One thing I would add:
CBT is not universally accepted as being helpful to people per se. Its proponents make enormous claims for its efficacy across the range of human problems.
BUT
It has key critics who are concerned at the basic presumptions informing CBT...
We mustn't assume that CBT is good per se. That claim itself is contested, independently of ME/CFS.
A year or so ago I spoke to a few people I know and respect who work on counselling and psychotherapy and asked for their views on CBT. Universally they confirmed the scepticism and I sensed a degree of resentment of the growing dominance of the approach. (And I wondered whether other approaches to psychology and psychotherapy might be important allies for us).
One point that rang true was that CBT is gaining popularity because it is supposedly more 'scientific' and more amenable to scientific analysis. I was told that the way it's structured makes that true, to a degree, but that it seemed likely that it wasn't particularly more effective than other forms of therapy but just easier to produce (apparent) evidence in its favour.
I still think the dutch actometer data should have been a potential killer blow, and it's the stand-out thing about PACE, for me, the way they failed to do that follow-up. Consider: the Dutch results seemed to show that when patients filled in questionnaires to the effect that "I am more active after the therapy", the actometer data showed this was not true. If PACE had confirmed this - which it easily could have done if the actometers had been retained after the study - that would have undermined the entire evidence base for questionnaire-based assessments of success. That wouldn't just affect CBT for ME/CFS, but potentially make all questionnaire-based assessments unsafe. It could perhaps have undermined CBT as a whole and a lot more besides.
No wonder they decided halfway through their study to change the protocol and not apply this test!
But in doing so they exposed their own manipulation of methodology in order to fit the results they want to demonstrate. And combined with the timing of the release of the dutch data just before the decision to abandon actomers (I think...I'd like to see all that worked up as a full chronology), that swings the way it looks from confirmation bias to conscious fraud, for me - not, perhaps, legally speaking, but in practice "this decision was for the participants own good because it wasn't fair to ask them to wear actometers again" strains credibility to breaking point.