The problem to me is that that machinery does not exist. It's not there, how can something (anything at all in them most extreme sense, a human, a 'system', an AI an animal the hidden machinery in the human brain, a bunch of particles or whatever) 'choose' when there's no alternatives? Real hard predeterminism implies that there's nothing to chose at all. Doesn't the theory of conservation of information tell us that information is never created or lost, so that on a grand scale, basically nothing is happening at all? All information that existed at the big bang is still there, and always will be. There's no place for hidden machinery to make more, or make choices. The 'result' of that choice must have existed/been known before, because otherwise information would have been created at that magical 'choosing' moment.
I've always found the notion of compatibilism untenable as the problem is deeper than 'the hidden machinery that is in the process of weighing alternatives'.
His argument, paraphrased for the layman (he wrote in fairly impenetrable terms) was basically:
Say that we do have a deterministic world, where human action is the result of each human being continually processing stimuli, rationally (or emotionally) evaluating it, and then deciding on a course of action. The meaning of "choice" is this process, and the meaning of "self" is all the hidden machinery that lets us go from a stimulus to a course of action. People who believe free will and determinism are incompatible are implicitly positing that there's some entity outside of this process of weighing alternatives and choosing one, but there is no evidence for such an entity and no need to invent one. Rather, our common-sense notions of choice are completely compatible with someone having received a bunch of sensory inputs and then used the internal circuitry of their brain to decide on a course of action, and our common-sense notions of self are compatible with different people having different dispensations to make certain choices under various circumstances. And we can make moral judgments based on our evaluation of whether the choices another makes will be helpful or harmful to our interests.
Interestingly (and this is me adding to the theory after 15 years of life experience), this perspective seems to mirror psychological development. I've found that as I've gotten older, I tend to settle into my own skin. When I was young, it was like I'd frequently "pop out" of my self and my actions and be like "who is this person?" As I've gotten older, it's more like "Yeah, I'm me, and I make the best choices I can with the information available to me, and ultimately I'm responsible for the consequences even if they don't work out." It's like there's less of a divide between that automatic stimulus -> processing -> response cycle and my objective perception of my life.
[1] https://www.st-andrews.ac.uk/~mnat/~ball0888/oxfordopen/hoba...