As a fan of craft knowledge, I think that there is both something in the evidence for the craft status of medicine and that nothing negative follows for its status. But I am not sure of the dialectical purpose of the craft versus science distinction as applied here.
One might argue that psychiatry is not reducible to science alone but involves skills that can only be characterised as craft skills. One way to put that would be to say that it irreducibly involves know-how as well as knowledge that. But if so, how does offering both these elements protect diagnosis from criticism? Won't diagnosis naturally count as part of what Jennifer Aniston would perhaps describe as the 'science bit'?
One way to bring the craft to bear conceptually on the diagnosis element itself might be to stress the latter's status as Kantian reflective rather than determinate judgement. Again, I like this idea. But the conceptual package brought to bear on an individual - the combination of concepts and laws - is again surely a science, or at least a science in aspiration, in the case of psychiatry? To deny that would be to concede diagnosis to anti-psychiatry not to defend it.
So might one think that not only is the subsumption of an individual (person, case, event, experience) under a concept a skill (reflective judgement, again) but that the derivation of inferential consequences of that subsumption is also a skill, something, eg., that resists codification? Perhaps. If so, it might be akin to making a moral judgement on Dancy's view. Nothing generally follows. What follows depends only on the particular circumstances. But not merely in the way that the application of codified Newtonian physics to a particular object depends on its context but rather in a way that resists any general codification (akin to Newtonian laws). Now that would be radical. Psychiatry as moral advice, as it were, though not only concerned with what is Good. Can this really be Tom's view?