I'm not sure I agree 100% with either side of this argument
at the moment.
If one discounts a whole area of possible causation, then
one introduces the possibility of assigning the wrong
cause to the witnessed effect.
If, on the other hand, one suggests possible causation that
cannot even be investigated one can impede the progression
of knowledge.
Take demons as an example of a possible causation. Whether you
find the idea plausable or not is irrelevant.
IFF one could provide specific predictions that could ONLY
be the result of demonic activity, and then test those
predictions -- would that be scientific or not?
If the opinion is that a demon could do anything, then that
is definitely non-refutable and so no further progress
can be made.