By un-fixed, most critics mean that there is no one definitive, final reading of a hypertext. There is no "fixed" text. Unlike a narrative or lyric poem, which may have many interpretations, a hypertext has multiple readings (each of which can support multiple interpretations). When I assign a print text to a class, I can assume that everyone has read the same text. When I assign a hypertext, I can make no such assumption. Everyone will have read the same opening page, but once the paths begin to diverge, I have no way of knowing which particular sections any one reader has read.
Reading a hypertext requires a significant role for the reader, in some ways a more active reading that a print text does, since an active reading is an absolute requirement for a hypertext. The text doesn't exist on screen until it is called up to be read. Only during the act of reading are the paths chosen. Prior to reading, the hypertext exists as a multitude of possibilities, of potential readings. The same may be said of other forms of literature, but with hypertexts the literal text is altered by the act of reading it.
Because of its unfixed nature, hypertext is a temporal form — although stories and poems (and especially plays) unfold over time, a hypertext is created by the temporal experience of reading it. If the text is unread, it is unformed; an unread hypertext exists as a set of potential reading paths (which can be visualized as a map), but the actual text as read is only one of many possible actual texts. One can finish a hypertext without having read every section; this cannot be said of a novel or a play.
The unfixed form of the hypertext consists of multiple reading paths (links) between sections of text (screens).
How is meaning created in such a structure?