"Newtonian laws of motion allow a prediction of the future based on knowledge of the present because these laws are of the formd\xi_k/dt = F_k(\xi_1, \xi_2,\cdots \xi_n) \qquad \qquad (k = 1, 2, \cdots n)...if the values of the "state variables" are known for the present instant of time t = 0, one can "predict" their values for any past or future time t. All laws of this kind are called "causal laws." The general "principle of causality" would claim that all phenomena are governed by causal laws which would have the [above] form... where \xi_1,\cdots \xi_n are any variables that determine the "state" of a physical system at the time t. ...belief in this general principle is supported by the special case of astronomy where \xi_k are the coordinates and velocities of mass-points and the functions F_k are known to be simple mathematical formulae derived from Newton's laws of gravitation. ...What caused the success was the simplicity of the laws in comparison of the complexity of the observed facts. If we regard the F_k as arbitrary functions... and admit complicated initial conditions, the causal law... may be "valid" but will not guarantee the same kind of success. It may be that the law is as complex as the observed facts. Then there is no advantage..."