The Architecture of Persuasion

The same thirty techniques from the other side. Each chapter names the threat pattern, gives the signals that arrive before the reasoning does, cites a documented detection case, and ends with a protocol.

The protocols share a property that is worth stating in advance, because it is what separates the ones that work from the advice that does not. None of them ask you to assess the person. Every method here produces an artifact, imposes a delay, or applies a rule set before exposure — because the mechanisms in Part 10 work by degrading exactly the judgment that a case-by-case assessment would rely on.

That is why the chapters recommend fixed rules rather than vigilance. The pharmaceutical gift literature is the cleanest demonstration: physicians' prescribing moves after a twenty-dollar meal, and their self-reports do not, which means introspection cannot detect reciprocity pressure and only a rule can. The same logic runs through the ninety-day interval in chapter 301, the mandatory delay in 303, the pre-written number in 309, the out-of-band callback in 319, the family safe word in 320, and the payment-method rule in 321 — which works precisely because it requires no judgment about whether your grandson is really in jail.

Four chapters carry findings with unusually direct experimental support, and they are the ones to read first.

Chapter 306: Asch's dissenter cut conformity by three quarters, and did not have to be right. Unanimity is the vulnerability of every manufactured consensus in this guide, and being the first to speak is the cheapest intervention available to anyone.

Chapter 313: Harsey and Freyd found that teaching people what DARVO is reduces its effect on their judgments. Naming the pattern is the countermeasure.

Chapter 318: replace is he lying — a coin flip — with unanticipated questions, reverse-order recall, and withheld evidence, all of which produce something checkable.

Chapter 328: inoculation gets there before the claim does, which is the only way to beat repetition, and technique-based prebunking generalises where fact-specific debunking cannot.

Chapters 311, 317 and 320 concern situations where getting it wrong is dangerous rather than expensive. They are written accordingly, and 350 is the chapter on separation.