Chapter 37 — Further Reading

⚠️ A synthesis chapter's further reading is mostly the book you have just read. ⚠️ What follows is therefore short, and is organized around the four things this chapter claims rather than around topics.

1. ⚠️ The claim that the answer is stable — check it yourself

⚠️ National dietary guidelines from three or four countries, in their current editions.

⚠️ Read the summary recommendations only, side by side. ⚠️ They are free, they are short, and the degree of agreement between countries with different food cultures, different agricultural interests and different political systems is the most direct test of §37.2's claim available to a reader.

⚠️ Then read an edition from twenty or thirty years ago, if you can find one.

⚠️ What you are checking: how much of the core changed, and how much of what changed was the removal of things that failed rather than the addition of things that were discovered. ⚠️ §37.2b predicts mostly the former.

⚠️ This is the single most useful hour on this page, and almost nobody does it.

⚠️ WHO population dietary guidance — ⚠️ for the version stripped of national agricultural interests, which is a useful comparison.

2. The convergence

⚠️ The two primary sources are in Chapters 35 and 36 and are not repeated in full here:

⚠️ Zeevi/Segal 2015 and the PREDICT work for route one (Ch 35's page). ⚠️ Poore and Nemecek 2018 and the EAT-Lancet Commission for route two (Ch 36's page).

⚠️ What to read them FOR, in this chapter's context, is different from why Chapters 35 and 36 cited them.

⚠️ Read each one's final dietary recommendation, ignore everything else, and put the two lists next to a national dietary guideline. ⚠️ Three columns, three fields, three methods.

⚠️ §37.2's argument is that comparison. It takes twenty minutes and it is either persuasive or it isn't.

⚠️ And read at least one critique of EAT-Lancet (Ch 36's page lists the directions), ⚠️ because §37.2 concedes the convergence is only partial and the critiques are where you can see how partial.

3. ⚠️ The history — did the exciting claims really fail?

⚠️ §37.2b makes a historical claim and you should check it rather than take it.

⚠️ Histories of the dietary fat debate — ⚠️ and read at least one from each direction, because this is a genuinely contested history with partisans, and reading only one side produces a confident and wrong account.

⚠️ The antioxidant supplement trials — ATBC and CARET (Ch 2, Ch 13). ⚠️ The clearest documented case of an exciting mechanism producing a failed and harmful intervention, and the template this book has used repeatedly.

⚠️ The industrial trans fat story — ⚠️ read it as a case where the exciting replacement for butter turned out worse than what it replaced, and was then eliminated. §36.13b's timeline.

⚠️ Retrospectives on withdrawn guidance — ⚠️ particularly dietary cholesterol. ⚠️ Worth reading for how a recommendation is removed, which is much slower and quieter than how one is added.

4. ⚠️ Why the information environment works this way

⚠️ Research on science communication and the press-release pipeline — ⚠️ including studies tracing exaggeration from journal to press release to news story. ⚠️ Chapter 1's five-step pipeline, with evidence.

⚠️ Work on publication bias and the file-drawer problem (Ch 3) — ⚠️ the reason null results, which are most of §37.2b's failures, are so much less visible than the claims they refute.

⚠️ Ben Goldacre's Bad Science and Bad Pharma — ⚠️ not about nutrition specifically, and the best general training available in the reasoning §37.7 formalizes.

⚠️ And any work on the "backfire" and continued-influence effects — ⚠️ on why refuted claims persist after correction, which is §37.6b's subject and is better studied in political misinformation than in nutrition.

5. ⚠️ For the position between the two extremes

⚠️ §37.11 says both comfortable extremes are popular and both are wrong. This is the reading that occupies the middle.

⚠️ Methodological critiques of nutritional epidemiology — ⚠️ including the strongest ones. ⚠️ Read them, and notice that most of the critics still recommend approximately §37.3's plate. ⚠️ That is a more informative fact than either the critiques or the defences.

⚠️ Defences of nutritional epidemiology from within the field — ⚠️ particularly those addressing the triangulation argument, which is §37.2's structure.

⚠️ Reading a strong critic and a strong defender back to back is the fastest way to locate the honest middle, and it is uncomfortable in a productive way.

6. ⚠️ Do not read

⚠️ Any book whose thesis is that one nutrient, one food, or one mechanism explains modern chronic disease.

⚠️ §37.2b lists eight of these that failed. ⚠️ The next one will be well written, will cite real studies, and will fail at §37.7's question 4 — it will REPLACE the boring answer rather than adjust it.

⚠️ Any book promising that the science has been suppressed. ⚠️ Chapter 17 §17.9 covers the structure, and Chapter 18 §18.12 shows what a genuine documented industry episode looks like — which is the standard to hold suppression claims to.

⚠️ And any product page citing this chapter's convergence as support for a personalized programme. ⚠️ §37.2 is an argument for the population answer, not for a subscription.

7. Where this goes

⚠️ One recommendation, if you read nothing else

Read Chapter 1 of this book again, now.

⚠️ It takes half an hour, it is free, and the four contradictory headline pairs that opened it will read completely differently — ⚠️ which is the only test of whether thirty-seven chapters did anything.

⚠️ If they still look like a contradiction to be resolved rather than a phenomenon to be classified, something did not land, and Case Study 1 is where to look.