Production updateA scoped discovery map is deployed in the Geopolitical Risk tab. It is deliberately narrower than the staged command center described in the original preview: no social evidence, inferred points, composite coloring, or discovery-driven scoring is included.

What Is Live Now

The production map accepts direct-publisher news only when the source supplies a valid publication time within 72 hours. A row must match existing country evidence by canonical HTTPS URL, normalized publisher, and normalized title. Matching happens before the newest 20 sources are ranked, so unrelated newer headlines cannot displace country-linked evidence.

  • Strict ISO_A2_EH joins attach matched discovery to country polygons.
  • Every highlight is labeled unconfirmed_discovery, score-ineligible, and confidence-free.
  • Missing, invalid, future, expired, stale, degraded, indirect, or mismatched rows are withheld.
  • No social rows, capital points, centroids, inferred geometry, or source-free markers are displayed.
  • Replay immediately removes live discovery evidence and blocks late responses from restoring it.
  • Vector, imagery, terrain, topographic, and OSM basemaps are available in embedded and full-screen views.

The release gate included 27 focused syntax, lifecycle, ranking, privacy, and route tests; an isolated Node 20 runtime smoke; a real Chromium desktop, full-screen, reduced-motion, and 390-pixel mobile smoke; and read-only compatibility checks against current public news and country-risk payloads.

The staged Geopolitical Risk panel is designed as a persistent global command surface rather than a single composite table. It combines country-level pressure with a bounded stream of news and social discovery signals while keeping each evidence family separately inspectable.

Core ruleMissing coverage is not calm. It is no data. Country-only reporting pulses a country polygon; it never receives an invented city or country-centroid point.

Eight Lenses, One Geometry Contract

The candidate panel exposes Overview, Civil, Prices, Banking, Sovereign, Markets, Controls, and Evidence lenses. Each lens declares its own metric and coverage. Civil, price, banking, and evidence views are intended to preserve the difference between an observed zero and a missing field. Sovereign and market values should enter observed scoring only when their source and freshness contracts pass.

The global join uses the bundled ISO_A2_EH boundary field. Duplicate or invalid country rows are withheld rather than guessed. Neutral polygons mean no data; they are not a zero-risk category.

Source-Native Points Only

A news or social record becomes a point only when it carries defensible source-supplied coordinates. Coordinates marked inferred, estimated, geocoded, centroid-based, or otherwise derived are rejected even if the numbers are geographically valid. If the same record has a strict country association, its country polygon can pulse instead.

This policy prevents a common map failure: placing a precise-looking marker at a national capital or centroid when the source only names a country. Geometry caveats remain visible in point popups.

The Newest 20

Up to 200 classified discovery rows are retained within a 72-hour window. The newest 20 receive stable ranks and visual pulses. Ranking uses timestamp first and source score only as a tie-breaker. News and social channels remain separate, and the animation respects reduced-motion preferences.

These rows are discovery aids. A pulse does not confirm an event, severity, attribution, or causal relationship with the country score.

Replay Uses the Captured Clock

The candidate implementation reconstructs evidence windows, age labels, model timestamps, and newest-20 ranks from the snapshot capture time. It does not use the viewer's current clock. Live refresh controls are disabled during replay, and both direct news loading and late in-flight responses are blocked from mutating captured evidence.

Scoring Corrections

The candidate country-risk backend is intended to emit unsupported countries as unscored with a null score and null confidence. An unknown NR rating should not contribute a default risk floor. Additive score breakdowns reconstruct the displayed score exactly, including a visible cap adjustment when a result reaches 100.

  • Synthetic CDS and debt baselines remain labeled context and are excluded from scoring.
  • CDS observations require a valid, non-future as-of date within the accepted freshness window.
  • Market rows require a trustworthy row or snapshot timestamp; stale: false is not enough.
  • Tax and fiscal diagnostics are not added again after they are represented inside sovereign-debt risk points.
  • Registered or reachable source inventory does not count as risk evidence.
  • SEC distress is labeled as a latest-filings sample when the feed is incomplete.
  • CISA 30-day additions are counted from the complete available catalog before display sampling.

Broader Candidate Verification Status

PASSOffline syntax and static contract checks pass against the staged files.
PASS174 bundled boundaries are present for strict country joins.
PASSStatic coverage includes eight lenses, five basemaps, newest-20 ranking, replay guards, and source-native geometry rules.
REVIEWDiscovery scoring, undated social rows, static-field scoring, and evidence metadata require correction before production.
PENDINGDesktop, mobile, reduced-motion, full-screen parity, and keyboard browser acceptance must run against final rebased files.
PENDINGProduction smoke testing, accepted hashes, account-state checks, and rollback rehearsal are not complete.

Reading the Live Map

Open the Geopolitical Risk tab and review the ranked source rail before interpreting a highlighted polygon. A highlight means one current direct-news row exactly matched country evidence; it does not confirm an event, severity, attribution, or causal relationship. No highlight means only that no eligible match survived the current source and freshness contract.

Related releaseConflict and Military OSINT Hardening