M. Glaum, J. Owen, G. Zames Most systems encountered in engineering and science, and in particular in economics and social science have highly uncertain dynamics, and their fundamental structure is very complex. However, the emphasis in robust control, especially from the point of view of assessing performance, has been on cases where uncertainty is either small or has a known parameterization. This gap between existing methods and the reality of many practical plants has created a clear need for a general approach to the design of feedback and identification schemes for situations where the plant uncertainty is large and its structure not known a priori. The goal of obtaining such a unified framework and design methodology constitutes the overall motivation for this research. Particular questions that such a theory should address, and which are the subject of ongoing research, are summarized as follows. i. How does one combine feedback and system identification to best overcome the effects of system uncertainty? ii. What are the key properties of a set of plant uncertainty which limit the ability of feedback to improve performance? iii. What does it mean for feedback to reduce system uncertainty?