Why source quality matters
The project moved away from social-media hunting because generative media, repost accounts and engagement incentives contaminate visual evidence.
NUFORC remains useful as a lead layer, but its own policy makes clear that reports are not treated as validated physical data. Older reports also lack the newer grading layer.
How rows are used
Tier A/B rows define the measurement target. Tier C historical rows define language and analog patterns. Tier D/E material is kept as lead or control data, not as proof.
| Level | Source type | Allowed use |
|---|---|---|
| A | DVIDS/AARO/FBI/DoD video or document; Altea 2015 direct witness geometry. | Phenotype and physical core. |
| B | Official IR/SWIR/FMV sensor cases without visible color. | Support for invisible modes, speed, low altitude, water/land and contrast behavior. |
| C | Historical cases with academic or primary source context. | Continuity of language and phenotype; not routes or physical proof. |
| D | Civil candidates with original files or strong manual review. | Candidate layer only; no global inference without verification. |
| E | NUFORC without media, social clips, viral reposts, AI risk. | Watchlist/quarantine; no inference. |