How AnomalyBrief works
We apply a structured seven-step research process to every case. The goal is to separate verifiable facts from claims, build a clear source chain, and score the completeness of the available evidence — without manufacturing conclusions.
The seven-step method
Gather from primary sources
We identify and collect only primary or official secondary source documents: government records, published agency reports, FOIA releases, peer-reviewed literature, or authenticated sensor data. No aggregated summaries, no anonymous tips, no content from anonymous forums.
Normalize metadata
Each case is assigned a standardized record: date, location, source agency, media types, evidence types, and status. We use consistent taxonomy across all cases so that filtering and comparison are accurate. Ambiguous or incomplete metadata is flagged as such rather than guessed.
Extract entities
From the source document, we extract: known facts stated by the reporting agency, possible explanations offered or ruled out in the record, and open questions not addressed by the primary source. These three categories are kept strictly separated in the case record.
Cross-check five data layers
We check every case against five independent context layers: weather data (NOAA), sky context (astronomical event catalogs), satellite passes (Celestrak TLE sets), aircraft traffic (ADS-B/FAA records), and astronomy cross-references (JPL Horizons, fireball databases). Each layer is marked with its current status — clear, needs review, active, pending, or planned for a future phase.
Preserve original links
Every case record includes direct links to the original source document. We do not rehost or excerpt documents in ways that could alter their meaning. Where a document is no longer available at its original URL, we note that and link to any archived version. Source links are verified at the time of publication.
Separate facts, interpretations, and open questions
The most important discipline in anomaly research is the separation of what is stated in a source document (facts), what could explain the observation (interpretations), and what remains unanswered by the record (open questions). These are presented as distinct sections in every case brief. We do not allow interpretations to migrate into the facts section.
Score the evidence
We compute an evidence score from 0 to 100 using five weighted factors. The score reflects how well-documented and cross-checked a case is. It does not measure the probability of any explanation. A score of 80 means the case is well-documented — it does not mean anything unusual happened.
Evidence score
The evidence score is a number from 0 to 100. It answers one question: how complete is the research record for this case? Higher is more thoroughly documented. Lower means there is less to work with.
Five scoring factors
Source completeness
Does the case have at least one primary source? Does it include the original document or official record, or only a reference to one?
Corroboration
Are there multiple independent witnesses, sensors, or records that agree on the key observations? Single-witness, single-sensor cases score lower.
Media and sensor presence
Does the case include physical sensor data (radar, infrared, sonar, photometric), or is it solely based on witness statements?
Metadata quality
How precisely are the date, time, and location documented? Cases with exact timestamps and coordinates score higher than those with approximate or missing metadata.
Cross-check coverage
How many of the five context layers have been checked? Cases with more completed cross-checks score higher than those still pending.
What we do not do
- We do not publish definitive conclusions about the origin or nature of any anomaly.
- We do not fabricate, generate, or infer case data from incomplete records.
- We do not publish AI-generated summaries as factual claims. AI may assist with drafting, but all published content is human-reviewed against the primary source.
- We do not scrape or mirror proprietary, paywalled, or classified data.
- We do not include cases that have not been reported by or to an official government body, scientific institution, or comparable authority.
- We do not create synthetic media, fabricated documents, or manipulated imagery.
- We do not allow interpretations or speculative explanations to appear in the "known facts" section of a case record.
- We do not assign extraterrestrial, supernatural, or classified-program explanations as likely or confirmed without a primary source making that designation.
Environmental context checks
AnomalyBrief checks the environmental conditions around a reported event to help identify possible false positives and to give researchers a complete picture of what conditions existed at the time and place of a report. This check runs against publicly available weather archives, space-weather data, and official agency feeds.
What we check
- Weather and visibility — cloud cover, precipitation, wind, pressure, and atmospheric visibility pulled from Open-Meteo historical archive and NOAA/NWS.
- Sky conditions — WMO weather codes translated to plain-language sky descriptions, including fog, overcast, clear, and storm conditions.
- Geomagnetic activity — Kp index from NOAA Space Weather Prediction Center indicating geomagnetic storm levels, aurora likelihood, and radio/GPS interference potential.
- Solar flares and CME events — solar flare class, timing, and active region data from NASA DONKI.
- Fireball events — bright meteor and bolide records from JPL CNEOS within ±2 days of the reported date.
- Active natural events — wildfires, severe storms, volcanic activity, and other natural hazards from NASA EONET within ~5° of the location.
- Sun & moon position — local altitude/azimuth and lunar phase computed from standard astronomical formulas. Identifies daytime vs twilight vs full night and flags bright-moon misidentification scenarios.
- Active meteor showers — IMO working shower list. Reports any major annual shower with an active radiant on the date plus days from peak and ZHR.
- Aircraft within ~50 km — live ADS-B transponder snapshot from the OpenSky Network. Anonymous tier serves live data only; historical lookups require authenticated access and are flagged as such. OpenSky intermittently rate-limits cloud-provider traffic, so this layer may report as unavailable from time to time.
- Rocket launches within ±2 days — global launch manifest from Launch Library 2 (The Space Devs). Upper-stage burns, fuel dumps, and re-entries are common prosaic explanations for high-altitude reports.
- International Space Station position — sub-point and observer elevation angle computed for the exact timestamp via wheretheiss.at. Identifies whether the ISS was above the local horizon and likely visible.
- Nearest aviation weather report — METAR observations from NOAA AviationWeather for the closest reporting stations (recent reports only).
- Confidence score — ratio of context providers that returned usable data, excluding providers that were not applicable for the request (historical-unavailable, not-US, not-configured).
All sources are free, publicly documented, and queried on demand. AnomalyBrief deliberately excludes paranormal or entertainment data feeds — every layer above exists to test prosaic explanations first.
Full disclaimer
AnomalyBrief is an independent research tool that aggregates and cross-references publicly available government records, scientific datasets, and official source material. We do not manufacture, fabricate, or editorialize case data. All information is sourced from publicly accessible documents. Evidence scores reflect research completeness only — they do not represent a probability of extraterrestrial origin, supernatural phenomena, or any specific explanation for any reported event.
Cases labeled "demo" are included for platform demonstration purposes. All data in demo cases is drawn from public sources or clearly labeled as illustrative. AnomalyBrief is not affiliated with, endorsed by, or operating under the authority of any government agency, including the Department of Defense, AARO, NASA, NOAA, or any intelligence community element. Use of this tool for journalistic, academic, or personal research purposes is permitted. Redistribution or republication of case briefs must include attribution and a link to the original source record.