Some of the most vexing questions about the exercise of official power arise when third parties are not at risk and when the welfare of choosers is all that is involved. In such cases, the objective of nudging is to “influence choices in a way that will make choosers better off, as judged by themselves” (italics in original).

It is important to underline the fact that many nudges are designed to reduce harms to others. In the environmental context, for example, that is a major goal of choice architecture (see Chapter 7), and if the goal is to reduce crime, choice architecture can help a great deal. But when people's decisions will affect only their own lives, and if we are concerned about welfare, autonomy, and dignity, we should begin by asking whether a nudge would make people better off by their own lights. If people believe that they have been made worse off, there is an excellent chance that they have, in fact, been made worse off. And if people's own tastes and values suggest that the nudge has harmed them, public officials should be concerned that they have failed to respect autonomy and dignity.

In some cases, a concern with welfare might lead in a direction different from that suggested by a concern with autonomy. But happily, the “as judged by themselves” standard can command support from both standpoints. In fact we might say that ordinarily, this standard is a reasonable test for all exercises of official power, at least when third parties are not at risk. Suppose that the state is imposing a mandate designed to protect people from unsafe food or unsafe workplaces. In such cases, it is sensible to start by asking whether the mandate will meet the “as judged by themselves” standard. In fact mandates – and incentives of all kinds – can be evaluated, and sometimes defended, with that kind of test.

In most real-world cases, the standard is fairly straightforward to apply (and hence this chapter will be relatively short). At the same time, the test sometimes raises reasonable concerns, both practical and conceptual.