It's important to consider the difference between falsifiability and testability. Something is falsifiable if it makes a prediction that could be proven wrong in principle; it's testable if the prediction could be proven wrong in practice. Our current inability to make a measurement to the necessary degree of precision doesn't make a hypothesis unfalsifiable or "unscientific," but it does mean that we have to put it on the backburner until we can figure out a way to test it.

It may very well be that proton decay is just extremely suppressed, but if that's the case, it does seem like we'll have to find some more easily testable prediction coupled to the prediction of proton decay in order to be confident that it happens.

Yes, in general, we can only invalidate a class of theories by ruling out some attribute common to every theory in the class. I'm not sure why that's a problem, persay.



Eliminating conservation of baryon number doesn't necessarily put us back to square one on MACHOs; it depends on how much the conservation is violated.

Yes, a continual shifting of the goal posts is generally a troubling sign for a theory, although it's not always. Sometimes, it may be that the goal posts were initially set in a bad spot, and people are hesitant to move them too far in one fell swoop.