The Earthquake-Lights Signal That Isn't
This feature has been withdrawn from the public site. It is retained in full in the TerraPulse backend.
Originally published 30 May 2026 · Withdrawn 28 August 2026
Why it was withdrawn
The analysis cross-matched 9,379 US earthquakes against 71,008 NUFORC sighting reports. NUFORC's republication terms have never been read — the site refuses automated requests — and reports written by members of the public are not automatically ours to republish.
The source in question is the National UFO Reporting Center (NUFORC). TerraPulse is a commercial platform, and data licensed for non-commercial use only cannot sit underneath work we publish. Rather than leave it in place and hope, we removed it.
This is not a retraction
The result holds. Observed coincidences (3) sat below the 10,000-permutation null mean (9.4), and the null survived every sensitivity check. Nothing about the statistics has changed; we simply cannot publish work built on data we have no clear right to redistribute.
What would bring it back
- Read NUFORC's actual republication terms and establish whether commercial reuse is permitted.
- If it is not, find a sighting-report corpus that is clearly licensed — the phenomenon under study is public reports of lights, so no other dataset substitutes.
- Re-run the permutation test against the replacement and confirm the null still holds.
TerraPulse ingests measured data only, and publishes only what it can publish cleanly. Withdrawing work we had already shipped is the cost of that rule, and we would rather pay it visibly than quietly.