What Bryan Johnson Says His Protocol Cannot Change

Bryan Johnson publishes more numbers about his own body than anyone in longevity. On October 23 he published a different kind of post: a carousel whose first slide lists what his protocol cannot do. It drew about 40,000 likes and 780 comments in under a day, the largest response to any figure we track that week. The list deserves to be taken at full length, because it is the rare claim from this world that a reader can actually check.

The concession, in full

Johnson says 2026 anti-aging technology cannot change five things: face age, hearing, baldness, teeth, and brain structure. Read the list slowly. These are not obscure biomarkers. They are the visible, audible, structural facts of aging that everyone around a person can see, the things no blood panel reports. The most measured man in the field is saying the measurements do not reach them.

Each item on the list is checkable in principle, and that is what makes the slide credible. Face age can be assessed by blinded raters. Hearing has audiometry. Baldness and teeth are visible. Brain structure is imaged routinely. A claim that can be tested can also fail, and a public figure with a commercial protocol has every incentive to publish only the claims that cannot. This one can. That is a meaningful signal about the man, whatever else is true of the marketing around him.

The second slide is a different species of claim

The carousel's second slide counters with percentile claims: biomarker results presented as top-percentile rankings. No panel is published beside them. A percentile is a comparison, and a comparison needs four things before anyone can check it: the panel itself (which markers, which values), the reference population (percentile among whom, at what age), the assay and lab (the same marker reads differently across methods), and the date (a percentile from one draw is a snapshot, not a state). Publish those four and the claim becomes as checkable as the concession. Without them, a percentile is a number about a comparison the reader cannot see.

This is not a Johnson-specific failing. It is the standard shape of biomarker marketing: the reference class stays invisible, so the number can never lose. Our own screening grades exist because of the same problem at one remove: a test earns its grade from whether acting on its result changes outcomes, not from how precisely it measures, and a percentile claim has not yet reached even the measurement question a reader can audit.

What neither slide can show

The concession does not show Johnson's protocol fails. It shows its boundary, which every honest intervention has. The percentile slide does not show the protocol works. It shows numbers he selected, ranked against a population he did not publish. Between the two slides sits the actual state of the evidence for nearly everything in his 63-item tracked regimen, most of which our library grades item by item from the same public record he publishes: some items rest on real human trials, many rest on mechanism and hope, and the protocol as a whole, the stack taken together for decades, has never been trialed anywhere, by anyone, including him. He is one person, and one person's panel cannot separate the protocol from the person.

The concession slide is the part of the post to keep. A field that published its boundaries as plainly as its percentiles would be easier to trust, and cheaper to follow.