How we read past the headline

A news page has a fold. Above it sits the headline, the part that goes viral. Below it sits the body: the fine print, the study, the caveats. Below the Headline is about the part below.

Who makes this

I'm not a scientist. I'm someone who got tired of scrolling past scary health and tech claims without knowing what the research underneath them said. So I started reading the studies, and the @BelowHeadline channel and this site are the result: one claim at a time, with every source linked.

The videos are faceless, and the narration uses an AI clone of my own voice. The articles on this site are written companions to the videos: same facts, same sources, more room to explain.

The method: five steps, every time

1. The claim. What went viral, paraphrased and labelled as such. I never name or show the people who posted it; the claim is the subject, not the person.
2. The grain of truth. The strongest version of the worry, stated fairly. Most scary posts start from a real study, and that study deserves a proper hearing.
3. The studies. What was actually measured, in what (cells, animals or people), how many, by whom, and what the authors themselves say it does and doesn't show.
4. How sure. A verdict on the four-level scale below, given after the evidence, not before it.
5. My take. What I personally do with it, clearly marked as opinion, not evidence.

How sources are chosen

• Primary first. The study, the standard or the regulator's own page, read directly wherever possible. Systematic reviews and measurement surveys by radiation-safety agencies carry more weight than single small studies.
• Secondary sources are labelled. When a figure could only be read in news coverage or a fact-check, the source list says so, and the article leans on it less.
• Every fact gets checked twice. Each script is fact-checked against its sources before recording, and again independently. Numbers that can't be confirmed at source are left out, even when they would make a better story.
• Funding and design are stated. If a company funds a study, or a result comes from a dish rather than a person, the article says so next to the result.

The Depth Gauge: how sure can we be?

Verdicts use four words. Each one describes how deep the evidence goes, not how dangerous something is, and each always comes with its one-line meaning. The gauge never uses red and green, and never relies on colour alone: the word, the depth and the pointer carry the meaning too.

• DISPUTED: Good studies disagreeGood studies disagree
• EARLY: Signals, not answers yetSignals, not answers yet
• STRONG: Consistent, well-designed evidenceConsistent, well-designed evidence
• SETTLED: Decades of converging evidenceDecades of converging evidence

A verdict always has an answer attached: "Strong evidence: no" means consistent, well-designed studies point to no effect; "Early" means there are signals, not answers yet.

Corrections policy

If something here is wrong, it gets fixed in the open. Corrections are made as soon as an error is confirmed, the article's "Updated" date changes, and every correction is listed with what changed and why on the corrections page. Small wording fixes that don't change meaning are not logged; anything that changes a number, a claim or a verdict is.

Not medical advice

These articles explain what research says about common claims. They are general education, not advice about your own health. If a claim touches a decision you're facing, take it to a doctor.