Chapter 21 — Intermittent Fasting: Protocols, Autophagy, Circadian Timing, and What Happens When You Match the Calories

The Hook: Theo's window

Theo Vasquez did 16:8 for eleven months. It worked for three of them.

Months 1–3: He ate between noon and 8pm. He lost 4.2 kg, felt sharp in the mornings, and told everyone. "I'm not even hungry until eleven. It's like my body was waiting for permission."

Months 4–7: The weight stopped moving. He kept the window.

Months 8–11: He gained 1.8 kg back. ⚠️ He was still doing 16:8, exactly as before, and had not broken a single rule.

Here's what actually happened, from his diary.

The protocol removed 550 kcal of morning eating. Then he learned to fit 3,290 kcal into eight hours.

⚠️ Nothing about the window changed. His compliance was perfect throughout. What changed was that he got good at it.

💡 This is the whole chapter in one table. ⚠️ A protocol that removes eating opportunities reduces intake until you adapt to it. And the "7:45 rush" — the eating that happens in the last fifteen minutes before a window closes — is a real, recognizable, entirely predictable behaviour that no protocol description mentions.

Intermittent fasting is a genuinely interesting research area with a genuinely inflated public claim. ⚠️ And unlike Chapters 18–20, this one has almost no industry behind it — nobody profits much from you not eating — which makes it a useful test of whether we can be sceptical about something without a villain to point at.

The question is precise: does when you eat matter, independent of what and how much?

🏃 Fast Track: §21.2 (four claims), §21.5 (the calorie-matched trials), §21.12 (⚠️ who should not do this), §21.14 (what to do). Twenty-five minutes.

🔬 Deep Dive: §21.6 (autophagy, and the human dose question), §21.7 (circadian timing — the most interesting part and the least discussed), §21.9 (what actually happens to metabolic rate).

21.1 The protocols

They are not the same intervention and the evidence differs by protocol.

⚠️ Notice how much of the popular claim gets made about "fasting" as a category while the evidence attaches to specific protocols with different windows, different energy intakes and different timing. Chapter 20's category problem, again.

21.2 Four claims

⚠️ Claim 2 says "comparable," and that word does the work. Comparable is not superior, and the public claim is almost always superiority.

21.3 What actually happens, hour by hour

Dose-labelled, because everything in this chapter is a dose of time. Chapter 6 §6.4's substrate mixing board is the underlying machinery.

💡 Aha moment. ⚠️ Look at where a 16-hour fast falls: in the band that everybody who sleeps eight hours and doesn't eat at 2am is already in every single night.

16:8 is not a metabolically exotic state. It's a normal overnight fast, extended by about four hours.

⚠️ This is worth holding on to, because a great deal of the marketing describes 16 hours as though it crossed a physiological threshold. It doesn't cross one. Nothing switches (Chapter 6's threshold — mixing board, not gearbox).

Which is not the same as saying it does nothing. It means whatever it does is a matter of degree, and the degree has to be measured.

Hunger doesn't work the way you think it does

⚠️ The single most useful physiological fact in this chapter, and it's rarely in the protocols.

Most people assume hunger accumulates monotonically — that it rises steadily the longer you go without eating, so a 16-hour fast means 16 hours of increasing misery.

It doesn't.

🔬 Claim → Evidence → Verdict

The claim: "Hunger comes in waves and is strongly entrained to habitual meal times, rather than rising steadily with time since eating."

What the evidence shows: Ghrelin — the principal hunger-signalling hormone (Chapter 3 §3.6) — shows a pattern of pre-prandial rises that anticipate habitual meal times, and it falls again after the expected meal time passes whether or not you eat. ⚠️ The rise is anticipatory and learned, not purely a consequence of an empty stomach.

This has been demonstrated by shifting people's meal schedules and watching the ghrelin peaks follow, over a matter of days.

Two practical consequences, and both are things people discover by accident:

1. ⚠️ Hunger at 8am on day three is largely a habit, and habits are re-trainable. Theo's "I'm not even hungry until eleven" was not a metabolic transformation. It was his ghrelin peaks moving, which takes days to a couple of weeks.

2. ⚠️ A hunger wave passes on its own in roughly 20–30 minutes whether or not you eat, which is why people report that pushing through a peak is much easier than anticipating one.

📉 Evidence quality: Well-characterized hormone physiology plus meal-schedule shifting studies.

Verdict: 🟢 Probably true, and it's the honest explanation for the "it got easier after two weeks" that essentially everyone reports.

⚠️ And note the symmetric implication, which the fasting community mentions less: if hunger entrains to when you habitually eat, then a person who habitually eats at 10pm will feel genuine hunger at 10pm. That's not a character flaw either — and it's re-trainable in the same way, in either direction.

21.4 Fasting is a dose of time

Chapter 18's threshold, applied to a variable that isn't a substance.

The five slots:

⚠️ The last row is the entire debate, and §21.5 is what happens when you control it.

21.5 What happens when you match the calories

The discriminating design, as in Chapters 18 and 19.

If IF works only by reducing intake, then at matched calories it should perform the same as continuous restriction. If timing has independent effects, it should outperform.

This has been tested repeatedly.

The general finding across randomized trials and meta-analyses: ⚠️ intermittent fasting produces weight loss comparable to continuous energy restriction, and generally does not outperform it when energy intake is matched.

Two trials worth knowing specifically:

The TREAT trial (Lowe et al., JAMA Internal Medicine, 2020) randomized people to 16:8 time-restricted eating or three structured meals a day for 12 weeks. ⚠️ Weight loss was modest and not significantly different between groups. A secondary finding — a signal of greater lean mass loss in the TRE group — was widely debated, and the measurement method's limitations were a large part of that debate. I'd treat it as a flag rather than a finding (§21.8).

A 12-month trial published in the New England Journal of Medicine in 2022 compared time-restricted eating plus calorie restriction against calorie restriction alone. ⚠️ No significant difference in weight loss between the groups.

That second design is the cleanest test available, because both arms restricted calories and only one restricted the window.

🔬 Verdict on Claim 2 — IF produces weight loss comparable to continuous restriction: 🟢 Probably true, well replicated.

🔬 Verdict on Claim 3 as a weight claim — timing is superior for weight loss: 🟠 Probably false. ⚠️ The calorie-matched trials are the discriminating design and they don't support superiority.

But hold on to Claim 3, because §21.7 is about a version of it that isn't a weight claim and that the trials above weren't designed to test.

The outcomes people actually ask about

Because "does it work?" collapses several questions, and the answers differ.

💡 ⚠️ Read the two columns against each other, because the difference between them is the whole public argument.

Column one is almost entirely 🟢 and it is what people experience. Column two is almost entirely 🟡 or 🟠 and it is what the science says about the mechanism.

Both are honest. ⚠️ "It worked for me" and "it's not metabolically superior" are simultaneously true, because column one includes the calorie reduction and column two removes it. Nobody arguing from column one is lying, and nobody arguing from column two is being dismissive — they are reading different columns and neither says which one they're in.

21.6 Autophagy

🔬 Claim → Evidence → Verdict

The claim: "Sixteen hours of fasting triggers autophagy — your body's cellular cleanup process — which clears damaged cells, prevents disease, and slows ageing. It won a Nobel Prize."

What's true, and it's a lot: ⚠️ Autophagy is real, fundamental, and beautifully characterized. Yoshinori Ohsumi received the 2016 Nobel Prize in Physiology or Medicine for elucidating its mechanisms, working largely in yeast. It is a genuine cellular recycling process, it is upregulated by nutrient deprivation, and its dysfunction is implicated in neurodegeneration, cancer and ageing. None of that is exaggerated.

Where the claim fails — three places, and the third is decisive:

1. ⚠️ The Nobel was for the mechanism, in yeast. It was not for a fasting protocol. Ohsumi studied how autophagy works. ⚠️ Citing a Nobel Prize for a claim it wasn't awarded for is a rhetorical move, not evidence.

2. Most of the fasting-and-autophagy data is from rodents and cell culture. ⚠️ And the translation problem here is unusually severe: a mouse's metabolic rate is several times higher per unit mass than a human's. A 24-hour fast in a mouse is a far larger physiological perturbation than 24 hours in a person — closer, by some estimates, to several days. Rodent fasting durations do not map onto human clock time.

3. ⚠️ THE HUMAN DOSE IS NOT ESTABLISHED. Where does "16 hours" come from? There is no human dose-response curve for fasting duration and autophagic flux in tissues that matter. Measuring autophagy in living humans is genuinely difficult — it requires tissue sampling and markers that are hard to interpret — and the studies that exist are small, in limited tissues, and do not establish a threshold.

📉 Evidence quality: Excellent on the mechanism. ⚠️ Absent on the human dose-response, which is the entire content of the claim.

Verdict: ⚗️ Untested, in the specific form the claim takes. ⚠️ Autophagy is real. "16 hours triggers it, and that prevents disease" is two unestablished steps stacked on a real foundation.

💡 ⚠️ This is Chapter 19 §19.5's broken-link problem in a new costume — and it's actually worse, because there the link had been tested and failed. Here it hasn't been tested. A real mechanism, a named process, a Nobel Prize, and no human dose — which is Chapter 17 §17.1's sixth tell exactly: mechanism named, never dosed.

21.7 Circadian timing — the interesting part

⚠️ This is where I think the real science is, and it gets a fraction of the attention that autophagy does.

Your body is not metabolically identical at 8am and 10pm.

Established circadian findings:

• ⚠️ Glucose tolerance is generally better in the morning and declines across the day. The same meal produces a larger glucose excursion in the evening than at breakfast.
• Insulin sensitivity follows a daily rhythm.
• Melatonin onset in the evening is associated with impaired glucose handling — which is a mechanistic reason late eating differs from early eating.
• ⚠️ Shift work involving circadian disruption is associated with metabolic disease, and IARC classifies it as Group 2A (Chapter 20 §20.11's category).

Which generates a prediction the weight trials weren't testing: ⚠️ an eating window placed EARLY should differ from the same window placed LATE, at identical calories.

And there's a trial that tested exactly that.

Sutton et al. (Cell Metabolism, 2018) used early time-restricted feeding — a roughly 6-hour window with dinner finished by mid-afternoon — in men with prediabetes, under eucaloric conditions (weight held stable), in a crossover design. ⚠️ It reported improved insulin sensitivity, blood pressure and oxidative stress markers, without weight loss.

⚠️ The sample was very small and it was a controlled feeding study, not a free-living one. So this is a promising, mechanistically coherent, calorie-matched signal — not a settled finding.

🔬 Verdict on Claim 3 as a metabolic claim — timing has effects independent of calories: 🟡 Unclear / it depends, and the "depends" is on WHICH timing.

⚠️ Early windows look more promising than late ones, and most people who practise 16:8 practise it as a LATE window — skip breakfast, eat until 8pm. Which is close to the opposite of the version with the better mechanistic and trial support.

🔬 And a related verdict that does hold up: 🟢 late-night eating is probably worse for glycemic control than the same food earlier, on consistent circadian evidence. (Note this is not Chapter 4's "eating late makes you gain weight," which was 🟡 and about energy. This is a glycemic claim, and it's better supported.)

💡 ⚠️ The practical irony is worth stating plainly. The most popular protocol is 16:8 with a late window, and the best mechanistic case is for an early one. If timing matters, most people doing intermittent fasting for timing reasons are doing the version least likely to deliver them.

21.8 What about muscle?

A real concern, honestly reported.

The mechanism is plausible: Chapter 8 established that muscle protein synthesis responds to protein doses distributed across the day, and that older adults show anabolic resistance — needing larger per-meal doses. ⚠️ Compressing intake into fewer meals reduces the number of stimulation events.

What the evidence shows: ⚠️ mixed and contested. The TREAT trial's lean-mass signal (§21.5) got substantial attention and substantial methodological criticism. Other trials have not consistently replicated it.

What's less contested:

• ⚠️ Total protein intake tends to fall when the eating window narrows, simply because there are fewer opportunities. This is a nutrition problem, not a fasting problem — and it's fixable.
• Resistance training substantially protects lean mass during weight loss regardless of meal pattern (Chapter 23).

🔬 Verdict: 🟡 Unclear / it depends. ⚠️ The mechanism is plausible, the trial evidence is mixed, and the practical fix is unambiguous: hit your protein target inside the window and lift things. A person doing 16:8 who eats 60 g of protein has a protein problem that the window caused and the window doesn't have to cause.

21.9 Does fasting "boost" or "reset" metabolism?

🔬 Claim → Evidence → Verdict

The claim: "Fasting resets your metabolism and boosts your metabolic rate."

The grain of truth, which is real: ⚠️ short fasts do not lower resting metabolic rate. Norepinephrine rises during a fast of 24–72 hours, and RMR is maintained or slightly elevated over that period. This directly contradicts "starvation mode" as popularly stated (Chapter 5 §5.6), and the fasting community was right about it before most of the diet world was.

Where the claim fails:

1. "Reset" describes nothing. ⚠️ Chapter 17's seventh shape — a dial described as a switch. There is no metabolic state that gets restored.

2. The RMR elevation is small, transient and irrelevant to weight loss. ⚠️ It does not survive sustained energy deficit — Chapter 5's adaptive thermogenesis is a response to sustained deficit and it happens on any diet, fasting included.

3. And "boosting metabolism" is Chapter 5 §5.8's aisle, arriving without a product attached.

📉 Evidence quality: Good on short-fast norepinephrine and RMR maintenance; the "reset" claim has no referent.

Verdict: 🟠 Probably false as stated — ⚠️ while conceding the real finding underneath it, which is that short fasts don't crash your metabolic rate and the popular fear of that is unfounded.

21.10 Why it works when it works

⚠️ Chapter 10 §10.4's finding, arriving in its cleanest form yet.

Intermittent fasting is a rule about when, and some people find rules about when much easier than rules about how much.

Reasons it suits some people:

• It requires no measurement, counting, or food-quality judgement. ⚠️ One binary decision, made once per day.
• It removes decision points. The mid-morning question disappears — Theo's 240 kcal of work snacking vanished without a decision.
• Some people genuinely aren't hungry in the morning, and the protocol legitimizes something they were doing anyway.
• It fits some work patterns and household routines very well.
• ⚠️ And it has a clear, unambiguous compliance criterion, which is psychologically valuable in a way that "eat moderately" is not.

Reasons it suits others badly:

• ⚠️ Some people compensate almost perfectly, and Theo eventually did.
• The "7:45 rush" — hurried, large, often poorly-chosen eating before the window closes.
• It's socially awkward. ⚠️ Breakfast with family, morning coffee meetings, lunch at fixed times.
• Hunger and irritability in the fasted window are real for many people, at least initially.
• ⚠️ And for a subset it maps uncomfortably onto restriction and control (§21.12, Chapter 34).

🔬 Verdict on Claim 1 — IF helps some people eat less: 🟢 Probably true. ⚠️ And this is the honest case for it — not a metabolic case, an adherence case. Which is the same argument Chapter 10 made for every diet, and it is not a lesser argument.

21.11 The claims that don't hold

Briefly, because they follow from the above:

One that deserves more than a table row: "fasting reverses type 2 diabetes"

⚠️ Because this one is partly true, is clinically important, and is the most consequential thing in the chapter to get exactly right.

What's genuinely established: ⚠️ substantial weight loss can produce remission of type 2 diabetes — normalized glucose without medication — particularly earlier in the disease course and with losses in the region of 10–15% of body weight. This is real, it is supported by structured weight-management trials, and it changed clinical practice. Chapter 26 covers it properly.

What's not established: ⚠️ that fasting is the agent. The trials that demonstrated remission largely used total diet replacement — formula low-energy diets — not intermittent fasting. The mechanism is the weight loss and the reduction in liver and pancreatic fat, not the timing of meals.

So: ⚠️ if intermittent fasting is how a particular person achieves substantial sustained weight loss, it can absolutely contribute to remission. It is a delivery mechanism, not the active ingredient — and any protocol works if it produces the same loss.

⚠️ And the safety point is not optional. A person with type 2 diabetes on insulin or a sulfonylurea who starts fasting without adjusting medication is at real risk of hypoglycaemia — and the better their glucose control gets, the higher that risk climbs. ⚠️ Remission attempts are a prescriber-supervised activity. This is the one place in this book where "try it and see" is genuinely dangerous.

21.12 ⚠️ Who should not do this

The most important section in the chapter, and the one the protocols mention least.

⚠️ This list is longer than most people expect, and several entries are not obvious.

⚠️ And a category that isn't a contraindication but should be named: people for whom food is already scarce. A protocol built on voluntary restriction reads very differently to someone whose fasting isn't voluntary, and the language of the movement handles this badly.

⚠️ On the eating-disorder entry specifically

Because it's the one people skim, and it's the one I'd least want skimmed.

Intermittent fasting is unusually well-shaped to be adopted by, and to conceal, disordered eating.

Four features, and each is a feature rather than a flaw of the protocol:

1. ⚠️ It provides a socially acceptable reason to refuse food. "I'm doing 16:8" ends a conversation in a way that "I'm not eating today" does not.

2. It converts restriction into compliance. ⚠️ Not eating becomes success rather than a symptom, and the better someone is at it, the more approval they get.

3. It has a countable streak. ⚠️ Apps that reward consecutive fasts are, for a subset of users, reinforcing exactly the wrong thing.

4. And the hunger is reframed as the mechanism (§21.14b). ⚠️ Discomfort becomes evidence that it's working — which is precisely the cognitive move that makes restrictive eating self-sustaining.

⚠️ This is not an argument that intermittent fasting causes eating disorders, and I want to be careful: for most people it is a neutral scheduling choice. It is an argument that for someone with a history — diagnosed or not — the protocol supplies structure, vocabulary, social cover and positive reinforcement to a pattern that needs none of those.

The honest test, and it's the same one Chapter 20 §20.13 gave for clean eating: ⚠️ is the window narrowing over time? Is eating with other people becoming harder? Is there guilt or anxiety when the rule is broken? Is the rule now more important than the reason for the rule?

Any yes is worth raising with someone. Chapter 34.

21.13 Ramadan, and a note on the literature

⚠️ The largest recurring natural experiment in human fasting is religious, and the fasting literature usually mentions it in a sentence.

During Ramadan, observant Muslims fast from dawn to sunset — including fluids — for a lunar month. Depending on latitude and season, that can be anywhere from about 11 to over 18 hours daily.

What the research generally finds: modest, largely transient changes in weight and metabolic markers, mostly reversing after the month. ⚠️ Effects vary substantially with daylight length, with what is eaten at iftar and suhoor, and with sleep disruption — which is often the more consequential variable.

Three things worth taking from it:

1. ⚠️ Fluid restriction is a real difference. Most secular IF protocols permit water; Ramadan does not. In long summer days and hot climates, dehydration is a genuine clinical consideration (Chapter 15).

2. Islamic jurisprudence contains established exemptions — illness, pregnancy, breastfeeding, menstruation, travel, and age. ⚠️ A religious tradition that has practised fasting for fourteen centuries built in medical exemptions. The wellness protocols mostly haven't.

3. And the sleep disruption is frequently the largest effect, which is a reminder that fasting studies in free-living people are studying a bundle (Chapter 17, shape 8).

⚠️ A note on framing: religious fasting is a spiritual practice, not a health protocol, and treating it primarily as a source of metabolic data is both scientifically limited and somewhat discourteous. It's included here because the data exists and is relevant — not because that's what it's for.

21.14 So what do you actually do?

1. ⚠️ Map your current window before changing it. Most people don't know theirs. The Project Checkpoint does this and it's frequently the whole intervention — a lot of people discover they're already at 13:11 and eating 700 kcal after 9pm.

2. If you want to try it, try 12:12 or 14:10 first. ⚠️ These capture most of the practical benefit with far less disruption, and 12:12 is close to a normal eating pattern.

3. ⚠️ If timing is your reason, shift the window EARLY rather than late. §21.7. Eating 8am–4pm has better mechanistic support than noon–8pm, and almost nobody does it.

4. Hit your protein target inside the window (Chapter 8), and do resistance training (Chapter 23). ⚠️ This addresses §21.8's concern directly and it isn't optional if the window is narrow.

5. Watch for the 7:45 rush. ⚠️ If your last hour before closing is hurried, large and poorly-chosen, the window is producing the behaviour — and a wider window will beat a compressed one.

6. Expect the effect to fade (Theo). ⚠️ This isn't failure — it's compensation, and it happens to every restriction that isn't measured (Chapter 4 §4.7). Know it's coming and you won't conclude you're broken.

7. ⚠️ Check §21.12 before starting, honestly, including the entries that don't feel like they apply to you.

8. And if it doesn't suit you, stop. ⚠️ There is nothing in this chapter that says you need it. Every documented benefit is available from eating less, eating earlier, or both — and Chapter 10's finding is that the method you'll actually keep beats the method that's theoretically superior.

"Does this break my fast?"

The most-asked practical question, and the answer is a dose question wearing a purity costume — ⚠️ which by now should feel familiar.

There is no biological event called "breaking a fast." There is a continuous, dose-dependent response to whatever you consumed.

⚠️ The pattern to notice: the answer people most want policed is milk in coffee, and the item that most obviously breaks the protocol is the 400-kcal butter coffee sold specifically to fasters.

Chapter 20's binary, arriving in a new place. A category boundary lets you finish; a dose gradient doesn't — so the question becomes "is it allowed?" instead of "how much, and does that amount matter?"

If your goal is intake reduction: calories are calories, and the butter coffee is the problem. If your goal is glycemic or circadian: the timing and composition of the first real intake is what matters, and the milk isn't it. ⚠️ Either way, nobody needs to adjudicate the milk.

What we don't know

⚠️ The human autophagy dose-response — the central claim of the whole movement, and it is genuinely unmeasured (§21.6).

Whether early TRE's calorie-matched metabolic benefits replicate at scale, in free-living people, over years. ⚠️ The Sutton-type finding is the most interesting thing here and it rests on very small numbers.

Long-term effects of any protocol over decades. ⚠️ There is essentially no such data, and there won't be for a long time.

And whether the lean-mass signal is real (§21.8).

⚠️ How firmly I hold these

21.14b The argument with no villain

⚠️ Worth a section, because this chapter is the control condition for Part IV.

Chapters 18, 19 and 20 all had a commercial structure. Sugar had an industry. Seed oils had a premium alternative at $350 a year. Organic had a 14.6% budget premium. Chapter 1 §1.5's question — what happens to their income if you decline? — did real work in all three.

⚠️ Fasting is free. Nobody profits much from you not eating. So if the sceptical apparatus this book has built only works when there's a seller, it should fail here.

It doesn't fail — and what drives the over-claiming instead is worth naming, because it operates everywhere:

1. ⚠️ Identity and community (Chapter 17 §17.11). Fasting has an unusually strong in-group — shared vocabulary, streak-counting, apps with milestones, a language of discipline. The protocol is something you are.

2. A perceptible signal. ⚠️ Hunger is the confirmation. You can feel the intervention happening, which Chapter 16's creatine could not offer and Chapter 11's fiber could not offer. The discomfort functions as evidence.

3. Restriction as virtue — Chapter 17's third shape, ⚠️ and the tell is that the difficulty is offered as the mechanism. "Your body needs the stress."

4. ⚠️ Genuine early results, reliably. Everyone gets Theo's first three months. A protocol that works for everyone initially and fades for many is the ideal shape for generating advocates, because the advocates form during month two.

5. And a real, prestigious, correctly-described mechanism sitting next to an unestablished claim (§21.6). ⚠️ Autophagy is the most respectable thing any wellness movement has ever had at its centre.

💡 ⚠️ The lesson generalizes beyond this chapter: commercial interest is one engine of over-claiming, and it is not the only one.

Identity, perceptible sensation, effort-as-evidence, reliable early results, and a prestigious mechanism will do it on their own — with nobody selling anything.

Which means Chapter 1 §1.5's question is necessary and not sufficient, and a reader who only asks "who profits?" will be systematically blind to exactly this case. ⚠️ Ask it. Then ask what else the claim is paying its holder in.

21.15 The two honest positions, side by side

⚠️ Both columns would give Theo the same advice: eat less, and preferably earlier.

Fifth Part IV chapter in a row where that's true.

🧾 What it costs

⚠️ Fasting is one of the few interventions in this book that is free by construction — and there is still a $250 kit for it. Tenth consecutive chapter where the best-evidenced option is the cheapest, and the first where the cheapest option is literally nothing.

Spaced Review

Answer before reading on.

1. (Chapter 4) Theo's compliance was perfect for eleven months and his weight went down then up. Explain it without using the word "willpower."

Compensation within an unmeasured restriction (Chapter 4 §4.7). ⚠️ The window removed 550 kcal of morning eating opportunities; it did not constrain what he ate inside the window. Over months he learned to fit 3,290 kcal into eight hours — larger meals, and the 7:45 rush. The rule was never about energy, so energy was free to return.

2. (Chapter 19) Compare the autophagy claim to the omega-6 ratio claim. ⚠️ Which is in worse evidential shape, and why?

Both are real mechanisms carrying unestablished conclusions. ⚠️ But the omega-6 chain was TESTED and the prediction failed (🟠). The autophagy claim's human dose-response has not been tested at all (⚗️). A failed test is more informative than no test — ⚠️ which is why ⚗️ is a different verdict from 🟠 and not a softer one.

3. (Chapter 8) Why does §21.12 flag older adults specifically?

Three things compound. Anabolic resistance means older adults need larger per-meal protein doses to stimulate muscle protein synthesis (Chapter 8 §8.6). ⚠️ A narrow window reduces the number of stimulation events. And total intake usually falls when the window narrows. Fewer events, smaller total, higher requirement — against a background risk of sarcopenia and frailty.

Project Checkpoint: Your Eating Window

Component twenty-one. Three days of observation before any decision.

Step 1 — Map what you actually do. ⚠️ Record only two things per day: the time of your first calorie and the time of your last. (Calorie. Black coffee doesn't count; the flat white does.)

My actual window: _ hours. My actual overnight fast: _ hours.

⚠️ Most people are surprised. A great many are already at 13:11 or 14:10 without having decided to be.

Step 2 — The evening question, which is usually where the finding is.

Calories consumed after 9pm, per day: ____

Was any of it hungry eating? ____

Step 3 — Before you change anything, check §21.12. ⚠️ Read the whole list. If any entry applies, the next step is a clinician, not a protocol. The eating-disorder-history entry does not require a diagnosis to count.

Step 4 — If you want to try something, choose ONE:

Step 5 — Set your protein floor and write it down, whatever you chose. ⚠️ A narrower window without a protein target is §21.8's problem, self-inflicted.

Protein target inside the window: ____ g/day (Chapter 8's checkpoint)

Step 6 — And write the compensation prediction. ⚠️ Before you start.

"If this stops working around month _, the most likely reason is _____."

Theo would have written "the 7:45 rush" if anyone had asked him. ⚠️ Nobody did, and he concluded he'd failed.

Next checkpoint (Chapter 22): your UPF audit — count what proportion of what you eat is ultra-processed, before anyone tells you what to do about it.

Chapter Summary

⚠️ Fasting is a dose of time, and the five slots apply: 12h and 48h are radically different exposures · "5:2 fast days" are 500–600 kcal and not fasts · almost no long-term human data exists · who you are is the most consequential slot · and "compared to what" is the whole debate.

Four claims:

Other verdicts:

⚠️ §21.5b's two columns are the whole public argument: against a person's previous diet, almost everything is 🟢. Against matched-calorie continuous restriction, almost everything is 🟡 or 🟠. "It worked for me" and "it's not metabolically superior" are simultaneously true — column one includes the calorie reduction and column two removes it, and nobody says which column they're in.

⚠️ §21.14b is the chapter's structural point: fasting is free, so Chapter 1 §1.5's "who profits?" finds nothing — and the over-claiming happens anyway, driven by identity, a perceptible signal (hunger), effort-as-evidence, reliable early results, and a prestigious mechanism. The who-profits question is necessary and not sufficient.

⚠️ §21.3's placement finding: a 16-hour fast sits in the band anyone who sleeps eight hours is already in every night. 16:8 is a normal overnight fast extended by about four hours. Nothing switches (Chapter 6).

⚠️ §21.6 is Chapter 17's sixth tell in its purest form — mechanism named, never dosed. Autophagy is real, Ohsumi's 2016 Nobel was for the mechanism in yeast, most fasting data is rodent — and rodent metabolic rates mean 24 hours in a mouse is not 24 hours in a person. The human dose-response does not exist.

⚠️ §21.7's irony: the best mechanistic case is for EARLY windows; almost everyone practises LATE ones. If timing matters, most people doing this for timing reasons are doing the version least likely to deliver it.

The one thing to remember: ⚠️ Theo's compliance was perfect for eleven months and his weight went down, flat, then up. He didn't fail. He adapted — and nobody told him that was the expected trajectory.

What's Next

Chapter 22 is ★ the chapter Part IV has been pointing at since Chapter 17, and it carries a threshold concept: processing is not a spectrum of badness — it's a question about what industrial formulation does to how much you eat.

NOVA and what "ultra-processed" actually classifies. ⚠️ The Hall ward trial — a randomized, controlled, calorie-matched feeding study that is the strongest single piece of evidence in this entire book. Why seed oils, sugar, additives and "chemicals" all turned out to be markers for the same thing. The classification's real weaknesses, argued properly. And whether the category is useful or is just "junk food" with a research grant.

⚠️ It also carries a 🪞 Learning Check-In, and it answers the question Chapter 20 Exercise M5 asked you to write down: what these arguments have actually been about.