If there is non-random laws, then the situation in the "next" moment is decided just by the laws, and the current situation. By induction, all future moments are decided just by the current situation, and the laws of physics. Thus, the system is deterministic.
But then the choosing must be done according to some process. That process is either deterministic or quantum-random. Either way, it's not free.
There is a difference between "free" and "completely determined and or caused by prior conditions." According to some dynamical systems theories, the ascription of causality fails: "Another instance in which causality in its classical interpretation appears to be challenged is the presence of feedback loops, in connection with the issue of self-reference. If acting alone a feedback loop appears to entrain not only effects fed on the causes, but also causes fed on the effects."
Foundations of Complex Systems : Nonlinear Dynamics, Statistical Physics, Information and Prediction.
One can, in principle, still speak of causility in such systems, but only by redefining it, because self-organizing emergent systems are "causal" only in the sense that causation is "the outcome of an interacting set of mechanisms or powers. These causal mechanisms arise from the extra
organization that appears at each level of structure of laminated entities. Therefore, actual laminated events are to be explained as the outcome of an interacting set of level-abstracted real causal powers. Where the higher-lvel entity has genuinely emergent causal powers,
these cannot be sufficiently eliminated from causal explanations by any reductionist strategy. Causal Power of Social Structures: Emergence, Structure, and Agency.
Most dynamical systems, even those whose trajectories cannot be determined and are inherently nondeterministic, can still be considered causal, because (as you noted earlier), the indeterminacy is
epistemic, rather than ontological. However, it is unclear if this is true for dynamical systems with complex, heirarchichal feedback mechanism. Additionally, for some systems types little is known concerning their nature. Take, for example, the so-called "Fuzzy Blue Sky Catastrophe." Dynamical systems which involve bifurcations resulting in qualitative, nondeterministic changes are complex enough. However,
fuzzy nonlinear systems of this type are even more challenging to assess, and current studies have only scratched the surface (there was a paper on the subject published in the proceedings of a conference on nonlinear science and complexity).
So how can causality apply to the type of systems I'm referring to? In other words, if a system can't be broken down into component parts, and the initial conditions only act on the component parts (the locality principle in physics), then what causes the emergent behavior of the system?