Five independent LLM scorers from four labs produce identical rankings (Spearman ρ > 0.8).

{
  "doc": "battery.md",
  "context": "Scorer bias validation: Claude Haiku, Gemini Flash, GPT-5.4, Grok 4, Kimi K2.5 all converge on same model ordering.",
  "norm_label": "Five independent LLM scorers from four labs produce identical rankings (Spearman ρ > 0.8).",
  "graphify_id": "finding_blind_ranking",
  "source_file": "battery.md",
  "imported_from": "/tmp/koan-debug/battery/graph.json",
  "extracted_type": "finding",
  "source_location": "§5.2, Table 6",
  "graphify_file_type": "finding"
}