You misunderstand. I am defining "free will" as the property that makes people moral agents --- ie. able to do otherwise in such a way that features in (some pe

You misunderstand. I am defining "free will" as the property that makes people moral agents --- ie. able to do otherwise in such a way that features in (some peoples') moral judgements. I am not claiming that we in fact possess this property (I think that's an open question). My point is that all conceivable systems of human morality are equally incapable of distinguishing between humans that possess "the property that makes them moral agents" and those that do not. If the universe is deterministic, then human morality is just another link in the chain. Morality cannot address the fundamental nature of our choices because it exists in a system where the only available input are the consequences of and the perceived motivations for human behaviors. Hitler didn't accidentally slaughter European Jews, his hateful feelings and aspirations for power drove him to towards those actions, even if he didn't choose to be hateful and power hungry. The state of determinism does not modify the criteria for moral judgement. Imagine you're trying to understand a codebase with as much legacy baggage as human thought. You do so by identifying modules, finding out what depends on each one and what it depends on, testing and later maybe improving on it, or discarding it. Free will is one of those modules and before we try to 'fix' it, we need to know what it does. In that analogy, I don't think you can describe free will as a module, contemplating free will is more like asking "is the x86 instruction set etched into the processor or is there a homunculus living in the chip that reads incoming instructions and executes them appropriately". The answer to that question has no bearing on whether or not we judge the "human thought program" to be malware (e.g. for corrupting the memory space of another program), what we do is first decide "what was the intention of this program?" and then assess if the memory corruption is a bug or the program's intentionally destructive payload. The absolute nature of the instruction execution mechanism doesn't really matter with regard to how we form judgments about which programs we can all agree are damaging to benign programs and thus a threat to the stability of the system as a whole.