Give me a concrete definition of free will. This is where such discussions fall over, inadequate definitions.
It's more difficult if we have to consider the internal, subjective state of the system. A radioactive particle probably doesn't experience anything we could identify with when it "chooses" to decay. And the experience we imagine to be free will could in fact be entirely deterministic, even if other aspects of the universe are nondeterministic. But if we can rule out any kind of nondeterminism in the universe, then we can rule out free will.
I mean, think about it for a second. You have a choice between A and B, and you freely choose A. From your definition it seems that if I rewound time to that moment, you might choose B instead. But it's the same situation! If I freely choose to take a walk this morning because it's nice out, I goddamn mean to go out. No matter how many times you rewound time, I would want to go out every single time. It isn't random: there's a reason why I'm doing this, and reasons are never random.
If free will is nondeterministic or random, then I don't want to have free will.
The key difference is whether that reason is entirely internal to you or it's the result of external influence. If you always take a walk, an observer could infer that by some arcane law of physics, the sun forces you to take the walk - and they might not be far off.
What I mean by nondeterministic/random is there is nothing someone could know about you, the sun, and the rest of the universe that would let them predict with certainty whether you'll go for a walk or not.
The supernatural ability to rewind time seems like cheating, to be honest. But if that were possible and everything always went the same, then perhaps we could rule out free will (depending on the nature of this time reversal mechanism).
Nothing someone could know about me? If someone knew everything there is to know about me, I sure hope they would be able to predict everything I will do. Free will doesn't mean freedom from my own nature.
The only way to be perfectly unpredictable is to act at random and without reason. If you act for a reason, then in principle, that reason could be known by someone else, and they could predict your actions. I think that's acceptable: I don't feel that in order for my choices to be free, my nature and my reasons must be unknowable to anyone but myself.
> But if that were possible and everything always went the same, then perhaps we could rule out free will
According to your definition, yes. Not according to mine. So I guess we made the OP's point ;)
It would mean the choice is mostly determined by their "self", but the definition you give isn't clear about whether that would count as free will or not.
The other side of this though is if the agent believes he or she has the ability to do or not do, but actually doesn't, is it still free will? An example being you enter your office with the intention of working for five hours. Without realizing it, the door locks behind you from the outside, leaving you effectively trapped. Even if you never attempt to leave the room, are you staying at your own free will?
But in simple model, of just a world, where one only has configurations of atoms (or configurations of some substance or etc), there's nothing uniquely determining what's internal and what's external, what's an agent and what's not. So "free will" winds-up achingly ill-defined/under-determined here.
I think your definition is further relevant in that a lot of arguments confuse an ontological model and with a social/political model. And this could well be natural - as social beings, it seems like we tend to both model the world and see agents within it and so saying "in this model, you have choice but in this other model, you and choice don't exist" is highly counter-intuitive to an average human.
This is really interesting. Thanks for the reply.