I think this is a key point.
It's also possible (and probably quite typical) for existing observations to trigger a hypothesis. But that hypothesis is useless if it can't generate predictions that can be confirmed. The unconfirmed hypothesis is always only a potential explanation for the observations that triggered it.
Michael is getting stuck right at the beginning of the process. He refuses, for bizarre reasons which he can't quite explain, to accept exotic matter hypotheses as testable explanations for observed phenomena. So far, AFIAK, these hypotheses have not been dis-confirmed, nor have they been adequately confirmed by predicted observations. We're still in "Q" mode.
There's a particularly ironic consequence of MM's approach, which can be shown by a historical example.
Lines^ of a new element were observed in the spectrum of the Sun, in 1868; the new element was given the name "helium".
The same lines^ were observed here on Earth, in 1882.
However, helium was not isolated, and studied, in controlled lab experiments until 1895.
By MM's logic, helium was an "invisible friend", or "exotic matter", from 1868 to at least 1882 (and, strictly, until 1895).
Worse, by MM's logic, 19th century scientists were not doing science when they went looking for evidence of this new element; instead, they should have limited themselves to examining only the spectra of already known elements.
What happens if, one day, a WIMP turns up in some lab? If MM had total authority over doing controlled experiments in labs on Earth, the only way such a thing could happen is serendipitously (all controlled experiments to look for WIMPs, here on Earth, would be verboten in the MM regime).
^
actually only one