Tim SpectorProfessor of Genetic Epidemiology, King's College London · founder of TwinsUK
Test it on yourself, there is no general answer
What it rests onPREDICT 1's finding that postprandial responses to identical meals vary severalfold between people, including between identical twins. In the same article he discloses that he is himself a regular breakfast eater.
Source: theconversation.com ↗
Valter LongoPhD · biogerontologist, Professor of Gerontology, USC
Keep breakfast, and do not delete it to lengthen a fast
What it rests onHis Cell 2022 review states that longer daily fasting periods involving breakfast skipping have been consistently associated with increased mortality; his foundation's 2024 booklet adds a gallstone-risk argument through reduced bile flow. His own booklet concedes the adult evidence is unsettled, which is why the claim grades Contested rather than Established.
Source: fondazionevalterlongo.org ↗
Neither side is addressed to the people for whom this is not a neutral experiment. Spector's practice carries flags for diabetes or glucose-lowering medication, eating-disorder history, pregnancy and adolescence, and all four are populations where "try it and see" is the wrong instruction. Longo's caution is aimed at gallbladder risk and at mortality associations in observational cohorts, which are population-level findings that do not tell an individual what will happen to them.
Both men are talking about generally healthy adults who are curious, and one of them, Spector, discloses that he does not do the thing he is inviting you to test.
More than the framing suggests. Both are against grazing across the whole day, both accept that an overnight fast of some length is useful, and their protocols nearly touch: Spector's 10-hour eating window is a 14-hour fast, which is exactly the ceiling Longo sets.
The disagreement is not about the length of the window, it is about whether the meal you delete to get there is allowed to be breakfast, and about whether the answer is general or personal.
A randomised trial of breakfast skipping with hard endpoints rather than weight, which neither side has. Short of that, per-person response data of the kind Spector's own PREDICT design could generate would settle whether the individual variation he invokes is large enough to overturn the population association Longo cites.
This is the rare disagreement where the two positions predict different things about the same person, so it is genuinely testable.
You do, and your call is dated, kept, and yours to change. Follow the people you already trust and what they publish turns into a short daily practice.
Free to sign up and use. We do not ask for a card.