Obedience to Authority

What is this book about?

In 1961 Stanley Milgram began a series of experiments at Yale that would unsettle his discipline. Subjects were told they were taking part in a study of punishment and learning, and were instructed to administer escalating electric shocks to a learner in the next room, beginning at 15 volts and rising one step with each error, up to 450 volts. The learner was an actor, who protested, cried out and finally fell silent. The experimenter, in a grey lab coat, pressed the subject on with four fixed prods. The results surprised everyone: in the baseline condition, 26 of 40 subjects went all the way to the highest switch. Published in 1974, the book first sets out the design and baseline result in full, then reports the variants that changed one situational variable at a time — the distance of the victim, the presence of the experimenter, the presence of a rebellious peer, conflicting authorities — and closes with a theory: the agentic state and the authority system.

Why read it?

Because it recasts evil from a question of character into a question of situation, and that recasting has direct consequences for institutional design. The valuable part is not the much-quoted 65 per cent but the variants: shift a single situational variable and the rate of obedience moves sharply, which shows how sensitive conduct is to structure. For managers, safety engineers, auditors and compliance officers it is an empirical handbook on why intelligent and decent people do wrong things inside a system. It also deserves reading as a methodological case: an experiment of extraordinary cleanliness that left a long-running ethical controversy.

Core Ideas

• Obedience is less a defect of character than a product of situation: once people define themselves as instruments executing another's wishes, they enter an agentic state and stop evaluating their conduct by their own moral standards.
• The power of authority comes from perceived legitimacy and the right to define the situation, not from coercion — the authority in the experiment was a stranger in a lab coat.
• Strain is nearly universal, but most people dissolve it by displacing responsibility, devaluing the victim or ritualising the procedure rather than stopping.
• Disobedience usually needs company: once a peer openly refuses, obedience falls sharply. Isolation is a key condition for maintaining it.
• Distance governs cruelty: the less visible and audible the victim, the easier it is to continue; physical contact and face-to-face presence weaken obedience markedly.
• Responsibility can be divided: break an act into one small step within a hierarchy and no single person feels responsible for the outcome.

What questions does this book try to answer?

• Why do ordinary people keep harming a plainly innocent person when no direct coercion is applied?
• How is individual responsibility divided, displaced and finally dissolved inside a bureaucracy?
• Can a laboratory situation be extrapolated to historical atrocities? Where are the limits of that extrapolation, and which conditions fail to match?
• Under what conditions can a person recover the right to judge his or her own conduct?

Who should read it?

Managers and organisational designers; anyone in engineering safety, clinical quality, audit or compliance — roles that depend on somebody being willing to call a halt; and readers interested in ethics, responsibility and social psychology. Those trying to understand how twentieth-century atrocities were carried out by ordinary clerks will find more explanatory power here than in moral condemnation.

How to Read It

Read chapters 1 to 4 first for the design and the baseline result; without them the variants have no reference point. Chapters 5 to 11 change one situational variable at a time and carry the highest information density in the book — worth tabulating as you go. Chapters 12 to 15 turn to theory (the agentic state, the authority system, strain and its adjustments) and can be left for last. Read it against Arendt's Eichmann in Jerusalem, noting that the two conclusions do not agree: Milgram stresses situation, Arendt stresses the absence of judgement. Pair it too with later archival work, such as Gina Perry's Behind the Shock Machine, on the disputed status of the original data.