Finiteness: Five Structural Reasons

The finiteness of the one-loop correction is not accidental. It follows from the axioms. Each axiom contributes one structural reason. Five reasons. Five locks. And you hold the key to each one — because the kill switches are yours.

4.1 — Axiom B: ultraviolet finiteness

The break has a minimum size. ε is the minimum viable splinter (Axiom B); the action-scale of the minimum record is ℏ (AP12 §1). There is no sub-ℏ physics — if you try to go smaller, there is nothing to go to. The cells cannot be subdivided below the Planck scale. The ultraviolet divergence does not arise because there are no arbitrarily short distances. The minimum record IS the Planck-scale cutoff, not imposed by hand but by Axiom B. Formally: K₁ = Σ_k a(k) has at most V/ℓ_P⁴ terms, each bounded by 1 (unitarity), the index set finite. The one-loop amplitude is finite; the induced correction scales as ℓ_P²/L² (§3.4).

4.2 — Axiom C: infrared finiteness

Propagation is bounded by c. The causal diamond of any observation has finite volume; correlations cannot extend beyond the horizon. The cosmological constant Λ (Lovelock, AP08 §9.3) sets a maximum causal volume — the de Sitter horizon at r ~ √(3/Λ). The sum does not extend beyond it. The infrared divergence does not arise because the manifold has a natural boundary. Formally: V ≤ V_Λ = c⁴/H² with H² = Λc²/3; the number of terms is bounded above by V_Λ/ℓ_P⁴. Finite.

4.3 — Axiom R: discreteness

Records are irreversible and the monoid is countable. The sum over virtual records is a sum over a countable set, not an integral over a continuum — there is no measure problem. Each amplitude contribution is a matrix element of the unitary evolution operator (§3.3), which is bounded. The Born rule acts on the total amplitude to give probabilities, providing the unique measure consistent with the involution σ (AP09, the Born-rule measure). At one loop the relevant sum is finite because the one-virtual-record index set is finite (§3.2) and each contribution is bounded; higher-loop convergence depends on the combinatorics (KS-27; D7).

4.4 — Axiom S: the probability measure is well-defined

The two sectors ℒ and 𝓟 are related by the involution σ. The Born rule P = ψψ* = Light × Dark (AP09 §6) gives the probability of each final state from the total amplitude. Both sectors must agree, so the measure is fixed uniquely. Without Axiom S the extraction of probabilities would be ambiguous — computed from which sector? With Axiom S there is no ambiguity: the probability is the product of both sectors, ψψ*. The 1:1 voting. The measure is canonical. You do not choose it; the axioms choose it for you.

4.5 — Lovelock in D = 4: no counterterms

The deepest structural protection. In standard quantum gravity the divergences are not merely large numbers — they are new operators (R², R_μνR^μν, R_μνρσR^μνρσ) absent from the Einstein-Hilbert action. Each loop order generates new terms; you need infinitely many counterterms; the theory is non-renormalisable.

In the record algebra the protection has two layers. Structural: the algebra constructs geometry as the accumulated record, and AP08 derives that this yields exactly G_μν + Λg_μν. Virtual records are the same algebraic objects as real records — they differ in commitment status (summed over vs. written), not in structure. The geometry they contribute to must therefore be the same geometry real records produce. Theorem layer: Lovelock’s theorem (AP08 §9) states that on a 4-manifold the unique divergence-free symmetric 2-tensor at most second-order in metric derivatives is G_μν + Λg_μν. AP10 derives N = 3 spatial dimensions, making D = 4 — and Lovelock — unconditional.

Higher-order curvature terms are fourth-order in derivatives. The record algebra avoids them because it constructs geometry directly from records rather than from an effective action: within its geometric construction, higher-derivative terms have no mechanism to arise, and Lovelock confirms no alternative second-order tensor exists in D = 4. A formal proof that the virtual-record sum inherits these constraints — that summing over non-committed records cannot generate higher-derivative contributions to an effective action — is outstanding (D7, §10). The structural argument is compelling but not yet a theorem.

The counterterms are not renormalised away. They are structurally absent within the algebra’s geometric construction (formal proof outstanding; D7). Here is the weapon: find the counterterm. If the virtual-record sum generates a higher-derivative operator, this argument collapses and KS-27 fires.