Methodology

1 · Detection

Every hour headlinez.news collects the current headlines from public Google News category feeds (World, Business, Technology, Science, Health, Sports, Entertainment). Headlines are normalized and clustered: two headlines belong to the same story when their significant-word overlap (Jaccard similarity) reaches 50%, or their character-level similarity reaches 62%. A cluster becomes a trend — and earns a permanent page — only when at least 4 independent news sources are covering it. This threshold is our quality gate: headlinez.news does not create pages for single-source stories.

2 · Velocity

Velocity measures how fast a story is spreading: velocity = (articles in last 6h + 0.25 × total articles) × √(distinct sources). Source diversity is weighted because ten newsrooms covering a story independently means more than one newsroom publishing ten articles. Velocity is snapshotted hourly; the charts on every page are drawn from these snapshots.

3 · Lifecycle

Each trend carries a status, recomputed hourly: rising (velocity climbing), peaking (at or near its maximum), cooling (below half of peak), and archived (no new coverage for 48 hours). Archived pages are never deleted — they become the historical record.

4 · Predictions — and self-grading

Once per day headlinez.news predicts, for every live trend, whether it will still be receiving coverage tomorrow (holds) or not (fades). The current heuristic: a trend holds if its velocity is within 80% of peak and at least 8 sources are on the story. The following day each prediction is graded against what actually happened, the result is stamped on the trend's page (✓ or ✗, permanently), and the global accuracy figure (currently 74% over 25,779 graded calls) is updated. headlinez.news cannot hide a bad call.

5 · Briefs and the no-invention contract

Before a brief can be written, headlinez.news attempts to retrieve readable pages from a diverse set of publishers in the cluster. SearXNG may help discover an additional relevant page, but a search result is never treated as evidence by itself: the destination must be fetched, readable and closely related to the story. At least two accessible source documents are required before drafting begins.

The drafting layer may use only that temporary research packet and the collected coverage metadata. Inventing numbers, quotations, people, causes or consensus is forbidden. A separate verification pass checks the draft against the evidence and removes unsupported claims; a final quality gate checks evidence coverage, clarity and minimum depth. If research is inaccessible, the verifier is unavailable or the draft fails a gate, the candidate remains unpublished and outside discovery surfaces for a later retry. The site does not publish a filler fallback merely to satisfy a schedule.

6 · Automation and editorial responsibility

headlinez.news is an automated newsroom-intelligence system, not a staff of human reporters. Each published trend page identifies its automated synthesis method, verification state, update time and the original coverage links used to understand the story. The system's health monitor checks its database, storage, cache, crawl endpoints, harvest heartbeat, research provider, publishing queue and media pipeline. Recovery actions are bounded to safe retries and cache repair; they cannot relax the evidence rules or force a failed candidate live.

7 · Media

Illustrative images are accepted only with a recorded source, credit and licence. Related YouTube or Instagram embeds are optional and appear only when the selected item's entities match the trend. Missing media is not replaced with a generic embed. Original platforms continue to host embedded material and may change or remove it.

8 · Data reuse and corrections

All headlinez.news metrics may be cited freely with attribution to "headlinez.news" and a link. Machine-readable data: https://headlinez.news/api/trends.json · AI-assistant orientation: https://headlinez.news/llms.txt. The editorial policy explains sensitive-topic rules, and the corrections page explains how to report factual, sourcing, attribution or media errors.