Editorial policy

No institution, agency, ministry, political party, campaign group, or commercial sponsor has review rights over our content, approval rights over publication, or the ability to require a change. If that ever changes for a specific piece of work, the arrangement will be disclosed on the page itself, naming the party and the nature of the arrangement.

We publish no sponsored content, advertorial, or paid placement, and we carry no advertising.

1. Commissioned. The reader question, the jurisdictional scope, what the page will not cover, and whether it is safety-sensitive are written down before research begins.
2. Researched. Sources are gathered from primary material first, and each is recorded with a note stating what it supports and what it does not.
3. Drafted. All required sections are written: definition, why it exists, how it works, common misconceptions, jurisdictional variation, and rights and accountability.
4. Editorially reviewed. Structure, neutrality, accessibility of language, and internal linking are checked.
5. Fact-checked.A separate pass, made against the source records rather than the author’s memory of them. Every checkable claim is matched to a source and confirmed at the level of specificity stated.
6. Safety-reviewed.Mandatory for law enforcement, investigations, forensics, and public safety. A different question from fact-checking: not “is this true?” but “what could this be used for?”
7. Published, then re-reviewed. Structural content at least every 24 months; content depending on current law at least every 12.

A claim must be scoped to the jurisdictions its sources actually cover. In practice this is the single most frequent edit we make.

Where we genuinely cannot scope a claim, we do not make it.

• Fact — sourced and checkable. The default register, rendered as plain body text.
• Analysis — our own synthesis. Labelled and rendered in a marked callout. Analysis may connect sourced facts; it may not introduce new ones.
• Attributed opinion— someone else’s position, always named. Never “it is widely accepted that…”

Evaluative adjectives applied to institutions — ineffective, robust, politicised — are analysis or attributed opinion, never fact.

1. Could this page have been written by that institution’s press office?
2. Could it have been written by that institution’s most committed critic?
3. Are contested positions attributed to identifiable holders?
4. Is any country used as the unmarked default, the implicit norm, or the recurring negative example?
5. Are evaluative adjectives sourced or marked as analysis?
6. Would a reader be able to infer the author’s political position?
7. Does the page state a power without its limits, or a limit without the power it constrains?
8. If the page covers a reform debate, are the strongest versions of the main positions presented — not the easiest to dismiss?

We take seriously that public-service work is difficult, carries real responsibility and real risk, and is performed under legal and resource constraints outsiders rarely see. We take equally seriously that these institutions hold coercive powers, that those powers have been misused, and that oversight and reform exist because of documented failures.

Both are content. Neither is a disclaimer attached to the other. We do not use “a few bad apples” framing and we do not use “systemically corrupt” framing — both substitute a conclusion for a description.

• Research methodology
• Source hierarchy and index
• Corrections policy
• Disclaimer