A minimal requirement for any new treatment is that its efficacy be greater than that of a placebo. When the treatment is directly compared to placebo in a randomized trial (or set of trials), demonstrating superiority to placebo is relatively straightforward. However, when the treatment is directly compared to an active control in a non-inferiority trial, demonstrating superiority to placebo requires an indirect comparison and is complicated by the assumptions of assay sensitivity and constancy. A key question is whether superiority to placebo is sufficient in these cases, or whether the existence of an effective therapy requires a higher threshold for any new therapy. In this presentation I will discuss the implications of imposing a higher threshold, and argue that they necessarily lead to logical inconsistencies that outweigh any advantages.