NEXUS Core Conceptual Layers v0
NEXUS is grounded in the idea that there is one underlying connected reality, while different people and purposes need different ways to localize, name, and work with that reality.
The system should therefore separate source truth, interpreted structure, semantic boundaries, and viewpoint.
1. Observed History
This is the acquisition and provenance layer.
It records what NEXUS actually encountered from:
- provider exports
- manual artifact additions
- later API capture
- later browser or session capture
This layer is append-only and epistemically conservative. It should say what was observed, referenced, captured, revised, or imported, without pretending to know more than the source provides.
Example event family:
ProviderArtifactReceivedRawSnapshotExtractedProviderConversationObservedProviderMessageObservedProviderMessageRevisionObservedArtifactReferencedArtifactPayloadCapturedImportCompleted
Purpose:
- preserve evidence
- preserve provenance
- support reparsing
- avoid premature ontology decisions
2. Graph Substrate
This is the normalized structural layer derived from observed history.
It represents:
- nodes
- edges
- claims or assertions
- provenance back to observed history
The graph substrate is the shared underlying reality inside NEXUS. It is where relationships can accumulate across providers, imports, artifacts, and later interpretations.
Important rule:
- not everything must be stored as graph first
- observed history remains the foundation
- graph structure is derived from that foundation
3. Concern Line
A concern line is a practical thread of work, meaning, or pressure that clearly matters even if its final semantic boundary is not fully settled yet.
Use it when:
- the area is clearly real
- the terms are still being worked out
- we do not want to pretend the final bounded-context shape is already stable
Examples:
- app startup behavior
- screen-path flow
- interaction composition
- business Event Modeling
Guidance:
- use concern lines to organize emerging work without freezing the final structure too early
- promote recurring concern lines to bounded contexts when vocabulary, rules, and responsibility become stable enough
4. Domain
A domain says what area of reality or work a part of the graph concerns.
Examples:
IngestionKnowledgeManagementSoftwareDevelopmentUXBusinessOperationsPersonalSystems
A domain is broader than a bounded context. It helps organize major areas of meaning, but does not by itself define exact language or rules.
5. Bounded Context
A bounded context defines a specific meaning boundary within a domain.
It determines:
- vocabulary
- rules
- responsibilities
- interpretation discipline
Examples within SoftwareDevelopment:
EventModelingRequirementsImplementationArchitecture
Examples near the foundation:
IngestionCanonicalHistoryArtifactReconciliation
Guidance:
- use
BoundedContext, not bareContext, in the model wherever possible
6. Lens
A lens is a way of viewing or working with the underlying graph for a purpose.
A lens does not change the truth underneath. It:
- aliases vocabulary
- groups related structures
- filters what is relevant
- emphasizes certain relationships
- supports a particular user perspective or task
Examples:
DeveloperLensBusinessLensUXLensPersonalKnowledgeLensEventModelingLens
Relationship Between the Layers
ObservedHistoryrecords evidence.GraphSubstrateexpresses normalized structure.ConcernLinehelps name an emerging thread of work before final context boundaries are stable.Domainsays what part of reality or work this concerns.BoundedContextsays what meanings and rules apply.Lenssays how we are looking at it right now.
Initial v0 Guidance
For the first importer foundation:
- implement
ObservedHistoryfirst - keep
GraphSubstratethin but planned - introduce
ConcernLine,Domain,BoundedContext, andLensconceptually now - avoid overcommitting the graph ontology too early
- let richer NEXUS semantics emerge through later interpretation and reparsing
Suggested Minimal Type Concepts
DomainIdBoundedContextIdLensIdNodeIdEdgeIdFactIdGraphAssertionFactProvenance
Working Principle
NEXUS should preserve one evolving underlying reality while allowing many bounded meanings and many valid lenses over that same reality.
The foundation should therefore favor provenance first, structure second, and interpretation as an explicit later layer rather than an implicit assumption baked into ingestion.
Related Direction
For the current longer-term direction from semantic substrate toward specs and running artifacts, see Graph, Spec, And Running Artifact Direction.
For the current ATLAS-oriented clarification about selectable lenses over a fuller model, see ATLAS, Lenses, And The Fuller Model.