1. The two reasons are linked, for reasons which I think are obvious. So I will be concentrating upon the second of these: that the actual content of these laws may exceed our capacity to confirm them with certainty. The case of momentum is one in which divergences between experimental evidence and (law-like) predictions eventually resulted in the “recognition” that the “law” was ill-expressed. Others, like the atmospheric laws relevant to weather prediction, are not so easily revised when faced with even greater divergences between expectations and fact. Our predictions, in most naturalistic or complex sciences, are simply too sensitive to slight variations in initial conditions to even tell us whether our favorite theory is “correct,” or merely a good approximation under “normal” conditions.
2. Dennet designates evidence which would invoke local fatalism as “a particular circumstance in the relevant portion of the past which ensured that the agent would not have done otherwise (during the stretch of local fatalism) no matter what he had tried, or wanted, to do.” (p.386) Further, he insists that local fatalism is “a phenomenon that is entirely neutral between determinism or indeterminism.” (p.386)
Hoy, Ronald C., and Oaklander, L. Nathan (editors). Metaphysics, Classic & Contemporary Readings. Wadsworth Publishing Company/ Belmont, California. 1991.