How we read research

Not all evidence is equal, and the gap between “this worked in mice” and “this works in people” is where most longevity marketing operates. Every substantive claim here is placed on the scale below, in the text, at the point the claim is made.

The counts beside each rung are live: they are recounted from the entries on every build, so this page cannot go on describing a labelling system that quietly stopped being applied. They also make the most useful admission this publication can make about its own subject. 44 of 56 graded claims sit on the bottom two rungs. That is not a flaw in the reporting. It is what the literature looks like once you stop grading it on a curve.

Where the numbers come from

A dose figure is treated as a claim like any other, which in this field is unusual enough to be most of the point. The question asked of every circulating number is not whether it sounds reasonable but where it came from, and the answer is frequently that it came from nowhere: it is a fraction of a vial, or a figure lifted off a label written for somebody else, or an amount that appears in no published study in any species. When that is the answer, the entry says so rather than omitting the number, because the number is what the reader arrived holding.

“I could not find it” is a result, and it is published as one. Searching four databases for a human trial and coming back empty is a finding about the state of the evidence, and reporting it plainly is more useful than the alternative, which is to fill the space with mechanism. Where no published human dose exists to compare a convention against, the entry leads with that rather than burying it.

The commonest failure is borrowed evidence

Most of what looks like a research base in this market turns out, on checking, to belong to something adjacent. It arrives in three recognisable forms, and entries are read against all three:

• Another molecule.A trial record gets attached to a compound that is not the one in the trial — a fragment standing in for the modified copy that was actually tested, or a seven-residue piece borrowing the safety history of the forty-three-residue protein it was cut from.
• Another route. A figure established by infusion into a vein under supervision gets carried across to an injection under the skin at home. That is a different exposure, and usually there is no published measurement by the route people actually use.
• Another population.A dose approved for one group is read as approved for whoever bought it — sometimes off a label whose own limitations name the buyer as a group the drug is not indicated for.

When something here is wrong

Figures are corrected in place and the entry records that it was revised. A revision date is set only when a fact changed — a figure corrected, a claim rescoped, a source re-read — and never for a typo or a reflow, because a trust signal that fires on cosmetic edits stops carrying information. If a claim here appears without a source, or a source does not say what it is reported as saying, that is an editorial failure and worth telling us about.

Sources are linked

Where a study is referenced it is linked directly, so you can read it yourself and decide whether the interpretation here is fair. The standard applied to this publication is the one it applies to everybody else in it: a claim you cannot trace is a claim you should not accept.

Conflicts

This publication earns through affiliate links on exactly one page: the vendor sourcing table, where every such row is marked, the disclosure sits at the top of the page rather than in a footer, and the findings printed beside a paid link are the same ones that would be printed without it. There are no affiliate links anywhere else — not on the compound cards, not in the entries, not in the newsletter — and nothing here is sponsored. The full accounting is on the about page.

What gets left out

No dosing guidance and no protocols. Not because they are uninteresting, but because publishing them responsibly requires a clinical relationship this publication does not have with you. That is a narrower exclusion than it sounds. Reporting what a trial administered, what circulates, and the distance between the two is reporting on the published record, and it is exactly what the comparison table is for. What is missing from it is the line every other page in this category leads with, which is a recommendation about you.

Sourcing is not excluded either, and an earlier version of this page wrongly said it was. What is in a vial, how to read a certificate of analysis, and what each vendor does and does not publish are covered openly — in the entries, and on the vendor page. How much of it to put in your body is the part that is not here.