What we measure
Casimo records public operator wording, changes between captured versions and collector reachability. We keep the exact source URL, a short excerpt, observation time and version identifier. A tracked brand is not an endorsement or a claim of first-person play testing.
Facts, dates and versions
A first snapshot is a baseline. Observation date, document publication date and effective date are separate fields. A hash identifies a capture, not the truth of its contents. We never invent earlier monitoring history.
Automation and publication
Collectors and deterministic rules structure public wording. Machine-extracted fields retain their excerpts and limitations. AI can assist with bounded tasks; it does not create user experiences or fill unsupported values. No page is claimed to have been individually reviewed by a human unless that actually occurred.
Conflicts and freshness
Conflicting source conditions remain visible. Failed or blocked fetches retain the last good observation and its date. State checks require sufficiently recent operator exclusions, and never turn “not listed” into a legal clearance.
Payouts and scores
Voluntary observations, public mentions, document-backed reports and independently confirmed data are different evidence classes. We do not infer a population complaint rate without a denominator or an arrival date from completed-only payouts. There is no unsupported composite trust score.
Independence
US affiliate monetization is disabled for this release. Source coverage, tool results and ordering do not depend on a commission. UK edition commercial arrangements are described within that edition.
Corrections
Use the correction form to identify the claim, source and proposed correction. Reports are held for verification. Material corrections preserve an audit record rather than silently rewriting history.