Skepticism is a method, not an identity
Critical analysis asks which explanation best fits the available evidence and what observation could prove that explanation wrong. It does not begin with ridicule, a verdict about a witness, or a promise that every case has the same answer. Ordinary explanations resolve many reports; other reports remain uncertain because the data are incomplete. Neither outcome establishes a rule for the next case.
Start with a precise claim. “A light was observed moving east for two minutes” can be tested. “An impossible craft appeared” already embeds conclusions about the object, performance, and available alternatives. Preserve the observer’s description, then rewrite it as separate statements about time, direction, apparent motion, shape, sensor behavior, and any measured values.
A repeatable hypothesis-testing workflow
- Fix the timeline and viewpoint. Record time zone, duration, observer or platform position, viewing direction, elevation, weather, and movement. Separate estimates from instrument readings.
- List competing explanations before choosing one. Include aircraft, balloons, drones, satellites, astronomical objects, atmospheric effects, biological objects, camera or display behavior, deliberate fabrication, and an unresolved remainder where appropriate.
- Derive predictions. A satellite candidate should match a pass time and path. An aircraft should align with traffic data, lights, altitude, or a flight corridor. A balloon should track plausible winds. A planet should remain fixed relative to the sky even if a moving observer experiences apparent motion.
- Seek discriminating data. Favor information that separates candidates: original media, exact geometry, weather profiles, star and satellite positions, air-traffic tracks, sensor manuals, and independent viewpoints.
- Test the strongest counterargument. Ask what the preferred explanation handles poorly and whether the mismatch is larger than the uncertainty in the inputs.
- State a bounded conclusion. Identify what is resolved, what is only likely, what remains unknown, and what new evidence would change the assessment.
This workflow owns the comparison of explanations. For detailed instruction on imagery, radar, provenance, and physical samples, use Evidence and Technology. For the longer end-to-end review, use Evaluate a UAP Claim.
Common candidates and the checks that matter
| Candidate class | Useful checks | Common error |
|---|---|---|
| Aircraft or drones | Air-traffic data, known corridors, navigation lights, range, altitude, and camera focus | Assuming a dim distant aircraft is nearby, then converting angular motion into extreme speed |
| Balloons and lighter-than-air objects | Wind direction by altitude, drift rate, changing illumination, shape over time, and launch information | Expecting a flexible or partly inflated object to keep one familiar outline |
| Satellites and astronomy | Exact time and location, sky position, pass predictions, twilight, star charts, and observer motion | Estimating distance from brightness or reading disappearance into shadow as abrupt acceleration |
| Atmosphere | Cloud type, haze, lightning, temperature profile, horizon, and local weather records | Treating a luminous or structured appearance as proof of a solid object |
| Perception | Viewing duration, reference points, lighting, fatigue, and whether independent instruments agree | Turning an honest perception into exact speed, size, or distance without external reference |
| Camera or display behavior | Original file, optics, focus, stabilization, gain, palette, compression, and known artifacts | Treating glare, autofocus, vibration, or image persistence as the object’s physical shape or wake |
These are hypotheses, not canned answers. NASA’s Spot the Station guidance illustrates why time and illumination matter: a bright, steady moving light may be the sunlit International Space Station, and it can disappear as the viewing geometry changes. FAA guidance documents false horizons and autokinesis—a stationary light can appear to move when stared at in darkness—without implying that every report is a perception error. Use these references to generate tests, then check the specific case.
Three sourced resolution classes
Distance and ordinary aircraft
In AARO’s Western United States case, lights initially reported as a possible restricted-airspace incursion were assessed as commercial aircraft on established corridors, as far as about 300 nautical miles from the sensor. The report says additional location and air-traffic data changed the estimated distance, while sensor vibration and autofocus affected apparent shape. The lesson is not “all distant lights are aircraft.” It is that range is a testable variable and shape on a sensor display can change for reasons outside the target.
Lighter-than-air objects
AARO’s Puerto Rico reconstruction assessed that two infrared targets drifted at wind speed and direction and did not enter the water. The office expressed high confidence that the objects showed no anomalous or transmedium behavior and moderate confidence that they were sky lanterns. That distinction matters: the analysis can strongly reject one performance claim while retaining less certainty about exact identity. AARO’s official imagery catalogue also publishes several 2022 European reports assessed with high confidence as balloons, based on morphology and wind-aligned performance.
Sensor artifacts and known traffic
In three “Atmospheric Wakes” reports, AARO assessed that the apparent wakes were sensor artifacts rather than anomalous propulsion. One object matched a military-aircraft track, another was assessed as a probable small aircraft, and a third was identified as a commercial aircraft using flight data. Analysts also noted similar trails behind other objects in the videos. The discriminating evidence was the behavior of the imaging system and matching traffic—not a judgment about the reporters.
A fourth class shows why a serious method keeps natural candidates on the list: AARO’s official imagery catalogue includes a 2023 Africa report resolved as migratory birds. Biological, atmospheric, and astronomical explanations should be tested with the same care as technological ones, not used as dismissive labels.
Unresolved means the test stopped short
A case can remain unresolved because the original file is missing, clocks are unsynchronized, geometry is unknown, a sensor was not calibrated for the question, or two explanations make the same prediction. “Insufficient data” is a finding about the record, not evidence of extraordinary origin. The correct next step is to identify the missing measurement, preserve the uncertainty, and avoid adding false precision.
Conversely, an ordinary explanation should earn its confidence. Do not call a target a balloon merely because balloons are common, or an aircraft merely because one was nearby. Show the time match, geometry, wind alignment, morphology, track, or artifact signature. Follow the site’s Sources and Methodology, retrieve originals through Primary Sources, and define technical terms through the glossary.
A calibrated conclusion scale
- Resolved: a specific explanation accounts for the material observations with documented evidence and no significant contradiction.
- Highly likely: one explanation is strongly favored, but exact identification or a material input remains incomplete.
- Plausible: the explanation fits, but competing explanations fit nearly as well.
- Disputed: analyses reach materially different conclusions; present their inputs and methods.
- Unresolved: available data cannot discriminate among candidates.
Always attach the confidence to the exact proposition. “No anomalous performance was demonstrated” is narrower than “the object was identified.” Review applied site examples in the Navy FLIR, Gimbal, and GoFast recordings and the Solway Firth photograph, then follow each briefing back to the strongest available record.