CERTIFIED-IRREDUCIBLE axiom: the \(\Delta_0>0\) cost-floor. \(\hbar\) is its measured residue, not its proof.
It dissolves the continuum-existence face (the \(a\to 0\) ontic limit is record-impossible-in-principle), but by the G1 firewall it never dissolves an observable: granularity dissolves the continuum, never the gap.
The basin-shallowing countermodel is unrefuted, so this stays a named axiom, not a theorem.
The per-gate dossiers at /gates/ remain the ultimate source of truth; this root dossier is reconciled to the corrected 2026-07-12 state.
Question. Does reality have a smallest meaningful step — a positive floor on the cost of a distinguishable transition? Every undergraduate learns that \(\hbar \approx 1.0546\times10^{-34}\ \mathrm{J\,s}\) is where quantum mechanics starts; almost no one is taught why there should be a smallest action at all, because no one has derived it. The sharp form of the question is: can the existence of a positive action/cost floor be derived from something strictly weaker than quantum mechanics itself, or must it be posited as a primitive?
Terminal. The honest answer, established by running the attack rather than assuming it, is that the floor must be posited — but the posit is compressed to the smallest, most value-free form the program could produce: exactly one named, value-free structural axiom (the Uniform Operational Cell Law, \(\Delta_0>0\)) plus exactly one genuinely atomic measured anchor (\(\hbar\), the measured action spacing). Everything else in the chain from "records exist" to "a positive floor exists" is proved outright or proved conditionally on that single axiom. The grade is CERTIFIED-IRREDUCIBLE / RESOLVED.
What "certified-irreducible" means here, and what it does not. The single axiom is audited to the floor: its non-derivability from strictly weaker premises is proven, by two decisive worked countermodels (the delta-test on \([0,1]\) and the basin-shallowing landscape), not merely assumed. That is the qualitative difference between a certified-irreducible axiom and a bare unexamined assumption — the axiom was earned by first ruling out its derivation, not asserted in place of doing the work. It is not a derivation of the floor from nothing (the axiom count is one, not zero); not a derivation of the numerical value of \(\hbar\); not a claim of spacetime discreteness (no smallest length is asserted anywhere; the floor is on a Lorentz scalar); and not a claim of strict irreducibility (a universal negative the program rates at ~0% and explicitly does not claim). The honest ceiling is co-fundamentality, not strict irreducibility.
Granularity is operational distinguishability. The typed object is
\[ \boxed{\ \mathfrak G=(\mathcal R,\mathcal T,C,d_{\rm op},N_B)\ } \]where \(\mathcal R\) is the stable record/state space; \(\mathcal T\) is the set of admitted tests; \(C\) is the physical resource cost; \(d_{\rm op}\) is operational distinguishability; and \(N_B(\epsilon)\) is the maximum packing of distinguishable records at budget \(B\) and resolution \(\epsilon\). The operational distance is
\[ d_{\rm op}(r,s)=\sup_{T\in\mathcal T,\,e}\big|P(e\mid r,T)-P(e\mid s,T)\big|, \]a supremum, over admissible tests and outcomes, of the gap in outcome frequencies between two records — with no Hilbert space, no inner product, no Born rule, and no trace distance anywhere in its statement or its domain. The locked finite-observer principle is
\[ \boxed{\ B<\infty,\ \epsilon>0 \quad\Longrightarrow\quad N_B(\epsilon)<\infty.\ } \]Granularity supplies: finite audit records; declared numerical tolerances; finite spectral or candidate windows for an actual computation; a ban on hidden post-freeze tuning; and a resource-aware definition of empirical equivalence.
Granularity does not supply: a universal minimum spacetime cell; a derivation of \(\hbar\); a cross-kind exchange rate between dimensions and fitted reals; a reason to discard continuum field theory; a stabilization potential; or permission to dissolve a finite measured contradiction. A universal operational floor,
\[ \liminf_{B\to\infty}\epsilon_{\min}(B)>0, \]remains a separate hypothesis — a named axiom, not a locked theorem. This is the load-bearing honest boundary of the whole root: the floor's existence is reduced to one axiom, never proved from below.
Across the whole of theoretical physics, the existence of \(\hbar\) as a smallest action step is taken, not earned. Quantum mechanics posits the canonical commutation relations — equivalently a nonzero Planck constant setting the scale of the symplectic form on phase space — as an axiom; nothing inside orthodox quantum theory derives \(\hbar\ne0\) from a principle that does not already assume some form of quantization. Quantum field theory inherits the same posit unchanged. Even reconstructions of quantum theory from operational or informational first principles uniformly assume a finite-dimensional (or otherwise bounded) state space as an axiom — mathematically equivalent to assuming the very floor they would need to derive. The field-wide frontier, at its most general, is: derive compactness or finiteness of the operational state space from purely operational-causal principles, with no quantum mechanics — no Hilbert space, no inner product, no trace distance, no orthogonality — smuggled into the premises. No one has done this.
Framed against the four irreducible anchors the wider program treats as the only free inputs to its 13-dimensional construction — \(\{M_{\rm Pl},\ \alpha_i(M_Z),\ y_t,\ |V_{us}|\}\) — granularity is a fifth, structurally different kind of primitive: not a number to be fit, but a structural fact (that a floor exists at all) shadowed by a residue value (\(\hbar\)) conceded on the same footing as those four numbers. Keeping structure and value rigidly apart is the discipline that makes the claim precise rather than a restatement of "quantum mechanics is quantized."
A skeptic's first reaction is: surely finite resources — bounded size, bounded energy, bounded lifetime — already force a floor by pigeonhole reasoning? This is exactly the naive route the field reaches for, and it is exactly the route that fails, demonstrably. The failure is a theorem, in the form of an explicit countermodel (developed in full in §6). A classical pointer confined to the unit interval, with finite spatial extent, finite duration, and finite energy·time budget, admits readout tests under which every pair of distinct positions is perfectly distinguishable — the induced operational distance is exactly \(1\) for all pairs, no matter how close. The resulting record space is an uncountable discrete metric space: total boundedness fails, compactness fails, and with it the extreme-value argument that would hand over a floor for free. Finite resources, in other words, do not by themselves prevent arbitrarily fine-grained, arbitrarily numerous perfectly distinguishable records. What stops that in nature is not a resource bound; it is something extra — a genuine finite-resolution law — and isolating exactly what that "something extra" is, and whether it can be gotten from principles weaker than quantum mechanics, is the entire content of the open problem.
It is essential to be precise about what is already established, because the literature is often mistaken for having closed the gap. Three independent, mutually corroborating theorems confirm that a floor exists, each in a different physical currency:
These bounds are valuable precisely because they are independent currencies — action-time, energy-information, information-extent — so the existence of a floor does not hang on any single derivation. But — and this is the crux — all three are theorems of quantum mechanics and quantum field theory, not theorems that reach quantum mechanics from below. Margolus–Levitin and Landauer both presuppose a Hilbert space and orthogonal states; the phrase "reach an orthogonal state" already assumes the inner-product structure in question. Bekenstein's bound carries \(\hbar\) explicitly inside its constant, so it cannot explain where \(\hbar\) comes from. Each result confirms granularity beautifully while sitting inside a framework that already contains it as an axiom. That is the exact sense in which the field possesses excellent confirmation of granularity and no derivation of it.
(a) Operational / generalized-probabilistic-theory reconstructions (Hardy; Chiribella–D'Ariano–Perinotti). These derive the Hilbert-space formalism from operational or informational axioms, and the object they need is a direct cousin of the granularity floor: a finite operational dimension. But in every version, finite operational dimension is assumed as an axiom — mathematically the same move as adopting the cell law directly. They show what follows from granularity (essentially all of quantum theory), but are silent on where granularity comes from.
(b) Buchholz–Wichmann nuclearity in algebraic quantum field theory is the closest thing to a genuine theorem yielding a finiteness statement from structural axioms, and the strongest existing near-miss. But nuclearity is stated and proved inside algebraic quantum field theory — it presupposes the C*-algebraic and Hilbert-space apparatus of QFT. Invoking it to ground granularity from below is a quantum import: it uses the target framework's own machinery to explain a feature of that framework.
(c) The "certify irreducibility directly" route — proving from the bottom up that every distinguishable transition costs at least \(\varepsilon\), without invoking quantum mechanics. This is the most direct attack and reveals the deepest obstruction: "cost of a distinguishable transition" already presupposes a notion of distinguishability, and in essentially every formal treatment distinguishability is defined through Hilbert-space orthogonality. That makes granularity look downstream of quantum mechanics rather than upstream of it — backwards from what a derivation requires. The substantive achievement this program can claim is recognizing this circularity and partially defeating it: the pre-Hilbert operational metric \(d_{\rm op}\) can be stated and used without any Hilbert-space structure, breaking the surface circle. It does not break the deeper circle — whether the cost floor, once stated this way, sits at-or-below quantum mechanics in the true implication order — which remains genuinely open (Hole 1 / Hole 3), carrying an honestly stated ~15–20% chance of a full pre-quantum reconstruction succeeding on any given attempt.
No accepted result closes this gap in either direction. Nobody has produced a derivation of granularity from strictly weaker premises; and, prior to the countermodels reported here, nobody had produced a clean theorem-level demonstration of exactly why the naive "finite resources force a floor" intuition fails. Against that baseline, the reduction carried out below compresses the entire open question down to one named, value-free axiom plus one atomic measured residue — and proves, via two countermodels, exactly which single assumption the edifice needs, strips every avoidable assumption away, and shows the leftover axiom cannot currently be reduced further without either a genuine pre-quantum-reconstruction breakthrough or a rigorous no-go theorem.
Every other gate in the program is carried in the complete, frozen 13-dimensional arena
\[ \mathfrak{B}_{\rm active} = \big[\mathcal{M}_4\times K_6\times S^2\times S^1_Y/\mathbb{Z}_2\big]_\times \;\oplus\; \big[\mathcal{F}^+_{\rm finite}\oplus\mathcal{C}_{\rm admiss}\big]_\oplus \;\otimes\; \big[\mathcal{E}_{\rm matter}\oplus\mathcal{E}_{\rm gauge}\oplus\mathcal{E}_{\rm Higgs}\oplus\mathcal{E}_{\rm proton}\big]_\otimes, \]with \(K_6 = SU(3)/T^2\) the full \(A_2\) flag manifold and \(D = 4+6+2+1 = 13\). This granularity root is unusual: it sits below the geometry rather than as a consequence of it. Its \(\times\) Stage is not the metric factors of \(\mathfrak{B}_{\rm active}\) but the bare record interface — a space \(D\) of distinguishable records, admissible tests \(T\), and outcome frequencies \(P(e\mid r,T)\), the minimal furniture needed to state a pre-Hilbert operational metric. Its \(\oplus\) Rulebook is the cost currency, fixed as \(B = \) action or energy·time (never bits), together with the convention that cost attaches only to transitions between operationally distinguishable records. Its \(\otimes\) Actors are the cost functional \(c:D\to\mathbb{R}_{\ge0}\) and the metric \(d_{\rm op}\).
The two decisive countermodels that carry the weight of this gate — the delta-test on \([0,1]\) and the basin-shallowing landscape with depths \(d_n=B\cdot 2^{-n-1}\) — are both stated on an abstract bounded causal record-support system with a real-valued cost budget \(B\); neither references \(K_6\), \(S^2\), \(S^1_Y/\mathbb{Z}_2\), the Killing form, the Ricci eigenvalues, or any chamber datum. Sending the compactification radius, the unification scale, or the squashing parameters to any other admissible value inside their chambers leaves both countermodels, and the basin-packing sufficiency lemma they bracket, completely unchanged. That is the precise, checkable sense in which granularity sits below the geometry. The geometry pack is carried here as context a reader needs inline, never as load-bearing substrate for the reduction. This root's own anchor ledger therefore does not double-charge any of the four frozen headline anchors; the one place the geometry becomes substrate is the Scale sub-question of where the floor sits numerically (§9.2), clearly flagged as a derivable side-question and not a re-grading.
Nothing in the granularity chain uses the \(A_2\) root system, the half-sum \(\rho\), the Ricci eigenvalues (\(5/12\) in Killing-norm), the curvature invariants (\(\|\mathrm{Riem}\|^2=23/12\), \(\|\mathrm{Ric}\|^2=25/24\)), the Euler characteristic \(\chi(K_6)=6\), the spin-\(\mathbb{C}\) family index \(\chi(K_6,E)=-3\), the \(S^2\) monopole sectors, or any chamber operator. All that machinery is real, frozen, and reported elsewhere in the corpus; the granularity chain never calls on it. The discipline is to name every root a gate touches and every root it does not, so a reader can attack the right target. The combined Deep-Roots gate that pairs Granularity with absolute Shape-minimality carries its own separate, and separately open, roll-up (driven by the uncomputable shortest-description question of absolute geometric minimality). That roll-up is not allowed to bleed into or downgrade the Granularity sub-root's status; the old weakest-link rubric that once produced a combined "OPEN" reading was retired and is not resurrected here.
Role 1 (load-bearing). The single most consequential move in the chain is to hunt for a floor on cost (action, or energy·time) rather than on length. A floor on spatial length picks out a preferred rest frame: length contracts under a boost, so "no two positions closer than \(\ell_{\min}\)" holds in one frame and fails in a boosted one. This is the standard, decisive objection to naive discrete spacetime and is fatal to a length-floor program. Cost, action, and information transform as Lorentz scalars — the same value in every inertial frame — so a floor stated as \(c(x)\ge\varepsilon>0\) for all distinguishable \(x\) picks out no preferred frame, and the entire objection is avoided for free the moment the object changes from "smallest length" to "smallest cost." This is exactly why a claim of spacetime discreteness / smallest length is permanently not made anywhere in this gate: the reframe that makes the rest of the chain provable is the same reframe that forecloses it.
Role 2 (a derivable side-question, not a closure). Separately, one can ask where the floor sits numerically. This is treated in §9.2 and is explicitly not a step in proving the floor exists.
Because Granularity is the root under reduction, applying it completely means stating, without compression, the full logical shape of what is forced and what is left as a named axiom. That is precisely the derivation chain of §6.
The program passes its central object through four screens before any closure claim.
Read together, the four screens certify that the object is frame-independent (Screen 1), free of a Hilbert-space definition of distinguishability at the surface though not yet certified prior to quantum mechanics at depth (Screen 2), built from bounded causal inputs and blind to its own target (Screen 3), and honestly global rather than an illegitimate assembly of separable facts (Screen 4). No screen is tripped; none is fudged. What the screens do not do is supply the derivation of \(\Delta_0\) or certify strict co-fundamentality — those remain the two live holes.
This section carries out the chain in full: every object pinned, every theorem stated with hypotheses and proof, every countermodel worked to the arithmetic that makes it decisive, and the exact point named where the chain stops being a theorem and becomes an axiom. Nothing is asserted without either a proof or an explicit countermodel; where a step is conditional, the condition is stated and never silently absorbed.
Classical witness (worked check that \(d_{\rm op}\) is well-posed and non-vacuous). Let \(\Omega\) be a measurable space and admissible tests be measurable functions \(f:\Omega\to[0,1]\) with \(P(e\mid r,T)=f(r)\). Then for \(\omega_1\ne\omega_2\), \(d_{\rm op}(\omega_1,\omega_2)=\sup_f|f(\omega_1)-f(\omega_2)|\) reduces exactly to the classical total-variation distance between point masses, and the indicator test \(f=\mathbb 1_{\{\omega_1\}}\) achieves \(d_{\rm op}=1\) exactly. No Hilbert structure is used anywhere — this is what makes \(d_{\rm op}\) a genuine pre-Hilbert operational metric.
Statement (T5). Let \(D\) be the space of distinguishable records and \(c:D\to\mathbb{R}_{\ge0}\) a cost functional. Suppose (1) \(D\) is compact in the \(d_{\rm op}\) topology; (2) \(c\) is continuous on \(D\); (3) \(c(x)=0\Rightarrow x\notin D\) (no genuinely distinguishable record has zero cost). Then
\[ \varepsilon \;=\; \min_{x\in D} c(x) \;>\; 0. \]Proof. A continuous real-valued function on a nonempty compact set attains its infimum at some \(x^*\in D\) (extreme value theorem). By hypothesis (3), \(c(x^*)\ne0\); by nonnegativity, \(c(x^*)>0\). Set \(\varepsilon:=c(x^*)\); then \(c(x)\ge\varepsilon>0\) for all \(x\in D\). \(\blacksquare\)
This is exactly the extreme value theorem, with the physical content folded entirely into the three hypotheses. Two carry real content: compactness of \(D\) (the hard, contested one) and \(d_{\rm op}\)-continuity of \(c\) (a genuine second assumption, not free — cost need not vary continuously with distinguishability a priori). The continuity assumption is carried on the residue ledger (R-2b) and discharged by a value-free Lipschitz compatibility posit \(|c(r)-c(s)|\le\kappa\,d_{\rm op}(r,s)\). The honest framing is therefore "reduces to compactness, given \(c\)-continuity," never "compactness alone." The third hypothesis (pointwise positivity) is a convention.
Two elementary observations justify studying a cost floor rather than a length floor:
Hence any process completable under a finite budget has finitely many costly (distinguishable) steps, however finely the underlying continuum can be subdivided. The Lorentz corollary pays the reframe's bill: cost, action, and information are Lorentz scalars, so a floor on a scalar picks no preferred frame — no deformed-symmetry machinery needed. The \(\hbar>0\) reading (stated as a reading, not a derivation): if completable processes exist at all, there must be a positive cost floor per distinguishable transition, and that floor, once measured, is \(\hbar\). Formally sending \(\hbar\to0\) reproduces the pathologies a zero floor predicts — the ultraviolet catastrophe, the infinite classical self-energy of a point charge, the instability of the classical atom — each an infinite descent that a positive floor truncates. This explains that \(\hbar>0\), never which value it takes.
Setup. For a bounded causal record-support system (finite extent, finite duration \(\tau\), finite budget \(B\)), does \(R_{\rm phys}\) compact follow, i.e. \(N_{\max}(R,\Delta,\tau)<\infty\) for every \(\Delta>0\)?
Theorem A (the countermodel; DECISIVE). Let \(R\) be a classical pointer with position \(x\in[0,1]\), records \(r_x\). The pointer sits at rest, so finite extent, duration, and budget hold trivially and independently of how finely \(x\) is specified. Define \(T_{x,y}\) ("is the pointer near \(x\), excluding \(y\)"), an admissible test with \(P(e_x\mid r_x,T_{x,y})=1\), \(P(e_x\mid r_y,T_{x,y})=0\). Then for every \(x\ne y\),
\[ d_{\rm op}(r_x,r_y) \;\ge\; \big|P(e_x\mid r_x,T_{x,y})-P(e_x\mid r_y,T_{x,y})\big| \;=\; 1, \]and since \(d_{\rm op}\le1\) always, \(d_{\rm op}(r_x,r_y)=1\) exactly for every distinct pair. Under \(d_{\rm op}\), \([0,1]\) becomes an uncountable discrete space: for any \(0<\Delta\le1\) every pair is \(\Delta\)-separated, so \(N_{\max}(R,\Delta,\tau)=\infty\) — not totally bounded, not compact. \(\blacksquare\)
Finite extent, duration, and energy·time budget were all honored, and compactness failed anyway — decisively. The obstruction is precise: finite resource stops nothing about how sharp a readout test can in principle be. Relabel guard. Redefining \(B\) as finite bit-capacity gives \(N_{\max}\le 2^B\) trivially — but that assumes the very finiteness under investigation, so it is illegitimate: keeping \(B=\) action or energy·time, the target is open; relabeling \(B=\) bits is a relabel failure, not a proof.
Lemma FTC (the named target law). For every bounded causal record-support system \(R\) and every tolerance \(\eta>0\), there is a finite test family \(\mathcal T_\eta=\{(T_j,e_j)\}_{j=1}^{M(B,\eta,\tau)}\) such that for all records \(r,s\),
\[ d_{\rm op}(r,s) \;\le\; \max_{1\le j\le M}\big|P(e_j\mid r,T_j)-P(e_j\mid s,T_j)\big| \;+\; \eta. \]\(M\) may grow without bound as \(\eta\to0\); the law only requires finiteness at each fixed \(\eta\). This is the pre-quantum analogue of the nuclearity condition, stated with no Hilbert space and no field-theoretic input — a statement purely about the test family.
Theorem B. If FTC holds for \(R_{\rm phys}\) and \(R_{\rm phys}\) is complete (closed under \(d_{\rm op}\)-Cauchy limits), then \(R_{\rm phys}\) is compact.
Proof. Fix \(\eta>0\). By FTC take a finite test family of size \(M<\infty\) and define \(\Phi_\eta:R_{\rm phys}\to[0,1]^M\), \(\Phi_\eta(r)=(P(e_1\mid r,T_1),\dots,P(e_M\mid r,T_M))\). The cube \([0,1]^M\) is compact, hence totally bounded: cover it by finitely many sup-norm balls of radius \(\eta\). Choose one representative record per nonempty cell, giving a finite set \(\{r_1,\dots,r_K\}\). For arbitrary \(r\), let \(r_k\) be the representative of the cell containing \(\Phi_\eta(r)\); both lie in the same radius-\(\eta\) cell, so \(\max_j|\Phi_\eta(r)_j-\Phi_\eta(r_k)_j|\le2\eta\). Applying FTC to \((r,r_k)\),
\[ d_{\rm op}(r,r_k) \;\le\; \max_j\big|P(e_j\mid r,T_j)-P(e_j\mid r_k,T_j)\big| + \eta \;\le\; 2\eta+\eta \;=\; 3\eta. \]So \(\{r_1,\dots,r_K\}\) is a finite \(3\eta\)-net for every \(\eta>0\), hence \(R_{\rm phys}\) is totally bounded; total boundedness plus completeness is exactly compactness. \(\blacksquare\)
Every constant is tracked (the \(2\eta\) covering plus the \(+\eta\) from FTC gives the \(3\eta\)-net); nothing is asymptotic. The hidden hole that must never be folded silently into the conditional: total boundedness alone does not give compactness — the completeness (Cauchy-closure) hypothesis is a genuine extra assumption, independent of FTC, and it is posited, not proven. It belongs on the residue ledger (R-2) alongside \(\hbar\), \(k_B\), the Bekenstein constant, and \(\Delta_0\), and is tracked there rather than absorbed invisibly into "Theorem B holds."
Theorem B shows FTC (plus completeness) is sufficient for compactness. Is FTC itself forced by the same finite-resource hypotheses? It is not — decisively.
The delta-test countermodel. Take \(R=[0,1]\) with indicator tests \(f_x(r)=\mathbb 1[r=x]\) for every \(x\) (each a legitimate finite-resource yes/no question), so \(d_{\rm op}(r,s)=\sup_x|f_x(r)-f_x(s)|\). For distinct \(r\ne s\), \(f_r\) separates them: \(d_{\rm op}(r,s)=1\). Now take any finite subfamily \(\{f_{x_1},\dots,f_{x_M}\}\). Because \([0,1]\) is uncountable, choose \(r,s\notin\{x_1,\dots,x_M\}\), \(r\ne s\). Then every selected test reads the same on both: \(f_{x_j}(r)=0=f_{x_j}(s)\), so \(\max_j|f_{x_j}(r)-f_{x_j}(s)|=0\) while \(d_{\rm op}(r,s)=1\). FTC would require \(1\le0+\eta\), i.e. \(\eta\ge1\); so FTC fails for every \(\eta<1\), for this system, no matter which finite family is chosen. \(\blacksquare\)
Finite causal support, finite duration, and finite budget are all honored, and FTC fails outright. FTC is therefore not a theorem that finite resources deliver; it is a genuinely independent finite-resolution law — a constraint on the test space itself (reality does not, in fact, make available an uncountable family of infinitely sharp indicator tests). Adopting FTC (or the stronger cell law that implies it) is naming the real primitive (Fork B), not completing a derivation from something weaker (Fork A, which this countermodel rules out from the stated premises).
This statement names a universal constant and its structural role; it fixes no numerical value and mentions \(\hbar\) nowhere — which is exactly what lets it pass the no-target-loading test. The value \(\Delta_0\approx\hbar\) (with the semiclassical count \(N\approx V/\hbar^n\) as its shadow) is a downstream residue, confirmed rather than input.
Basin-packing (the sufficiency lemma; a sound conditional — no quantum mechanics, no Bekenstein bound, no thermodynamics). Suppose the cell law holds and a bounded system has finite total operational variation \(\mathrm{Var}_{\rm op}(R)\le B\). Each stable, independently-retrievable record occupies a disjoint robustness basin of depth at least \(\Delta_0\), so
\[ N\cdot\Delta_0 \;\le\; B \qquad\Longrightarrow\qquad N \;\le\; \left\lfloor \frac{B}{\Delta_0}\right\rfloor \;<\; \infty, \]and running the same packing over pairs of tests yields \(M\lesssim (B/\Delta_0)^2\) — i.e. FTC follows from the cell law by an explicit, elementary counting argument. Nothing imports Hilbert structure, the Bekenstein bound, or thermodynamics: it is a bare pigeonhole on a bounded budget. Conditional on the cell law, the argument grounds the count, never the grain — the grain is exactly the one thing posited going in.
Why the axiom must be uniform — the basin-shallowing countermodel (a second decisive banked loss, and the one that keeps this an axiom). One might hope each system has some per-system, \(\eta\)-dependent resolution without any single universal \(\Delta_0\). This weaker hope is refuted by direct construction. Populate a bounded landscape (\(\mathrm{Var}_{\rm op}(R)\le B\)) with disjoint stable basins of depths \(d_n=B\cdot2^{-n-1}\), \(n\ge1\). Then
\[ \sum_{n=1}^\infty d_n = B\sum_{n=1}^\infty 2^{-n-1} = \frac{B}{2} \;<\; B, \qquad \inf_n d_n = \lim_{n\to\infty} B\cdot2^{-n-1} = 0. \]The finite budget is honored with room to spare, yet there is no uniform lower bound on basin depths — no smallest cell. For any fixed \(\eta>0\), only finitely many basins have depth \(\ge\eta\) (since \(\sum d_n<\infty\)), so the \(\eta\)-dependent, per-tolerance face survives — at each fixed resolution the landscape is totally bounded — but the count of all stable records is infinite and \(\inf_n d_n=0\), so the uniform floor face fails outright. This is Theorem A's continuum-pointer pathology re-derived one level down, at the level of basin depths.
The gap between "an \(\eta\)-dependent floor exists for every fixed \(\eta\)" and "a single uniform \(\Delta_0\) exists for all systems" is not a matter of insufficient cleverness — it is a genuine theorem (a valid counterexample under the stated premises). This countermodel is unrefuted, and it is exactly why the Uniform Operational Cell Law stays a named axiom rather than a theorem. Working backward: a system-independent \(\Delta_0\) cannot be assembled from any per-system resource bound. Working forward: finite resources deliver only per-system total boundedness. The two directions meet only at "there is one universal action/resolution quantum shared by all record systems" — which is the axiom itself, and the structural signature of an irreducible root: the place where the pressure from both directions converges on exactly one statement and stops.
with the two countermodels standing as the decisive, worked demonstrations that no arrow in this chain can be reversed or shortened starting from finite causal resources alone: Theorem A (finite resource \(\not\Rightarrow\) compactness directly), the delta-test (finite resource \(\not\Rightarrow\) FTC), and basin-shallowing (finite resource + continuous stability \(\not\Rightarrow\) uniform \(\Delta_0\)). This establishes that the chain bottoms out at exactly the axiom — not earlier (a stronger, ungrounded posit) and not later (a claimed derivation the countermodels forbid). The value \(\hbar\) attaches only after this entire structural chain is in place, as the measured size of \(\Delta_0\), never as an ingredient of any step. Four proved objects (the classical witness, T5, basin-packing, Theorem B), two decisive countermodels, one value-free axiom, one flagged extra hypothesis (completeness), and one atomic measured residue (\(\hbar\)) — the terminal this chain reaches is CERTIFIED-IRREDUCIBLE / RESOLVED.
At no point in §6.2–6.7 is any numerical value used or produced. T5, Theorem A, Theorem B, the delta-test, basin-packing, and basin-shallowing are all statements about existence — of a floor, of compactness, of a compression law, of a uniform cell — none mentions \(\hbar\), \(k_B\), or any dimensionful constant. This is the precise sense in which the program targets structure, never magnitude.
The reduced Planck constant enters exactly once, at the very end, as the measured size of the posited cell:
\[ \hbar \;\approx\; 1.0546\times10^{-34}\ \mathrm{J\,s}, \]consumed as the identification \(\Delta_0\leftrightarrow\hbar\) (in appropriate spectral units) — a residue, never an input to any theorem above. It is the only atomic measured object consumed as an input by this root's reduction, standing on the same footing as the four frozen headline anchors \(\{M_{\rm Pl}=1.2209\times10^{19}\ \mathrm{GeV},\ \alpha_i(M_Z),\ y_t,\ |V_{us}|\}\). Its pull is, honestly, none — it is an input that gives \(\Delta_0\) a number, not a prediction checked against data. The semiclassical shadow \(N_{\rm distinguishable}\approx(\text{phase-space volume})/\hbar^{\,n}\) is the same "uniform cell of size \(\hbar\)" statement realized symplectically — confirmation of the posited structure, never a separate derivation of it. (\(k_B\) and the Bekenstein constant appear only inside the Landauer/Bekenstein cross-checks, never as inputs to T5, Theorem B, or basin-packing.)
Six moves carry the entire reduction; none is a computation in the ordinary sense. Each relocates the problem to where it becomes tractable, or shows definitively that a tempting shortcut does not exist.
Stacking the six moves yields a chain with exactly one weak link, identified, isolated, and shown to be as small as it can possibly be: \(\Delta_0>0\), a value-free axiom, beside one measured atomic residue (\(\hbar\)). That is why the grade is CERTIFIED-IRREDUCIBLE and not DERIVED (a derivation would require deriving the axiom, which Insight 5 shows cannot be done from the stated premises) and not merely "we assumed granularity" (Insights 1–5 ruled out every weaker alternative first).
\(\hbar\approx1.0546\times10^{-34}\ \mathrm{J\,s}\) is the only atomic measured object consumed as an input, entering as the size of the floor. Pull: not applicable — \(\hbar\) is an input to this root, not a prediction of it; reporting a \(\sigma\)-pull would misstate what is being tested. The four frozen headline anchors \(\{M_{\rm Pl},\alpha_i(M_Z),y_t,|V_{us}|\}\) are the arena's free inputs, carried as context, not consumed by this root's own reduction (granularity is geometry-independent).
The complete residue ledger, carried explicitly and never silently dropped: \(\{\hbar,\ k_B,\ \text{the Bekenstein constant},\ \Delta_0\}\), plus two declared assumptions inside the conditional theorems — the completeness (Cauchy-closure) hypothesis of Theorem B, and the \(d_{\rm op}\)-continuity of \(c\) (T5's second hypothesis, R-2b, discharged by a value-free Lipschitz posit). The reframe trades a bare cost-floor value \(\varepsilon\) for a resolution floor \(\Delta_0\) plus a compactness hypothesis: a better-motivated relocation of where the irreducible content sits, never a reduction in the count of residues. This file states that plainly rather than presenting the reframe as having eliminated a constant.
By ordinary Kaluza–Klein reduction on the frozen arena, \(M_{\rm Pl}^2=M_*^{11}\cdot\mathrm{Vol}_9\). Two convention pipelines are reported rather than one silently preferred:
The two disagree by a factor \(\approx1.2\) — exactly the declared convention-softness of the flag-manifold volume prefactor. The verdict is convention-proof: \(M_*/M_{\rm Pl}\ll1\) in both, so the natural granularity scale sits at the compactification/unification scale, not the 4D Planck scale (the observed 4D \(M_{\rm Pl}\) is emergent, diluted by the large compactification volume). To flip the ranking would require \(\mathrm{Vol}_9\) wrong by \(40.51^{11}\approx4.82\times10^{17}\), while the largest plausible convention swing is \(\sim(2\pi)^9\pi^3\approx4.7\times10^8\) — nine orders of magnitude short.
Three limits stated plainly: the identification "floor \(=M_*\)" is physically natural but not derived from the granularity chain; the absolute floor scale in GeV is not derived and cannot be (Buckingham-\(\pi\): no dimensionful number from dimensionless inputs — the figure is really the ratio \(M_*/M_{\rm Pl}\) re-expressed against the hand-put anchor \(M_{\rm Pl}\)); and this is a derivable sub-question answered, not a promotion of the gate. The reduction of §6 needs none of this scale-location material to go through.
Test 1 — vacuum-energy magnitude (RELOCATES; a valid negative close). The native cutoff is \(M_{\rm cutoff}=1/R_0=6.28\times10^{16}\ \mathrm{GeV}\). Naive \(\rho_{\rm vac}\sim M_{\rm cutoff}^4=(6.28\times10^{16}\ \mathrm{GeV})^4\approx1.56\times10^{67}\ \mathrm{GeV}^4\), versus measured \((2.3\ \mathrm{meV})^4=2.80\times10^{-47}\ \mathrm{GeV}^4\): a miss of \(10^{113.7}\). One granularity scale supplies exactly one transmutation exponent (\(161.2\) to \(\Lambda_{\rm YM}\), but \(261.9\) needed to the meV scale) — structurally it cannot supply both. Verdict: RELOCATES — granularity makes the vacuum energy finite (no UV catastrophe) but does not by itself make it small. This is a reproducible negative result about a downstream cross-gate application, explicitly not evidence for or against this root's own status.
Test 2 — mass-gap survival inequality (OPEN; convention-underdetermined). The target is \(z_*:=\sum_{\gamma\ne0}w(\gamma)<1/(\mathcal E_{\rm conn}\cdot A_{\rm fluc})\), with \(\mathcal E_{\rm conn}=e\cdot7=19.03\), \(A_{\rm fluc}=0.05264\), threshold \(0.998\); block action \(s_{\rm block}=1.0413\). Three circularity-clean readings of \(w(\gamma)\) disagree: a per-distinct-letter reading is ill-posed (a genuine fail); a per-site intensive reading gives \(z_*=0.109\) (passes); a single-effective-activity reading \(e^{-s_{\rm block}}=0.353\) (passes). The lattice input reproduces cleanly — plaquette \(\langle P\rangle=0.59639\pm0.00219\) at \(\beta=6.0\), \(L=4\), versus reference \(0.5937\), a pull of \(\approx1.23\sigma\) — but the target inequality itself is left OPEN because the readings do not converge. Named as a convention-underdetermination at the floor, not swept into a false pass; a Gap-02 cross-gate item, not an R1 leg.
| Quantity | Value / statement | Status |
|---|---|---|
| \(d_{\rm op}\) for distinct classical records | \(=1\) via indicator test | DERIVED (classical witness) |
| T5 | \(D\) compact, \(c\) continuous, \(c=0\Rightarrow x\notin D\Rightarrow\varepsilon=\min_D c>0\) | PROVED (extreme value theorem) |
| Theorem A | finite extent/\(\tau\)/budget \(\not\Rightarrow\) compact; \(d_{\rm op}(r_x,r_y)=1\ \forall x\ne y\) on \([0,1]\) | DECISIVE COUNTERMODEL |
| Theorem B | FTC + completeness \(\Rightarrow\) compact; net radius \(d_{\rm op}\le3\eta\) | VALID PROOF, conditional |
| Delta-test | \(f_x(r)=\mathbb 1[r=x]\) on \([0,1]\): FTC fails for every \(\eta<1\) | DECISIVE COUNTERMODEL |
| Uniform Operational Cell Law | \(\Delta_0>0\), system-independent | NAMED AXIOM (count = ONE) |
| Basin-packing | \(N\le\lfloor B/\Delta\rfloor\), \(M\lesssim(B/\Delta)^2\) | SOUND CONDITIONAL |
| Basin-shallowing | \(d_n=B\cdot2^{-n-1}\) (\(n\ge1\)), \(\sum d_n=B/2| DECISIVE COUNTERMODEL (unrefuted) | |
| \(\hbar\) | \(\approx1.0546\times10^{-34}\ \mathrm{J\,s}\) | MEASURED ANCHOR, atomic, no pull (input) |
| \(M_*\) (main pipeline) | \(6.010\text{–}7.467\times10^{16}\ \mathrm{GeV}\) | DERIVED ratio; convention-soft value, convention-proof ranking |
| \(M_{\rm cutoff}=1/R_0\) | \(6.28\times10^{16}\ \mathrm{GeV}\) | FROZEN |
| Naive \(\rho_{\rm vac}\) vs \((2.3\ \mathrm{meV})^4\) | \(1.56\times10^{67}\) vs \(2.80\times10^{-47}\ \mathrm{GeV}^4\); miss \(10^{113.7}\) | CROSS-GATE, RELOCATES (valid negative) |
| Plaquette \(\langle P\rangle\) | \(0.59639\pm0.00219\) vs reference \(0.5937\) | REPRODUCED, pull \(\approx1.23\sigma\) |
| \(z_*\) target inequality | threshold \(0.998\); readings \(0.109\) (pass), \(0.353\) (pass), per-letter (ill-posed) | CROSS-GATE, OPEN (convention-underdetermined) |
A reader auditing whether the "certified-irreducible" grade is doing honest work should check that the program has live negative controls it did not explain away. Three are on record.
What would falsify the reduction itself (a bounded, checkable bet, not a vague hedge): either (a) a pre-quantum reconstruction theorem deriving FTC from a distinguishability test-space plus an additive-cost axiom set — no Hilbert orthogonality, no trace distance, no Bekenstein bound, no imported nuclearity — with Margolus–Levitin, Landauer, and Bekenstein falling out as theorems; or (b) a demonstration that the cost floor and quantum mechanics are strictly mutually derivable, certifying co-fundamentality directly. Either is finite, target-blind, and falsifiable; either, if produced, is the specific named event that moves this root's grade — not a reviewer's re-reading of existing material.
The root closes at CERTIFIED-IRREDUCIBLE / RESOLVED: one value-free axiom (\(\Delta_0>0\)) plus one atomic measured anchor (\(\hbar\)). That is the honestly reached floor of this attack, reached only after running the derivation as far as it will go and running the reduction attempt (Fork A) into two decisive countermodels. The holes below are not "unstarted work"; each has been engaged, has a named obstruction, and carries a stated odds estimate from having tried and failed in a specific, diagnosable way. Holes 1–4 are geometry-independent (pure operational topology / resource theory); Holes 5–6 are cross-gate points where the floor meets the frozen geometric spectrum.
What a specialist should attack first. Holes 1 and 3 are the same object and carry by far the largest leverage — either would move the grade, both rated ~15–20% odds of a flip on any given framework attempt (a real, fundable, hard-but-tractable target). None of Holes 1–6 can retroactively reopen the terminal itself; CERTIFIED-IRREDUCIBLE / RESOLVED is the correctly and honestly reached floor of what bounded-resource reasoning plus the two decisive countermodels can deliver today. These holes are the map of exactly where a future result could move that floor, and in which direction.
The canonical closure taxonomy distinguishes two RESOLVED-family terminals: DERIVED / DISSOLVED / MEASURED-ANCHOR (the leg reduced to a theorem, dissolved as a non-question, or pinned to a measured number — no structural posit remains) and CERTIFIED-IRREDUCIBLE (the leg reduces to exactly one named structural axiom, and that axiom is proven non-derivable from strictly weaker premises — the standing of an external wall or a measured anchor, because there is provably no lever left to pull). Against these sits ANCHORED: a leg reduced to an axiom whose non-derivability is assumed, not proven, marking an un-discharged obligation.
The granularity root is in the CERTIFIED-IRREDUCIBLE class for a checkable reason: its single axiom has had its non-derivability proven twice, by two decisive countermodels (delta-test; basin-shallowing), each a fully worked counterexample under the stated premises, not a conjecture. The obligation "either derive the axiom or prove it underivable from these premises" has been discharged in the second direction. That discharge is what makes the terminal a legitimate RESOLVED closure rather than an un-audited assumption.
The honest, load-bearing caveat — and the reason the status header is worded as it is — is that "proven non-derivable from the stated premises" is not "proven non-derivable in any future framework." The basin-shallowing countermodel refutes the resource-only route and is unrefuted; but no one has proved that no deeper principle could ever sit below the floor, and no one claims to. That is why this stays a named axiom: the honest ceiling is co-fundamentality, not strict irreducibility (§15.4). A closed gate here reopens only on a named finite measured contradiction, a missing finite blocker, a wrong anchor assignment, a full-13D calculation error, or a theorem failure — and it may be strengthened toward DERIVED only by the one named move of §11 (closing Hole 1 / Hole 3), never softened by a re-reading.
Every gate touches this root in one of three distinct ways, and the distinction is load-bearing — over-inheritance (claiming to inherit \(\Delta_0\) when only the finite-record discipline is used, or claiming \(\Delta_0\) closes a wall it does not) is exactly the unearned closure this program forbids.
UQF-5C (graviton UV-runaway leg of the shared quantum-gravity wall). Inherits the same \(\Delta_0\) cost-floor applied to a Lorentz-scalar proper-time/action record, no new posit. The floor blocks the \(a\to0\) UV runaway (the tower of ever-shorter-distance divergences) by cutting off the infinite descent through ever-cheaper distinguishable states — the (L2) pathology run on the graviton propagator. Terminal: DERIVED-GIVEN-Granularity / RESOLVED. Ceiling: \(\Delta_0\) dissolves the \(a\to0\) divergence tower; it does not build the strong-coupling completion of quantum gravity. The strong-coupling completion remains the shared global quantum-gravity wall and is explicitly not closed by granularity — dissolving the infinities is not the same as building the theory. UQF-9 / UQF-14 inherit this transitively (UQF-14's above-cutoff unitarity question is the same wall, not a granularity debt).
Gap-02 (Yang–Mills mass gap). Inherits the continuum-existence half: a finite-record substrate has no continuum limit in which the gap could vanish as an artifact; the finite-gap value is a measured anchor. Explicitly not a Clay-problem solution. The cross-gate mass-gap inequality is not closed here — it is left honestly OPEN on the convention-underdetermination of the certificate weight \(w(\gamma)\) (three readings disagree: \(0.109\) pass, \(0.353\) pass, per-letter ill-posed). The plaquette anchor \(\langle P\rangle=0.59639\pm0.00219\) (\(\approx1.23\sigma\)) reproduces cleanly; the inequality does not close. Honest pointer: this terminal rests on the existence-half dissolution plus the measured gap anchor — not on a finite-gap certificate; the zero-weight multiplicity feeding the \(K_6\)-spectrum arithmetic is a caught fabrication, flagged and not relied on.
Λ-catastrophe (cosmological constant). Inherits granularity as bookkeeping + negative control only: the huge naive \(\rho_{\rm vac}\) is a smooth-space artifact, and a cost-floor supertrace at the compactification cutoff is shown to relocate, not solve, the smallness problem. CLOSED-NEGATIVE for "granularity explains \(\Lambda\)." Ceiling: granularity makes \(\rho_{\rm vac}\) finite (no UV catastrophe) but does not make it small; \(\Lambda\)'s value is a measured anchor elsewhere. Never conflate the \(113.75\)-OOM miss (vs \(M_{\rm cutoff}=1/R_0\)) with the \(\sim122\) figure (vs \(M_{\rm Pl}\)).
Black-hole singularity. Zero-radius singularities / infinite-curvature points dissolve as continuum-extrapolation objects, not finite observable records: an "infinite curvature at a point" is an infinite descent through ever-finer records with unbounded cost, exactly the (L2) pathology the floor truncates. Terminal: DISSOLVED-GIVEN-Granularity + scale-regular core. Ceiling: dissolves the singularity as a continuum object; the scale-regular interior geometry is a Scale-root contribution, not a granularity posit.
Many gate dossiers cite Granularity as the finite-record discipline (e.g. gap01, gap05, gap10, deeproot-scale, deeproot-shape). These use the discipline; they do not inherit the \(\Delta_0\) axiom as a resolved leg and must not claim to. Two same-batch sibling former-floors carry different certified-irreducible axioms and do not inherit \(\Delta_0\): Born (axiom: non-contextuality of the probability assignment) and SG-2 (axiom: forces realized as isometries of the internal metric). The "four floors" share a grade and a regrade date, not an axiom.
This root leaves a small, fully enumerated set of residuals. None reopens the terminal — they are the honest ledger of what is posited-not-derived (which is what a certified-irreducible closure means) plus the cross-gate debts of sibling gates.
| # | Residual | Type | Terminal now | Exact closing condition |
|---|---|---|---|---|
| R-1 | \(\Delta_0>0\) (Uniform Operational Cell Law) — the one structural axiom | Certified-irreducible axiom (count = ONE) | CERTIFIED-IRREDUCIBLE | Hole 1/3: a pre-quantum reconstruction deriving FTC (no Hilbert/trace/Bekenstein/nuclearity import, no pre-fixed \(\hbar\)) — or a rigorous no-go. ~15–20% odds per attempt. |
| R-2 | Completeness (Cauchy-closure) of \(R_{\rm phys}\) — second assumption inside Theorem B | Declared posit on ledger | Carried (not folded) | A proof that \(R_{\rm phys}\) is Cauchy-closed from strictly weaker premises — or a permanent documentary residue. Never silently absorbed into "FTC \(\Rightarrow\) compact." |
| R-2b | \(d_{\rm op}\)-continuity of the cost functional \(c\) — T5's second hypothesis | Declared assumption, carried not folded | Carried (not folded) | A proof of \(c\)-continuity from a stated compatibility axiom (cleanest: a Lipschitz bound \(|c(r)-c(s)|\le\kappa\,d_{\rm op}(r,s)\)) — or keep as a permanent declared assumption. T5 needs compactness AND \(c\)-continuity. |
| R-3 | \(\hbar\approx1.0546\times10^{-34}\ \mathrm{J\,s}\) — the floor's value | Measured atomic anchor (input, zero pull) | MEASURED-ANCHOR | Not a hole: no reconstruction manufactures a dimensionful number from dimensionless structure. Same footing as \(M_{\rm Pl}\)'s value. |
| R-4 | Uniformity clause of \(\Delta_0\) (universal, not per-system) | Sub-residue of R-1, with banked countermodel | Documentary (default) / derivable | Hole 4: a deeper substrate law forcing one universal \(\Delta_0\), checked to exclude basin-shallowing by construction — or keep on ledger. Leverage is structure-only; never touches the value. |
| R-5 | Cost-currency convention \(B=\) action/energy·time | Declared rulebook convention | Convention (charged) | Not a derivation debt: the alternative \(B=\) bits is a relabel failure (trivializes via \(N\le2^B\)). Named, not owed. |
| R-6 | \(\Lambda\) vacuum-energy magnitude (cross-gate) | Negative control (RELOCATES) | RESOLVED / CLOSED-NEGATIVE for "granularity explains \(\Lambda\)" | Belongs to \(\Lambda\)-catastrophe, not this root. Closing = the honestly-completed regulator-free \(\tfrac12\Sigma\lambda\) over the frozen \(K_6\) spectrum (expected to confirm RELOCATES). |
| R-7 | Yang–Mills mass-gap inequality (cross-gate) | Convention-underdetermined | OPEN (Gap-02 item, not R1 leg) | Hole 6: resolve the \(w(\gamma)\) convention by an independently-motivated block-spin construction, then verify the spectral-gap window uniform in lattice spacing with no fitted prefactor. |
The reframe trades one bare cost-floor value \(\varepsilon\) for a resolution floor \(\Delta_0\) (R-1) plus a completeness hypothesis (R-2): a better-motivated relocation of where the irreducible content sits, not a reduction in the count of residues. R-1 through R-5 are the root's own; R-6, R-7 are cross-gate debts of sibling gates, carried here for completeness. No residual reopens the terminal; each is exactly what a certified-irreducible closure looks like when stated honestly rather than inflated to a derivation.
The one named value-free axiom (the Uniform Operational Cell Law, \(\exists\,\Delta_0>0\)), which is the entire structural price of the reduction, mentions \(\hbar\) nowhere; the one atomic measured anchor (\(\hbar\approx1.0546\times10^{-34}\ \mathrm{J\,s}\)), charged with zero pull as an input; the completeness (Cauchy-closure) hypothesis of Theorem B, tracked in full; the \(d_{\rm op}\)-continuity of \(c\); and the declared cost-currency convention (\(B=\) action/energy·time, not bits). The full residue ledger \(\{\hbar,k_B,\text{Bekenstein constant},\Delta_0,\text{completeness},d_{\rm op}\text{-continuity}\}\) is carried together and never silently dropped. The count of irreducible residues did not shrink; what shrank is the number of structural posits — from an unexamined blanket assumption ("quantum mechanics is quantized, full stop") down to one named, checkable, value-free law. What was won rather than paid for: the surface distinguishability circularity is broken, not paid around, by the pre-Hilbert metric — a genuine, checkable, zero-cost theorem (the classical witness), not a hidden anchor.
Every object this root's chain uses — \(D\), \(T\), \(P(e\mid r,T)\), \(d_{\rm op}\), \(c\) — is defined without reference to \(\mathcal M_4\), \(K_6\), \(S^2\), \(S^1_Y/\mathbb Z_2\), the Killing form, or any curvature/Casimir/heat-kernel datum. Both countermodels survive with zero modification if the frozen geometry were changed, reparametrized, or replaced entirely. So this root's ledger does not double-charge any of the four frozen headline anchors. The one place they enter is the separate, explicitly-flagged Scale-location screen (§9.2), a derivable sub-question already answered, not a further cost charged to the reduction.
The honest ceiling is co-fundamentality, and it is a ceiling to be claimed with confidence rather than apologized for. The claim the program does not, and structurally cannot, make is strict irreducibility: that no principle anywhere, in any future theory, could ever sit below this floor. That is a universal negative over an open-ended domain — the same logical shape as "no theory will ever explain the fine-structure constant" — unprovable in principle for any root, in any field, at any point in the history of science. The program rates its own odds of ever establishing strict irreducibility at approximately 0%, and that is not a weakness specific to this reduction — it is the correct epistemic status of every claim of ultimate irreducibility ever made in physics. This is the unicorn that genuinely dissolves here, exactly once, and it dissolves as a limit on all knowledge, not a gap in this program's knowledge.
What survives is checkable and falsifiable: the cost floor sits at-or-below quantum mechanics, thermodynamics, and gravity in the true order of logical implication — at least as fundamental as each, never a mere downstream consequence of any one — though it has not been shown (and may not be showable) to sit strictly below all three simultaneously. The named, finite path to strengthening it is closing Hole 1 (~15–20% per attempt), with a rigorous no-go theorem counting as an equally valid, equally publishable close in the other direction (terminally confirming the named-axiom route). Both outcomes are wins for the honesty of this ledger; only a quiet claim of derivation without either outcome having been produced would be a loss.
Weighing the scaffolding proved outright (T5, the classical witness, the finite \(3\eta\)-net inside Theorem B, the basin-packing sufficiency lemma) against the one honest gap that remains genuinely open (Hole 1/3), the terminal this root has earned, stated plainly and without softening, is CERTIFIED-IRREDUCIBLE / RESOLVED.