Our Methodology

What the record describes, what we cannot infer, how evidence is reviewed, and the limits of comparison.

What this record describes

We record what was reported, what instruments observed, and what remains unresolved. We distinguish a reported occurrence from a claim or denial. Independent corroboration concerns the reporting behind an event; it is not a judgement that every detail is settled.

What we cannot infer

A source we did not poll, a source we could not reach, a successful empty response, and an arriving report are different observations. Silence alone is never evidence that an event occurred. A missing language observation is not English, and an unresolved location is not an approximate pin.

Evidence and review

Related reports are grouped, attributed and assessed before publication. The held record remains distinct from the published record. Observed geography and historical association do not predict a particular attack or establish causation.

Limits of comparison

Comparisons need adequate observation history and comparable source and receiver coverage. Changes in publishers, ownership, language coverage, outages and taxonomy can change the record without a change in the world. Insufficient history is a result, not a zero.

What we cannot see

The vessel subscription records a connection, not a geographic observation footprint. Open water without a supported coverage source cannot be assessed as quiet. We cannot infer coverage from where ships arrived: that would define an unobserved quiet lane by the very traffic whose absence we are trying to measure.

Earlier versions

Before immutable revision capture begins, we do not hold an earlier version of this page. Current wording cannot reconstruct what a reader saw before that cutover.