Chapter 35 — Exercises

Three tiers. 🔄 retrieval; 🪞 reflection.

⚠️ A note carried from §35.12: several exercises here involve self-monitoring products. ⚠️ If Chapter 34 §34.9's questions or §35.12's warning signals describe you, do the analytical exercises and skip anything involving a device or a log. That is a complete way to do this chapter.

Tier 1 — Recall and Comprehension

R1. ⚠️ State the three-part structure the chapter says every product in it shares. ⚠️ Which part is doing the selling?

R2. ⚠️ What did the Weizmann glucose-response work establish, and what did it NOT establish? ⚠️ How does it complicate the glycaemic index?

R3. ⚠️ Give §35.2's five-step chain and the verdict on each. ⚠️ Which step is doing all the work?

R4. ⚠️ Why is §35.2's structure the beta-carotene template? (Ch 2.)

R5. ⚠️ Give the verdicts on the six genetic claims in §35.3. ⚠️ State the pattern that runs through the table.

R6. ⚠️ What is the reproducibility finding for DTC genetic tests, and why is it damning?

R7. ⚠️ Describe the DIETFITS trial. ⚠️ Give its six design features and its four results. ⚠️ Why does result 2 matter as much as result 3?

R8. ⚠️ What does DIETFITS NOT show? Three things.

R9. ⚠️ Give four problems specific to consumer microbiome tests. ⚠️ Where IS microbiome analysis clinically useful?

R10. ⚠️ State §35.5's three honest positions on CGMs in non-diabetic people.

R11. ⚠️ Give the four interpretation errors in §35.5. ⚠️ Why does "flatter is better," pursued honestly, recommend a worse diet?

R12. ⚠️ Give the three rules that make CGM use safer.

R13. ⚠️ What did the blood-type diet study find, and why is it "the cleanest available demonstration"?

R14. ⚠️ Give §35.7's nine evidence-based bases for personalization. ⚠️ Which is the biggest lever, and which chapter established it?

R15. ⚠️ Name the four things that produce apparent "responders" in a trial. ⚠️ What design separates them?

R16. ⚠️ Give the seven rules for a proper n-of-1. ⚠️ Which does nobody do, and what does it separate?

R17. ⚠️ List §35.10's convergent recommendations. ⚠️ Give the three reasons the chapter offers for why they converge.

R18. ⚠️ Give §35.11's six things that would change the author's mind. ⚠️ Which one does almost every trial in the sector fail?

R19. ⚠️ Distinguish precision nutrition as a research programme from personalized nutrition as a consumer sector. ⚠️ What two developments does §35.11b say to watch for?

R20. ⚠️ Give §35.12's five warning signals. ⚠️ Why is a rule generated from your own biology harder to argue with than a conventional diet rule?

Tier 2 — Application

A1. 🔄 ⚠️ Find the join. Take three real advertisements for personalized nutrition products.

For each: a) ⚠️ What does it claim your body does differently? b) ⚠️ What does it claim to measure? c) ⚠️ What outcome does it claim will improve? d) ⚠️ Where is the join — which claim is unevidenced and doing the selling?

A2. ⚠️ Score a product. Pick one and run §35.11's six criteria.

⚠️ Most score zero or one. If yours scores three or more, bring it to a discussion — that would be interesting.

A3. ⚠️ Strip the personalization. Find a published example of a personalized report's recommendations.

a) ⚠️ Which are in §35.10's convergent list? b) ⚠️ Which are genuinely individual? c) ⚠️ Apply Yolanda's rule: what do two such reports DISAGREE about? d) ⚠️ Could the individual advice have been generated from age, sex, activity and one sentence about diet? (The circularity check.)

A4. ⚠️ Design an honest study. You want to know whether personalized glucose-based advice beats good conventional advice.

a) ⚠️ Population, arms and control condition. (Careful with the control — §35.11.) b) ⚠️ Primary outcome, and defend it against "why not a glucose curve?" c) ⚠️ Duration. d) ⚠️ What result would falsify the personalization hypothesis? e) ⚠️ Why has nobody run this?

A5. ⚠️ The n-of-1. Design one for a short-latency question that matters to you — caffeine and sleep, a food and symptoms, timing and training.

a) ⚠️ One variable, one pre-defined outcome. b) ⚠️ ABAB schedule with block lengths and a washout. c) ⚠️ Your prediction, written first. d) ⚠️ The result that would make you STOP. e) ⚠️ Can you blind yourself? If not, what does that cost you?

A6. ⚠️ Cost it. Take one product.

a) ⚠️ Annual cost including any supplements it recommends. b) ⚠️ How many weeks of Chapter 32 §32.13's food is that? c) ⚠️ What would you buy with the same money to improve your diet?

Tier 3 — Analysis and Synthesis

S1. ⚠️ The convergence. §35.10 says personalized outputs converge on conventional advice, for three reasons.

a) ⚠️ Which of the three do you find most important? b) ⚠️ Reason 2 — hard-coded conventional constraints — is uncomfortable. If a "personalized" system must be prevented from recommending cheese over lentils, what is being personalized? c) ⚠️ Does convergence disprove personalization, or is it what you'd expect from a young field?

S2. ⚠️ DIETFITS as a model. §35.3b calls it a hypothesis killed honestly.

a) ⚠️ What made the pre-specification so important? (Ch 3.) b) ⚠️ Both arms ate a good diet. How would the result have differed if one arm had eaten badly? ⚠️ What would have been concluded, and by whom? c) ⚠️ Individual variation was large and unexplained. What are the candidate explanations, and how would you test them?

S3. ⚠️ The CGM question. §35.5 gives three honest positions and holds all three.

a) ⚠️ Is "educational but unevidenced" a coherent recommendation? b) ⚠️ Theo said the money bought conviction rather than information. ⚠️ Is that a legitimate purchase? Price it. c) ⚠️ Should CGMs be available over the counter to people without diabetes? Argue both, and say what evidence would settle it.

S4. ⚠️ Unfalsifiable frameworks. Corinne's "you haven't eliminated enough yet."

a) ⚠️ What structural features make a framework unfalsifiable? b) ⚠️ Find two others in this book. (Chapters 17 and 21.) c) ⚠️ Is any part of THIS book unfalsifiable? ⚠️ Look properly before answering no.

S5. ⚠️ The ethics of selling ahead of evidence. §35.11b says the products arrived about a decade ahead of the evidence they cite.

a) ⚠️ Is that fraud, optimism, or normal commercial behaviour? b) ⚠️ What regulation would you write, and what would it cost in innovation? c) ⚠️ Compare with the supplement industry (Ch 16). ⚠️ Same problem or different?

S6. 🔄 ⚠️ The Chapter 34 overlay. §35.12 maps these products onto Chapter 34 §34.7's vector list.

a) ⚠️ How many of §34.7's seven rows does a CGM subscription commit? b) ⚠️ "Rules that are unfalsifiable because they are yours." ⚠️ Why is this worse than a conventional diet's rules? c) ⚠️ Should these products carry a warning? Draft it in under fifty words.

🪞 Reflection

M1. ⚠️ Have you bought, or wanted to buy, any of these? ⚠️ What were you hoping to find out? Would the answer have changed what you ate?

M2. ⚠️ §35.7 says preference, culture, budget and schedule are the biggest personalization lever. ⚠️ Does that feel like a real answer or like a consolation prize? Interrogate the reaction.

M3. ⚠️ Priya's report was accurate and circular. ⚠️ Where else do you accept a description of yourself as an explanation of yourself?

M4. ⚠️ Theo checked the device before eating rather than after. ⚠️ Is there anything you check before eating — an app, a number, a rule? What is it doing?

M5. 🔄 ⚠️ §35.10 concludes that the most sophisticated personalization available produces recommendations that look like the boring ones. ⚠️ How does that land? Disappointing, reassuring, or suspicious? ⚠️ Chapter 37 is going to make an argument out of your answer.