AbstractThe problem with decision-making in an institutional setting is that it is often more important for one's reputation and job security to appear to be logical than it is to be truly effective. Indeed, a far lower burden of proof is applied to ideas and suggestions that accord with simple economic axioms than to those that are oblique, counterintuitive or employ second-order thinking. Hence, the best new ideas generated by behavioural science will and should often lie outside the bureaucrat's natural comfort zone; any difficulties encountered by behavioural scientists should not surprise us: like the sting of an antiseptic, they are proof that the treatment is working.