Chapter 42 — Exercises

How to use these. Thirty-six items, no answers provided. Items marked † are extension work: they take longer, most require you to go and find something in the world rather than reason from the chapter, and several have no single right answer.

Two standing rules for every item in this set.

Name nobody. When an exercise asks you to analyze real content, describe it by role and structure — "a fitness creator with an affiliate link," "a remote prescribing service" — never by name. This is not squeamishness. A structure you can describe generalizes to the next case; a name you can post does not, and posting it makes you part of the mechanism the chapter describes.

Incentive is not proof of dishonesty. Any answer whose conclusion is "therefore this person is lying" has skipped the analysis. The conclusions this chapter licenses are about which claims get made, not about who is sincere.

Part A — The funnel and its economics (§42.1, §42.2)

A. Reproduce the five-step funnel from §42.1 from memory. For each step, write one sentence naming who is paid and what they are paid for. Then check against the chapter and note which step you got wrong or omitted — the step people forget is diagnostic of how they think about the structure.

B. §42.1 argues that persuasion is concentrated at the top of the funnel and money at the bottom. Explain why that asymmetry is economically rational for the entity at the bottom, and what it predicts about the amount of care taken at step 3.

C. Define friction twice: once as a conversion optimizer uses the word, once as a patient-safety concept. Give two concrete examples of friction in conventional care that are protective, and one that genuinely is waste.

D. List four genuine access benefits of remote prescribing and four structural features that create volume pressure. Then write one paragraph holding both without resolving the tension. If your paragraph resolves it, you have written an opinion piece rather than an analysis.

E. † Find a remote prescribing service — any condition, any country — and, using only its public pages, answer the diagnostic question from the 💊 In the Clinic callout: what would have caused this service to decline to prescribe? Write down every disqualifying criterion you can find stated. Note where you had to guess, and how much of the page was devoted to speed.

F. §42.2 argues that vertical integration removes checks without anyone behaving badly. Explain the mechanism in your own words, then name one non-medical industry where the same integration is regulated or prohibited, and say what harm that rule was written to prevent.

G. Write the strongest version of the objection "this chapter is singling out telehealth unfairly," then write the chapter's best reply. Which is more persuasive to you, and what evidence would move you?

H. † Design a remote service you would consider genuinely well built. Specify clinician compensation, follow-up ownership, and the cancellation path. Then identify which of your choices would make it less profitable than its competitors, and by what mechanism.

Part B — Affiliates, disclosure, and selection (§42.3)

I. List the seven affiliate payment forms from §42.3. For each, state in one clause which audience behavior the money is attached to: curiosity, registration, purchase, or persistence.

J. Explain the difference between a payment structure that rewards starting a product and one that rewards continuing it. What content would you expect each to produce, and which is harder to recognize as commercial?

K. State the three things disclosure does not fix. Rank them by how much distortion you think each produces, and defend the ranking in a paragraph.

L. The selection effect is described as invisible within any individual piece of content. Explain why that is true, and why it follows that auditing individual videos for accuracy — even a very large number of them — cannot detect it.

M. † Pick a consumer category outside health entirely: kitchen equipment, running shoes, software, whatever you know well. Estimate how many negative assessments of a heavily promoted product you have seen from creators who take sponsorships in that category, versus how many positive. Then write down what your estimate is and is not evidence of.

N. Rewrite the ⚠️ Hype Check claim ("I only recommend things I actually use") into the strongest, most honest version a creator could truthfully say. Then identify what remains unaddressed even in your improved version.

O. † Draft a disclosure format you think would actually work: placement, wording, timing, and what it must survive (clipping, reposting, muted playback). Then state honestly which of the three unfixable problems from §42.3 your format still fails to touch.

Part C — Content and platform economics (§42.4, §42.5)

P. Take a well-formed evidential sentence from an earlier chapter — one with a population, an endpoint, a comparator, and a duration — and run it down the five-rung compression ladder yourself. At each rung, write which qualifier you dropped and why dropping it improved the content's performance.

Q. Explain, without using the word "algorithm," why a hedged claim is penalized by a system that ranks on watch time.

R. §42.4 says a transformation narrative "never has to state a causal claim." Explain how the causal inference gets made anyway, and why that matters for whether the content could be described as deceptive.

S. Trace the mechanism by which moderation exerts selection pressure toward imprecision, in four steps, from "a keyword is penalized" to "the circulating claim is less checkable."

T. Explain why the §42.5 argument does not depend on the diffusion research in the 🔬 Read the Study callout. What kind of claim is each — architectural or empirical — and why does the distinction change how confidently you can hold each?

U. † Find one long-form piece of health content and one clip taken from it by a third party. Compare the claims. Write down exactly what was removed, and rate the claim as it appears in each version using this book's system. Note whether the two versions deserve different ratings, and what that implies about rating "a claim" at all.

V. Chapter 5 ranks confidence, novelty, and personal narrative low as evidence. §42.5 says engagement optimization selects for exactly those three. Coincidence, or common cause? Argue for one.

Part D — Pipelines, regulation, and the compounding case (§42.6–§42.8)

W. Explain what a displayed credential is doing in a monetized channel, distinguishing carefully between (i) signaling real expertise, (ii) transferring authority beyond its scope, and (iii) deceiving anyone. Which of the three requires intent?

X. Set out the three options the chapter describes for a clinician-creator and the cost of each. Then say what you would do, and what your answer costs.

Y. Reconstruct the §42.7 table from memory: five entity types, what each is legally selling, and how tightly its efficacy claims are constrained. Then state the chapter's thesis sentence about why the gaps exist.

Z. "The absence of an approval made the advertising less constrained, not more." Explain the mechanism to someone who finds this obviously backwards. Use no more than 150 words.

AA. † Take the six-step template from §42.8 — demand spike, shortage, legal opening, supply, manufactured demand, opening closes — and apply it to a non-peptide category where a regulatory gap opened and closed. Assess how well the template transfers and where it breaks.

AB. For each of the five marketing terms examined in §42.8's vocabulary layer, write the accurate replacement phrase: the wording that says exactly what is true and nothing more.

AC. State the constraints under which patients using compounded preparations were, per §42.8, behaving rationally. Then explain why an account that calls those patients gullible will also make bad predictions about what happens next.

AD. † §42.8 predicts that a marketing layer migrates rather than dissolving when a regulatory opening closes. Design a test of that prediction using only public information, and state in advance what result would falsify it.

AE. Explain the "invisible denominator" from the 🩺 Safety and Risk callout in terms a non-scientist would follow, using an example from outside medicine entirely.

Part E — The read, and the dossier

AF. Apply the five-question read to a piece of health content you encountered this week. Write the five answers. Then write the sixth thing the read does not entitle you to conclude.

AG. † Apply the five-question read to a piece of health content you agreed with before you started. Report honestly what it cost you.

AH. Distinguish the genetic fallacy from a legitimate inference about a selection effect, using a worked example of each. Then state which of the two errors — treating a conflict as a refutation, or treating an absence of conflict as a warrant — you are personally more prone to, and how you know.

AI. † Build the interest map for one dossier compound, following all three steps of the 📋 Your Evidence Dossier section. Complete Step 2 in full: find three claims made by parties with no financial interest, and at least one result published against the interest of whoever produced it. Record how long Step 2 took compared with Step 1.

AJ. † Write the paragraph you would send to a family member who has just started a treatment through a funnel you consider poorly designed. Constraints: no accusation, no naming, nothing that tells them to stop a medication, and short enough that they actually read it. This is much harder than it looks, and the difficulty is the exercise.