Sample
Publisher country is provenance.
A publisher's location tells you where the sampled outlet is based. It is not treated as national public opinion or as the view of everyone in that country.
← Today's stories
Methodology & data sources
What this site measures — and what it does not.
America Through the World compares coverage within a defined sample of news outlets. Its rankings and comparisons are ATW-derived measurements of that sample, not claims about what an entire country, population, or global media system thinks.
Ranking
Attention is measured inside the sample.
ATW prioritizes events using outlet breadth, source-country breadth, bounded article volume, and recency. The resulting rank is a sample-attention measure, not an objective measure of a story's importance.
Comparison
The Attention Gap is an outlet-share difference.
For matched events, ATW compares the share of sampled foreign outlets covering the event with the share of sampled U.S. outlets covering it. It does not measure audience interest, ideology, sentiment, or public opinion.
Interpretation
ATW avoids national-opinion claims.
Reader-facing language describes patterns among sampled outlets. A headline or framing choice from one publisher is not presented as “what a country thinks.”
Data and discovery
ATW is evaluating the GDELT Project as a foreign-news metadata and discovery source. GDELT publishes global news-derived datasets that ATW can use for provenance and deterministic analysis. GDELT's data rights do not transfer copyright or other rights in the underlying publisher articles.
During the prototype period, ATW also uses selected direct publisher feeds for controlled collection and comparison. A feed being publicly accessible is not treated as blanket permission for commercial republication or automated reuse.
Publisher content
ATW is designed to rely primarily on provenance, event-level analysis, source counts, and other derived measurements rather than reproducing publisher content. Full article text and publisher images are not part of the default product model.
Publisher names and links identify the reporting source. Linked reporting remains with its respective publisher, and inclusion in ATW does not imply partnership, sponsorship, affiliation, or endorsement.
How the pipeline works
ATW normalizes and deduplicates source records, groups likely reports of the same event, ranks events using deterministic sample-attention rules, compares wording and source-country evidence, matches sampled foreign events to sampled U.S. coverage, and calculates the Attention Gap.
The core ranking and matching pipeline is deterministic code rather than an opaque AI ranking. Future AI-assisted features must follow the same source-rights and provenance rules.