Simon argues that it is not the an analysis of the optimal solution to a problem posed by a particular environment and a particular set of goals that matters, but what is needed is the best solution in general to a wide variety of tasks in a diverse environment given the constraints of the architecture. As Simon points out, in Anderson's Rational Analysis, it is the assumptions about the computational costs that do most of the work. The issue lies in how much these assumptions reflect the actual architecture and thereforehow dependant the final analysis is on a specific architecture. Furthermore, Simon argues, that although such an analysis may predict the path to a solution a human architecture may take, there are many different methods, or algorithms, that an architecture can implement to achieve the same path. Each method of course has it's own distinct properties which themselves are constrained by the need to interact to solve a wide variety of goals in multiple environmnets. An analysis of this type is termed 'bounded rationality'.