Been reflecting on the role of hierarchy in knowledge construction - is a strict hierarchy too restrictive?
Looking at the richness of representations, the need to accommodate dynamism and imposed restrictions of outlines - it seems 'natural' to gravitate towards maps that support 'cross-links' and cyclic connections. Top-down directionality in a concept map robs us of important insights and fails to portray an important class of relationships. In a way, this is similar to the arc only restriction of mind mapping.
This article by Safayeni and Canas takes a deeper look at the issues and concludes both static and dynamic connectors are necessary to represent knowledge.
In many ways the arguments spiral around ways to record and portray context. For me, context is about clustering, spatial continguity and labeled arcs. Cyclic maps, we should note, quickly become very complex as the magnitude of a variable leads to different paths.