At the bottom of this page there is a "Webliography" of some examples of "native hypertext," documents that really take advantage of the medium. But most WWW texts are still longish linear documents, often prepared for paper publication and then posted for convenience of downloading. In its most extreme form, this can take the form of documents posted in formats such as rtf, which must be downloaded and converted before it is readable, or even postscript, which can only be read if printed in hardcopy on a postscript printer.
Even documents posted in HTML and broken into separate nodes often contain "next" links at the bottom of each page to encourage the reader to put them right back in linear order again, which makes one ask why they should have been broken up in the first place. This reminds me of tales of early automobiles equipped with buggy whip sockets.
Some of this may simply be a result of people not having become used to a new medium yet. Stuart Moulthrop suggests two more complex reasons for the presence of these buggy whips. The drag of the past is not simply a result of a few individuals being slow to move. It is endemic to any profound change of medium:
Even after we have given up on print, the majority of "really electronic" text will be hopelessly contaminated with the old ways of knowing. What we must carry forward is a strong sense of foundational irony. The past is always present; we are Gutenberg creatures no matter how hard we play at revolution; there is no such thing as "non-sequential writing." We make our way by recursion, by folding a new order back upon and into its predecessor.
("Getting over the Edge")
Elsewhere he sees a more political motive:
Compare the print paper to a truly multi-linear hypertext where the writer explores with equal discursive value the ambiguities and alternative hypotheses that derive from her research. Depending on their interactions with the text, subsequent readers might form very different conclusions about the researcher's findings, making it impossible to represent her article in terms of a simple, particular proposition. If the article cannot be thus represented, it cannot effectively contribute to the stream of citations from which technical consensus emerges. Its intertextuality is simply too pronounced, or its focus too weak, to serve the dialectical process of science-in-action. This, then, is why hypertext researchers do not work primarily in hypertext: because the work we do and the institutions in which we work are both hierarchical, and because a fully realized or "native" hypertext is incompatible with hierarchical discourse.
("Shadow of an Informand")
Well, maybe. Or maybe the medium is simply not very friendly to argumentative rhetoric.