Actually this was a cool thread, at least as far as I'm concerned. Thanks to arthwollipot for bringing this up.
I'd always thought placebo was the mind-body thing. A psychosomatic kind of effect that is, in this case, beneficient. (It can also be negative, in which case nocebo.) At least that was my impression.
I see now, basis this thread, that placebo's basically one of three things, not just the one. First, a catch-all for all kinds of uncontrolled variables. Two, the mind tuning out, so that the baleful effects of some ailment appears to have improvement, even though there's been no real improvement as measured objectively. And three, the psychosomatic thing. (Although: Is the psychosomatic thing even a thing at all? Does that actually happen, the mind thinks you're getting better, so the body does get better, to an extent? The comments on this thread seem to indicate not.)
Also, two observations:
(1) Like I'd said earlier, the design of the study referenced in the OP seems glaringly shoddy. (Said as an absolute amateur, open to being corrected if someone knows better.) They've got three groups, two of which improved somewhat as a result of post-op physio apparently (one with real surgery, one with sham surgery), compared to the third group that had no surgery and no post-op physio. Whoever designed this thing, shouldn't they have had the wits to fit in some equivalent physio in the third group as well (call it something else, since it can't be called "post-op" physio), so that you're comparing apples with apples?
(2) Somehow in these studies, they're kind of blinkered, it seems to me. I mean, intervention gives better result than no intervention, but the intended intervention gives no better result than the placebo intervention. So what do you do? Conclude your intervention is no good, and move on. But shouldn't you be doing something else? Given that there has been some improvement vis-a-vis no intervention, then shouldn't you be exploring how that improvement happened? Call it placebo, call it catch-all variables, whatever, but shouldn't you be exploring that, whatever the hell it is? Because the ultimate aim is surely not just to test some specific intervention, but to actually get some intervention that works, right?
... in this case the 'sham' surgery was just as effective as the 'real' surgery. Or IOW, the 'real' surgery was just was just as ineffective as the 'sham' surgery.
The message being missed here is that some 'real' medicine is actually quite useless. Considering all the downsides of surgery, this suggests that a lot of harm is being done to placate patients' desires to get their medical problems 'fixed' when in many cases the body will repair itself given time.....
Like I was saying, while sure, one takeaway is that the 'real' surgery is no better than the sham surgery, but another takeaway, and the more important takeaway, is, or should be, that surgery (whether real or sham, and everything that goes with it) is somehow somewhat better than no surgery at all. Therefore what that "somehow" is, that makes this better (probably physio, in this case), surely
that is what is really interesting here? And surely investigating that should be the next step now, in cases like this?
I speak without first-hand knowledge of the research process, only a cynical but TBH uninformed impression about how these things happen, but I'd guess the people pouring research dollars in are usually interested only in how some surgery or some molecule performs. More mundane stuff like those "other variables" don't promise good returns, and so aren't bothered with --- while, conceivably, similar improvements recorded with the intended intervention would've got them all excited and happy. (If true, this suggests a change might be called for as far as how these research goals are set, basically.)