Methodological starting point
The method starts with a bounded question and an explicit description of the reality to be understood. Undefined terms, assumed actors and hidden objectives are identified before model construction.
Reality decomposition architecture
Entity and state decomposition
Entities are identified together with their possible states, attributes, authority and temporal behaviour. Each element must have a reason for inclusion in the model.
Evidence and governance loop
Relationship construction
Relationships record dependence, influence, ownership, authority, sequence, composition, conflict, evidence or uncertainty. A graph becomes meaningful only when relationship semantics are explicit.
Cybernetic operating loop
Evidence specification
The method defines what can support, contradict or update each important proposition. Evidence provenance and quality are represented separately from the proposition itself.
Hypothesis and causal discipline
Correlation, mechanism, prediction and causation must not be collapsed into one label. Causal assertions require defined mechanisms and domain-appropriate validation.
Simulation and scenario analysis
Models may be used to explore scenarios and system interactions, but simulation outputs remain conditional on assumptions, data quality and model boundaries.
Validation
Validation compares model statements and predictions with observed reality, alternative explanations and domain expertise. Failed validation triggers correction rather than concealment.
Governance
Every model should declare its purpose, owner, version, data basis, known limitations, validation state and authorized uses.