Twenty-seven years ago, Ole Therkildsen’s study of water projects in Tanzania contributed to a compelling argument about the importance of uncertainty in development interventions. This chapter provides some reflections on the history of this publication and its implications for the future of development assistance.

It begins by recognising some significant differences between the 1980s debate and the discussions now under way. It then points to substantial continuities, which raises the question of whether the prospects are better now than they were then for widespread adoption of the recommended adaptive approach to programming. The answer offered is that they are significantly better.

This is because, in several respects, the intellectual context is more favourable and because a body of evidence is building up on the achievements that are possible when aid-funded interventions become both context-sensitive and adaptive.