Limits to the modifiability of theories- the key to progress

Examples where theories have been modified when new observations have been made are often (ab)used in false generalizations for claiming that any theory could be modified to explain any observation, which is not the case. For example, the "paradigmists" talk a lot about how the geocentric model was modified with epicycles to explain the retrograde planet movements, and claim that it "proves" that their "Occam's razor" (preferring the "simplest" explanation, whatever the definitions of "simple" and "complex" happens to be) should be the only reason to reject geocentrism. They totally ignore the fact that it would have been totally impossible to explain the observations made by space probes geocentrically, no matter what epicycles was made up. Another example is the discovery that the speed of light is the same in all directions despite the movement of the Earth (the Michelson-Morley experiments), which the lumniferous aether theory could not explain. So although theories can sometimes be modified, there are absolute limits to the modifiability of theories. Of course ultraparadigmists can do their false generalizations and claim theories to be infinitely modifiable and say something on the lines of "the theory gives many useful predictions, so we ignore what it cannot explain", and even citing fallacies "disproving" things and generalizing them into falsely claiming that "everything can be proven wrong by some observation or logic" simply by conflating the fallacies with genuine evidence (one example is when they claim that "you can disprove thermodynamics by pointing out that there are no closed systems", which is a fallacious statement). But that ultraparadigmist rhetoric is full of fallacies. Absolute limits to the modifiability of theories refers to limits that cannot be exceeded without fallacies. "Occam's razor", or any other "prefer that-or-that a priori" rules, are recipes for stagnation of science. To get science going again, the top priority must be to create new kinds of observations and experiments with the potential of finding absolute limits to the modifiability of theories. That can generate true progress in theoretical physics, i.e. the kind of change that generates new predictions that can be used for purposes that the old predictions could not. One example would be a theory that explains why gravity is much weaker than the other forces of nature, so that it could be adressed (useful for modifying space-time). That is, as opposed to merely modifying the formulas in ways that makes no novel useful predictions, which is just about all what has been done since the tradition of considering pre-publication peer review to be the definition of scientific evidence began.