Warconomy refreshes figures manually: a human re-checks the cited source and updates the value with new as-of and review dates. There is no scraping and no scheduled job. Each value has a cadence and a freshness state — current, due-soon, overdue, or stale — and the data-needs and review-actions queues, together with source-health, drive what gets re-checked. Some values are intentionally kept historical and not force-refreshed.
- Manual, source-checked refresh — no scraping, no jobs.
- Four freshness states: current, due-soon, overdue, stale.
- Data-needs, review-actions, and source-health drive the work.
Freshness states
| State | Meaning |
|---|---|
| current | Reviewed within its cadence window; no action needed. |
| due-soon | Approaching the end of its cadence window; queued for re-check. |
| overdue | Past its cadence window; a re-check is recommended before heavy reliance. |
| stale | Long past its window; flagged prominently and prioritized in the review queue. |
Principles
Manual, source-checked refresh
A human re-reads the cited source and updates the value, its as-of date, and its review date. There is no scraping, no scheduled job, and no automated ingestion.
Cadence per value
Each value has an expected refresh cadence based on how often its source publishes (e.g. daily benchmark vs annual estimate). The cadence sets the freshness window.
Queues drive the work
The data-needs backlog lists missing or unverifiable values; review-actions lists what to re-check. Source-health summarizes per-source freshness so attention goes where it is needed.
Historical by intent
Some values should stay as they were: a frozen snapshot, a one-off estimate for a past period, or a figure whose source no longer updates. These are kept historical, not force-refreshed, and labeled accordingly.
Related
Freshness dashboard: /freshness · freshness methodology: /methodology/freshness · data needs: /data-needs · review actions: /review-actions · source health: /source-health.