One way to express the critique of Aristotle’s physics in his own terms is to say that, while he was right that every action or change involves act and potency, he thought this meant there were agent-patient relations in the inanimate when in fact there are only interactions.
Take an interaction between two inanimate things A and B. To posit an agent and patient is to add something to this interaction, namely that there is an order from the one to another. Such an order requires a way to determine which is potential and actual, and therefore which is perfecting and which is being perfected. If we are speaking of local motion, it requires that one impart a motion (and so be in motion) absolutely and not in a way that is essentially dependent on the hypothesis of something being at rest. If we are speaking of a qualitative change (let’s take heating) it means we need an absolute way to know that the interaction between A and B is one where A is heating B as opposed to being a case of B cooling down A.
Now Aristotle obviously knew that one thing couldn’t push another without itself being pushed by it, but he thought that there must be some way of determining which was pushing per se and being pushed per accidens. This is one reason why he posited a center and boundary edge to the universe, since they were fixed locations that made it in principle possible to determine what was pushing and what was being pushed. The case of heating was a bit trickier: here Aristotle was committed to the claim that heating was more perfect than cooling, and so heating was occurring per se and cooling only per accidens. The same would have to be true of chemical reactions: making water from gas would have to involve either a perfecting or corruption per se, and the opposite per accidens.
While it is perhaps logically possible for a theory one day to give absolute place values and orders of perfection to heating, chemical reactions, and other inanimate interactions, this does not seem to be the smart way to bet, and we seem to learn a lot more about nature by assuming that such things aren’t there. If, for example, there are no absolute place values, then local motion is always relative to a background, which is the charter for both inertial reference frames and the identity of acceleration and rest in a gravitational field. Again, seeing chemical reactions as a matter of the per se generation of more perfect substances seems to point in the direction of some most perfect inanimate substance – like gold – which gives us something more like alchemy than chemistry.
Adding agent-patient relations on top of physical interactions did give us the ordered cosmos, with a hierarchy of movers and mobiles, along with all the grand-unified ideas of the common good of the whole universe. But there was a dark side to all of this too, as it tended to point in the direction of fatalism and astrology. All of St. Thomas’s struggles with astrology were based on the idea that there had to be a first agent mover of all motion, and it’s unclear to me whether his response to it could really survive a robust defense of the causal closure thesis.