← All findings

Global Hum

4,448 reports. Era-split reveals two fundamentally different reporting regimes.

Decision Chain: Global Hum

Step 1Raw Data Ingestion

Ingest Global Hum reports. 4,448 reports of unexplained low-frequency humming from the World Hum Map.

4,448 reports ingested.

Step 2Hawkes Process Decomposition

Fit self-exciting point process to the full dataset.

Full-dataset branching ratio 0.873. Appears highly self-exciting.

Step 3Era-Split Analysis

Split dataset at structural break. Early period dominated by media coverage; late period shows different reporting dynamics.

Early period: media-driven contagion. Late period: branching ratio drops to 0.274 — mostly independent reports.

Step 4Media Influence Assessment

Early-period reports cluster around media coverage events. Late-period reports arrive independently.

Early reporting wave explained by media exposure. Late reporters are genuinely independent.

Two eras, two stories

The Global Hum dataset splits cleanly into two periods. The early era (branching ratio 0.873) is dominated by media-driven reporting — people hear about the Hum and then report hearing it. The late era (branching ratio 0.274) shows a fundamentally different pattern: most reports are independent, arriving without connection to prior reports or media waves. Whatever the late-period reporters are hearing, they aren't getting the idea from each other.