Begin with the claim, not the conclusion
Write the claim in one sentence without promotional language. Identify whether it concerns an observation, an object’s identity, extraordinary performance, government knowledge, a physical sample, or a wider hidden program. Different claims require different evidence.
1. Locate the earliest available source
Find the original file, full interview, agency report, hearing statement, dataset, or contemporaneous record. A later article may accurately quote it, but each layer of retelling can remove qualifications and context. Record the creator, date, publication venue, and stable identifier.
2. Separate observation from interpretation
“A light crossed the frame” is an observation. “The object accelerated beyond known technology” is an interpretation that requires distance, timing, geometry, sensor behavior, and uncertainty. List what the source directly records before discussing possible causes.
3. Check provenance and continuity
For images, video, documents, and samples, ask where the item came from, who handled it, whether the complete original exists, and what transformations occurred. Re-encoding, cropping, enhancement, missing metadata, or an undocumented chain of custody does not automatically make an item false, but it limits what can be concluded.
4. Test independence
Multiple witnesses are strongest when their accounts were recorded separately and before coordination. Multiple articles repeating one witness are one source, not many. Two sensors may also share a platform, clock, tracking solution, or processing pipeline.
5. Establish geometry and baseline behavior
Apparent speed and size depend on distance and field of view. Camera motion, parallax, autofocus, infrared glare, tracking gates, stabilization, rolling shutter, and compression can create surprising effects. Compare the observation with known aircraft, balloons, satellites, astronomical objects, atmospheric effects, and sensor artifacts under similar conditions.
6. Read the strongest counterargument
Look for the best ordinary explanation, not the easiest one to dismiss. Then ask whether it accounts for the full record and whether its assumptions are stated. A confident identification should explain why competing interpretations fail; an unresolved case should say which missing data prevents a conclusion.
7. Calibrate confidence
Use conclusions such as documented, well-supported, plausible, disputed, weakly supported, unresolved, or contradicted. Avoid forcing every case into “proved” or “debunked.” Confidence should follow the quality and completeness of the data.
8. Preserve falsifiability
State what new evidence would change the assessment: original telemetry, a second camera, a contemporaneous log, a verified chain of custody, a calculable range, or a document from an independent office. A claim that explains every possible outcome cannot be meaningfully tested.
Quick checklist
- What exactly is being claimed?
- What is the closest original source?
- What does the source directly show?
- What is inferred?
- Is the provenance documented?
- Are the witnesses or sensors independent?
- What ordinary explanations were tested?
- What information is missing?
- How confident is the conclusion?
- What evidence could change it?
Continue with the Primary Sources guide, Sources and Methodology, and Critical Analysis hub.