About

Every investigation on this site follows the same rules. Every claim is assigned a tier: T1 for facts stated in primary source documents, T2 for statistical associations that survive null models, T3 for plausible inferences with mundane alternatives tested, T4 for speculation. Every finding above T2 carries a kill condition — a specific, falsifiable statement of what evidence would destroy it. If a finding dies, we publish the autopsy.

We built a registry of over six million entities drawn from public datasets: offshore corporate leaks, suspicious activity reports filed with the US Treasury, diplomatic cables, congressional stock trades, sanctions lists, court records, FAA registrations, and dozens more. An entity resolution engine links records across these sources. A convergence engine surfaces where the connections cluster. Network null models, temporal permutation tests, and base-rate analyses separate signal from coincidence.

Most of what the system finds, we kill. The banking overlaps that turned out to be structural rather than suspicious. The temporal patterns that collapsed under placebo dates. The entity matches that were common-name collisions. We publish what survives, and we show what didn’t. That’s the difference between investigation and conspiracy: a willingness to be wrong in public.