Form and function are, of course, intimately related, or at least should be.
Most studies of hypertext form grow out of assumptions appropriate to either hypertext fiction or information retrieval. In the former case, form is intended to maximize associative links, to present a story from multiple points of view, in short, to instantiate in textual form the assumptions of postmodern literary criticism about how people really read books that appear to have a linear form. In his discussion of Michael Joyce's hypertext novel Afternoon, Jay David Bolter provides an excellent example of the forms that might grow out of this function. He suggests various forms of complex looped and multi-levelled narratives that refract the plot into multiple points of view and multiple possibilities (Writing Space 127-29).
The rhetorical function of information retrieval, the original purpose of hypertext that dates back to Bush's memex, has naturally generated the largest supply of literature on rhetorical form. When the goal is to piece together the facts that one needs on a particular subject as quickly and effortlessly as possible, forms tend to be those which emphasise ease of navigation from one node to another, minimize surprise and maximize coherence, Fragmentation of information is used to help the reader reassemble it in the most immediately functional form:
Research results suggest that many short documents are preferable to a smaller number of long documents. An experiment was performed by students at the University of Maryland using the Hyperties system in which the same database was created as 46 short articles from 4 to 83 lines long and as 5 long articles of 104 to 150 lines. Participants in the study were given 30 minutes to locate the answers to a series of questions by using the database. The 16 participants working with the short articles answered more questions correctly and took less time to answer the questions.
(Shneiderman and Kearsley, "Document Size" node)
Even Landow's elaborate webs of commentary, background and signals of part-to-part relationships, in the centre of which hang classics of Victorian literature, tend to fall into this category, though the goal is more leisurely exploration than quick retrieval of facts. Much of his advice on rhetorical form boils down to navigational advice.
Argumentative discourse represents an entirely new set of rhetorical problems linked to a different set of rhetorical functions. David Kolb has, in my opinion, thought more deeply on this question than anyone else. He asks "What new tasks for thought will the medium help us discover?" and provides the following highly provocative list by way of answer:
(Socrates, "Phil HT Actions" node)
It is far from clear what kinds of form might help us with all of these functions. How will we build webs that walk the thin line between providing a linear default path for the reader--in which case the document might as well be written in straight text--and enveloping the reader in a cloud of mystery? If we can't figure out what these forms are, how will we help our students enter a discourse community that is coming to use these forms as part of its knowledge-making processes?