There’s a newer, cleverer version of the dry-fasting claim. It abandons the indefensible multiplier and retreats to something that could, in principle, be measured: maximum autophagy takes three days on a water fast, but only one day dry. It’s a genuine improvement. It also kept the same number — and that’s the tell.
The claim compares two curves, neither of which has ever been measured in a human. The dry-fast autophagy curve doesn’t exist. And the water-fast one — the famous “72 hours to peak” — is itself largely extrapolated from cell culture and animals. You cannot compute a ratio between two unmeasured quantities. Meanwhile the mechanism it needs — a hydration sensor feeding mTOR — still doesn’t exist, and still hasn’t been proposed.
Give it its due, because it earned some. The earlier claim — “three times the autophagy, three times the bad cells removed” — was unfalsifiable nonsense: no assay counts cells cleared by autophagy in a living person, so the figure had no possible source. Somebody noticed. This version retreats to a claim about time to reach a threshold, which is a real kind of claim. Thresholds can be measured. Curves can be compared. In principle, this one could be tested.
That’s a meaningful upgrade, and it deserves a real answer rather than a reflex.
Track the claim across its rebuilds. Version one: 3×, for water weight and toxins. Version two: 3×, for autophagy and cells removed. Version three: 3×, for time to peak autophagy. Three different biological claims, requiring three completely different experiments — and one ratio that never moves.
When a claim retreats to a better version and keeps the same number, the number is what’s being defended. Real findings don’t survive the replacement of their own subject matter — change what you’re measuring and the number changes with it, because the number came from the measurement. A figure that rides through three complete reconstructions intact was never attached to any of them.
Now the specific failure, and it’s more interesting than the earlier one. The claim says: 72 hours on water, 24 hours dry, therefore 3× faster. To have that, you’d need both curves.
The dry-fast curve has never been measured in humans — that much you’d expect by now. But here’s the part people miss: the water-fast curve hasn’t either. The “72 hours to peak autophagy” figure that circulates everywhere is itself largely extrapolated from cell culture and animal models. Human autophagy kinetics are poorly characterised — measuring autophagic flux in a living person requires serial tissue biopsies and flux blockade, and that study, at that duration, hasn’t been done at scale either.
So the claim divides one unmeasured number by another unmeasured number and reports the quotient with confidence.
“When researchers have looked at how long it takes…” — which researchers? That phrase is doing all the work and no citation ever follows; it’s the grammatical form of evidence without the substance. And notice the study now required is harder than before, not easier: serial human biopsies at multiple timepoints across both conditions, to establish where each one plateaus. The rebuild made the claim more defensible in shape and more expensive in evidence — while acquiring none.
Reaching the threshold faster still requires dehydration to accelerate the signal. And the signal is mTORC1, whose inputs are mapped: amino acids via Sestrin2/GATOR, energy charge via AMPK, insulin via PI3K/Akt, oxygen via REDD1. There is no hydration sensor. Not a contested one — an absent one. For water status to speed up autophagy, it would need a physical route into that pathway, and nobody has proposed what it might be.
The rebuild changed the claim’s shape. It didn’t give it a mechanism.
Be fair: “faster to threshold” is a coherent hypothesis shape. If dehydration accelerated the AMPK/mTOR signal, then reaching peak autophagy sooner would follow naturally, and the claim would be sensible. That’s a real argument structure, and the earlier version didn’t have one. The problem is purely that the premise is unsupported — and note that the obvious candidate route is already closed: dehydration doesn’t increase fat oxidation either, because metabolic water production is capped by energy expenditure. The most plausible bridge was checked, and it doesn’t hold.
This is a useful specimen precisely because it’s the strong version. It shows the tests aren’t tuned to one phrasing — they catch the sophisticated rebuild too. Test one flags the fixed ratio. Test two finds no study, and a harder one now required. Test three finds no sensor, unchanged.
And the honest limit stands, exactly as before: this doesn’t prove dehydration does nothing to autophagy timing. It shows nobody knows, that the number offered was not obtained by finding out, and that the mechanism required has never been proposed. That’s a bad bet carrying a real cost — not a proven zero. Keeping that distinction is the whole discipline.
The clever version of the claim — “max autophagy takes 3 days on water, 1 day dry” — is a genuine upgrade: it retreats from an unmeasurable multiplier to a measurable-in-principle threshold. But it kept the same 3× through a complete rebuild of its own subject matter, which is the tell: the number is the thing being defended, not the thing being reported. And it compares two curves neither of which has been measured in humans — the dry one doesn’t exist, and the famous water one is itself extrapolated from cells and animals. You cannot compute a ratio between two ghosts. The mechanism problem is untouched: reaching threshold faster needs dehydration to accelerate mTOR signalling, and there is still no hydration sensor — nor a proposed one, nor the obvious bridge (dehydration doesn’t raise fat oxidation either). Better argument. Same verdict.
This article is for educational purposes only and is not medical advice, diagnosis, or treatment, and does not recommend any fasting protocol for any individual. It examines the structure of a public claim, not the practice of any named individual or organisation. Extended fasting is not appropriate for everyone and can be dangerous in the context of certain medical conditions, kidney disease, diabetes, pregnancy, older age, eating-disorder history, or many medications — including GLP-1 receptor agonists, which blunt thirst as well as appetite and carry a documented acute kidney injury signal driven by volume depletion. Dehydration can cause acute kidney injury, electrolyte disturbance, arrhythmia, and death, and is not reliably signalled by how you feel. Anyone considering an extended fast should do so only under the supervision of a qualified physician who can monitor renal function and electrolytes.
This lesson relates to these health systems — health works as a connected system, not isolated topics.
Prerequisite: The Floating Number: How to Tell a Measurement from a Slogan