Nudge is a semantically multifarious concept that originates in Thaler and Sunstein's (2008) popular eponymous book. In one of its senses, it is a policy for redirecting an agent's choices by only slightly altering his choice conditions, in another sense, it is concerned with bounded rationality as a means of the policy, and in still another sense, it is concerned with bounded rationality as an obstacle to be removed by the policy, when the latter has a benevolent aim. The paper centres on the interrelations, both semantic and factual, of these three nudge concepts. It argues that the first and second are basically disconnected on Thaler and Sunstein's major examples of nudges, and that this has gone unnoticed to them because they wrongly equate the second with the third concept, and also because they overestimate the explanatory power of behavioural economics, compared with that of classical rational choice theory, to account for successful interventions. After completing this analysis, the paper moves to some of the normative issues raised by Thaler and Sunstein. Their thought-provoking claim that liberalism and paternalism can be reconciled within one and the same doctrine of social ethics - libertarian paternalism – has been subjected to thorough philosophical criticism. Rather than following this abstract line, the paper takes the shortcut of arguing that Thaler and Sunstein lose their best defence of libertarian paternalism after the nudge concepts are disentangled. They had effectively based their case on the view that slight interventions could have powerful effects through a clever use of bounded rationality, and it has been shown that the latter is not really at work in the interventions they consider. The paper finally concludes that the three nudge concepts are worth pursuing, though independently of each other, and in particular that the third one, which involves correcting the pitfalls of bounded rationality, should receive sustained attention from policy analysts.
