How we check a claim.
The more consequential a claim, the stronger the evidence it needs.
Start at the source
For specifications, use the exact model’s manufacturer documentation. For a feature, check the supported version, platform, region and plan. For research, distinguish a laboratory result from a commercial product. For an announcement, separate what was announced from what is actually available.
Dates and context
We verify dates and preserve the distinction between publication date and event date. We do not treat a search snippet, AI answer or social post as conclusive evidence. Important uncertain claims need corroboration or an explicit limitation.
Evidence of popularity
The Popular on SignalFold collection at /trending uses recent recorded article page loads on this site. It is not a claim about Google Trends or global popularity. Repeated visits can count; simple bot filtering is imperfect. We do not invent traffic statistics.
Public affairs, social media and festivals
For public affairs, trace records and statements to their source, corroborate consequential claims and distinguish election results from projections or allegations. For online trends, inspect the original post and surrounding context, and state the platform and observation date. For cultural coverage, identify regional differences and verify dates and timings for the stated year and location. AI output and a screenshot without provenance do not establish a claim.
Practical checks
Tutorials should describe conditions and expected results. Research-based guides should identify their sources and avoid unsupported personal-test claims. Product prices, if added, need a verification date and a reminder to confirm the merchant’s current price.
Report an error
Send the article URL, the disputed statement and supporting evidence through Contact. We assess the report and update the article when warranted.