Despite two decades of work, learning classifier systems researchers have had relatively little to say on the subject of what makes a problem difficult for a classifier system. One focus of our work has been the issue of what makes a problem difficult for XCS -- Wilson's recent accuracy-based classifier system. This document outlines the approach taken, provides some initial results and outlines possible directions for future work.
[This work is subsumed by "What makes a problem hard for XCS?". A chapter in Advances in Learning Classifier Systems. P. L. Lanzi, W. Stolzmann, S. W. Wilson, (eds.), pp. 80–99. April 2001]