02 — Information-Piece Measures & Map Value (Workstream A)

Goal: separate low-value counts from meaningful map counts, so Registries-Pivot stops being "an IU counter" and becomes an information map.

The principle

A measure has map value when its drilldown lands on governed, cross-checkable substrate that a human or DOT can act on — a document to build, a topic to route, a workflow to trigger. A measure is low map value when it only answers "how many rows" with no further navigable meaning. Both are tracked; only high-value measures get axis treatment.

Measure classification

#	Measure	Countable?	Map value	Verdict	Existing substrate	Missing substrate	Suggested pivot	Layer behavior	Gov/birth implication
A1	Information pieces (raw IU count)	yes (219)	LOW	FINAL, keep as basic L1	information_unit	none	PIV-311 (live)	drill by unit_kind (law_unit 187 / design_doc_section 32) → SATISFIED with a small group pivot	none (already born)
A2	Documents built from information pieces	yes (5,710)	MEDIUM-HIGH	CANDIDATE L1	knowledge_documents (status, category, tags, parent tree)	a pivot + birth of the pivot def	PIV-310 (new)	drill by category / status, then by topic → documents	new pivot row owner-gated; source already governed
A3	Topic / business-subject axis	yes (nodes)	HIGH	CANDIDATE → governed axis	taxonomy FAC-08 (empty), iu_metadata_tag topic:* (7)	populate FAC-08; axis_registry; reconcile tags	PIV-320..324 (new)	dynamic-depth graph (doc 03)	topic nodes are born/governed objects (doc 04)
A4	Workflow ↔ information/topic relations	yes	HIGH (automation)	CANDIDATE	workflows (PIV-004), workflow_steps, universal_edges	edges topic↔workflow; pivot	PIV-327	drill: topic → workflows that consume it	edges governed; register edge type
A5	Agent/DOT/event ↔ topic/IU/document relations	yes	HIGH (control)	CANDIDATE	dot_tools (PIV-007), agents (PIV-011), event_type_registry, universal_edges	edges; pivot	PIV-328	drill: topic → DOTs/Agents/events touching it	edges governed; Điều 35/45

What Registries-Pivot should count for information pieces (answer to macro Q1)

Three tiers, in increasing map value:

1. Substrate tier (low map value, always tracked): raw IU count (PIV-311=219), documents count (PIV-310=5,710). These prove the corpus exists and are independently cross-checkable, but they are not a map.
2. Axis tier (high map value): topic axis (FAC-08) node/relation counts (PIV-320..324) — the actual map. A business user navigates topics, not row counts.
3. Relation tier (high map value, automation): documents-by-topic, IU-by-topic, workflows-by-topic, DOTs/Agents/events-by-topic (PIV-325..328) — the drill edges that connect the map to action.

Meaningful vs low-value, tracked-anyway (answer to macro Q2)

• Meaningful: A2 (documents), A3 (topic axis), A4/A5 (relations). These get axis registration, governance, and drill.
• Low value but still tracked: A1 raw IU count. It stays as PIV-311 (a basic L1 verification number) but is explicitly demoted in the UI — it is a substrate count, not the headline of the information map. The headline becomes the topic axis.

Coverage reality check (honest, not faked)

Topic coverage today is 16 of 219 IUs (7%) and 0 of 5,710 documents carry a governed FAC-08 topic. So the topic axis is defined but nearly empty. The pivot map (doc 05) therefore reports topic counts honestly as small/zero with NEEDS_LAW_DEFINITION or CANDIDATE status — never faked, never back-filled by AI auto-promotion. Populating FAC-08 and raising coverage is a governed, post-ratification activity (doc 04 lifecycle), not a counting trick.

Workstream A completion: IU measures are no longer vague — five named measures, each with map-value class, substrate, gap, and pivot.