Data review queue

Review by category

The same sitewide review queue, grouped by public category. Category review is derived from the same static observations — it does not change any sitewide totals. Stale means a re-check is recommended, not that a value is wrong.

Sanctions review

13 live sanctions indicators across 4 topics (14 sources); 7 high-priority, 7 stale, 4 due soon. Policy thresholds are policy parameters, not market prices; this is an economic-impact reference, not legal or compliance advice. See the sanctions dashboard for the cross-topic overview.

• U.S./G7 Russian crude oil price cap — as of December 5, 2022
• Russian refined products price cap (premium-to-crude) — as of February 5, 2023
• Shadow-fleet share of Russian oil exports — as of September 25, 2023
• Common High Priority Items (export-control list) — as of February 23, 2024
• G7 ERA loans to Ukraine (total) — as of October 25, 2024

Review queue

Live/static indicators, highest priority and most overdue first. Recommended-review-by is derived statically from source cadence and each value’s asOf date.

By page

Each canonical page’s live/sample split, oldest live value, and the worst review status across its live rows.

By source

Every source behind a live observation, with its cadence, last-reviewed date, recommended-review-by date, and status.

Sample-row promotion queue

Sample rows are not live data. They are illustrative placeholders to be replaced by a source-linked observation when an authoritative value can be verified directly.

• Russia–Ukraine war — 2 sample rows. Look for an official UN/FAO trade source for grain volumes and a market benchmark for prices.
• Strait of Hormuz — 1 sample row. Look for official EIA transit-volume statistics for the strait.
• Red Sea shipping corridor — 1 sample row. Look for an authoritative, openly published freight-rate value (e.g. a public UNCTAD or non-paywalled index figure).
• Global shipping risk — 1 sample row. Replace the composite with a single transparent transit/trade metric rather than an opaque score.

How review dates are computed

• Recommended-review-by dates are deterministic and static, derived from each source’s cadence and the value’s asOf date relative to a curated reference date — never the runtime clock.
• Cadence thresholds: monthly (review due ~30d, overdue ~45d, stale ~90d), annual (~12mo / ~18mo / ~30mo), irregular/event snapshots (~90d / ~180d / ~365d).
• Review status is not data correctness. A stale or overdue value can still be a historically valid figure; the status only signals that it should be re-checked against its source before being cited as current context.
• Category review (sanctions, conflicts, chokepoints, dashboards) is the same sitewide queue grouped by public subject type — derived from the same static observations, so it never changes the sitewide totals.
• Refreshing is a manual workflow: open the cited source, confirm the latest value and date, update the observation and its lastReviewed date, and bump the reference date. There is no scraping, scheduled job, or runtime fetch.

Frequently asked questions

What is the data review queue?

A deterministic list of which source-linked values to re-check next, with recommended-review-by dates derived from each source's cadence. It is review status, not real-time monitoring.

What does category review show?

The same sitewide queue grouped by category (sanctions, conflicts, chokepoints, dashboards). Category counts reconcile with the sitewide totals.

Does stale mean the figure is false?

No. Stale means a manual re-check is recommended before citing the value as current context; dated policy effective dates and event snapshots read stale by design.

Related

• Data coverage — the live/sample split sitewide and per page/source.
• Source registry — every public source, with per-source detail pages.
• Source quality — citation readiness for the sources backing high-priority rows.
• Data maintenance — the periodic-source watchlist and next manual checks by cadence.
• Review actions — the prioritized manual-action checklist (refresh, append, preserve, replace, verify).
• Methodology — confidence levels and language policy.