How it's made

The method

Nothing here hides how it was made. The seam between machine and human is the brand.

The seam: AI × human

Every article is researched and drafted with AI running under a strict source filter, then curated, sourced and verified by a human. We mark that seam openly on each piece. Original photographs, drawings and notes are always labelled human work, kept plainly separate from anything the machine produced.

Why it beats the usual three

The tiers of evidence

When a claim is about what something does (a benefit, an effect), we never flatten the evidence. Each is sorted into one of three tiers, and the tier is stated on the piece:

Tier 1Traditional use — documented human use over generations. Strong as history; not, by itself, proof of mechanism.
Tier 2Laboratory & preclinical — evidence in cells and animals. Shows something is plausible; it routinely fails to translate to people.
Tier 3Human clinical — controlled trials in actual humans. The hardest to earn, and the only tier that really settles a claim.

"Evidence reached: Tier 3" on an article means the strongest tier any claim in it is supported by — and we tell you when that Tier 3 is one small trial, not a mountain.

Read across languages

We read sources in their original language wherever it matters — and tell you which. Passages drawn from machine translation are flagged; soon, whole articles will be mirrored in Russian (which can be verified by hand), with the original phrase shown beside the translation so you can re-check it yourself.

Sourced in the open

Every rabbit hole ends in a colophon: the tier of evidence reached, the languages consulted, the open questions left standing, and the full list of sources — each one a real link, grouped by type. The raw research document is never published as-is; the article is the edited, human layer on top of it. The vetted source spine lives at the Registry →