Version v1.0-2026-10
How we grade evidence
Every rating on this site is a published claim about science. This page is how you check it. It is versioned, and the version that produced any given rating is shown next to that rating.
Four measurements, never one score
We do not publish a single number. A compound is not “82% effective”, and any site that tells you so has invented a precision the evidence does not support. Instead we publish four separate measurements, each on its own scale.
Human evidence
Strong · Moderate · Limited · Very limited · None identified. Derived from the design, size and consistency of human studies on record. Animal work never raises this rating.
Research maturity
Approved · Phase III · Phase II · Phase I · Preclinical. Where a compound sits in development, independent of whether the results were good.
Safety knowledge
Established · Moderate · Limited · Unknown. How well characterised the safety profile is — not how safe the compound is. Those are different questions.
Community interest
Computed from search and discussion volume. It is never editorial and it carries no scientific weight. It exists so the gap between interest and evidence is visible.
Source tiers
Every source is tiered, and the tier is visible wherever the source is shown.
What we will not do
- Publish a rating without a visible rationale.
- Let animal evidence raise a human-evidence rating.
- Present an advisory recommendation as a change in legal status.
- Sell a compound, or take money from anyone who does.
- Let a commercial relationship change a rating. Ever.
- Fabricate a citation. A citation that does not resolve is removed before display.
Corrections
Ratings are superseded, never silently edited. When a rating changes, the previous version and the date remain on record. If you believe something here is wrong, tell us and we will publish the correction with its date.