Classical economics assumes that people weigh up their options and choose whatever serves them best. Behavioural research complicated this picture by showing how heavily choices depend on presentation: making pension enrolment automatic, with the option to withdraw, raises participation far more than campaigns urging people to save. Governments were quick to adopt such nudges, which are cheap and compel nobody. The enthusiasm has since cooled. Several celebrated findings have proved smaller than first reported, and effects that hold inside a trial often fade once a programme is scaled up. More fundamentally, critics argue that redesigning the moment of choice can substitute for asking why saving was difficult in the first place. Nudges remain useful; the mistake lay in treating them as a replacement for structural policy.