Methodology

How this desk reaches a judgment

ICD 203 applied

Every assessment separates what is known from what is judged, states the reasoning, flags what would change the judgment, and distinguishes the underlying reporting from the analysis built on it. Judgments are written to be falsifiable: if the facts move, the assessment moves with them and says so.

Confidence levels

High confidence
Multiple independent, reliable sources agree, and the evidence is direct.
Moderate confidence
The sourcing is credible but thin, single stream, or partly circumstantial.
Low confidence
The reporting is fragmentary, unconfirmed, or the situation is still moving.

The citation rule

Nothing is published without at least one real, reachable source link. Every event in the archive carries the publisher's own URL and the publisher's own date, unchanged. Numbered citations in an assessment link straight to the article they rest on, and they appear the same way in the email, the RSS feed, the export and the weekly PDF, including on white-labelled editions run by partners.

Source policy

Unclassified open sources only: government advisories, court filings, agency releases, vendor research published by the vendor, and named reporting. Feeds are pulled from each publisher's own channel, never a mirror or an aggregator. No classified or otherwise non-public material is ingested, stored or displayed. No event is ever invented, seeded or rewritten by a model.

The daily integrity check

Twice a day the platform audits itself: collection freshness, the health of each source, whether the briefing is fully cited, whether the figures on the home page agree with the archive, and whether anything is rendering empty. The result is public.

Read the current integrity check