Narrative Layer
Time-stamped projections of the dialectical graph suitable for human reading
The narrative layer is the human-readable projection of a Dialectical Graph: documents, summaries, and explanations generated from the underlying inference layer. Narratives are time-stamped and versioned; they represent what the graph said at a particular moment.
This separation matters because knowledge evolves but documents persist. If you treat a document as the source of truth, it can become stale or misleading as the graph updates. By treating narratives as snapshots, Coscientist makes explicit that prose is derived, not primary.
Every sentence in a narrative should be traceable back through claims, evidence spans, and sources. Without that chain, a sentence is just a sentence, not verified knowledge.