Knowledge-graph value experiment

Opportunity: Test whether a maintained research graph improves retrieval, handoff accuracy, duplicate detection, and research selection enough to justify its cost.

Unknowns: Causal value, maintenance burden, ontology stability, and graph-vs-search advantage.

Origin and evidence: CURRENT-STATE.md open question; knowledge-platform/knowledge-genome.json; Repository-Architecture.md; Search-Architecture.md; this analysis’s dependency graph.

Dependencies: RFR-004 and RFR-009; stable corpus and identifiers.

Method: Predefine representative navigation and synthesis tasks; compare file/search baseline with graph-assisted workflow; measure correctness, time, missed provenance, duplicate proposals, and maintenance time.

Outputs and success: Minimal graph prototype, benchmark tasks, cost-benefit decision. Success requires material, replicated gains that exceed ongoing maintenance cost.

Recommended agent: Knowledge-engineering researcher. Effort: medium-large. Expected gain: potentially compounding research coordination; high ontology-debt risk.