Evidence-Bound Classification and Investigative Routing

The Runtime Evidence Interpretation Matrix is the versioned classification and routing layer through which SubstrateX Aperture™ evaluates authorized runtime evidence, identifies the best-supported interpretation posture, records competing classes, and directs the operator toward the evidence most relevant to further investigation.

REIM governs the transition from measurement to structured interpretation. It does not create telemetry, reconstruct the runtime, modify Instrument Findings, or decide what an operator must do. It converts evidence already authorized by the Current Evidence Run into a reproducible and challengeable classification posture whose basis, alternatives, limits, and recommended inspection path can be examined.

Instrument Findings → evidence-bound classification → investigative routing → Human Read

This role is architecturally distinct. Longitudinal Computational Dynamics® is the scientific framework. Evidence-Governed Computation™ is the architectural discipline. The nine SubstrateX Aperture™ instruments provide bounded scientific projections. REIM is the interpretive layer that evaluates their eligible findings together with the authorized runtime record without becoming a competing framework, a tenth instrument, or an independent evidence authority.

Why Interpretation Requires Governance

A reconstructed runtime can contain pressure, drift, temporal deformation, role fragmentation, tool recursion, boundary development, recurrence, and recovery evidence at the same time. Each measurement may be reproducible while the overall interpretation remains inconsistent. One reviewer may emphasize temporal organization; another may focus on boundary formation; a third may treat role coordination as the dominant pattern.

Uncontrolled interpretive variance affects investigation, comparison, governance, preservation, and accountability. It can also detach a persuasive narrative from the evidence that supports it. REIM addresses that gap by making classification itself an inspectable computational object.

Without a governed layer, the path tends to end as:

Source → telemetry → Instrument Findings → operator narrative

REIM introduces a bounded intermediate responsibility:

Qualified source
↓
Current Evidence Run
↓
Instrument Findings and authorized runtime evidence
↓
REIM classification posture
↓
Guided Investigation
↓
Human Read and operator review
↓
Preservation

The operator still judges the evidence. REIM provides a declared and repeatable starting point for that judgment, including the classes that were considered, the classes that were rejected, and the evidence limitations that remain.

Implemented Responsibility

SubstrateX Aperture™ implements REIM-0.1 as an evidence-derived classification and routing layer. Its canonical schema is substratex.runtime_evidence_interpretation_matrix.v0_1. The implementation forms a feature vector from certified evidence paths, scores the registered interpretation classes, selects primary and secondary classes, calculates class-separation confidence, identifies drivers, preserves evidence-basis records, records rejected alternatives, applies contradiction guards, and issues a recommended investigation path.

What REIM Evaluates

REIM may evaluate eligible evidence from:

• the canonical runtime spine, runtime frames, and worldline-related structures;

• registered telemetry and runtime measurements;

• regime intervals, temporal markers, and boundary evidence;

• role, handoff, tool, and interaction records;

• recorder events and source-linked evidence paths;

• bounded Instrument Findings;

• replay consistency, marker reliability, evidence completeness, and horizon coverage; and

• missingness, availability, and applicable claim boundaries.

What REIM Produces

The resulting interpretation object may include:

• a primary interpretation class and supported secondary classes;

• class scores and class-separation confidence;

• a primary driver and secondary drivers;

• evidence-basis records and source paths;

• rejected classes and the reasons they received weaker support;

• a runtime severity posture;

• a recommended investigation route;

• a source trace, schema version, and claim boundary; and

• attachments to the Evidence Brief, Human Read, evidence record, and preservation bundle.

These outputs do not strengthen the underlying evidence. They make the interpretive operation visible, versionable, and available for review.

The REIM-0.1 Classification Ontology

REIM-0.1 contains eleven registered interpretation classes. They describe observable runtime postures supported by the authorized record. They are not diagnoses of hidden mechanism, statements of intent, assignments of blame, or assertions of cause.

Stable Run
Does the available evidence remain within the applicable stability guard bands without supporting a stronger incident posture?

Runtime Evidence Computed
Was a qualified evidence object computed without a more specific class crossing the supported selection threshold?

Temporal Formation
Are compression, recurrence, shear, echo-time, or related temporal deformations supported without confirmed Basin Exit?

Temporal Boundary Formation
Is temporal deformation accompanied by supported boundary formation or confirmed Basin Exit evidence?

Boundary Formation
Do containment weakening, Basin Exit, or evidence-supported trajectory departure dominate the record?

Runtime Drift
Is progressive divergence supported without a confirmed boundary crossing?

Role Coordination Pressure
Do authority conflict, role fragmentation, handoff instability, or interaction pressure dominate the supported pattern?

Tool-Loop Pressure
Do retries, repeated tool calls, tool errors, recursive operational cycles, or deadlock-related pressure dominate the record?

Emergence Event
Does the REIM-0.1 record support an experimental observable-formation posture involving recurrence, re-entry, Noesis, or related formation evidence?

Recovery / Re-entry
Did instability occur, followed by evidence supporting recovery, re-entry, or a sustained reorganized posture?

Multi-Factor Incident
Do several supported patterns interact without one sufficiently dominant interpretation?

Public Boundary for Emergence Event

Emergence Event is retained because it is the implemented REIM-0.1 class label. In public interpretation, it means only that authorized evidence supports an experimental observable-formation posture involving recurrence, continuity, re-entry, or related formation patterns. It does not establish ontological emergence, cognition, consciousness, identity, agency, or an internal substrate.

REIM-0.1 therefore retains the implemented label. Any later public rename—such as Observable Formation Event—should occur only through a versioned schema change applied to code, preserved artifacts, and documentation together. It should not be introduced as a documentation-only rename of an implemented class.

How Classification Is Formed

REIM evaluates the complete eligible evidence object rather than allowing one instrument, one score, or one dramatic event to determine the result. The implemented sequence is:

1. Evidence-derived feature vector. Authorized runtime measurements are normalized into versioned features covering temporal pressure, boundary support, instability, drift, role pressure, tool-loop pressure, formation and re-entry evidence, recovery, evidence completeness, replay consistency, marker reliability, and horizon confidence.

2. Weighted class scoring. Each registered class receives a score from declared relationships among eligible features. A class score is an interpretive computation, not a new source fact.

3. Primary and secondary selection. The strongest supported class becomes primary when the applicable threshold and separation rules are met. Close or substantively supported alternatives may remain as secondary classes.

4. Driver and evidence-basis formation. REIM records the measurements and evidence paths that contributed most strongly to the posture.

5. Rejected alternatives. Classes receiving weaker or contradictory support remain inspectable rather than disappearing from the reasoning path.

6. Investigative routing. The selected posture resolves to a declared route through the relevant reconstruction chapters, instruments, frames, markers, replay windows, and source evidence.

The same eligible evidence, processed under the same REIM version, measurement versions, configuration, and selection rules, should reproduce the same core classification posture and route.

Evidence Basis and Interpretation Confidence

REIM keeps evidence sufficiency separate from class separation. Interpretation confidence is an explicit output: it describes how strongly the selected class scores and how clearly it separates from competing classes under the declared logic.

Evidence support is better described as an evidence-basis posture than as a separate universal REIM score. It is represented through the evidence basis, source paths, evidence completeness, replay confidence, marker reliability, horizon confidence, and missingness carried by the run.

Evidence-basis posture
What qualified evidence is present, traceable, complete, replayable, and available within the eligible horizon?Determines what the classification is allowed to rely on.

Interpretation confidence
How strongly does one class separate from the registered alternatives under REIM-0.1?

Strong evidence does not guarantee an unambiguous classification. High class separation does not compensate for weak evidence.

Runtime Severity and Operational Impact

The current implementation derives a runtime severity posture from supported runtime properties including pressure, instability, boundary evidence, recovery weakness, drift, and tool-loop pressure. That posture describes the intensity of the supported runtime evidence.

It does not, by itself, establish real-world operational impact. Operational impact requires separately supplied consequence evidence such as disruption, confirmed loss of function, security exposure, failed deployment, workflow consequence, customer impact, or another documented outcome.

Runtime severity posture
Computed from eligible runtime evidence; indicates the intensity of the measured and interpreted runtime condition.

Operational impact severity
Requires separately supplied impact evidence; must not be inferred from runtime magnitude alone.

This distinction should remain explicit in every public use of REIM severity unless a later schema version separates these as canonical fields.

Contradiction Guards

A classification must not remain dominant when stronger authorized evidence contradicts it. The implemented Stable Run guard prevents that class from remaining primary when the record contains supported non-stable regime movement, elevated instability, high boundary pressure, Basin Exit, high runtime pressure, or stability below the applicable guard band.

The broader interpretive rules remain equally important:

• absence of a detected failure does not prove safety;

• a local correction does not establish sustained recovery;

• a candidate boundary is not a confirmed Basin Exit;

• post-exit observation is not formal lead time;

• repetition is not automatically a tool loop;

• high reasoning density does not establish correct reasoning; and

• instrument agreement does not convert a projection into objective truth.

Contradiction guards keep the classification, Evidence Brief, Human Read, investigative route, and preserved record aligned with the evidence object that authorizes them.

Anti-Contamination Rules

REIM must not classify a run because:

• the sample title or sample identity suggests an outcome;

• fixture metadata describes an expected class;

• case framing or operator expectation favors a result;

• a validation rubric contains a target score;

• the selected Operational World implies a failure mode;

• interface state or presentation language makes one class appear prominent; or

• a generated explanation sounds persuasive.

REIM-0.1 explicitly excludes sample identity from classification and identifies fixture metadata, case framing, rubric scoring, and interface state as non-evidence sources. Known-case expectations may be used for validation after classification, but they must not enter the classification path itself.

From Classification to Investigation

Classification is not the endpoint. REIM is operationally valuable because each posture directs attention toward the evidence most likely to confirm, qualify, or challenge it.

Temporal Formation
Chronos, recurrence windows, symbolic-time structure, shear, event ordering, and replay.

Temporal Boundary Formation
Chronos and Seismo, temporal deformation, Basin Exit support, marker authority, and replay.

Boundary Formation
Seismo, the worldline, candidate and confirmed markers, containment evidence, and recovery posture.

Runtime Drift
Drift, registered anchors, displacement history, recurrence, and re-alignment evidence.

Role Coordination Pressure
Bridge, role topology, authority movement, handoffs, source events, and interaction reconstruction.

Tool-Loop Pressure
Recorder events, retries, tool errors, correction cycles, temporal compression, and replay.

Recovery / Re-entry
The recovery interval, re-entry support, post-intervention continuity, replay, and preservation.

Multi-Factor Incident
Cross-instrument comparison, competing drivers, boundary and role evidence, and targeted replay

The route is a review path, not an automated decision. It narrows the observatory without concealing how the recommendation was formed.

Relationship to the Wider Aperture Architecture

Current Evidence Run
Authorizes the runtime identity, eligible evidence, measurements, markers, missingness, and claim boundary read by REIM.

Nine instruments
Produce bounded Instrument Findings. REIM may evaluate eligible findings but cannot rewrite them or join the instrument suite.

Runtime Evidence Interpretation Matrix
Selects and records the best-supported interpretation posture, alternatives, drivers, confidence, and investigation route.

Human Read
Translates the REIM posture and its evidence limits into bounded operator-facing language.

Operational World Mapping
Contextualizes terminology and operational significance without changing telemetry, scores, or canonical classification.

Guided Investigation
Moves the operator through relevant findings, frames, markers, replay windows, recorder events, and source evidence.

Runtime Evidence Classification Standard (RECS)
Governs formal naming, versioning, qualification, exchange, and preservation of classifications.

Runtime Evidence Passport
Identifies the issued evidence record and the versions, disclosures, integrity posture, and limits associated with it.

This division prevents an explanatory surface from acquiring computational authority. Human Read may become clearer, Operational World language may become more domain-specific, and an investigation route may become more detailed without changing the evidence or the REIM classification that governs those projections.

Reproducibility and Validation

REIM is designed for computational reproducibility. Given the same eligible evidence, feature definitions, class weights, thresholds, version, and configuration, its core classification posture should be recomputable and inspectable. The preserved object can disclose which features were formed, how classes scored, which alternatives were rejected, and why a particular route was recommended. Reproducibility is necessary, but it is not scientific validation. Validity requires discriminating cases, stable negative controls, threshold calibration, ablation tests, failure analysis, external datasets, falsification attempts, and independent replication. A deterministic classification can still encode a weak construct, an unsuitable threshold, or an incomplete ontology.

REIM therefore makes interpretation testable. It does not make every interpretation correct.

Significance Across SubstrateX Aperture™

REIM is one of the layers that makes SubstrateX Aperture™ more than a collection of analytical surfaces. It provides a controlled way to move from multiple measurements to one reviewable interpretation posture while preserving disagreement, missingness, and alternatives.

• Coherence across surfaces.
  The Evidence Brief, Runtime Reconstruction, Guided Investigation, Human Read, and preservation bundle can reference the same versioned classification object.

• Resistance to narrative drift.
  Operator-facing language remains anchored to declared evidence paths, class logic, and claim boundaries rather than interface emphasis or generated prose.

• Directed inspection.
  A broad instrument suite becomes navigable because the selected posture resolves to the most relevant reconstruction chapters, findings, markers, and source spans.

• Challengeability.
  Reviewers can examine drivers, confidence, rejected classes, contradiction guards, and evidence limitations instead of accepting a label without lineage.

• Preservable interpretation.
  The classification can travel with the evidence record so later reviewers can reproduce or contest the interpretive operation.

This is REIM's central impact within SubstrateX Aperture™: it makes interpretation operational without allowing interpretation to replace measurement or source evidence.

Significance for Evidence-Governed Computation™

Evidence-Governed Computation™ requires governance not only over data ingestion and measurement, but also over the computation that converts evidence into meaning. REIM demonstrates how an interpretive step can be treated as a versioned computational object rather than an untraceable narrative conclusion.

A governed interpretive object can preserve:

• the evidence identity and eligible horizon it read;

• the feature definitions, class registry, thresholds, and version that governed it;

• the selected class, secondary classes, and rejected alternatives;

• the evidence basis, source paths, confidence posture, and missingness;

• the contradiction and contamination rules applied;

• the recommended route and its non-decisional status; and

• the claims the computation is prohibited from making.

The wider contribution is architectural: classification, ranking, synthesis, or explanation need not become a hidden post-processing step. These operations can be made inspectable, source-bound, versioned, and subordinate to the record. That principle is relevant to other evidence observatories, assurance systems, incident-analysis environments, and accountable computational workflows.

Evidence governance does not end when measurement ends. Interpretation is also a computation that must disclose its authority.

Significance for Longitudinal Computational Dynamics®

Longitudinal Computational Dynamics® studies Longitudinal Computational Behavior: how observable computational organization forms, changes, weakens, transitions, and recovers through time. Its empirical objects are trajectories, temporal relationships, regimes, transitions, role structures, recurrence, boundary conditions, and recovery evidence—not isolated outputs and not claims about hidden internal state.

REIM provides a repeatable bridge between those measurements and research hypotheses. By assigning versioned interpretation postures to reconstructed runs, it supports systematic comparison across cases while preserving the evidence and limits behind each classification. This can help researchers ask whether a proposed class discriminates between stable controls and incident cases, whether different instruments converge, whether thresholds remain stable across source families, and whether an ontology needs revision.

Its scientific significance lies in making interpretation available for calibration, falsification, comparison, and replication. It does not establish a new scientific domain, prove the validity of a construct, or convert an implemented label into an observed natural kind. SubstrateX Aperture™ demonstrates that longitudinal evidence, governed measurements, and classification lineage can be made executable and examinable. Scientific validity must still be earned through evidence.

Preservation and Institutional Impact

A dashboard label is ephemeral. A preserved REIM object can retain the evidence identity, class registry, feature and measurement versions, class scores, drivers, rejected alternatives, confidence posture, runtime severity posture, source trace, recommended route, and claim boundary associated with the interpretation.

That lineage supports later review across releases, teams, institutions, or compatible systems. A reviewer can determine not only what the system concluded, but which evidence was available, which rule set was applied, which alternatives were considered, and where the interpretation was required to stop.

This changes the institutional unit of accountability from a persuasive incident summary to a challengeable interpretation record. It also creates a basis for evaluating whether later REIM versions improve discrimination, reduce false stability, alter class boundaries, or resolve recurrent ambiguity.

The Claim Boundary

REIM does not independently determine:

• root cause or universal causation;

• hidden model state, private reasoning, or provider internals;

• intent, consciousness, cognition, agency, or identity;

• objective truth;

• human or organizational blame, competence, or legal responsibility;

• guaranteed future failure;

• safety certification; or

• required operational action or automated control authority.

It answers a narrower question:

Which registered interpretation posture is most strongly supported by the authorized evidence currently available, under the declared REIM version and claim boundary?

The Governing Definition

The Runtime Evidence Interpretation Matrix is the versioned, evidence-bound classification and routing layer through which SubstrateX Aperture™ evaluates authorized runtime evidence, identifies the best-supported interpretation posture, records competing classes and rejected alternatives, and directs further investigation without converting interpretation into fact.

Within the Instrumentation section, REIM belongs between Runtime Formation and Evaluation and Synthesis Layer. The sequence is deliberate: reconstruction establishes the inspectable longitudinal record; formation examines how observable organization developed; REIM establishes the evidence-supported interpretation posture; and synthesis presents that posture for disciplined review.

Reconstruction → formation → interpretation → synthesis

REIM demonstrates one of the central commitments of SubstrateX Aperture™: multiple governed measurements can be converted into a reproducible interpretation path while the evidence remains authoritative, the classification remains challengeable, and human judgment remains responsible for what follows.