You might think that with the spread of the World Wide Web, hypertext has already joined the list of communication media in which people spend time making and reading complex communicative artifacts.
True, but there is not yet enough hypertext on the Web that demands and shapes a new literacy. The Web was originally designed to link relatively self-sufficient documents (scientific papers), and that is still what it does most easily.
There are technical reasons why it is difficult to create complex linked structures or experiences on the Web, where links are all of one type, and they cannot themselves be linked to or commented upon. Most Web links suggest binary movements: from index entry to fuller detail, from question to answer, from statement to ironic comment, and so on.
Furthermore, the Web's horizon is very close: there is no convenient way to map areas and show context beyond one link away. So it is difficult to have a textual gesture or trope that extends over multiple pages.
Another difficulty: the way Web browsers mark links encourages scanning rather than reading. Web browsers also distinguish those links that lead to already visited pages; this encourages a way of reading that tries to cover all the linked pages, without paying much attention to the link patterns. This also makes it difficult to construct linking paths which rely on repeated exposure to the same material in different contexts. In this text the link patterns have you keep re-travelling "used" links; this feels wrong.
If we are to have hypertexts that do more than go with the easiest flow, we will need tools that encourage complex and interesting link patterns.