My tired brain may not be analysing this well, but I don't think that Wessely's claim is even entirely logical. It fails to take into account the degree to which common/general beliefs and fact can differ. If this difference is very large, it can overcome the fact that a patient does not expect a treatment to work, if it is actually an extremely effective/ineffective treatment, which will work/not work despite their strong belief that it wouldn't/would.
Many of us have experiences that surprise us and are not in line with our prior expectations. Yes - our prior expectations will have some effect, but they are not the only determinants. And the degree to which they determine outcomes will vary between people, just as some people are very susceptible to hypnosis and others are not.
Not sure if I'm explaining this well or even whether it makes sense! And I am not of course intending to suggest that CBT is an effective treatment for ME.
I think it depends on the nature of the trial, I would agree that in some cases such as if you are measuring very big effect sizes, or using objective measures, or the trial is blinded so participants don't know what to expect, then the levels of expectation would not really be of any importance.
In an un-blinded trial though when measuring relatively small effects using subjective measures, then the relative size of the placebo and other biases influencing each of the results likely to be very significant, possibly the only thing separating the groups, so levels of expectation are of critical importance to control.