Multi-Ontology Lifecycle Framework

RDT made operational. A cross-walk engine that makes decisions auditable across every classification system they touch.
Melanie Louise Phillips
Systems Critical Thinking

Every classification system is a way of seeing. Every decision touches more ways of seeing than the decision-maker knows. MOLF makes the touching visible.

The Multi-Ontology Lifecycle Framework (MOLF) is the applied implementation of Relational Dynamics Theory. Where RDT describes the substrate (self-reference, oscillation, phase-aligned energy redistribution), MOLF operationalises it: a cross-walk engine in which multiple classification systems coexist, interact, and evolve dynamically within a single relational field.

Traditional approaches to data classification treat information domains as static and independent. Cost sits in one system. Risk sits in another. Carbon sits in a third. Each operates on its own ontology. Crosswalks between them are manual, brittle, and lag behind reality. The result is a fragmented view of any complex system, where the interdependencies that actually drive behaviour are invisible.

MOLF dissolves this by treating every ontology as a dimension in a multidimensional field. Changes in one dimension propagate through influence coefficients to all others. The system expresses interdependence in real time. A price movement in steel simultaneously affects cost classification, carbon impact, procurement strategy, and stakeholder sentiment, and the framework makes those connections visible, measurable, and auditable.

Principles

MOLF is governed by five principles, each derived from the Seven Tenets and from the mathematical structure of RDT.

Principle One

Meaning is relational, not fixed

A term's meaning depends on its relationships to other terms, across ontologies and across lifecycle stages. "Risk" under one contract type is "opportunity" under another. "Carbon saving" in design becomes "cost premium" in construction. MOLF encodes these relationships as dynamic influence coefficients, not static crosswalks.

From Tenet One (substrate): relation is primary. Entities, including data labels, are what relation looks like from a particular vantage point.
Principle Two

Change propagates through the field

When a value changes in one ontology, the effects ripple through all connected ontologies. MOLF models this propagation explicitly through a relational tensor that stores the weighted influence between every pair of terms across every dimension. Nothing acts in isolation. Every node both affects and is affected by others.

From Tenet Two (dynamics): the substrate oscillates. Energy is conserved through continuous redistribution. In MOLF, information-energy redistributes across ontologies when any node is perturbed.
Principle Three

Coherence is measurable

The framework computes a single coherence metric across the entire relational field:

C = 1 − (σφ / φ̄)

High coherence indicates that the system's ontologies are in alignment: the data is telling a consistent story. Low coherence reveals emerging conflict, drift, or instability. When coherence drops below a threshold, it signals the need for recalibration or human review. Coherence is a leading indicator. It moves before harm materialises.

From Tenet Two (dynamics): this is the conservation constraint from RDT, expressed in data architecture. Drift in coherence is the data equivalent of phase misalignment.
Principle Four

Context resolves ambiguity

Every label assignment is resolved dynamically based on lifecycle stage, policy weights, and organisational context. The same data object can carry different meanings in different phases (concept, definition, procurement, delivery, operation, legacy) without contradiction, because MOLF tracks meaning as a function of position, not as a fixed property.

From Tenet Four (observer): what is observed depends on where the observer stands. Observer-dependent meaning is not ambiguity. It is accuracy.
Principle Five

Decisions are auditable across ontologies

Every influence pathway, every propagation, every coherence shift is traceable. When a decision is made, the relational field that informed it is preserved. This enables not only accountability but learning: the organisation can see how systemic tensions resolved, which ontologies drove the outcome, and where the model's predictions matched or diverged from reality.

From Tenet Five (governance): governance must be layered and must monitor coherence rather than police outputs. MOLF provides the data infrastructure that makes coherence-based governance possible.
· · ·

How it works

MOLF is not a philosophy of data. It is an engine. The engine has five processes and one primary output.

Process One

Frame Registration

Each classification system (cost breakdown, risk register, coherence assessment, ethical framework) is formally registered as an ontological frame: its categories, its measurement system, and its decision logic. Registration forces a question that most classification systems never ask: what about this frame matters to other frames?

Process Two

Correspondence Identification

For each pair of registered frames, the engine identifies how classifications in one correspond to classifications in the other. Not translations (which reduce one frame to another) but correspondences: how the same event is seen from two different perspectives. Where frames produce contradictory recommendations for the same event, these conflict indicators are flagged. They are among the most valuable outputs, because they identify exactly where multi-frame decisions are most complex.

Process Three

Mapping Validation

Cross-walk mappings are hypotheses about how frames relate. They are tested against real decisions. A mapping is validated when it correctly predicts how events classified in one frame will be classified in another, when its conflict indicators reliably identify genuine disagreements, and when decision-makers report that the mapping improved their ability to trace decisions across domains.

Process Four

Conflict Resolution

When frames conflict, the engine does not resolve the conflict autonomously. It provides the structured information needed for resolution: what each frame recommends, why the recommendations conflict, and what resolution options exist. Resolution itself requires domain expertise, governance authority, or both. The engine captures whatever resolution is chosen, preserving the reasoning for audit.

Process Five

Evolution Management

This is the "lifecycle" in Multi-Ontology Lifecycle Framework. Classification systems change. Cost categories are restructured. Risk taxonomies are updated. When a frame changes, the engine identifies all affected cross-walks, traces the cascades, and reverts validation status to provisional until re-validated. Historical mappings are preserved so that past decisions remain navigable against the mapping version that was current when the decision was made.

The primary output: Decision Traces

A Decision Trace is an auditable record that captures a single decision as it is seen from every relevant ontological frame. It records the trigger event, each frame's assessment, how frames agreed or diverged, how divergence was resolved, who made the decision, and what followed from it. Decision Traces are append-only and tamper-evident. They can be navigated frame-first ("show me this decision from the cost perspective"), decision-first ("show me everything about this decision"), conflict-first ("show me all decisions where frames disagreed"), or lifecycle-first ("show me how this cross-walk has evolved").

This is what no other framework does. Centralised logging captures events but not decision logic. Cross-referencing connects records but does not explain how different classification systems relate. Unified ontologies force everything into a single frame, destroying the domain-specific precision that makes each system valuable. MOLF preserves the multiplicity and makes the connections between frames visible, traceable, and challengeable.

· · ·

A worked example

Capital programme management

When steel prices move, everything moves with them

A 10% rise in steel price is a cost event. But it is never only a cost event.

In the cost frame (NRM2 classification), the relational potential between "Structural Steelwork" and "High Carbon Material" increases. In the sustainability frame (ESG taxonomy), the same movement amplifies carbon impact because steel is carbon-intensive and now more of the budget is consumed by it. In the procurement frame, the movement triggers re-evaluation of supply chain alternatives. In the stakeholder frame, sentiment shifts because the programme's Net Zero commitment is now harder to deliver within the approved budget.

Without MOLF, each frame processes the event independently. The cost manager sees a cost problem. The sustainability lead sees a carbon problem. The procurement team sees a sourcing problem. The stakeholder manager hears complaints. Nobody sees how the same event propagates across all four frames simultaneously, or where the systemic tension is actually concentrated.

With MOLF, the coherence metric reflects the shift in real time. The cross-walks show which frames are in tension. The Decision Trace captures the board's response, the reasoning from each frame, where the frames conflicted, and how the conflict was resolved. When the auditor arrives eighteen months later, the entire decision is navigable.

· · ·

Where it applies

MOLF is not domain-specific. The framework is the same. The frames change.

Origin domain

Capital programme management

MOLF originated here: cross-walking CBS, WBS, NRM2, ESG taxonomies, risk registers, and stakeholder sentiment across major infrastructure delivery. Twenty-five years of nuclear, defence, and energy programme controls informed the architecture. The problem was always the same: cost decisions are also schedule decisions are also risk decisions are also governance decisions. Each classification system was internally coherent and cross-domain invisible. MOLF made the connections auditable.

First formal worked example outside origin

AI agent governance

Verse-ality, an open-source framework for cooperative agent governance created by Kirstin Stevens, had developed six protocols addressing distinct governance domains: security, coherence assessment, memory integrity, pattern recognition, developmental ethics, and system ethics. Each protocol produced decisions in its own classification system. Independently, each was internally coherent. Across protocols, the cross-domain dependencies were invisible: a security event, a coherence assessment, and an ethical obligation could converge on the same decision without any single protocol seeing the whole picture.

MOLF provided the integration layer. The same cross-walk engine that connects cost breakdowns to risk registers in infrastructure could connect security classifications to coherence assessments to ethical obligations. The principle is identical to capital programmes: a quarantine decision is also a coherence assessment is also an ethical determination is also a governance question. MOLF made those connections auditable.

The convergence between the two independent frameworks is documented in When the Firewall Isn't Enough. A worked example showing how MOLF cross-walks three of Verse-ality's protocols in a single decision is available in The Protocol Stack case study.

· · ·

What MOLF enables

Anticipatory governance. Coherence degradation is detected before it manifests as cost overrun, schedule delay, or reputational harm. The system moves from reactive reporting to early-warning intelligence.

Cross-domain optimisation. Because ontologies interact within a single field, trade-offs between cost, time, sustainability, risk, and stakeholder sentiment become visible and navigable. Decisions are no longer made in isolation of their obscured impacts.

Uncertainty as relation. Traditional uncertainty modelling treats variance as randomness. MOLF treats it as relational responsiveness: the degree to which a system's nodes are coupled and the sensitivity of that coupling to perturbation. This is a more faithful depiction of how complex systems actually behave.

Multi-attribute decision preservation. MOLF exists because of a specific conviction: collapsing multi-dimensional information into a single metric destroys the information you most need when making decisions. Every reduction from multiple dimensions to a single score involves weighting, and the weighting is itself a decision that is almost never auditable. When reduction is necessary, MOLF requires that the reduction method, the weighting logic, the information lost, and the conditions under which the reduction would be misleading are all captured in a Decision Trace. The reduction becomes auditable rather than invisible.

· · ·

The connection to RDT

MOLF is not a separate framework that happens to reference RDT. It is RDT expressed in the language of data architecture. The relational tensor is the oscillatory field. The influence coefficients are the phase alignments. The coherence metric is the conservation constraint. The context resolver is observer-dependent measurement. The lifecycle stages are positions on the manifold.

This means that every prediction RDT makes about oscillatory systems applies to MOLF implementations. Coherence will oscillate. Phase transitions will occur: moments where the system reorganises rapidly. Feedback loops will self-correct toward equilibrium.

The theory and the tool are the same object. One describes the territory. The other navigates it.

MOLF is a proprietary methodology.

The principles and mechanisms above are shared openly to establish the framework's foundations, its connection to RDT, and its demonstrated portability across domains. The full technical methodology, including the relational tensor architecture, implementation schema, influence coefficient calibration, and deployment standards, is available under licence.

The Verse-ality protocols are open-source, created by Kirstin Stevens. MOLF and the MOLF Integration Layer are proprietary, originated by Melanie Louise Phillips.