Chapter 17 — Myth-Busting: Detoxes, Cleanses, Superfoods, and the Claims That Sound Good but Aren't Supported by Evidence

The Hook: The envelope

Theo brought the folded sheet back to the clinic seventeen months after he wrote it.

It had been in a kitchen drawer, under the takeaway menus, exactly as instructed — ten things he believed about food, where each one came from, how confident he was, and whether he'd ever checked. He'd written it before Chapter 2, before the food diary, before any of it.

He read it out. Then he stopped halfway through the fourth item and said: "Oh, that's embarrassing."

Here's what was on it, with what he'd rated each and what he thinks now.

Ten beliefs. He now thinks eight were wrong or substantially overstated.

And look at the confidence column. Four of the five he rated a 5 — maximum confidence — are on the wrong list. The one he rated a 2, superfoods, he'd been right to doubt.

His confidence was inversely correlated with his accuracy, across his own beliefs, written by himself, without anyone tricking him.

That is the most useful thing in this chapter, and you're going to do it yourself at the end.

But there's a limit to going through myths one at a time, and it's worth naming before we start: there are always more. I could give you fifty verdicts and by the time you finish reading, three new claims will have appeared, and you'd have no way to evaluate those.

So the verdicts here are not the point. They're the worked examples. What this chapter is really doing is dissecting the shapes — the recurring structures that make particular ideas sticky — so that you can recognize the next one before it has a name.

🏃 Fast Track: §17.1 (the shapes, sharpened), §17.10 (the ones that turned out true — this is the section that stops you over-correcting), and §17.13 (what to do about someone you love). Twenty-five minutes.

🔬 Deep Dive: §17.9 (the contamination cluster, where genuine concerns hide inside false claims) and §17.11 (the anatomy of stickiness) are the ones that generalize furthest.

17.1 The six shapes, sharpened

Chapter 1 §1.6 introduced these. Now that you've seen sixteen chapters of examples, they can be stated more precisely — and each has a tell, which is the useful part.

Two additions from what we've seen since Chapter 1, because sixteen chapters produced patterns that the original six didn't quite capture:

7. The dial described as a switch (Chapter 6, Case Study 2). Mode, stuck, reset, activated, unlocked, flipped. Tell: binary language about a continuous variable.

8. The bundled intervention (Chapter 3, Case Study 1). A protocol changes eight things at once and credits one. Tell: count the changes. If it's more than three, nobody will ever learn which mattered — including the person selling it.

💡 Aha moment. Notice that every tell is structural rather than topical. Not one of them requires you to know any nutrition.

Is a dose specified? Is a comparator specified? Is the language moral? Is the difficulty being offered as evidence? Is it unfalsifiable? Is the past specified? How many things change at once?

You can run all seven questions on a claim about a nutrient you have never heard of, which is the entire reason this chapter is organized around shapes rather than around a list.

Interlude: what the first sixteen chapters already retired

Before adding new verdicts, it's worth seeing how many are already behind you — because a myth-busting chapter is a strange thing to arrive at in Chapter 17. Most of the work is done.

Twenty-four, before this chapter starts. And note the distribution: only six are ❌. Most are 🟠 or 🟡 — overstated, mis-worded, or true-in-a-narrower-sense, which is what most nutrition myths actually are.

💡 This is the single most useful pattern in the table. ⚠️ The dominant failure in nutrition is not fabrication. It is over-extension.

Someone took a real finding — protein synthesis plateaus, fat oxidation varies with intensity, saturated fat raises LDL-C, refined carbohydrate makes overeating easy — and stretched it past what it could carry. Which means the right response is almost never "that's nonsense." It's "what's the narrower true version?"

A person who learns to ask that question will out-perform a person who learns fifty verdicts.

17.2 Detoxes and cleanses

Chapter 3 §3.5 gave the verdict; here it is as a category rather than a product.

The claim: your body accumulates toxins; a detox, cleanse, or "reset" removes them.

The physiology: your liver runs phase I and phase II biotransformation continuously; your kidneys excrete; your gut, lungs and skin all contribute. None of this has an off switch, and none of it accumulates a backlog.

The unfalsifiability: ⚠️ the toxins are never named. Named toxins — lead, mercury, solvents, specific drugs — are measurable, have specific medical treatments, and are not addressed by juice. An unnamed toxin cannot be measured before or after, which means the claim cannot fail.

Shapes: purity and contamination · restriction as virtue · the hidden simple fix · bundled intervention.

And the honest part, from Chapter 3's case study: people genuinely feel better. Theo did. What changed during his ten days was alcohol to zero, sleep from six hours to seven and a half, ultra-processed food to nothing, vegetables up, caffeine down, and a thousand-calorie deficit. Every one of those is free and permanent, and the $289 bought him permission rather than detoxification.

🔬 Verdict: ❌ Not supported. Chapter 3 §3.5. ⚠️ Colon cleansing carries real harms — electrolyte disturbance, dehydration, and rarely bowel perforation — and some marketed detox products have been found by regulators to contain undeclared pharmaceutical ingredients (Chapter 16 §16.2).

17.3 Superfoods

🔬 Claim → Evidence → Verdict

The claim: "Certain foods — açaí, goji, kale, spirulina, moringa, wheatgrass — have exceptional health properties that make them qualitatively different from ordinary foods."

Where it comes from: Real observations, over-extended. Some foods genuinely are more nutrient- dense than others. Some contain high concentrations of particular compounds. And antioxidant capacity assays — ORAC in particular — produced impressive-looking numbers that were enthusiastically marketed.

What the evidence actually shows: "Superfood" is not a scientific or regulatory category. It has no definition, no criteria, and no threshold. The EU has restricted its use on packaging without an authorized health claim since 2007, precisely because it means nothing.

The ORAC story is instructive: the USDA withdrew its ORAC database in 2012, stating that the values had been misused by industry and that antioxidant capacity measured in a test tube does not predict effects in humans — which is Chapter 13's isolated-nutrient graveyard arriving from a different direction.

And when specific superfoods have been tested against ordinary ones, the results are unremarkable. Blueberries are good. So are blackberries, and they're cheaper. Kale is nutrient-dense. So is cabbage, at a fifth of the price.

📉 Evidence quality: No category to assess; individual foods are ordinary foods.

Verdict: ❌ Not supported as a category. The individual foods are usually fine — that's the trap. Nobody is wrong to eat blueberries. They're wrong to think a blueberry is doing something a vegetable isn't, and to pay a premium for the belief.

🧾 Cost check. Açaí powder ~$45/lb · goji berries ~$18/lb · wheatgrass powder ~$60/lb · spirulina ~$40/lb — against **cabbage at $0.60/lb, carrots at $0.90, tinned tomatoes at $1.10, and frozen spinach at $1.40. The nutritional case for the second group is stronger per dollar by more than an order of magnitude, and Chapter 10 §10.5's list of what successful dietary patterns share contains none of the first group.

17.4 The alkaline diet

🔬 Claim → Evidence → Verdict

The claim: "Modern diets are acid-forming, which acidifies your blood and causes disease. Eating alkaline foods restores pH balance and prevents cancer."

Where it comes from: Two real things. Foods do differ in their acid or alkaline load — the concept of potential renal acid load (PRAL) is genuine, and diets high in protein and grains produce more acid load than diets high in fruit and vegetables. And urine pH genuinely does change with diet, which provides a visible, testable-looking confirmation.

What the evidence actually shows: Blood pH is regulated within roughly 7.35–7.45 by the lungs (adjusting CO₂ within minutes) and the kidneys (adjusting bicarbonate over hours). ⚠️ It does not change with diet. A person whose blood pH moved meaningfully outside that range would be acutely ill — acidosis and alkalosis are medical emergencies with causes like respiratory failure, sepsis and renal failure, not breakfast.

The urine pH change is real and is the opposite of evidence for the claim — it demonstrates that your kidneys are doing their job, excreting the acid load and keeping blood pH constant.

On cancer specifically: tumours can create locally acidic microenvironments through their own metabolism. That is a consequence of tumour metabolism, not a cause of cancer, and it cannot be altered by diet.

📉 Evidence quality: Refuted by basic physiology. No supporting clinical outcome evidence.

Verdict: ❌ Not supported. ⚠️ And this one has caused documented harm — Robert Young, a prominent alkaline-diet promoter, was convicted of practising medicine without a licence after treating cancer patients.

The honest residual, because it explains why people report benefit: an "alkaline diet" in practice means more fruit and vegetables and less processed food and meat. That's a genuinely better dietary pattern (Chapter 10 §10.5). It works, for reasons entirely unrelated to pH — which is Chapter 3's bundled-intervention problem in yet another costume.

17.5 The single-food cure

Celery juice. Lemon water. Apple cider vinegar. Bone broth. Turmeric lattes.

The specific food rotates every eighteen months; the structure never changes.

🔬 Claim → Evidence → Verdict

The claim: "Sixteen ounces of celery juice on an empty stomach every morning heals chronic illness, autoimmune conditions, digestive problems and skin conditions."

Where it comes from: The celery juice claim specifically originates with Anthony William, who writes as the "Medical Medium" and who states that his information comes from a spirit voice. This is not an interpretation — it is his own stated account of his source. He has no medical or nutritional credentials, and his books have been enormous bestsellers.

What the evidence actually shows: There is no clinical trial evidence for celery juice treating any of the conditions claimed. Celery is a perfectly good vegetable — water, some fiber (mostly removed by juicing), potassium, vitamin K, some phytochemicals.

And people do report feeling better, which by now you can decompose without help: they've added a morning routine, increased vegetable intake, often replaced a sugary drink, are drinking more fluid, and are paying attention to their health for the first time in years. Bundled intervention, five changes, one credited.

📉 Evidence quality: Rung 3 testimonials only. Origin is explicitly non-empirical.

Verdict: ❌ Not supported. ⚠️ The specific harm here is not the celery. It is that these claims target people with chronic illness — autoimmune disease, thyroid conditions, long-term fatigue — who have often been failed by conventional care and who may delay or abandon effective treatment. That is the cost, and it falls on the people least able to bear it.

17.6 "Boosting" your immune system

🔬 Claim → Evidence → Verdict

The claim: "This supplement/food/protocol boosts your immune system."

Where it comes from: A real relationship — nutritional deficiency genuinely impairs immune function, and correcting it genuinely restores it. Zinc, vitamin A, vitamin D, protein and iron all have documented roles, and in malnourished populations correcting deficiency substantially reduces infectious disease. That is real and important.

What the evidence actually shows: The word "boost" describes a model that doesn't exist. Your immune system is not a dial with a volume setting. It is a vast, layered, tightly-regulated network that must distinguish threat from self, and that is calibrated to respond proportionately.

⚠️ An immune system that is "boosted" indiscriminately is not a healthier one — it is autoimmunity, allergy, or a cytokine storm. The therapeutic goal in most immune disease is suppression, and immunosuppressants are among the most consequential drugs in medicine.

So the claim is not merely unsupported. It describes a state you would not want.

What genuinely helps immune function: correcting an actual deficiency · adequate sleep · not smoking · managing chronic stress · physical activity · vaccination, which is the only intervention that meaningfully and specifically improves your defence against particular pathogens.

📉 Evidence quality: The deficiency-correction relationship is well established. "Boosting" in the replete is unsupported and conceptually incoherent.

Verdict: ❌ Not supported as usually meant. 🟢 Correcting a genuine deficiency does improve immune function — which is Part III's entire pattern, one more time.

17.7 Negative-calorie foods and metabolism folklore

The claim: some foods — celery, cucumber, grapefruit, ice water — take more energy to digest than they contain.

The arithmetic: a stalk of celery contains roughly 6 kcal. The thermic effect of food for a carbohydrate-and-fiber food is around 5–10%, so digesting it costs perhaps 0.5 kcal.

Net: about +5.5 kcal. Not negative. Not close.

The ice water version is at least arithmetically checkable: warming 500 mL of ice water to body temperature costs on the order of 17 kcal. Real, tiny, and not a strategy.

🔬 Verdict: ❌ Not supported. Chapter 4 §4.5 and Chapter 5 §5.8 cover the machinery. The general defence, from Chapter 5: convert the claim to calories and compare it to one beer. It defeats this entire category in about ten seconds.

17.8 The blood type diet

🔬 Claim → Evidence → Verdict

The claim: "Your ABO blood type determines which foods suit you. Type O should eat like a hunter; type A should be vegetarian; eating the wrong foods for your type causes disease."

Where it comes from: Peter D'Adamo's 1996 book, which proposed that blood type antigens interact with dietary lectins, and that blood types emerged at different points in human history corresponding to different diets. It is enormously popular and has sold millions of copies.

What the evidence actually shows: The historical premise is wrong — ABO blood types long predate the divergence D'Adamo describes, and are found across human populations and in other primates.

And the diets have been tested directly. A study published in PLOS ONE in 2014 examined whether adherence to blood-type diets was associated with cardiometabolic risk markers, and whether that association differed by actual blood type. It found that adherence to some of the diets was associated with better markers — and that the association was completely unrelated to the person's blood type.

Read that carefully, because it's an unusually clean refutation. The "type A" diet is largely a plant-based diet; the "type O" diet is largely a lower-carbohydrate, higher-protein diet. Both are reasonable diets, and both produced benefits in people of every blood type. The diets work approximately as well as their content predicts and the blood type contributes nothing.

📉 Evidence quality: Pre-specified test of the central hypothesis. Negative.

Verdict: ❌ Not supported. A rare case where a popular claim was tested exactly as it should be and failed exactly as predicted — and where the diets themselves still "work," which is why it survives the refutation.

17.9 The contamination cluster — where real concerns hide inside false claims

Three claims that need separating carefully, because each contains a genuine issue wrapped in an inaccurate one. This is the hardest section in the chapter and the most useful.

Microwaves

🔬 Claim → Evidence → Verdict

The claim: "Microwaving destroys nutrients and changes food at a molecular level."

Where it comes from: Microwave radiation sounds alarming; the word "radiation" does most of the work. And it's true that heating food does degrade some nutrients — particularly vitamin C and some B vitamins.

What the evidence actually shows: Nutrient loss during cooking depends on time, temperature, and contact with water. Microwaving typically uses shorter times and much less water than boiling — which means it often preserves water-soluble vitamins better than boiling does. Studies comparing cooking methods for vegetables have repeatedly found microwaving and steaming retaining more vitamin C than boiling, which leaches it into water you then pour away.

Microwaves are non-ionizing radiation — they cause water molecules to rotate, generating heat. They do not make food radioactive and do not break chemical bonds in the way ionizing radiation does.

📉 Evidence quality: Direct comparative studies of cooking methods and nutrient retention.

Verdict: 🟠 Probably false — and backwards. For many vegetables, microwaving preserves more nutrients than boiling. ⚠️ (The genuine cautions are mundane: uneven heating can leave cold spots where bacteria survive, so stir and check; and don't microwave in containers not designed for it — which is §17.9's next item.)

Non-stick cookware

🔬 Claim → Evidence → Verdict

The claim: "Non-stick pans leach toxic chemicals into your food."

Where it comes from: A genuinely legitimate concern, imprecisely stated. PFOA — a perfluorinated compound formerly used in manufacturing non-stick coatings — is persistent, has been detected widely in human blood, and has been associated with health effects. It was phased out of US production by 2015 under an EPA stewardship programme.

What the evidence actually shows: Two things need separating. PTFE (the coating itself) is chemically stable at normal cooking temperatures and, if ingested as flakes from a scratched pan, passes through largely inert. ⚠️ The real hazard is overheating — an empty pan heated above roughly 260°C can release fumes causing polymer fume fever, a transient flu-like illness, and these fumes are lethal to pet birds.

And the broader PFAS issue is genuine and much bigger than cookware. Per- and polyfluoroalkyl substances are persistent environmental contaminants found in water supplies, food packaging, firefighting foam and textiles, and regulatory bodies have been progressively tightening limits. That's a real public health issue, and your pan is a small part of it.

📉 Evidence quality: Strong on PFAS as environmental contaminants; weak on intact non-stick cookware as a meaningful exposure route at normal temperatures.

Verdict: 🟡 Unclear / it depends — and the "depends" is on which claim you mean. "Your non-stick pan is poisoning you" is not supported. "PFAS are a genuine and under-addressed environmental contaminant" is well supported. ⚠️ Practically: don't preheat empty non-stick pans, replace badly scratched ones, ventilate, and don't keep birds in the kitchen.

Plastics and BPA

🔬 Claim → Evidence → Verdict

The claim: "Plastic food containers leach hormone-disrupting chemicals. Never microwave in plastic."

Where it comes from: Bisphenol A (BPA) has demonstrated oestrogenic activity in laboratory systems, it does migrate from some plastics into food (increasing with heat, acidity and wear), and it is detectable in the urine of most people tested. None of that is fringe.

What the evidence actually shows: This is the one item in the chapter where expert bodies genuinely disagree, which is why it gets a 🟡 rather than a verdict.

EFSA substantially lowered its tolerable daily intake for BPA in 2023, to a level far below its previous figure — a change that implies typical exposure in some populations may exceed it. The FDA has maintained that BPA is safe at currently occurring levels, citing its own review of the evidence. Two competent bodies, the same literature, materially different conclusions.

Meanwhile "BPA-free" replacements — BPS, BPF — are structurally similar, less studied, and there is emerging evidence of comparable activity. "BPA-free" is a label change, not necessarily a safety improvement.

📉 Evidence quality: Substantial mechanistic and exposure data; genuine and unresolved disagreement about the risk at real-world exposures.

Verdict: 🟡 Unclear / it depends — a live regulatory disagreement, not a myth. ⚠️ Practically, and cheaply: don't heat food in plastic (use glass or ceramic), replace scratched or cloudy containers, and don't put plastic in the dishwasher repeatedly if you're minimizing. These cost nothing and are reasonable under uncertainty — which is a different posture from either alarm or dismissal.

💡 Aha moment. These three belong together because they share a structure that the rest of the chapter doesn't.

Detoxes, superfoods, alkaline diets and blood-type diets have no true version. The claim is wrong at the root.

The contamination cluster is different: each contains a real concern that has been converted into an inaccurate one. PFAS genuinely matter; your intact pan mostly doesn't. BPA is genuinely unresolved; microwaving your leftovers in glass costs nothing. Microwaves genuinely change nutrient content; they do it less than boiling.

And this is the harder skill. ⚠️ Dismissing the false version can make you dismiss the true one — which is exactly the failure a sceptical reader is most prone to, and why this section is the one I'd most want you to reread.

17.10 The ones that turned out to be true

A chapter of pure debunking would fail the test this book keeps applying to everyone else. So here are four folk beliefs that the evidence supports — including two that are counterintuitive.

🔬 Claim → Evidence → Verdict

The claim: "Cooked tomatoes are better for you than raw ones."

What the evidence shows: True, for lycopene specifically. Cooking breaks down plant cell walls and converts lycopene to a more bioavailable form. Studies measuring blood lycopene after cooked versus raw tomato consumption consistently find higher absorption from cooked, and higher still when eaten with fat, since lycopene is fat-soluble (Chapter 3 §3.4).

Verdict: ✅ Well supported — and a useful corrective to the assumption that raw is always superior. (It doesn't generalize: vitamin C is heat-labile and does decline with cooking. Different nutrients, different directions, which is why "raw is better" is as empty a generalization as its opposite.)

🔬 Claim → Evidence → Verdict

The claim: "Ginger helps with nausea."

What the evidence shows: One of the better-supported traditional remedies. Randomized trials and systematic reviews have found ginger effective for nausea and vomiting in pregnancy, where it is included in several clinical guidelines as an option, and there's reasonable evidence for postoperative and motion-related nausea. Effects on chemotherapy-induced nausea are more mixed.

Verdict: 🟢 Probably true, particularly for pregnancy-related nausea. ⚠️ Discuss with a clinician in pregnancy; dose and formulation vary, and "natural" doesn't mean unlimited (Chapter 16 §16.5).

🔬 Claim → Evidence → Verdict

The claim: "You should rinse rice before cooking it."

What the evidence shows: True, and for a better reason than most people know. Rinsing removes surface starch — which is the culinary reason — and rinsing plus cooking in excess water meaningfully reduces inorganic arsenic content. Rice takes up arsenic from soil and water more than most crops, and inorganic arsenic is a recognized carcinogen. Regulatory bodies including the FDA and EFSA have issued guidance, with particular attention to infants and young children, for whom rice products can be a large share of intake relative to body weight.

Verdict: 🟢 Probably true — rinse, and cook in plenty of water and drain if you want to reduce it further. ⚠️ Particularly relevant for rice-based infant foods and for people eating rice as a staple several times daily.

🔬 Claim → Evidence → Verdict

The claim: "Chicken soup helps when you have a cold."

What the evidence shows: Weakly but genuinely supported, and not for mystical reasons. Hot fluid and steam ease congestion; the fluid addresses losses from fever; it's palatable when appetite is poor; it supplies sodium and some protein; and there is a small experimental literature suggesting mild effects on neutrophil migration in vitro. The effect sizes are modest and the mechanism is mostly boring.

Verdict: 🟢 Probably true, modestly. It does not shorten the cold — but "makes you feel better while you have it" is a legitimate outcome, and one this book should be willing to count.

💡 Notice what these four have in common, because it's the mirror image of §17.1's shapes.

Each makes a small, specific, dosed claim about a mechanism that can be checked — lycopene bioavailability, nausea, arsenic content, symptom relief. None claims to cure anything. None requires a category of person. None has a suppressed-truth story.

Modest claims survive scrutiny at a much higher rate than grand ones — which is not a coincidence and is close to a general rule.

17.11 The anatomy of stickiness

Why these ideas, when thousands of equally wrong ones never catch on?

1. They resolve uncertainty. Nutrition is genuinely confusing (Chapter 1). A myth offers what the science can't: a clear answer. "Eat alkaline" is actionable; "dietary patterns matter more than individual foods, and the evidence is moderate-certainty" is not.

2. They come with a visible confirmation. Urine pH strips. Ketone meters. Beta-alanine tingling. Feeling better on day three of a cleanse. A perceptible signal that something is happening beats an invisible statistical benefit every time.

3. They provide agency. Chronic illness, ageing, and risk are frightening precisely because they feel uncontrollable. A protocol restores the feeling of control, and that is genuinely valuable even when the protocol isn't.

4. They confer identity and community (Chapter 10 §10.10). Something to be, and people to be it with.

5. They're morally legible. Clean and dirty. Virtue through sacrifice. This maps onto intuitions far older than nutrition science, and it's why the vocabulary is religious rather than technical.

6. Every advocate has a genuine personal result. Because these interventions are bundled, they work — through mechanisms nobody credits. Nobody in this chapter is lying about how they feel.

7. And they're unfalsifiable, or nearly. Unnamed toxins. Suppressed truths. "You didn't do it properly." A claim that cannot fail cannot be corrected.

17.12 How myths die — and why they mostly don't

Occasionally one does, and the mechanism is worth knowing.

Trans fats died (Chapter 9 §9.3) — evidence accumulated, regulation followed, the substance left the food supply. Note what did the work: regulation, not persuasion.

Protein combining died, slowly, over forty years, after its own author retracted it (Chapter 8 §8.5).

Ephedra died after documented deaths and a ban.

And the eight-glasses rule hasn't died, despite a definitive debunk in 2002, because nothing forced it to.

The pattern: myths die when a regulator removes the substance, or when the harm becomes visible enough to be counted. They very rarely die from being refuted.

Which is Chapter 12's asymmetry again, and it sets the realistic expectation for §17.13: you are probably not going to argue anyone out of anything.

17.13 What to do when someone you love believes one

The practical section, and the most requested thing in my clinic.

First, decide whether it matters. ⚠️ A ranking:

Then, if it matters, the approach that actually works — in my experience, which is rung 3 evidence about myself:

Don't debunk. Compare. Chapter 11's Case Study 1 — the winning argument with Walt wasn't "berberine doesn't work," it was a table showing the same outcome for a fifth of the cost with better evidence. Comparison beats debunking because it doesn't require the person to have been wrong.

Ask about the mechanism, not the conclusion. "How does it work?" — genuinely curious, not rhetorically. People frequently discover the gap themselves, and a gap you find yourself is durable in a way a gap you're shown isn't.

Decompose the bundle. "What else changed when you started?" Theo got to "probably the beer" on his own (Chapter 3, Case Study 1) and acted on it because he'd arrived there himself.

Concede the true part first, and mean it. Almost every claim in this chapter has one. Alkaline diets do mean more vegetables. The immune claims do rest on a real deficiency relationship. PFAS are genuinely a problem. A person who has heard you concede something will listen to the rest.

Never make it about intelligence. Theo has a folder, two highlighter colours, and eight of ten beliefs wrong. Chapter 1's whole argument is that being careful in a hostile information environment produces exactly this — and saying so out loud is the single most useful sentence available.

A worked version, because the principles are easy and the execution isn't. Your sister has started an alkaline protocol for fatigue and is spending about $140 a month on it.

❌ "That's nonsense, your blood pH doesn't change." True, useless. She now has to choose between the protocol and agreeing she was stupid, and you've given her no reason to prefer the second.

✅ "How's it going? What's actually changed since you started?" Then listen properly.

✅ "That's interesting — because that list is mostly more vegetables and less takeaway. Which of those do you think is doing it?" Decomposing the bundle, without contradicting anything.

✅ "The pH part I don't think holds up — your kidneys keep blood pH fixed, that's why the urine strips change. But the food part is genuinely good, and it's the part that costs nothing." Concede, correct narrowly, separate the $140 from the benefit.*

Notice the structure: nothing she experienced is denied, and the only thing challenged is the mechanism — which is the part she has the least investment in, and the part that's costing her money.

And accept that you will usually fail (§17.12). What you're actually doing is leaving a marker — so that when their own experience eventually contradicts the belief, there's an alternative explanation already sitting there. That's a slower and more realistic goal than winning.

What we don't know

We don't know why some people update and others don't, and it doesn't track intelligence, education, or numeracy in the way you'd expect. It appears to have more to do with identity investment and social context (Chapter 10 §10.10) than with evidence literacy — which is uncomfortable for a book that has spent seventeen chapters teaching evidence literacy.

And I don't know how much this chapter will change anyone's mind, including yours. Theo updated — but he updated over seventeen months, in a clinical relationship, with a food diary and lab results that contradicted him personally. That is not what reading a chapter provides, and I'd rather say so than pretend otherwise.

17.14 The ten-minute check — for a claim that doesn't exist yet

The chapter's actual deliverable. A protocol you can run on anything, in about ten minutes, without access to a journal database.

Minute 0–1: Restate the claim so it could be wrong.

Most claims arrive in a form that can't fail. "Seed oils are inflammatory." Restate it: who, consuming how much, of what, instead of what, with what measured outcome, over how long?

⚠️ If you cannot restate it in that form, stop — that's the finding. A claim that can't be restated as a testable comparison isn't a weak claim; it isn't a claim. Roughly a third of what you encounter dies here, which is why it's the first minute rather than the last.

Minute 1–3: Run the tells (§17.1).

Dose · comparator · moral vocabulary · difficulty-as-evidence · falsifiability · specified past · mechanism-without-dose · count the changes. You are not deciding whether it's true. You are deciding how much scrutiny it's worth, which is a different and much faster question.

Minute 3–5: Locate the rung (Chapter 2 §2.4).

Where does the underlying evidence sit? Mechanism? Rodent? Testimonial? Cross-sectional? And critically: what rung does the claim need? "Prevents cancer" needs rung 6–8. "Increases lycopene absorption" needs rung 5. ⚠️ The gap between the rung supplied and the rung required is the single most diagnostic number available to a non-specialist.

Minute 5–7: Check the denominator and the direction (Chapter 7 §7.7, Chapter 12 §12.4).

Compared to what? Replacing what with what? And who is the comparison group — are they people who do lots of other healthy things (Chapter 2's healthy-user bias), or are they genuinely comparable?

Minute 7–9: Ask three questions about the source (Chapter 1 §1.5).

What happens to their income if you decline? Do they ever say "I don't know" or "I was wrong"? Do they name a dose? ⚠️ The second is the most predictive of the three, and the cheapest to check — scroll their last twenty posts. Someone who has never publicly updated is not a source; they're a position.

Minute 9–10: Assign the verdict — including the two you'll want to skip.

✅ 🟢 🟡 🟠 ❌ ⚗️. And be honest about ⚗️ (untested) and 🟡 (it depends), which are the correct answer far more often than either enthusiasts or sceptics find comfortable. A ❌ is a claim about the evidence. Reaching for one when the honest answer is 🟡 is the same error as the claim, with the sign flipped.

The asymmetry to keep

⚠️ This protocol is much better at rejecting than at accepting, and you should know that about it.

Ten minutes can establish that a claim is unsupported, over-extended or unfalsifiable. Ten minutes cannot establish that something is true. The four claims in §17.10 that hold up are supported by literatures you did not read in ten minutes; I'm asking you to take those on the same trust you extend to any other reference work, and to check the further-reading if it matters to you.

So the honest description of what you now have is a good filter and a weak detector — which is enough, because the base rate is heavily against the claims arriving at you, and a filter is what that situation calls for.

Spaced Review

Answer before reading on.

1. (Chapter 3) Why do cleanses reliably make people feel better despite the mechanism being unsupported?

Bundled intervention. Theo's ten days changed eight things at once — alcohol to zero, sleep 6 → 7.5 hours, ultra-processed food to zero, ~1,200 kcal/day, caffeine down, no late eating, and sustained attention. A protocol that changes eight things can be believed in but not evaluated, and the active ingredients are free and permanent.

2. (Chapter 13) The "boost your immune system" claim rests on a real relationship. Which one, and where does the inference fail?

Correcting genuine deficiency does improve immune function — zinc, vitamin A, vitamin D, protein, iron. The inference fails at Chapter 13's threshold: supplementing someone who isn't deficient doesn't help, and the dose-response curve plateaus. Plus a conceptual failure — "boosting" an already-calibrated immune system describes autoimmunity, not health.

3. (Chapter 2) The blood type diet study is described as an unusually clean refutation. Why?

Because it tested the central hypothesis directly: not "do these diets work?" but "does the benefit depend on blood type?" It found benefit from some diets unrelated to the person's actual blood type — which isolates and refutes the specific claim while explaining why adherents report results. That's a well-designed test of a mechanism rather than of an outcome.

Project Checkpoint: Your Myth Audit

Component seventeen. Go and get the Belief Inventory you sealed in Chapter 1.

If you didn't write one — that's information too. Write down ten things you believe about food now, and note that you'll be grading them without a baseline, which is a weaker version of this exercise.

Step 1 — Read it before you grade it. All ten, out loud if you can. Notice your reaction to each one before you decide anything.

Step 2 — Grade each belief.

"Still don't know" is a legitimate grade and you should expect several. This book has not covered everything.

Step 3 — The correlation. This is the point of the exercise.

Of the beliefs I rated 5, how many were wrong or substantially overstated? _ / _ Of the beliefs I rated 1–2, how many were wrong? _ / _

For most readers — and for Theo — confidence and accuracy are inversely related. If that's true for you, sit with it, because it's a fact about how you form beliefs, not about nutrition, and it will still apply to claims that don't exist yet.

Step 4 — Trace the sources. Go back to the "where I got it" column. Is there a pattern? One outlet, one person, one era? Most people find two or three sources responsible for most of the errors.

Step 5 — The prediction check. In Chapter 1 Exercise M3 you predicted which belief was most likely to be wrong. Were you right? Most people aren't — and the direction of the error is usually that they suspected the unfamiliar belief and were let down by the familiar one.

Step 6 — Reseal, and add a date. Add any beliefs you now hold that you didn't then — including ones you got from this book. Write today's date on it and put it back in the drawer.

Then set a reminder for three years. ⚠️ Some of what this book told you will be wrong by then, and the exercise works exactly the same way with me as the source.

Next checkpoint (Chapter 18): added-sugar tracking — find your top three sources and decide about each.

Chapter Summary

Eight recurring shapes, each with a structural tell that requires no nutritional knowledge:

This chapter's verdicts:

⚠️ The hardest section is §17.9, where genuine concerns hide inside inaccurate claims. Dismissing the false version can make you dismiss the true one — the failure a sceptical reader is most prone to.

And §17.10's four true beliefs share a shape: small, specific, dosed, checkable claims about a mechanism. Modest claims survive scrutiny at a far higher rate than grand ones.

Myths die from regulation and counted harm, almost never from refutation — so when someone you love believes one: triage by harm · compare rather than debunk · ask about mechanism · decompose the bundle · concede the true part · never make it about intelligence · and expect to fail while leaving a marker.

The one thing to remember: four of the five beliefs Theo rated maximum confidence were wrong. His confidence was inversely correlated with his accuracy — and that's a fact about belief formation, not about nutrition.

What's Next

Chapter 18 takes on the first of Part IV's genuine debates — and the difference from this chapter matters. These weren't arguments. Sugar is.

Is sugar toxic? Added versus natural. Sucrose, glucose, fructose and HFCS compared without theatrics. What the liver actually does with fructose, and where the "sugar is like alcohol" analogy holds and breaks. Sugar-sweetened beverages as the one genuinely distinct case.

And the Sugar Research Foundation episode — real, documented industry influence that is routinely asked to carry far more than it can.

It carries the threshold concept the whole of Part IV depends on: the dose makes the poison.