Kill Switches

Twenty-four kill switches are engaged in this paper. Each is operationally testable. Each has a specified recovery position — if the kill switch fires, the work retreats to a named earlier position rather than collapsing entirely. Numbering follows the master registry: KS-V.1 through KS-V.24. Legacy labels (F0, F1–F3, G1–G3, B2, FC1–FC5, FD0–FD4.4b) are annotated at first mention and remain attached to the formal blocks in the body of the Papers.

KS-V.1 is the global kill switch. If it fires, the entire structure fails — AS is not well-defined and every subsequent result rests on a quantity that has no meaning. KS-V.2 through KS-V.7 target the selection postulates and the gravitational rate limiter. KS-V.8 targets the selection-irreversibility ordering. KS-V.9 through KS-V.13 target the agency apparatus of Paper C. KS-V.14 through KS-V.24 target the coupled-viability apparatus of Paper D.

KS-V.1 — Operational invariance of AS (F0)

Claim. AS values computed from any two co-admissible coarse-grainings of the same system agree within experimental tolerance. The Actualization State is a property of the system and its physical coupling, not an artefact of the analyst’s choice of basis. This is the global kill switch (legacy label F0, formalised at D5 in Paper A).

Test. Prepare a calibration state. Compute AS in two distinct physically realizable coarse-grainings 𝓞₁ and 𝓞₂ that both satisfy D1 for the same physical coupling. The values must agree within the experimental noise floor δ_exp. Failure: persistent disagreement beyond δ_exp between two co-admissible coarse-grainings of the same physical system.

Status. LIVE — EMPIRICAL. GLOBAL.

Recovery. If KS-V.1 fires, AS is not a well-defined operational quantity and the entire structure fails. There is no graceful retreat — every theorem in Papers A through D is stated in terms of AS or quantities that inherit from AS’s well-definedness. The work would have to be rebuilt on a different operational measure of record-structured irreversibility, or the project’s central claim — that irreversibility can be measured operationally — would be retracted.

KS-V.2 — Pointer-basis targeting (F1)

Claim. The objective-actualisation channel respects the record algebra 𝓞. Selection lands on the pointer states selected by the system-environment coupling, not on bases orthogonal to the pointer basis. Legacy label F1.

Test. In systems where the pointer basis is not position — superconducting qubits, photon polarisation, spin ensembles — measure where definiteness consistently appears. Failure: definiteness consistently appears in position despite 𝓞 ≠ position (or, more generally, in a basis not selected by the coupling).

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.2 fires, the objective-actualisation-channel postulate (P in A4.2) fails. The work retreats to decoherence-without-objective-selection: the ensemble dynamics of Paper A’s irreversibility results survive, but the trajectory-level selection account in Paper B has to be reformulated. Paper C’s agency apparatus depends only on the existence of a single realized record sector and survives the retreat.

KS-V.3 — Born violation (F2)

Claim. Ensemble statistics of realised branches deviate from the Born distribution {p_i} predicted by the pre-selection density matrix. The objective-actualisation channel preserves Born statistics at the ensemble level. Legacy label F2.

Test. Run repeated selections on prepared states. Compare the empirical distribution of realised branches against the Born-predicted distribution. Failure: systematic deviation from the Born distribution beyond statistical noise across many trials.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.3 fires, the boundary condition BC1 (B3.6) and the ensemble-consistency requirement (B3.5) fail. The work retreats to a position where selection produces individual realizations whose statistics do not match the Born prediction. AS as a quantity survives; Paper A’s irreversibility theorems survive; only the postulate that the objective channel respects the Born rule fails.

KS-V.4 — Context dependence (F3)

Claim. Selection depends on observer intervention rather than objective dynamics. Whether and when selection occurs is a function of whether an external system has been brought into interaction with the target system, beyond what decoherence-with-environment-coupling already supplies. Legacy label F3.

Test. Run identical decoherence dynamics with and without explicit observation. Measure whether the timing or basis of selection differs in ways not predicted by the standard decoherence-environment-coupling account. Failure: detection of selection signatures whose timing or basis depends on observer intervention beyond what the environmental coupling predicts.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.4 fires, the objective-dynamics requirement on the selection channel fails. The work retreats to a position closer to the Copenhagen interpretation, where selection is observer-relative. Paper A’s record-algebra-based irreversibility results survive at the ensemble level; Paper B’s individual-realization account would require observer-relative formulation.

KS-V.5 — Selection rate exceeds gravitational bound (G1)

Claim. Selection rates obey the bound λᵢⱼ ≤ ΔE_G[i,j]/ℏ for gravitationally distinguishable record sectors. The gravity-limited hypothesis (postulate G in A4.3, formalised in B4) sets the rate ceiling. Legacy labels G1 / F_G1.

Test. Create spatial superpositions of mesoscopic masses with computable ΔE_G. Measure selection timescales. Compare against the predicted bound τ_min = ℏ/ΔE_G. Platform: levitated nanoparticles (tungsten, R ≈ 100 nm, τ_min on the order of seconds) in cryogenic vacuum. Failure: observation of selection with τ < τ_min for systems with well-defined ΔE_G.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.5 fires, the gravity-limited rate hypothesis fails. The work retreats to the weaker position that selection is rate-limited by some universal limiter not yet identified. Papers A and C survive intact. Paper B’s structural requirements (B3) survive; only the gravitational-specific identification of the limiter in B4 fails.

KS-V.6 — Selection in gravitationally degenerate regime (G2)

Claim. Selection does not occur between records with ΔE_G = 0. In gravitationally degenerate configurations, multiplicity persists indefinitely (or at least at timescales much longer than the gravity-limited prediction would allow). Legacy labels G2 / F_G2.

Test. Prepare decohered superpositions with ΔE_G = 0 — for example, nitrogen-vacancy centres in diamond, or nuclear spin states with identical mass distributions. Monitor for single-sector stabilisation versus persistent multiplicity. Failure: rapid selection in the absence of any alternative limiter satisfying B4.2.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.6 fires, the gravitational identification of the rate limiter fails — selection is happening when ΔE_G = 0, so something else is driving it. The work retreats to the search for the actual limiter. Papers A, C, D survive; B retreats to structural requirements on selection without identifying gravity as the specific limiter.

KS-V.7 — Non-gravitational rate scaling (G3)

Claim. Selection rates do not scale universally with non-gravitational parameters across macroscopic records. If selection rates depend systematically on, for example, electromagnetic coupling strength or thermodynamic state across regimes where the gravity-limited bound is not saturated, the universal-limiter requirement fails. Legacy labels G3 / F_G3.

Test. Compare selection rates across systems with similar ΔE_G but different non-gravitational characteristics (charge distribution, temperature, internal coupling). Failure: systematic scaling of selection rates with non-gravitational parameters at rates where the gravitational bound is not yet saturated.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.7 fires, gravity is not the universal limiter. The work retreats to a multi-mechanism account or to a different universal limiter not yet identified. Same retreat as KS-V.5 and KS-V.6 — Papers A, C, D survive; B’s gravitational-specific claim fails.

KS-V.8 — Pre-irreversibility selection (B2)

Claim. Selection cannot precede operational irreversibility. The selection deviation δ_sel(ρ) is zero on the pre-irreversibility manifold of states. Definiteness emerges only after the record algebra has become operationally definite. Legacy label B2 (B2.5).

Test. Look for selection signatures (single-trajectory variance, individual-realisation distinguishability) in states that have not yet undergone operational decoherence to the record algebra. Failure: detection of selection signatures prior to operational irreversibility — δ_sel(ρ) ≠ 0 on states where the dephasing map Δ_𝓞 has not yet acted.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.8 fires, the post-irreversibility activation requirement (B3.1) fails. The work retreats to a position where selection and irreversibility may be coincident rather than sequentially ordered. The cost-of-selection result (B2.4) survives at the ensemble level; the structural ordering between Papers A and B reverses or merges.

KS-V.9 — Agency increase without control (FC1)

Claim. Agency cannot increase without control expenditure. If reach grows without active control input — for example, a non-motile mutant maintaining position in a gradient — agency as defined in C1 has been violated. Legacy label FC1.

Test. Observe a system where reach(x) grows or remains stationary in a drifting environment without active control. The canonical case is a non-motile mutant (u_max = 0) maintaining position in a nutrient gradient. Failure: reach grew or held without control expenditure.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.9 fires, the definition of agency as reach-within-viability-under-admissible-control fails. The work retreats to a position where agency includes passive structural-position effects (e.g. anchoring) alongside active control. Papers A and B are unaffected. Paper D’s coupling apparatus would require reformulation for the agency-transfer subsections.

KS-V.10 — Irreversible loss reversed (FC2)

Claim. Once a system has passed a no-return surface, it cannot return — even with unlimited control input. Irreversibility is operational: no admissible control trajectory exists from the post-Σ_NR region back to the viability kernel. Legacy label FC2.

Test. Identify a system trajectory that crosses a no-return surface (as defined in D9) and then demonstrates return to the viability kernel through control. Failure: observation of passive or controlled return from beyond a verified no-return surface.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.10 fires, the no-return surface concept fails operationally — what looked irreversible was reversible under sufficient control. The work retreats to a position where no-return surfaces are defined only relative to specific control budgets. Paper C’s operator-horizon theorem (T2) survives at bounded-control regimes; the absolute-irreversibility claim weakens.

KS-V.11 — Stable control past no-return surface (FC3)

Claim. Control authority approaches zero as the state approaches a no-return surface (C1.2). Stable controlled behaviour cannot be sustained on Σ_NR or beyond. Legacy label FC3.

Test. Demonstrate a system maintaining stable controlled behaviour in a region the formalism identifies as past a no-return surface. The system holds its position or completes controlled trajectories from the post-Σ_NR region. Failure: stable controlled behaviour from beyond the predicted no-return surface.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.11 fires, the control-authority result (lim ℳ(x) = 0 as x → Σ_NR) fails. The work retreats to a position where some no-return surfaces are operational only against specific drift regimes, not universal.

KS-V.12 — Free lunch (FC4)

Claim. Control expenditure carries non-zero cost (C5). A system with finite control budget B₀ cannot exceed the survival time bound T ≤ B₀/c_min derived in theorem C5.1. Legacy label FC4.

Test. Observe a motile cell with finite ATP reserve persisting indefinitely in a persistent nutrient gradient flow. Or any system that exceeds the predicted survival-time bound while verifiably operating with finite control budget. Failure: persistence beyond T_max with verified finite B₀.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.12 fires, the control budget formulation fails. The work retreats to a position where some control regimes are cost-free in the relevant operational sense (perhaps under specific energy-harvesting boundary conditions). Theorem C5.1 retreats to a conditional statement scoped to non-harvesting regimes.

KS-V.13 — Resurrection (FC5)

Claim. Once an agent is in the ruin set (x ∉ Viab(R), absorbing per C4.2), no admissible control trajectory returns the agent to the viability kernel. Legacy label FC5.

Test. Observe an agent in the ruin set returning to the viability kernel through control. The ruin set is operationally absorbing; failure means it isn’t.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.13 fires, the absorbing-ruin-set definition fails. The work retreats to a position where ruin is reversible under specific control regimes. Most of Paper C’s results survive at the bounded-control level; the irreversibility-of-ruin claim weakens.

KS-V.14 — Multi-agent persistence violation (FD0)

Claim. Multi-agent persistence requires satisfaction of structural conditions N1–N4 (D3.2a). A multi-agent configuration cannot persist indefinitely while violating any of these conditions, under the regularity and monotonic-alignment assumptions. Legacy label FD0 (global D-paper falsifier from D5.2).

Test. Identify a multi-agent configuration that persists for a duration much longer than the drift timescale while clearly violating at least one of the N1–N4 conditions. The configuration must satisfy the monotonic-alignment and regularity assumptions. Failure: long-run persistence with verified N-condition violation.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.14 fires, the necessary-conditions theorem (proposition D3.2a) fails. The work retreats to a position where some N-condition violations are tolerable under the alignment assumption. Sufficient conditions (D3.2b, without alignment) may still hold for narrower scopes.

KS-V.15 — No free survival (FD1)

Claim. Record-writing actions by one agent in a shared viability domain change the viability measure available to other agents. The Geometric Exclusion Principle (D1.3) holds: there is no record-writing action that leaves all other agents’ viability strictly unchanged. Legacy label FD1.

Test. Identify a record-writing action by agent A that leaves both μ(K_B) and ℳ_B unchanged for all other agents B operating in the same shared viability domain. The action must be non-trivial (it must have actually written records). Failure: existence of a non-trivial free-survival action.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.15 fires, the Geometric Exclusion Principle fails. The work retreats to a position where some record-writing actions are externality-free. The structural-filtering (D4.2) and hierarchy (D4.3) results would need to be reformulated; cooperation and deterrence (D4.4) survive if recast in terms of strict-Pareto rather than generic externalities.

KS-V.16 — Additivity under coupling (FD2.1)

Claim. Joint agency is non-additive under coupling. In a coupled system with non-zero coupling terms (non-orthogonal shared constraint coordinates), ℳ_joint ≠ Σ ℳ_i. Legacy label FD2.1.

Test. Measure ℳ_joint and Σ ℳ_i in coupled versus uncoupled configurations. If they coincide across coupled systems with non-orthogonal shared coordinates, additivity holds and the claim fails. Failure: ℳ_joint = Σ ℳ_i across non-trivial coupling regimes.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.16 fires, the non-additivity result fails and the analysis of impedance, anti-resonance, and cascade effects loses its structural ground. The work retreats to a position where coupling effects are treated additively at first order with corrections fitted empirically. Paper D’s emergent-order results (D4) would require reformulation.

KS-V.17 — Impedance-independent efficiency (FD2.2)

Claim. Coupling efficiency — viability transfer per unit control effort — degrades as the impedance ratio |Z_i/Z_j| deviates from unity. Impedance mismatch is a real cost. Legacy label FD2.2.

Test. Measure viability transfer rate at impedance ratios 1:1, 2:1, 5:1, and 10:1 under matched conditions. Failure: efficiency does not degrade with impedance mismatch — transfer rate is independent of ratio.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.17 fires, the impedance-matching subsection fails. Paper D retreats to a position where coupling efficiency is governed by mechanisms other than impedance. The deadline-mismatch and absorption-bottleneck failure modes named in D2.2 would require alternative structural accounts.

KS-V.18 — Anti-resonant optimality (FD2.3)

Claim. Joint viability margin is maximised away from phase-aligned (resonant) coupling and approaches a minimum at phase offsets that align constraint excursions. Anti-resonance — the offset that distributes constraint excursions — outperforms resonance. Legacy label FD2.3.

Test. Measure joint viability margin at phase offsets 0, π/4, π/2, 3π/4, π in coupled oscillatory systems with shared constraint. Theorem D2.3 (toy model) predicts the optimum is off-resonance. Failure: maximum joint viability at phase offset 0 (perfectly resonant).

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.18 fires, the anti-resonance optimality claim fails. The work retreats to a position where resonant coupling is optimal — which would invert the design intuition for stable multi-agent systems. Most of Paper D’s framework survives; the specific phase result fails.

KS-V.19 — Cascade non-propagation (FD3.3)

Claim. In a chain A → B → C with appropriate coupling and shared constraint, termination of A leads to degradation of B and C on predictable timescales. Cascade failures are real and geometrically constrained. Legacy label FD3.3.

Test. Measure survival times T_B and T_C after termination of A in a 3-agent chain. Compare against predictions from D3.3. Failure: B and C do not degrade after A’s termination — the chain does not cascade.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.19 fires, the cascade-failure account in D3.3 fails. The work retreats to a position where multi-agent chains are more resilient than the structural account predicts — each link is somehow buffered. The structural-filtering claim (D4.2) may need to be scoped to non-chained configurations.

KS-V.20 — Order indistinguishable from noise (FD4)

Claim. Persistent multi-agent configurations satisfy structural conditions (N1–N4 or similar) at rates significantly above what null-model noise produces. Emergent order is structurally filtered, not random. Legacy label FD4.

Test. Compute empirical p-value for slack correlation across observed multi-agent configurations against a null model. Failure: p ≥ 0.05 for all candidate systems — observed configurations are statistically indistinguishable from null-model noise.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.20 fires, the structural-filtering claim fails — what looked like emergent order is indistinguishable from random distribution of configurations. The work retreats to a position where Paper D’s results describe the space of possible configurations but do not predict observed-frequency biases.

KS-V.21 — Persistent violators (FD4.2)

Claim. A multi-agent configuration cannot persist indefinitely while violating one or more of N1–N4 conditions, under the monotonic-alignment and regularity assumptions. Legacy label FD4.2.

Test. Identify a multi-agent configuration that persists at timescales much longer than the drift timescale while a single N-condition is verifiably violated. Failure: long-run persistence with verified single-condition violation under the alignment assumption.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.21 fires, the necessary-conditions theorem (D3.2a) is weaker than stated — some N-condition violations are individually tolerable. The work retreats to a stricter statement: persistence requires joint satisfaction of the conditions in some specific logical structure (e.g. at least three of the four) rather than all four.

KS-V.22 — Hierarchy inversion (FD4.3)

Claim. In impedance-asymmetric systems, the lower-impedance agent gains more reach per unit record-writing action than the higher-impedance agent. Hierarchies emerge as a consequence of impedance asymmetry, not from imported authority. Legacy label FD4.3.

Test. Measure Δμ(K_i) and Δμ(K_j) per unit record-writing action in impedance-asymmetric systems. Failure: high-impedance agents gain more reach than low-impedance agents per unit action — hierarchy inverts from the geometric prediction.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.22 fires, the hierarchy-as-impedance-geometry claim fails. The work retreats to a position where hierarchies may have geometric origins but the impedance-asymmetry mechanism is not the principal driver. Paper D’s structural-filtering and cooperation results survive; the specific hierarchy mechanism would need reformulation.

KS-V.23 — Cooperation nonexistence (FD4.4a)

Claim. In systems with mutually positive externalities, joint agency exceeds the sum of individual agencies — cooperation is structurally favoured. Legacy label FD4.4a.

Test. Measure ℳ_joint and Σ ℳ_i in systems with verified mutually positive externalities. Failure: ℳ_joint ≤ Σ ℳ_i — cooperation produces no joint-agency benefit in systems where the structural account says it should.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.23 fires, the cooperation-as-geometric-outcome claim fails. The work retreats to a position where cooperation in observed systems is sustained by mechanisms other than joint-agency benefit (e.g. reciprocal altruism, evolved norms). Paper D’s deterrence and hierarchy results survive; the cooperation-specific result fails.

KS-V.24 — Deterrence exit (FD4.4b)

Claim. Unilateral decoupling from a stable deterrence configuration reduces the decoupling agent’s individual agency. Deterrence is a geometric fixed-point, not a threat: leaving it costs the leaver. Legacy label FD4.4b.

Test. Measure ℳ_i before and after unilateral decoupling from a verified stable deterrence configuration. Failure: the decoupling agent’s ℳ_i increases or remains unchanged after exit.

Status. LIVE — EMPIRICAL.

Recovery. If KS-V.24 fires, the deterrence-as-geometric-fixed-point claim fails. The work retreats to a position where deterrence configurations are sustained by mechanisms other than joint-agency-loss-on-exit. Paper D’s other results survive; the deterrence-specific result fails.

Why these twenty-four

The kill switches engage every load-bearing claim in Papers A through D. KS-V.1 is global — if AS is not operationally well-defined, every theorem downstream rests on a quantity without meaning. KS-V.2 through KS-V.4 engage the postulate that selection respects the record algebra and the Born distribution. KS-V.5 through KS-V.7 engage the gravitational rate-limiter postulate. KS-V.8 engages the structural ordering between irreversibility and selection. KS-V.9 through KS-V.13 engage the agency apparatus. KS-V.14 through KS-V.24 engage the coupled-viability apparatus.

Each kill switch has a specified recovery position. If a kill switch fires, the work does not collapse — it retreats to a named earlier position with the specific theorems that fall explicitly identified. KS-V.1 is the exception: there is no graceful retreat from a framework-level failure of AS’s operational well-definedness.

Smaller claims throughout the paper — specific numerical estimates in the experimental-feasibility appendices, the worked qubit example in Paper A’s F appendix, the specific platform candidates in Paper B’s B5, the chemostat and operator-required-charger worked examples in Papers C and D — could each be tested independently. The work does not list them as formal kill switches because they are illustrative rather than load-bearing. If a specific worked example fails on its specific parameters, the paper still stands. If KS-V.1 through KS-V.24 hold, the load-bearing structure holds.