The fact that we cannot include all the information about our residual function into our refinement results in some parameters of the model oscillating, other becoming stuck, and the requirement that we run many, many cycles. Until more powerful methods of minimization become available the crystallographer must be on guard.