Methodology
This page explains how teebeenews works — how we rate sources, how our summaries are written, and how stories get ranked. We'd rather be clear than impressive.
How we rate source lean
Every outlet we read is rated for where its coverage tends to sit, on a five-point scale:
Left · Lean Left · Centre · Lean Right · Right
- Ratings are set by our editors, not by an algorithm.
- Each rating cites a public source for that rating, and we show the citation with the outlet and in the coverage bar. We use publicly available ratings and our own editorial judgement. We do not copy or scrape licensed datasets (such as AllSides, MBFC or Ad Fontes) — we cite public ratings instead.
- A lean rating is a guide, not a verdict. Outlets are complex and diverse; a rating describes the overall tendency of an outlet, not every article or every reporter.
- If you think a rating is off, tell us via corrections.
Every outlet's rating is on our media bias chart.
No outlet is perfectly neutral, and that includes us. Unbiased news, explained covers why we show every side instead of picking one.
How our AI summaries are written
We don't write news with AI. Every story we cover was reported and published by human journalists at real outlets. Our AI reads those reports and organizes them:
- It groups articles about the same event into one Story.
- It writes one short, neutral headline and a brief — in our own words, never copied from an outlet's lede or opening line.
- Every AI-written summary is labelled "Summary by teebeenews AI from N sources" and links to the source articles underneath, so you can read the original reporting yourself.
- AI also assigns topics and judges how significant a story seems, which feeds into ranking.
- Editors can override anything our AI writes. A human can edit or replace any headline, summary or tag.
We keep AI summaries short on purpose: they give you the gist in seconds, not a substitute for reading the reporting behind them.
How "hot" ranking works
The home page and topic pages are ranked by a heat score, recalculated every 30 minutes. Roughly, a story's heat comes from:
- Breadth of coverage — how many distinct outlets are covering it.
- Velocity — how fast new coverage has appeared in the last 6 hours.
- Recency — newer coverage counts more.
- Significance — how important the LLM judges the story to be.
- A small engagement term from reads, follows and poll votes.
We also cap the number of lead stories per topic, so one story doesn't crowd out everything else.
Heat is about what's changing right now, not what's most liked. That's why a slow-burn story can outrank a popular one, and why a fast-developing story climbs quickly.