Linear decision rules for controlling complex systems are often obtained by matrix inversion, but transform methods offer an alternative approach that yields insights into the structure of the decision problem of maximizing expected payoffs under constraints.