The interface can frame historical questions more broadly than the loaded evidence. It does not backfill unsupported centuries or carry population-health estimates beyond their source boundary.
| Domain | Observed range | Outside the range |
|---|---|---|
| Organized conflict | 1946-2025 annual; candidate events through August 2026 | No atlas observation before 1946 |
| Civilian casualties | UCDP GED 1989–2025; candidate events through August 2026; default view 2017–2026 | No interpolation or indirect-death estimate |
| Pump gasoline and diesel | Source-dated September 2026 snapshot at source-native resolution | No estimate for geography without a published observation |
| V-Dem governance | Country-years through 2025 in v16 | No automatic forecast |
| Mortality and total fertility | 1980-2023 | Shown as unavailable, never extrapolated |
| Life expectancy | 1543-2023; country availability varies, with broad coverage in recent decades | Shown as unavailable, never interpolated or extrapolated |
| Crude birth rate | 1960-2024 | Shown as unavailable, never extrapolated |
| Total population | 1960-2025 | Shown as unavailable, never extrapolated |
Known interpretive limits
- Absence from a loaded dataset is not evidence that an event did not occur.
- Fatality estimates retain low, best, and high source values where available.
- Country joins can conceal historical boundary changes; source names remain available in raw records.
- Map geometry is reference geometry, not a territorial adjudication.
- Associations among conflict, regime, fertility, and mortality are descriptive until a causal enclosure is specified and supported.
Corrections
Corrections should identify the page, source row or identifier, proposed change, and supporting source. Open an issue in the public repository so the review and resulting change remain inspectable.