But this begs the same question: if mental illness really is what psychologists say it is, and if treatment is a learnable skill, then the practitioner shouldn't matter that much assuming his training was good.

But most evidence suggests that some "je ne sais quoi" has to exist in the therapeutic relationship.

In other words, Freud was right about Transference as a necessary ingredient to psychotherapy (and probably about a lot else that is still too controversial to talk about or pass IRB muster).