Paul (and everyone), if my initial presentation gave the impression that I proceeded with symptomatic treatment without basing intervention on an initial formulation (or conceptualization) of Gary and his problems, then I may not have expressed myself clearly. I agree wholeheartedly that "CBT does not involve a superficial grab at a few problems followed by teaching relaxation." I proceeded with symptomatic treatment because I had already formed my initial formulation and it led me to the conclusion that it was appropriate to begin therapy with symptomatic treatment and then to work on Gary's long-standing tendency to lack confidence, put excessive pressure on himself, and worry excessively.
Regarding your objections to the use of personality disorder diagnoses, I agree that if the diagnosis of personality disorder is made "when the therapist does not like or agree with the patient" or if it is used to distance the therapist from the client, this use of diagnostic categories is counter-therapeutic. (Note that the same criticisms have been made of all use of diagnostic categories.) However, instead of concluding that we should avoid using personality disorder diagnoses, I conclude that we should not use personality disorder diagnoses in these ways.
Your example of the misuse of the label "Borderline Personality Disorder" is an excellent one. If a therapist labels a client BPD because therapy is going badly, because the client is hard to diagnose, or because the client gives them a hard time, the use of this label will be destructive. If the therapist overlooks the client's "borderline" characteristics and proceeds with standard CBT, a number of reports suggest that there is an increased risk that treatment will be ineffective or counterproductive. If the therapist attempts to "start from scratch" in developing a formulation of the client's problems, this will be a complex and time-consuming task. If the therapist makes an accurate diagnosis, then they do not have to start from scratch in developing a formulation and choosing a treatment strategy. Several CBT formulations of BPD have been published and at least one has significant empirical support (Linehan's). They will still need to develop an individualized formulation but this task is much simpler if they can make use of the formulations and treatment strategies which have already been developed
You ask "Why not instead [of using a personality disorder diagnosis] consider his evidence for his beliefs?" Are the two mutually exclusive? Of course I'll consider his evidence for his beliefs, accept the beliefs which turn out to be true, and help him modify the beliefs which turn out to be untrue. However, I normally do that a bit later in therapy, we're only at the sixth session.
Finally, you may be interested in knowing that the scientific basis for CBT as a treatment for PPD isn't quite "zilch." Dan Turkat and his colleagues have done some interesting single-case-design research on CBT with personality disorders (including PPD) and there is a fair amount of other relevant research available. We could use a lot more research but there is more empirical support for CBT as a treatment for personality disorders than is generally recognized.