The article Hypertext and the Art of Memory by Janine Wong and Peter Storkerson makes an important remark on the ideas of hypertext as being essentially decentralising. They comment upon Vanevar Bush's idea of the Memex as an ideal tool for knowledge formation, because it works by virtue of building and retaining associations. However, Wong and Storkerson say, this notion of knowledge gaining builds upon a strong Anglo-American tradition of philosophy which missed one major point: that of intelligibility. They argue that:

"When we build breadth of knowledge, we must also build depth; in order to make sense of all of the data we have collected, we need to provide concepts that orders it. This barrier has dogged association theory."

Association theory fails to deal with these issues and perceives having knowledge as growing directly out of data possession rather than concept formation. Furthermore, this notion of knowledge building, they say, falsely presumes that a complex interlinked system of data is the way to some sort of 'universal knowledge or truth'. This critique reminds me of the argument I made earlier on the high level of confusion between form and content in a lot of hypertext theory. Concept formation works on the levels of both form (relations between words and lexias) and content (narratives within lexias); whereas association works on the level of form only. Of course, organising principles depend on both author and reader, and in this sense hypertext is multivocal, but by similar reasoning, any text can potentially be multivocal. Logically, providing for concept formation in our postmodern age can then only be done by assuming some kind of partly subjective and partly objective standpoint while closely situating and positioning oneself and one's goals. This means that building connections alone is not enough; projects should have some overall conceptual view.