт]
The full Fano grammar in a foundation (T-236)	designed: a transplant — the blueprint realizes the grammar by construction; whether mathematics natively produces it is exactly the Fano-foundation problem	[О] + [Г]

Accordingly, the phrase "monism closed over mathematics with three computed instances of one cocycle" — used once in working notes — is downgraded by this audit to its correct form: two natural instances [Т] plus one constructed blueprint [О], with the naturalness question open [Г]. This is the difference between a theory crystallizing and a theory flattering itself, and the corpus chooses the former.

K-scores. K1 ✓ — delete the floor and T-230/T-231/T-234 vanish: real theorems about real foundations (the four-rung bound on intensional depth of charted foundations is not expressible elsewhere). K2 ✓ — inward: the tower arithmetic came from the organisms floor; outward: T-234's binding principle returned to the organisms floor as a design law for federations, and T-233's strict/pseudo mechanism is now the named candidate for closing the blueprint's own completeness gap. K3 ✓. K4 ✓ — exhaustive machine verifications (class computations, the 1344-group, the ladder to $m \leq 4096$); the blueprint itself is a checkable artifact (formalizable in Verum/Agda). K5 ✓ — fibrancy failures, $\tau$-syndromes, covering radii of display codes have no organism-floor or axiom-floor formulation. K6 ✓ — register below.

Holes (K6). H3.1 ΣQ1′ (does genuine intensionality force $d_{\min}\geq3$?) — closed on the frame by T-246: faithfulness is a corollary of division (T-244) — distinct defects superpose to a third defect $e_ie_j=e_{i+j}$, never an equivalence, so no confounded pairs and $d_{\min}=3$; on $\tau=1$ finite-signature fibers finiteness follows from former-localized uniform failure, forcing the Fano frame (residual $=$ the N$=7$ selection per signature; T-231 bounds $\tau=0$); H3.2 syndrome completeness of the blueprint — closed by T-237 (strict gauge; the anticipated pseudo-relabelings proved unnecessary); H3.3 coincidence of the designed toggle metric with the canonical display grading — closed at the blueprint level by T-237, the general-fiber question living inside H3.1; H3.4 naturalness: does any natural foundation realize the grammar the blueprint constructs? — the Fano-foundation problem, the floor's central open object; answered affirmatively by T-243: the frame is the canonical pivotal structure of the octonions (the doctrine's duality group $(\mathbb{Z}/2)^3$ graded, seven quaternionic axes $e_g^2=-1$, lines $=$ associative $\mathbb{H}$). The selection is then forced (T-244): $-1$ $=$ anisotropy $=$ division $=$ Hurwitz $N=7$ — one condition; a $+1$ axis is a null-direction zero-divisor that fails diagnosability, so a perfectly diagnosable foundation carries all seven signs $-1$. H3.4 closed.

Verdict. Legitimate, with the designed/discovered split stated wherever the floor is presented.

§6. The hole register of the vertical

The consolidated register — what the vertical does not yet know, in one place:

#	Hole	Floor	Status	What would close it
H1.1	Universal $\rho^*$: internal terminality over sub-holons	I	[С] (T-248)	internal terminality (Universe terminal over its sub-holons, T-222 Pareto-optimum the sink), conditional on part→whole inclusions being resource morphisms + the Markov hypothesis; a computed cosmological T-222 analogue would give [Т]
H1.2	The Universe's stage $P$ neither derived nor measured	I	[Г]	Λ-deficit programme output or an independent cosmological estimator; by T-244 the stage is the frame's degree of anisotropy, so the estimator target is sharpened (cosmic anisotropy ↔ $P$) — the floor's sharpest open edge, and the residue of H1.3
H1.3	Scale-freeness of the grammar	I	closed on the viable carrier (T-247)	the coinductive carrier types every level as a seven-axis system; D1–D3 $=$ nondegeneracy $+$ nontriviality (T-244/T-246); holons are viable by definition, so the grammar is transmitted downward — residue $=$ H1.2 (which structures are viable holons)
H2.1	Identity of the two ceiling mechanisms (purity vs coding)	II	closed in the negative (2026-07-11, T-239)	mechanisms provably distinct: $V = \{1,2,3\}$ monotone vs $K = \{1,3\}$ non-monotone, unique disagreement at the Vasil'ev height $m = 2$, cardinality $3 \neq 2$ excludes every identification; witness identity at $m = 3$ stands
H2.1′	Common root of the value agreement $\max V = \max K = 3$	II	closed (T-242)	mechanisms proven independent — the purity ceiling is nonconstant in the contraction base $b$, the coding depth is $b$-independent; they coincide only at $b = 3$, the $\mathrm{PG}(2,2)$ line order that also fixes $N = 7$ (one geometric input, not one mechanism)
H2.2	(P)-genericity for actual Rich-metatheories	II/III	closed (2026-07-11, T-240)	verified against R1–R5 with two corrections: strict pullback → iso-comma, equality → provable isomorphism of $\mathsf{Q}$-images; consistency (R3) inherited from either leg; glue canonical on intensional fibers ((P) a theorem there), a real hypothesis outside
H2.3	K-P1–K-P4, Σ-P1/Σ-P2 unverified	II	falsifiable	the experiments (E4–E7 battery; Γ-tomography)
H3.1	ΣQ1′: charts forced by intensionality?	III	closed on the frame (T-246)	faithfulness is a corollary of division (T-244): distinct defects superpose to a third defect $e_ie_j=e_{i+j}$, never an equivalence, so no weight-$2$ codeword and $d_{\min}=3$; on $\tau=1$ finite-signature fibers finiteness follows from former-localized uniform failure — the Fano frame is forced; residual $=$ the N$=7$ selection, discharged per signature
H3.2	Blueprint syndrome completeness	III	closed (2026-07-11, T-237)	closed by strict gauge: Wilson-loop soundness + exhaustive gauge completeness; the anticipated pseudo-relabelings proved unnecessary
H3.3	Designed metric = canonical display grading?	III	closed at blueprint level (T-237)	both metrics $\{0,1\}$-valued with the same zero set ($d_{\min}=3$, $\rho_{\mathrm{cov}}=1$ + T-234); general-fiber question remains inside H3.1
H3.4	Natural Fano foundation	III	closed (T-243 + T-244)	naturalness: the frame is the canonical pivotal structure of the octonions (T-243); selection: the $-1$ sign is forced by nondegeneracy — anisotropy $=$ division $=$ Hurwitz $N=7$; a $+1$ axis is a null-direction zero-divisor failing diagnosability (T-244). Per-foundation anisotropy is a checkable instantiation

§7. The anti-numerology register

A theory whose constants are few and whose geometry is rigid will produce numerical coincidences. The rule of this register: a resonance may be upgraded only by exhibiting a morphism of derivations — never by the equality of numbers. Current inventory:

Resonance	Relation between the sides	Status
$N = 7$: number / structure / closure / diagnosability	four independent derivations of one constant — the opposite of numerology	[Т]×4
$21 =$ coherences $= \binom{7}{2} = \dim \mathfrak{so}(7)$	literally the same object	[Т]
$14 = \dim \mathfrak{g}_2 =$ number of rate sum-rules	literally the same decomposition ($W$-twist, T-226)	[Т]
Ceiling $3$: purity ladder vs tower ladder	mechanisms provably distinct and arithmetically independent (T-242): the purity ceiling responds to the contraction base, the coding depth does not; both inputs are $\mathrm{PG}(2,2)$ parameters, so the shared $3$ is one geometric datum in two roles	[Т]+[Т], dichotomy [Т] (T-239), independence [Т] (T-242)
Golay $t = 3$ vs $\mathrm{SAD}_{\max} = 3$	annotated curiosity; no derivation either way	[Г]
$3$ generations ($	QR(7)	$) vs$3$ organisms/levels
$2/3$: Lüders rate vs Koide ratio	no relation claimed; inference forbidden	—
$1344 = 2^3 \cdot 168$: monomial stabilizer vs blueprint rigidity group	the same group, same computation reused	[Т]
Duality group $(\mathbb{Z}/2)^3$ of the depth-$3$ doctrine vs the blueprint's flux $\mathbb{F}_2^3$	given a common natural realization by the octonions (T-243): $\mathbb{O}$ is graded by this group, the seven imaginary units are the axes, the pivotal sign $-1$ is the holonomy	native plane [Т], octonionic realization [Т] (T-243); selection [Г]

§8. The standing canonization discipline

1. Gate. Any new organizing reading passes K1–K6 before entering the docs; the pass/fail is appended to this document. A reading that fails any criterion may still be discussed — in research notes, marked [И]/[Г] — but not canonized as structure.
2. Register duty. Readings enter with their hole register. Canonization of Floor I specifically is authorized by this audit only together with H1.1–H1.3.
3. Crystallization over accretion. When a new result fits two framings, prefer the edit that merges framings over the edit that adds one. Precedents to imitate: T-229 (repaired the conjecture rather than layering a caveat), T-232 (gave the ceiling echo a shared skeleton instead of a second echo), T-235 (same-day retraction with the audit note kept in place — the error made structural).
4. The bare-mathematics test, permanently. Every floor-level sentence in canonical documents must survive the question "what breaks if this is deleted?" — and the answer must name a theorem, a verification, or a prediction.
5. Inter-section coherence. The theory must be fully consistent between its sections: every fact has one canonical home (single source of truth), and every change sweeps all claim-sites corpus-wide, both languages. A contradiction between sections is a defect of the same rank as a broken proof — the v2.2 audit moral, promoted to standing law.
6. Non-redundancy, with its two exceptions. One canonical statement per fact; all other occurrences reference it. Two kinds of duplication are not redundancy and are encouraged: indexes (the status registry's rows summarize and point — they do not re-derive) and independent verifications (a second derivation in another formalism or language strengthens; a second statement in another section weakens). Anything else duplicated must be merged into the canonical home.
7. Impeccable formulation. Every formulation carries its status inline; every quantifier and domain is explicit; level and context of every equivalence are named (the T-235 lesson: a gauge-level fact stated as a fiber-level fact is not an imprecision but an error). Detail is not optional decoration — worked examples and explicit tables are part of the proof surface, and twice this arc they caught real errors (the guardian slip in the fingerprint §5; the level conflation in T-235).

Verdict of this audit. The three-floor vertical is legitimate: Floor I as the axioms read aloud (conditional on its register), Floor II unconditionally, Floor III with the designed/discovered split declared. The planned overview document «Вселенная как Голоном» is authorized under discipline items 2 and 4.

Canonization log (append-only).

• 2026-07-11. Floor I canonized: The Universe as Holonom. Passed K1–K6; register H1.1–H1.3 embedded in place (§2–§4) with a floor-local index (§6); bare-mathematics self-audit table included (§7). License conditions of this audit discharged in full.

Where this leads

• Status registry — the micro-level discipline this document extends to the meso-level.
• Articulation hygiene — the sentence-level test that K1 generalizes.
• Falsifiability — the registry K4 points into.
• Σ-calculus and the Fano fingerprint — the floors II–III material audited in §4–§5.
• The corpus practice of audit chronicles — the same-day retraction inside T-235 is the standing precedent for §8.3.
• The Universe as Holonom — the first canonization performed under this discipline (see the log in §8).