I outline concerns over the scientific legitimacy of psychiatric diagnoses. I outline Kantian accounts of constructing entities. I portray categorical and dimensional psychiatric diagnoses as constitutive principles, a form we apply to scientific data to generate inductive and causal probabilities. A good categorical or dimensional diagnosis would be one which plays a role in a system of science which incorporates as much phenomena into the system whilst constructing as few entities as possible. I show how categorical diagnoses can play a role even if the DSM was made dimensional or if RDoC and HiTOP produced relatively homogeneous dimensional diagnoses.