Methodology
How we assess public evidence
Each item should tell you what changed, what supports that account and which conclusions remain uncertain.
What supports a claim
Official documents establish what an institution has announced or enacted. They do not necessarily establish whether a policy has worked, a capability is operational or an attributed event occurred as described. Reporting and specialist analysis provide additional context; contested claims retain their attribution.
| Source type | Use |
|---|---|
| Official statement | Government, military, international organization, court, regulator, or central bank. |
| Established media | Major news outlets with editorial standards, corrections processes, and named reporting. |
| Specialist analysis | Think tanks, defense analysts, regional experts, trade publications, and technical research. |
| OSINT observation | Public imagery, geolocation, flight tracking, maritime tracking, and other observable public data. |
| Adversary or state media | Useful for official narratives, propaganda signals, and stated positions; treated cautiously. |
| Unconfirmed claim | An early claim that may matter but is not treated as fact. |
| Social discussion signal | Discussion from public platforms. Useful for discovery, never confirmation by itself. |
| Strategic document | Defense strategies, white papers, budgets, regulations, legislation, court filings, and official plans. |
Reporting and interpretation
“What changed” describes the reported development. “Why it matters” is analysis, with uncertainty preserved. “Watch next” identifies an observable event that could strengthen or challenge that assessment. A topic's general importance is not evidence for a particular implication.
Confidence
Confidence describes the evidence available for an assessment. It is not a numerical probability or a guarantee. Source authority, corroboration and unresolved contradictions limit how strongly a conclusion can be stated.
| Label | Meaning |
|---|---|
| High | Multiple reliable sources or official confirmation; limited contradiction. |
| Medium | Credible reporting or official claim with partial corroboration. |
| Low | Single-source, unclear, early, or contested reporting. |
| Unconfirmed | Not treated as fact; included only as a signal requiring verification. |
Automation and editorial responsibility
Models assist with selected inputs after filtering. They help draft and organize material; they do not grant publication authority. The Daily Brief uses automated screening and content-quality checks. An editor does not approve each automated edition individually.
Reviewed Techno-Economic Watch records and the retained Crisis chronologies pass human review. The Watch also has a separately labeled automated preview of official Federal Register actions that pass deterministic source and relevance checks. Those previews are not individually reviewed and are kept separate from reviewed analysis. When present, separately labeled official-document timeline records reproduce a narrow set of screened agency documents. Their publication and effective dates describe the document; they do not verify a military occurrence or imply human approval. Private candidates are not public evidence. Autonomous products require separate verification and operating acceptance before activation.
Uncertain attribution, casualty claims, nuclear issues, escalation and incomplete or weak evidence require additional scrutiny. A failed or incomplete evaluation is different from a completed evaluation that finds no qualifying development.
Dates, archives and corrections
Daily Brief editions enter the public archive after a 24-hour delay. Their original publication dates remain visible. Older Watch records and timeline entries retain their dates; a page visit does not make them current.
Missing analytical fields in older editions are left blank rather than filled with generic commentary. Published source claims are not rewritten to make an old edition appear fresh.
An old edition is a record of the analysis at that time. It is not a current assessment. A notice about a failed publication check describes a gap in publication; it does not mean the world was quiet or that every source was checked successfully. Retained Watch and Crisis records preserve their dates even when update mechanisms are inactive.
Report an error or missing context with the page link, the disputed sentence and the evidence supporting a correction. A substantive correction should identify what changed and preserve the earlier record. Source publication dates, reported event dates and our own publication dates serve different purposes; an unknown time stays unknown.