Surprisingly, even problems constructed specifically to be random (and hence, intuitively, not to have grouped solutions) exhibit a large degree of clustering among their solutions. These clusters can be exponentially large, and the clustering can be exploited in a number of ways.
The key insight needed to address these issues is the recognition that all of these concepts can be described in terms of the distribution of problem solutions in the overall search space. Solutions tend to cluster; the larger the cluster in which any particular solution appears, the more robust the solution is likely to be. Solutions appearing in different clusters are qualitatively different; limiting factors correspond to situational changes that move one outside of a solution cluster entirely.
Important problems include finding mechanisms for identifying solution clusters, for recognizing when a search is in a failed cluster (a cluster of non-solutions) and moving outside the cluster, and for moving within a cluster to a more desirable (e.g., more robust or more optimal) representative.