We fit a mathematical model from earthquake science to every database. It measures how much of each dataset is genuine independent events vs social contagion cascades (one report triggering more reports). The results sort into three tiers.
These phenomena are primarily social media phenomena. When one person reports a sighting, it triggers a wave of copycats.
| Dataset | Domain | N Events | BR | Independent % | Notes |
|---|---|---|---|---|---|
| BFRO Bigfoot | cryptid | 3,659 | 0.988 | 0.7% | Only ~25 independent reports |
| Reddit Glitch | anomalous | 1,475 | 0.986 | 1.4% | Social media maximum |
| Crop Circles | anomalous | 2,891 | 0.971 | 2.9% | Media-driven reporting |
| NUFORC UFOs | anomalous | 79,638 | 0.959 | 4.1% | Largest UFO database |
| UFO Stalker | anomalous | 218,066 | 0.953 | 4.7% | MUFON-sourced |
| BackwoodsCreepy | paranormal | 3,200 | 0.953 | 4.7% | Reddit horror |
| Loch Ness | cryptid | 1,150 | 0.945 | 5.5% | Classic cryptid |
| Dataset | Domain | N Events | BR | Independent % | Notes |
|---|---|---|---|---|---|
| UCDP Conflicts | conflict | 349,733 | 0.770 | 23% | Retaliation cycles |
| Autoganzfeld | psi | 354 | 0.706 | 12% | Lab scheduling artifact, NOT psi contagion |
| Storm Events | weather | 639,465 | 0.670 | 33% | 1996 measurement expansion |
| Mass Hysteria | anomalous | 63 | 0.663 | 30% | Not robust (year-level jitter) |
| NDE Reports | anomalous | 388 | 0.629 | 23% | Media-driven REPORTING, not experience |
| Earthquakes | geophysics | 50,000 | 0.492 | 52% | VALIDATION: matches published aftershock models |
People report these experiences regardless of media waves. The timing is personal, not social.
| Dataset | Domain | N Events | BR | Independent % | Notes |
|---|---|---|---|---|---|
| Marian Apparitions | religious | 386 | 0.385 | 61% | Personal religious experience |
| Stigmata | religious | 127 | 0.281 | 72% | Extremely private phenomena |
| OBERF OBE | anomalous | 1,200 | 0.244 | 76% | Out-of-body experience dates |
| NDE Experience Dates | anomalous | 388 | 0.231 | 77% | When it HAPPENED, not when filed |
| Volcanoes | geophysics | 11,178 | 0.112 | ~90% | Physical triggering only |
| Fireballs | astrophysics | 1,052 | 0.007 | ~100% | NEGATIVE CONTROL: validated |
| Meteorites | astrophysics | 1,100 | 0.005 | ~100% | NEGATIVE CONTROL: validated |
The most important thing about this project is what didn't survive.
We tested whether independent reports are written differently than triggered reports. 19 linguistic features: sentence length, negation rate, emotion words, sensory vocabulary, question rate. First result: 16 out of 19 were significantly different. Independent reports were quieter, less dramatic, more negation-heavy.
Then we caught three errors in our own methodology:
After fixing all three: 16 features dropped to 6.
Then we asked: are those 6 features about the experience, or about the platform? A government form sounds different from a Reddit post regardless of what the person saw. After controlling for reporting platform and source:
0 out of 6 survived.
How people write about anomalous experiences is determined by where they file the report, not by whether the experience was genuine.
So we stopped asking how people write and started asking what they describe. We built structured content inventories: 300+ specific concepts across 5 domains. Not word counts. Specific things: shapes, physical sensations, sounds, smells, entity descriptions.
Then we crossed everything with everything. 83 databases. 1.35 million reports. Every concept checked against every dataset.
4,356 recurring content motifs. 3,335 crossing domain boundaries. A vibration-vestibular mega-cluster appearing independently in UFO, ghost, NDE, and sleep paralysis reports at 9,120x the expected rate. A ghost “haunting signature” of goosebumps + temperature change + felt presence at 1,202x lift.
These still need to survive the final control stack (checked against 500K storm reports, 44K civilian distress reports, and the triggered copycats we already filtered out). Results pending.
Solar wind signal weakens after 100+ population normalization
Mass psychogenic illness frequency is not a robust independent stress proxy
Armed conflict × geomagnetic activity: validation gate fails under robustness checks
OMNI × UFO multicollinearity audit: most apparent associations collapse after PCA/partial control
Global Hum late-period independents show no single industrial or geographic signature
FAERS weekend death premium persists within top indication strata (2024Q2–Q4)
Solar wind × supercentenarian deaths is trend-sensitive and likely confounded
Washington per-capita UFO hotspot is not explained by NUFORC HQ proximity alone
Send me your event data. I'll run the full pipeline and tell you how much is real.
bo@projectaletheia.org