Coordination Metrics

Documented coordination connections overwhelmingly form within industry boundaries in the documented record; the cross-sector share is far below what the corpus’s own industry composition would produce at random.

Documentation accessibility is itself sector-asymmetric — under the coverage-confidence framework?Coverage-confidence frameworkWEO grades how documentable each domain is; Security is graded medium/low because classification suppresses public documentation. Domain grades: the Coverage Confidence panel on the events page · methodological basis: the Methodology Manual’s structural visibility asymmetry provision. the Security domain is graded medium/low, and is additionally constrained by national security classification across all blocs. A within-industry concentration in the documented record is therefore not, on its own, evidence of a within-industry concentration in the world. Connection discovery itself also begins from events that appear related, and apparent relatedness correlates with shared industry — so some portion of the same-industry concentration may reflect how connections are found rather than how coordination is structured. Each connection is individually evidence-qualified, so cross-sector counts are conservative floors; the concentration figure describes the documented network as constructed and searched.

Baseline: computed against the 150-CC corpus, 28 July 2026 · methodology: Towards Coordination Science

Current figures: computed against the 150-CC corpus as at 28 July 2026 — recomputed and appended to the observation series at every corpus deployment.

Most connection types are mutual and are counted once per pair (123 of 150). The 27 connections with a genuine direction are listed separately below, preserving their from→to arrows.

Reporting rule (provisional, from Towards Coordination Science): cells below 3 are displayed but no finding is built on them — at counts this small a single correction could erase the cell. Corridor language — a recurring transmission route between two industries — is reserved for cells of 5 or more; at this baseline exactly one cross-sector corridor qualifies — AI Software ↔ Cloud & Data Centres, at 6.

Boundary condition. The Towards Coordination Science paper recommends at least 150 connections before matrix-level interpretation. The corpus sits at exactly 150 — the recommendation floor is precisely met, not comfortably cleared.

All 11 categories render, including the four with no primary-tag incidence (Trade & Logistics, Automotive, Quantum, Telecoms) — absence is information. Telecoms is never assigned as primary by taxonomy design; its row is structurally empty.

Cross-sector transmission matrix, 28 July 2026 · row order fixed to the Methodology Manual’s 11-category industry taxonomy · row labels carry each industry’s total connection incidence.

Baseline: computed against the 150-CC corpus, 28 July 2026 · methodology: Towards Coordination Science

Current figures: computed against the 150-CC corpus as at 28 July 2026 — recomputed and appended to the observation series at every corpus deployment.

Each deploy recomputes these metrics and appends a dated observation. Values are comparable only within a taxonomy epoch: a revision to the industry taxonomy breaks the series, and any trend must be scoped to a single epoch rather than drawn across a break. Successive observations record whether documented coordination remains within industry boundaries or increasingly crosses them as the corpus matures — the series exists to answer that question, not to presume it.

One observation on the series. The change-over-time graph ships when the series carries enough observations to plot; this container and its epoch contract are already in place.

Machine-readable: /api/metrics · MCP tool get_metrics — the same stored observations, free on both tiers.