Xerith Blog
Editorial standards
How we choose questions, use evidence and correct mistakes.
Start with a reader's question
We prioritize questions with a clear use: checking eligibility, understanding a change, comparing like-for-like costs or reading a claim correctly. A trending topic is a research lead, not evidence that a story is true or that readers need another summary. We do not fill a fixed number of category slots with weak material.
Use evidence at the point it matters
Original announcements, official data, research and product documentation take priority. Every article links its sources. Company claims remain attributed to the company. We distinguish estimates from observations, proposals from decisions, allegations from findings and research prototypes from available products.
Interpretations, calculations and hypothetical examples are identified. We preserve a statistic's population, period, units and important limitations. A practical checklist is our suggested way to compare evidence, not a promise about a product, outcome or legal entitlement.
Add something useful
A publishable explanation needs more than a rewritten announcement. It should resolve the main question and add relevant analysis, a comparison, a calculation or a verification process. We do not use an arbitrary word count or repeat background just to make an article longer.
Be clear about AI and expertise
AI tools assist with research, writing, organization and checks. They are not primary sources. Our publication does not claim that an automated check is independent human review. We do not invent authors, qualifications, tests, interviews or first-hand experience.
Illustrations generated with AI are labeled. They are conceptual artwork rather than documentary photographs. A drawing of a package, person or laboratory does not establish the appearance of an actual product, participant or experiment.
Use honest dates and make corrections visible
Publication and substantive update dates are separate from announcement dates and the periods covered by reports. We do not change a date to make an old story seem new. A substantial revision includes a note explaining what changed.
To flag a problem, use Contact the publisher. Include the article address, the claim in question and any supporting source. A correction request does not automatically prove a claim wrong; we examine the evidence.
Review comments before publication
Comments wait for publisher review. We may decline or remove spam, threats, harassment, discriminatory abuse, private information, unsupported harmful claims or material unrelated to the discussion. Disagreement is welcome when it is relevant and expressed respectfully. Link-free text is required to reduce spam.
Approval is not an endorsement of a reader's claim. Use our private form to report a comment. We do not promise that every submission will be published or that every message will receive an individual response.
Commercial interests
There are currently no active ads, affiliate links or paid placements. If this changes, we will identify commercial relationships and update the privacy information before advertising or tracking starts. Paid placement will not be disguised as independent testing.
