Beyond Structural Invariants: Admissibility of Invariant Preservation

Status

Research state: FORMALIZATION_CANDIDATE
Novelty claim: NOT_ASSERTED
Execution authority: NONE
Canonical owner: StegVerse-Labs/admissibility-wiki
Goal id: admissible-structural-invariants-v0.1

This note separates two predicates that are often collapsed:

1. whether a property remains unchanged through transformation; and
2. whether preserving that property remains authorized, purpose-consistent, recoverable, and admissible.

The distinction is motivated by continuity-governance cases in which perfect preservation can sustain the wrong objective, obsolete authority, a purpose-inverting boundary, or a relationship that remains coherent but is no longer legitimate.

Source Observation

A public Continuity Science discussion proposed that identity persists because structural invariants survive admissible transformation. A response question asked whether an invariant is temporal, environmentally conditioned, or subject to decay. The author clarified that invariants need not be timeless; they are preserved within a continuity thread that may begin, evolve, or terminate.

This repository does not treat that exchange as proof of novelty or as an adopted external specification. It is recorded as a research stimulus. The StegVerse contribution under examination is the stronger separation between invariant preservation and the admissibility of preservation.

Baseline Model

Let a system state be (S_i), a transformation be

[ T_i : S_i \rightarrow S_{i+1}, ]

and a candidate invariant be (I_k). A conventional preservation condition is

[ I_k(S_i)=I_k(S_{i+1}). ]

Equivalently, where the notation is meaningful,

[ T_i(I_k)=I_k. ]

This proves preservation. It does not by itself prove that continued preservation is legitimate.

Second-Order Admissibility Predicate

Define a reconstructed commit-time context

[ C_i=(A_i,P_i,E_i,B_i,R_i,G_i), ]

where:

• (A_i): authority and delegation state;
• (P_i): governing purpose and policy state;
• (E_i): evidence available at the decision boundary;
• (B_i): boundary and constraint state;
• (R_i): recoverability posture;
• (G_i): affected entity or multi-entity governance state.

Define the preservation-admissibility predicate

[ \operatorname{AdmPres}(I_k,T_i,C_i) \in {\text{ALLOW},\text{DENY},\text{REVIEW_REQUIRED},\text{FAIL_CLOSED}}. ]

A candidate continuity claim requires both:

[ I_k(S_i)=I_k(S_{i+1}) ]

and

[ \operatorname{AdmPres}(I_k,T_i,C_i)=\text{ALLOW}. ]

Preservation is therefore necessary only when the governing specification declares it necessary; it is never sufficient by itself.

Admissible Structural Invariant

Admissible Structural Invariant — A property whose preservation across a specified transformation remains authorized, evidence-supported, purpose-consistent, boundary-recoverable, and non-inverting for all materially affected entities at the applicable commit-time boundary.

This definition is scoped. It does not imply that the property is absolute, timeless, universal, or permanent.

Domain of Admissibility

Rather than model an invariant as merely decaying with time, define an admissibility domain

[ D(I_k)={C \mid \operatorname{AdmPres}(I_k,T,C)=\text{ALLOW}}. ]

An invariant can remain mathematically unchanged while the current context leaves (D(I_k)). In that case, the property did not decay; its authority to remain controlling expired, was superseded, or became purpose-inverting.

This distinction supports four different outcomes:

Preservation	Admissibility	Interpretation
preserved	ALLOW	continuity candidate survives this gate
preserved	DENY	stable but illegitimate continuation
changed	ALLOW	governed invariant succession or identity-preserving replacement may be possible
changed	DENY / FAIL_CLOSED	neither preservation nor governed succession is established

Invariant Succession

Continuity may require replacing a lower-order invariant while preserving a higher-order governing relation. Let (I_k) be superseded by (I'_k). A succession claim requires a receipt-bound relation

[ \Sigma(I_k,I'_k,C_i), ]

showing at minimum:

• the supersession authority;
• the reason for replacement;
• the governing purpose retained or intentionally changed;
• the evidence and policy references;
• the affected entities;
• the recoverability path;
• the commit-time determination;
• the receipt connecting predecessor and successor.

The higher-order continuity claim is not “nothing changed.” It is “the change was governed, reconstructable, and admissible.”

Failure Classes

Preserved-but-obsolete

The property remains stable after the authority or policy that justified it has expired.

Preserved-but-purpose-inverting

Maintaining the property prevents convergence toward the purpose the property originally served.

Preserved-but-unrecoverable

The property can be maintained only while operator authority or system coherence remains high; degradation removes a safe recovery path.

Preserved-but-relationally-inadmissible

An invariant valid for one entity imposes an unauthorized or nonrecoverable constraint on another affected entity.

Preserved-but-unreconstructable

The property appears stable, but the evidence needed to reconstruct why its preservation was authorized is unavailable.

Multi-Entity Extension

For entities (e_1,\ldots,e_n), define entity-relative admissibility determinations

[ \operatorname{AdmPres}_{e_j}(I_k,T_i,C_i). ]

A system-wide determination must not silently collapse disagreement. A conservative aggregate is:

[ \operatorname{AdmPres}^{*}=\text{ALLOW} ]

only when every required standing class returns ALLOW and all quorum, conflict, and evidence rules are satisfied. Otherwise the result is DENY, REVIEW_REQUIRED, or FAIL_CLOSED according to the governing contract.

This changes continuity from a single-object identity question into a coupled governance problem: can the transition preserve or succeed the relevant structures without defeating another entity's standing, authority, recoverability, or legitimate purpose?

Relationship to the StegVerse Governance Triad

Layer	Question applied to invariants
Transition Governance	Is preservation or succession structurally constructable?
Admissibility Governance	May this invariant remain controlling at commit time?
Continuity Governance	Can the preservation or succession determination be independently reconstructed later?

The triad prevents three invalid substitutions:

preservation != legitimacy
stability != authority
reconstruction != admissibility

Minimal Decision Record

A preservation-admissibility record should contain:

record_id
continuity_thread_id
transition_id
candidate_invariant_id
pre_state_reference
post_state_reference
preservation_result
invariant_scope
affected_entities
authority_references
policy_references
evidence_references
purpose_consistency
boundary_recoverability
multi_entity_dispositions
commit_time
decision
reason_codes
supersession_reference
receipt_chain_reference

The machine-readable candidate record is maintained at:

static/research/admissible-structural-invariants.v0.1.json

Testable Propositions

ASI-P1 — Preservation insufficiency

There exists a transition for which (I(S_i)=I(S_{i+1})) while preservation admissibility is DENY.

ASI-P2 — Governed succession

There exists a transition for which (I(S_i)\neq I(S_{i+1})) while a receipt-bound invariant succession relation permits a higher-order continuity claim.

ASI-P3 — Context-conditioned domain

A candidate invariant can move outside its admissibility domain without changing its represented value.

ASI-P4 — Multi-entity conflict

An invariant can be admissible relative to one entity and inadmissible relative to another; a system-wide ALLOW requires an explicit aggregation and standing rule.

ASI-P5 — Reconstruction independence

Preservation, commit-time admissibility, and later reconstructability are independently variable properties.

Literature-Review Program

The novelty question remains open. Research must compare this formulation with primary literature in:

• invariant preservation, refinement, and simulation relations;
• temporal and dynamic authorization;
• runtime assurance and safety envelopes;
• deontic and defeasible logic;
• institutional and constitutional rule change;
• policy versioning and revocation;
• multi-agent norm conflict;
• identity criteria, persistence, and process ontology;
• category-theoretic morphisms and structure-preserving maps;
• provenance, auditability, and independently reconstructable decisions.

The review must distinguish exact precedent, partial analogue, vocabulary overlap, and a genuinely missing composition. No public novelty claim is admissible until sources, comparison records, and counterexamples are installed and reviewed.

Current Determination

Structural invariant preservation: NECESSARY IN SOME SPECIFICATIONS
Structural invariant preservation: NOT SUFFICIENT FOR GOVERNED CONTINUITY
Admissibility of preservation: FIRST-CLASS CANDIDATE PREDICATE
Novelty: NOT DETERMINED
Publication posture: RESEARCH NOTE
Execution authority: NONE