Discussion Guide — Chapter 11

1. "Insulin was proven without a randomized trial. What does that tell us about trials?"

Open here, and expect the wrong answer first. Listen for: "we over-rely on trials." Then run the scaling principle and ask them to name modern conditions where the same reasoning would hold. The list is short — untreated conditions with near-total short-term mortality and dramatic reversal — and building it is more persuasive than being told.

2. "Is hypoglycemia a dosing failure?"

Listen for: the shift from "bad technique" to "structural." Then the payoff question: if it's structural, what would actually fix it? Students arrive at glucose-responsiveness — which is exactly what "smart insulin" would provide, and why it is the one genuinely different innovation on the horizon rather than an incremental one.

3. "Sort these six therapies into replacement and override."

Run it as a group. Listen for: the arguments about the ambiguous cases — insulin in type 2 is the interesting one, and testosterone replacement produces good disagreement. Then ask each group to predict discontinuation behavior. This is the most valuable fifteen minutes in the chapter for students who will read Part III.

4. "The DCCT confirmed what everyone already believed. Was it worth running?"

Listen for: students initially saying no. Then introduce CAST and ask what would have distinguished them in advance. Then point at the hypoglycemia finding. Students who come out of this able to say "you cannot know in advance which case you are in, and the cost was unmeasured until somebody measured it" have understood something they will use.

5. "They sold the patent for a dollar. So why does insulin cost what it costs?"

Listen for: the simple story, then let the room correct itself with the three complications. The teaching moment is that the correction does not exonerate anyone — the harm is real, the deaths are documented, and the accurate account locates the cause more precisely and therefore suggests fixes the simple version cannot. Model correcting a story you are sympathetic to.

6. "Which of insulin's five features will the GLP-1 drugs share?"

Close on Case Study 2's forward question. Have them write predictions down. Listen for: whether anyone notices that GLP-1 drugs do not share the "cannot be stopped" property — and what follows. The insight worth reaching: insulin's rationing deaths are terrible and legible; the GLP-1 version may be larger and almost entirely invisible. Chapter 12 checks the prediction.