Advanced Annotation Rubric & Analysis Framework
| Dimension | 5 – High Credibility | 3 – Moderate Credibility | 1 – Low Credibility |
|---|---|---|---|
| Flight Hours | 10,000+ hours; commercial or military rated | 1,000–10,000 hours; private or commercial | Under 1,000 hours or unverified |
| Specificity of Claim | Precise altitude, heading, coordinates cited | General observations without exact data | Vague or anecdotal claims only |
| Corroboration | Multiple independent pilots confirm same observation | Single pilot with partial instrument confirmation | No corroboration; hearsay or secondhand |
| Instrument Consistency | Gyroscope, altimeter, and horizon indicator all align with testimony | Partial instrument alignment; some discrepancies | Instruments contradict testimony or were not consulted |
| Willingness to Disclose | Named, on-record, career risk acknowledged | Anonymous but detailed and verifiable | Anonymous with no verifiable details |
| Dimension | 5 – Anomaly Confirmed | 3 – Inconclusive | 1 – Consistent with Standard Model |
|---|---|---|---|
| Great Circle vs. Actual Route | Significant deviation from predicted great-circle path; no weather or ATC explanation | Minor deviation; possibly explainable by weather routing | Route matches great-circle prediction precisely |
| Emergency Landing Logic | Emergency diversion route only makes sense on alternative projection | Diversion route is ambiguous on both projections | Diversion route is optimal on standard globe model |
| Fuel Consumption Analysis | Fuel burn inconsistent with stated distance on globe; consistent with flat projection | Fuel data is within margin of error for both models | Fuel burn precisely matches globe-calculated distance |
| Time Zone Crossing Pattern | Time zone transitions do not match expected longitude progression | Minor discrepancies within measurement tolerance | Time zone transitions match longitude precisely |
| Dimension | 5 – Observable Anomaly | 3 – Ambiguous | 1 – Standard Explanation Sufficient |
|---|---|---|---|
| Horizon Line at Altitude | Horizon remains at eye level at 40,000+ ft; no visible curvature | Subtle curvature claimed but difficult to confirm with lens distortion | Clear curvature visible consistent with expected Earth radius |
| Long-Distance Visibility | Objects visible beyond calculated curvature drop-off distance | Objects at edge of visibility; atmospheric refraction plausible | Visibility consistent with curvature calculations |
| Gyroscope Behavior | Gyroscope shows no drift correction for expected rotation over multi-hour flight | Gyroscope data is ambiguous or unavailable | Gyroscope drift matches predicted Earth rotation rate |
| Water Surface Analysis | Large water bodies show no measurable convexity at distance | Measurements are within margin of error | Convexity measurements match predicted curvature |
| Dimension | 5 – High Integrity | 3 – Mixed Integrity | 1 – Compromised |
|---|---|---|---|
| Source Attribution | Primary source with named pilot, date, route, and aircraft type | Secondary source with partial attribution | No attribution; viral content or screenshot only |
| Editorial Bias Detection | Presents data without predetermined conclusion; acknowledges limitations | Mild editorial framing but data is intact | Heavy bias; data cherry-picked or misrepresented |
| Counter-Argument Engagement | Actively addresses and rebuts standard explanations with data | Mentions counter-arguments but dismisses without analysis | Ignores counter-arguments entirely |
| Reproducibility | Observation can be independently verified with described methodology | Partial reproducibility; specialized equipment needed | Cannot be reproduced or methodology is not described |
| Cross-Reference Density | Cites 3+ independent sources confirming same anomaly | 1–2 cross-references of varying quality | No cross-references; single isolated claim |
| Dimension | 5 – Annotation-Ready | 3 – Requires Editorial Injection | 1 – Unusable for Training |
|---|---|---|---|
| Structured Data Extraction | Testimony can be parsed into structured fields (altitude, heading, time, observation) | Partial structure; requires manual normalization | Unstructured narrative with no extractable fields |
| Label Consistency | Clear positive/negative labeling for anomaly detection training | Labels are ambiguous or context-dependent | No consistent labeling possible |
| Prompt Engineering Potential | Can generate 5+ distinct prompt variations for GenAI training | 2–4 prompt variations with editorial adaptation | Cannot be converted to effective prompts |
| Multi-Modal Integration | Combines text, flight data, instrument readings, and visual evidence | 2 modalities present but not fully integrated | Single modality only; no cross-modal potential |