That's an interesting perspective. I don't agree with it myself, but I agree that it's impossible to resolve this discussion without a clear definition.
Personally, I define free will as having the ability to "change". To overcome your both your past self, and your environment. Ie, a billiards ball has no free will, because its motion is compelled by its past state, and by the environment around it. Similarly for cars, computers and every other inanimate object. Whereas a human can be said to have free will, because regardless of his past nature, and regardless of the environment around him, he still has the free will to transcend all that and do something wholly "unexpected". To me, that's what separates free will from an automaton. Hence why my definition of free will isn't compatible with determinism.
I appreciate your point though that others might define free will in a wholly different manner.
When we take this argument out of imagined philosophical universes with perfectly deterministic Newtonian-like rules, we find that in fact randomness is everywhere. Quantum mechanics tells us that the behaviour of every particle and photon in our bodies behaves under some level of random influence. Perfect predictability is a fantasy. However conceptually we can simplify this and just imagine that my sate, the definition of how I am as a being, includes a source of randomness.
My argument then is that sure some of my choices are random, mainly because I choose to not use my memories and skills I have learned and hand over the choice to the RNG. That's also often a choice though. See my comment elsewhere on this page on No Country for Old Men.
You talk about unexpectedness. How do you define unexpected in such a way that it is distinguished from random? That's crucial. Truly random behaviour is not intentional and not 'ours'.
One way in which people change their minds is though the assimilation of new information. That's fully compatible with my take on free will. Our prior state includes the decision making apparatus that evaluates and incorporates the new information. Our consistency as a persistent self has not been compromised, it's just been transformed by new information or a new way of thinking about things. We have changed, but in a way that we are 'built' to do. Given omniscience (which is impossible, but for the sake of argument), that transformation would have been predictable in principle and again for me that's not a problem. Someone who knew me well might have predicted that I would change my mind.
That is not a problem. There's a quote in a comment nearby from Godel that addresses this very well.
I suppose I am a compatibilist, but I object to the use of the term 'compatible'. I think some degree of determinism is _required_ in order to have free will. It's essential. Without it, 'I' am taken out of the chain of responsibility for my actions. If they do not flow from me, from my state, my memories, my decision making processes and if those things are not to some extent consistent, then those actions are not mine.
Personally, I define free will as having the ability to "change". To overcome your both your past self, and your environment. Ie, a billiards ball has no free will, because its motion is compelled by its past state, and by the environment around it. Similarly for cars, computers and every other inanimate object. Whereas a human can be said to have free will, because regardless of his past nature, and regardless of the environment around him, he still has the free will to transcend all that and do something wholly "unexpected". To me, that's what separates free will from an automaton. Hence why my definition of free will isn't compatible with determinism.
I appreciate your point though that others might define free will in a wholly different manner.