How we check
The pictures can exaggerate.
The conclusions can't.
Animation needs metaphors, simplification and a few jokes. But every "what" and every "how much" has to hold up. These are the rules we follow, from reading the sources to putting a line in a video.
- 01
Read first, write second
Before a series starts we read the relevant guidelines, meta-analyses, key trials and bestselling books, and write a note on each one: what it says, on which page, who wrote it, any conflicts of interest, and which of its claims don't hold up. So far that is 200 sources.
- 02
One card per fact
Every fact a script uses gets a claim card: the claim in one sentence, the type of evidence, our confidence, who it applies to, the actual numbers, the caveats, and every source that supports or contradicts it. When sources disagree, both sides go on the card.
- 03
Popular claims get their own list
Many widely repeated claims, including ones from bestsellers, don't survive a close look. We list them as myths, with what the evidence actually says and a more accurate way to put it, and we make sure our videos don't repeat them.
- 04
Human review
We use AI tools to help find, read and organise the literature, and to help make the animation. That is exactly why every claim card has to be checked against the original. Cards marked ✓ Reviewed by a human have been checked line by line against the source; cards marked Not yet reviewed haven't been yet. We'd rather show you that status than hide it.
- 05
When we're wrong, we fix it
If a video gets something wrong, the corrections page says which episode, which line, what was wrong and what is right, and the claim card is updated.
Levels of evidence
Not every "studies show" is equal
The coloured dot before each source shows how much this kind of evidence can carry on its own.
Clinical guidelines
Recommendations from professional bodies after weighing all the evidence. The closest thing to medicine's current consensus.
Meta-analyses / systematic reviews
Many studies pooled together. They show whether an effect holds up consistently.
Randomized controlled trials
People randomly split into a treatment group and a control group. They can answer "did this cause that?"
Observational studies
Large groups followed over time to look for associations. Big samples, but correlation isn't causation.
Animal and mechanism studies
They explain why something might be so, but a mouse isn't a person.
Popular books, podcasts, web pages
Good for leads and analogies. We trace their numbers back to the original papers.
Confidence
The three bars on every card
Several high-quality studies agree, or a guideline clearly recommends it.
The evidence points one way, but the samples, populations or study designs have limits.
Little evidence, animal studies only, or studies that conflict. The video says so.
There are currently 20 claim cards on this site, 0 of them reviewed by a human. Educational content, not medical advice.