How the barometer works
This dashboard tracks escalation through four labeled layers: verified events, analytical assessments, interpretive signals, and structured market data. It exists to inform, not persuade.
Score logic
The headline score is a weighted synthesis of vectors. Each vector receives a score, a change value, a confidence label, and a plain-language primary driver. The final score is rounded and expressed as an index out of 100.
Status labels follow threshold bands:
- 0–39: Low / Watch
- 40–59: Elevated
- 60–74: High
- 75–89: Severe / Critical
- 90–100: Threshold / Extreme
Content classes
Verified covers sourced events, actions, statements, or datapoints. Assessment covers analytical interpretation of what those signals may mean. Interpretive covers theological, symbolic, or worldview-based significance. Data covers structured market or macro inputs.
Confidence model
Each card or vector carries one of three confidence tags:
- High — multiple credible confirmations or strong direct sourcing.
- Moderate — one credible source or a narrower evidence chain.
- Low — speculative, thinly sourced, or provisional.
Publishing workflow
The dashboard is designed for semi-automated updates. A scheduled backend job drafts a proposed snapshot, which is then reviewed before promotion to the live snapshot. If an update fails, the site continues serving the last known good state.
- Fetch source inputs and review packet content.
- Generate
proposed.json. - Review proposed changes and score movement.
- Promote the proposal to
latest.json. - Archive the previous published snapshot.
Source policy
Key cards should expose a source label, source type, publication time, retrieval time, and a source URL when available. Public visitors only load static assets and JSON; no browser-side model calls are used.