Standards

Methodology

How we collect, check and rate open-source information, and how to tell us when we get something wrong.

About 01 — Who we are

Research from a detection company

DroSonic OSINT is the research arm of DroSonic, a company building passive acoustic drone detection. We publish what we learn about the drone threat and the counter-UAS response because good detection starts with understanding the threat. We have a commercial interest in counter-UAS, and we say so; the analysis aims to be useful whether or not you ever buy from us.

About 02 — Sources

Open sources only

We use only publicly available information: government and military statements, court and regulatory records, academic and technical papers, reputable news reporting, manufacturer documentation, and verifiable imagery or video. We do not use classified, controlled or leaked material, and we do not publish detail that would meaningfully help someone attack people or infrastructure.

  • Primary sources first. Where a news report quotes an official statement, we link the statement.
  • Every claim is cited inline, next to the sentence it supports, and listed again at the end of each report.
  • Imagery is checked for location and date where possible before we rely on it.
About 03 — Confidence

How sure we are, stated every time

Every report, key judgment and logged incident carries a confidence level. It describes the strength of the evidence, not how likely or how serious an event is.

  • High: confirmed by authorities, corroborated imagery, or several high-quality independent sources with no credible contradiction.
  • Moderate: several independent sources agree, or an official acknowledgement exists, but key details remain unconfirmed.
  • Low: a single source, an unverified claim, or reporting with significant gaps. Treat it as a lead, not a fact.

Where a number is contested, we show the competing figures and who produced them, rather than picking one.

About 04 — Language

Estimative words mean the same thing everywhere

Term Approximate likelihood
Almost certainly 95% or more
Very likely 80 to 95%
Likely 55 to 80%
Roughly even chance 45 to 55%
Unlikely 20 to 45%
Very unlikely 5 to 20%
Close — corrections

When we get it wrong

If we get something wrong, we fix it and note the change with an updated date. Send corrections, with a source, to sudo@drosonic.com.