Well, theories of market failures belong to the rare species of economic theories that actually are falsifiable and thus scientific in popperian sense, so one could argue that answer to your question is, at that point when someone thought that it actually is good to add a piece of science into economics.