NPDev Trends reads the Google Trends RSS feed for each configured country — currently 15: United States, United Kingdom, Germany, France, Italy, Spain, Netherlands, Poland, Indonesia, Thailand, Vietnam, Philippines, Malaysia, Singapore, India. Each country's feed carries a short list of searches, each with the publisher's own approximate interest label, and up to three related headline titles with their publishers and links. That is the whole input.
Every explanation on this site is written from that stored metadata and nothing else. The feed supplies headline titles, not article text, so every page labels its coverage headline-only and links every source it cites. This site never fetches an article page, never rehosts anyone's text, and never invents a source. A search that trends in two countries is two separate records: one page per country, each attributed to the coverage that country's feed carried.
An automated check refuses an explanation that cites a source that was not supplied, states a number or date the supplied material did not contain, claims why something happened, gives medical, legal, financial or voting advice, asserts that a named person is guilty of something, or names an AI provider or model. A refused explanation is recorded and that search's coverage set is left alone rather than retried forever or padded with invented text.
Explanations are published at most once an hour, and only when a search has coverage that has not already been written about — an hour with nothing new publishes nothing, so the archive stays a set of grounded, attributed pages rather than a wall of thin ones. One hourly opportunity is shared across every country rather than added per country: the country that has waited longest for a publication is chosen first, so the cadence does not multiply with the size of the country list. Every read of a feed is stored as a dated observation, and every headline keeps its original wording, so a search stays look-up-able here — by term, by country and by UTC day — after it leaves the feed. Searching this site reads its own database only; no search, and no page view, ever starts a generation.
Limits. Explanations on this site are written by an AI model from the publisher's own headline metadata and are labelled as such. They are summaries of the linked coverage, not reporting: verify the sources before relying on anything, and treat every claim as the source's, not this site's. A summary of a headline can be wrong about details the article would have corrected. Trending lists are noisy and can reflect curiosity, not importance. A country whose feed is temporarily unavailable is reported as unavailable for that country and does not affect the others. Where the supplied headlines are thin, the correct answer is a short page that says so, and that is what this site publishes.