I fear that the requirement by the US National Institute of Mental Health for investigators to come up with a hypothesis as an initial part of the drug-discovery process (Nature 507, 273 and 288; 2014) will not end well.

The discovery of all early and most recent drugs has hinged on testing the effects of small molecules on the phenotype of cells and organisms (M. J. Keiser et al. Biochemistry 49, 10267–10276; 2010). A hypothesis is not needed (D. B. Kell and S. G. Oliver BioEssays 26, 99–105; 2004) and finding a drug's likely targets can be postponed until it is known whether the drug works.

It is now more fashionable to start the drug-discovery process with a hypothesis on how a drug might act on (typically) a single target, then to test how effective it is. But if the hypothesis turns out to be wrong, which is not unusual (D. B. Kell FEBS J. 280, 5957–5980; 2013), money and effort will have been wasted.