Discussion Guide — Chapter 5

1. "Confidence is free. What actually costs something?"

Open with this. Listen for: money, years, and — the one students reach last and that matters most — a real chance of being proven wrong. A claim that cannot fail costs nothing to make. Build toward the observation that the willingness to specify a falsifier is what distinguishes a position from a belief.

2. "Is a hundred animal studies worth one human trial?"

Listen for: students separating research activity from evidential weight. Push on the follow-ups: are they independent? Same lab, same model? Do they test the claim being made? And — the hardest — how many were run and not published? The absence of an animal-study registry is the point, and students who grasp it will read Part III completely differently.

3. "Why can't the honest version of a trial result be a headline?"

Genuinely open and worth twenty minutes. Listen for: the recognition that this is structural rather than a matter of anyone's bad faith. The honest sentence is three times as long and contains two numbers and a population qualifier. Then ask what would fix it — and watch students discover that most of the proposed fixes require the reader to do more work, which is exactly what does not happen at scale.

4. "CAST: the trial was undertaken to confirm what everyone already knew. Should it have been run?"

Listen for: the ethical dimension — there was real debate about whether a placebo arm was defensible, and in hindsight the placebo arm was the safer arm. Then the harder question: how do you know, in advance, which confident belief is the one that needs testing? There is no clean answer and students should not be given one.

5. "For a compound with no patent, no regulatory obligation, and an established market — who funds the trial that could show it doesn't work?"

The most important question in the chapter and the one that decides how students read Part III. Listen for: the recognition that self-correction is not automatic; it requires that somebody be able to afford it. Then guard against the cynical overcorrection — the answer is not "therefore all evidence is corrupt," it is "therefore the absence of a trial is sometimes about economics rather than about the compound." Both readings are available and only one is useful.

6. "You wrote a rating for something you hope works. What happened?"

Only in a group with enough trust, and only if you have made clear that honesty is what is being assessed. Listen for: the falsifier problem — students discovering they could not specify what would change their mind. That discovery is the chapter's real deliverable. Do not resolve their discomfort. Note that nothing about it means the hope is wrong; it means they now know which of their beliefs are load-bearing.