In "Hypertext: An Introduction and Survey," Conklin defines cognitive overhead as "the additional effort and concentration necessary to maintain several tasks or trails at one time" (40). A well-designed hypertext system will minimize cognitive overhead.
Even at its easiest, adding a new node is a non-trivial task. One must select anchor text in the current document, pick an appropriate menu option, choose the node class or type, name the node, and then, once it is created, type in the text. Few of the decisions in this sequence are easy to make, even for experienced authors. One must take into account the local network and how it may be apprehended by future readers. A stray thought lacking direct connection to the current document may be lost while the author struggles to pick a suitable title or an appropriate anchor.
Writing about gIBIS, Begeman and Conklin said that "it would be nice to have a 'proto-node' in which to record ideas, snippets of text, and perhaps graphical sketches before having to structure them" (260).
There are several software solutions to this problem. The ability to make quick non-connected and ill-defined nodes is essential if the system is to be used as a brainstorming tool. These nodes without links could be automatically gathered into one place for later review and contextualization. StorySpace allows notes to be quickly attached to nodes; these are automatically titled and stored in a cluster called "Notes."
The ability to display multiple windows simultaneously is almost essential. Authors need direct access to several different nodes to provide sufficient context for further writing. Organizational nodes (such as maps and outlines) can then be left on the screen as roadmaps while content nodes are visible.