Editorial policy
Every report states its test conditions: the dataset, the sample size, who scored it and how. Findings are about approaches to a task, never endorsements of a product. Failed and inconclusive runs are published alongside clean ones. Corrections are dated and left in place, and a report whose conditions change is re-tested and re-filed rather than quietly edited.
Corrections
Write to the desk and we will correct the entry itself. Corrected entries carry an updated date.