Hype - Hyphen - Hypertext
Hypertext Fiction in the Classroom
by Dean Tacuich

One of my goals in my recent English 202, "Texts and Contexts," which took as its context Postmodernism(s) was to look at the mechanisms by which meanings are created in a literary text. I chose a variety of texts from the Norton Anthology of Postmodern Poetry and The Norton Anthology of Postmodern Fiction with several goals in mind, one of which was to generate a questioning critical attitude about prescribed meanings (including the rather deeply engrained idea that texts are puzzles to be solved) and seemingly unmediated language and image.

I chose texts which require a degree of reader involvement and reader participation in the making of meaning, and which make the medium of meaning-making visible, often by using unfamiliar mechanisms or by rendering the familiar mechanisms unfamiliar. Hypertexts, I felt, might be useful in this regard, and they were, but accidentally so. A hypertext is one way of approaching these issues, but the internet is not. The Norton anthology makes available, on its website, two password-protected adaptations of hypertexts (one must buy a copy of the anthology to get the password): a selection from Michael Joyce's afternoon: a story, and Jane Yellowlees Douglas' "I Have Said Nothing."

The WWW is not hypertext

I placed the hypertexts fairly early in the term, at the end of a section on literature and performance; the idea was to highlight the participatory aspect of the hypertexts. In this, the online versions of the hypertexts were successful: students simply had to interact with the onscreen text, even if that just meant choosing from a set of links, as is the case with the Douglas hypertext. The performative aspect of hypertextual fiction is fairly clear, I think, but as many of my students pointed out, this kind of interactive reading reminded them of the "Choose Your Own Adventure" stories many of them had read as children. In other words, interactivity is nothing new.

Of course, interactivity is in fact very old: oral literature is interactive, and in a sense electronic texts do seem to have a kind of orality to them. Not spoken, certainly, but fluid, as stories in oral tradition are, and participatory: the readers, like an audience at a small-scale event, are not mere spectators, but are in fact part of the performance. (I introduced this performative idea to the class with Jerome Rothenberg's "New Models, New Visions: Toward a Poetics of Performance" essay).

This participatory reading of hypertext fiction does create an interesting pedagogical problem: the lack of a "common" text. As George Landow notes in his introductory essay in Hypertext Theory ("What's a Critic to Do?"), hypertexts are "abundant"█ in many cases, they cannot be read through, and as each reader chooses her own path through the text, there is no one reading that an instructor can assume all have completed. My reading of Joyce's afternoon, for example, included a lexia which consisted of a Robert Creeley poem. Great, I thought, we've discussed Creeley; here's a point for class discussion. No one else found the Creeley poem█and in fact, I couldn't get back to it either.

In practice, this lack of a common text fit in with the material we had been covering in the course. Rothenberg's essay on performative paradigms includes the stipulation "no more masterpieces." Because of the transitory nature of a performance-based art, one which requires (not merely allows) reader participation, the concept of a unitary masterpiece dissolves: there are no masters because there are no servants. Or, conversely, everyone is a master (which also destroys the concept by democratizing it out of existance).

In a hypertext fiction, a single "author"itative reading is not possible, since it would require every link to be followed and every possible combination of linked pages to be accounted for. George Landow, in his introduction ("What's a Critic To Do?") to Hypertext Theory points out that linking of any kind "blurs a fundamental difference█fundamental in the Gutenberg world█between texts designed to be read through and those designed only for consulation" (18). In this sense, hyperfiction may have something more in common with instruction manuals and electronic dictionaries than with traditional print fiction.

Still, the lack of a common text is not new. Like the "choose your own adventure" model or the indexed reference book, this paradigm can be found in print texts as well, though transparent literary realism, straightforward narrative structures, and our own reading and viewing habits tend to obscure this lack and comfort readers (especially critical readers) with the illusion that we have, with patience and effort, mastered a text.

Some texts, including the ones I used in the Postmodern course, complicate mastery by providing an excess of detail, or refusing to provide the structure in which one expects such details to be embedded. William Gass' "In the Heart of the Heart of the Country," for example, is a kind of print hypertext, in that the similarly titled sections are "linked" (via the reader's experience of having read them). No dominant narrative line emerges, though plausible ones can be constructed. Joyce Carol Oates' "Turn of the Screw" is another print hypertext: the title links to Henry James' story, and students who are familiar with the James fiction will, in a sense, read the Oates story through that filter. The Oates story itself is presented as two columns of text, not quite independent (the narrator of one side is watching the narrator of the other, who eventually becomes aware of being observed). There are several "paths" through the dual columns: most students chose to read them day by day, contrasting daily events from each point of view. But it would be possible to read each side individually, or to break the reading of one side off at any other arbirtrary point and begin reading the opposite column.

The interactivity of such print texts is even clearer in the hypertext model. Obviously, a hypertext fiction requires something more than a passive reader: paths must be chosen, links clicked. But most hypertext fictions do have a default path (often "chosen" via the "Return" key). This default action is the hypertext equivalent of turning the page. But the default is not the only action possible. Hypertext fictions, while variable in design, always allow more choice, and more direct interaction with the textual material, than print text does. The interface is more complex, and, at least for the moment, less transparent than print. The printed page is, after all, an interface, just as a button-laden screen is. As the web becomes ubiquitous, however, hypertext may lose this visiblity or opacity.

In his keynote address at the Hypertext '98 Conference, Stuart Moulthrop stated "hypertext is not (just) the Web," and went on to note that "the trouble with the World Wide Web is that it may shortly become the social equivalent of air conditioning: indispensable, reasonably reliable, and generally invisible to its users." Such invisibilty in print text is sometimes called transparency or (to use Charles Bernstein's term) "absorption." An absorptive text is one in which the reader feels immersed; the page fades from one's attention; disbelief is suspended and finally unnoticed; the constructedness and fictionality of the text, the materiality of the page and of the language itself, vanishes.

The transparency or visibilty of a text strongly depends upon the reader's familiarity with the conventions of the medium█it is learned. To a reader unfamiliar with the conventions of traditional narrative fiction, a short story may be quite opaque (poetry as well, perhaps even moreso). In "Of Theory, To Practice," another essay I used in the Postmodern course, Ron Silliman (commenting on "difficult" poetry) notes that "Once reading strategies catch up to those of writing, a lot of complexity is going to dissolve" (in Postmodern American Poetry, 662). His concern is similar to mine here: readers are already familiar with the point-and-click interface of the web; hypertext fiction offers a good deal more complexity, but even these complex structures can█and will█become familiar. As browsers (the people, not the software) learn to navigate hyperspaces, the at times frightening openness of hyperfiction will probably begin to close. But the web model does not present itself as a model, and the hiddenness of its structure is usually only revealed when soemthing goies wrong. Unlike hypertext fiction, the web strives for transparency.

Hypertext fictions are decidedly not transparent. To illustrate, consider Michael Joyce's standalone hyperfiction Woe. While there is no typical hypertextual fiction, among hypertext author or scriptors, Michael Joyce is perhaps the best known, most likely due to his work on one of the standard hypertext interfaces, Storyspace, which is the interface Woe uses. Upon opening the hypertext reader (just as "browser" refers to a software interface, so here does "reader"), we are asked, first, if we want to start a new reading or continue a previous one, then we see the program loading.  If we choose a new reading, we see a credits page. We can choose to view some directions by clicking in a designated spot, or just click in the general field to begin and bring up the first "page" or "space"█Mandala. As you can see from the screen capture, there is a map visible behind the page. This map can be selected by clicking in it, or by clicking the square in the center of the directional "rosette" of arrows.

The important thing, I feel, is that the map is always visible, always available. Furthermore, by selecting the ? mark in the toolbar, one is presented with an array of options, in which one can view the links to and from the current space, as well past spaces and set any of them as the current space. The options, numerous and confusing as they are, are not hidden. The Storyspace reader provides an excess of information; if the reading were to proceed in a linear fashion (as can almost happen if one simply hits "Return" or chooses the <--> arrow), most of these options would not be needed. The effect of the various options, the visible map and such, is to complicate the reading; an active reading is called forth by having to choose. Even the simplest, most straightforward path through the space will often force the user to select from two or three possible options; several times while trying to move directly through Woe using Returns, I was asked to choose which link I wanted to follow. There is no overall default path; eventually, even the most passive reader will have to choose (and choosing to end the reading, to quit, is also a choice).

The many paths around or through Woe can be illustrated by looking around the space of the first "page". From "Mandala," one can click the <--> arrow in the toolbar (this is a default option), and arrive in a space called "she"but if one selects the --> arrow in the toolbar, a page called "Relic"appears. Clicking on various words will bring one to different spaces. The word "mountain," for example, takes one to a page or screen called "I. Mountain"

The overall structure around "Mandala" is quite complex; most of the first level pages have other pages within them. The section called "Relic" for example, expands into several other spaces.  The page called "relic," however, is not within the Relic structure, but is on another branch (called "second son in NE"), as you can see in the background of the "relic" screen capture.

Another Joyce hypertext available for the Mac platform from Michael Joyce's homepage, like Woe, is Lucy's Sister.  A lengthy area within Lucy's Sister seems to be modeled on internet chat, complete with typos and complaints about scrolling through messages. The reader arrives in the middle of an ongoing discussion, much as in an actual internet chat, and comments seem curiously oblique, as though some messages aren't getting through, or arent't getting read.  Eventually, characters sign off.  Once Ned signs off, the only available link leads to "Eco's Secrete," an image or map of the (entire?) space decorated with a rose ( bringing to mind Umberto Eco's Name of the  Rose). The only link out of "Eco's Secrete" is called "An Actual Ending."  It in turn leads back (through the only possible route) to "Omnet," the opening page of Lucy's Sister.

Lucy's Sister seems a much more limited hypertext fiction, with a clear ending space which leads, of course, back to the beginning space from which one may choose a different path. In fact, one may be forced onto a different path if the reading is a continuous one, since Storyspace keeps track of the paths one has followed in each reading. But the activity of navigating it still retains the visible map in the background (clearly visible in the chat session). With this network of links always visible, not merely available, one can never believe that the pages represent actual internet chat. There is no suspension of disbelief, no confusion as to the medium in which these words are presented.

As a counter-example, consider those web-based ads that often appear in one's browser, ads which are meant to look like, and are animated to imitate, Netscape or IE download progress windows or error messages. Hyperspace in general (including the web) has very permeable barriers (hence the fear of email viruses, trojan horses, and such); it would be easy to compose a hyperfiction in which one believed, for example, that the characters were actual people chatting away in an AOL room or on IRC. But Lucy's Sister does not. The map is a constant reminder of the fictionality, the constructedness of the space. Such constructedness is, I believe, important when considering hypertext as literature.

In contrast to hypertext fiction and the Storyspace model, the web model, in its actual and perhaps especially in its ideal form, strives for transparency. One of hypermedia's idealists and visionaries, Ted Nelson, sees contemporary hypermedia, especially the web, as anathema to the ideal of hypermedia. As founder of Project Xanadu in 1960, Nelson foresaw much of what today is called hypertext, but as he states on his current homepage,

contrary to legend, Project Xanadu was NOT trying to create the World Wide Web. The World Wide Web is precisely what we were trying to PREVENT. We long ago foresaw the problems of one-way links, links that break (no guaranteed long-term publishing), no way to publish comments, no version management, no rights management. All these were built into the Xanadu design.        (http://www.sfc.keio.ac.jp/~ted/XU/XuPageKeio.html)

In an interview given in 1996, Nelson describes HTML as "kind of format where you've got all these warty little knobs and boogers in it that are formatting codes █ this is absolutely contrary to the Xanadu idea that you have clean data undefiled." Of course, browsers hide these codes: the web model is

FTP with lipstick so that you can look at these things and the jump addresses are hidden and the formats and you have paragraph levels and stuff and it's basically what people needed and frankly I think it's much better than word processing. I'm really happy now that I'm planning to switch from Microsoft Word to HTML just because there's no need not to. It's a perfectly good format, and it makes everything simpler to browse in.       (http://www.ics.uci.edu/~ejw/csr/nelson_pg.html)

Such a model is as transparent as a typical word processed document: the formatting codes are invisible unless called up by a command. The typical browser-user (notice here how the browser is not the person browsing, but is the tool they are using) sets preferences once and doesn't need or want to think about them again. The interface stays out of the way. Transparency is what the web model strives after. And in doing so, the web takes control away from the browser-user, the reader. Choices are avoided; the experience is as simple as shopping (in fact, if the web develops as some plan, that's all it will be). Choices will appear as the standard consumer choices: where would you like to go today? The underlying questions (What's the mechanism that brought me here? Where else could it have taken me? Who is in control?) are not asked once one is immersed in the web model.

Since the web model takes control away from the reader, it is less interactive than hypertext itself. One can, however, create a hypertext fiction which functions on the web by utilizing some aspects of HTML which are not available in Storyspace. Stuart Moulthrop's Hegirascope is an example of a hypertext fiction which balances the viewer's sense of control against a pre-programmed "refresh" every 30 seconds. That is, the viewer can interact (via a choice of four links on most screens), but after 30 seconds, the screen "refreshes" and brings up a new page (apparently distinct from the four available choices. This means there are five links from each page: four visible ones and a default). The effect is a bit annoying at first, particularly since the first several pages have refresh times set at 3 seconds, so they dissappear before (or just as) one finishes reading them. These initial 3 second screens have no links, no choices available. There is no navigible map; in fact, one screen bears the message "Where you're going there are no maps." It is about information being pushed in that it is information being pushed--through time, past into future. There are no maps.

Even the pages with visible links are often sarcastic: such statements as "Think fast" and "Read faster" are commenting upon the 30 second time constraint built into the program. Other screens in Hegirascope have hidden links (black text against a black background). One, for example, says, in yellow letters "-click-" but it isn't a link. The links surround it, black on black, revealed only as the mouse cursor moves over them. (The cursor, in Netscape anyway, changes shape when it is over a link.) Clicking also reveals the link, as it changes color during the click itself. And then it's off to another screen. Another screen is totally black; clicking around reveals various links, one asking "Are you pissed off yet?"

Such flaunting of convention, of course, foregrounds the conventionalities themselves. In denying a certain amount of control, a control that most web-surfers have come to expect, Hejirascope makes the control visible. Consider that the most common use of the "Refresh" HTML tag that is usually encountered is the one that takes the viewer to an updated location of an old page (replacing the even more common "Error 404: File not Found." Allowing the program to carry the viewer along a stream of screen refreshes is unnerving; one almost instinctively tries to stop it. As the screen titled "Catalog of Dreams #1004" states (in its second paragraph, so the 30 seconds are almost up by the time a reader will get to this)

Since the universe is infinite, you could push that button indefinitely, which means this dream might never end. This is an interesting thought, but also kind of scary. So you decide not to push the button. Which is when you discover that it's too late...

And, of course, by the time this is read, it is "too late"; the 30 seconds are up and the screen has refreshed. The browser's "Back" button does, however, allow one to return to the previous page.
The hypertexts I chose for the 202 class were not written for the web, but adapted for web presentation on the Norton Postmodern Fiction website. The Douglas piece is presented in a format using standard hypertext links. The Joyce hypertext, on the other hand, attempts to implement a "guard" function (part of the Storyspace reader, the guard function prevents one from visiting certain lexia until others have been viewed) via Java and Javascript, and so required at least a 3.0-level browser. And it did not work correctly on many of those.

The Jane Douglas piece, "I Have Said Nothing," is fairly straightforward in its use of links: readers are presented with several options on most pages, and a map of the present position is available. The maps, however, aren't interactive; they are not clickable as the Storyspace maps are. Eventually, the Douglas story cycles, and the final screen provides a link back to the (or a) beginning screen, from which readers can choose another path. The most interesting aspect of the Douglas hypertext may be the use of graphical elements. Several of the pages are not actually text, but .gif images of text, using (for example) an old broken typewriter font, or vivid color against color. These images take awhile to load, but the slowing down of the experience is actually valuable, since the clickable-link interface is something that readers are quite used to by now.

The Joyce piece, however, is more complex in its use of technology. One of the opening pages (called "a hypertext") gives some of the necessary background on Storyspace. But this version does not use Storyspace; instead the web version attempts to use Java and javascript. One of the features that requires such advanced programming is Storyspace's "guard" function, which prevents a reader from visiting certain spaces unless other spaces have already been visited. In Storyspace, for example, the ? option allows one to view spaces already visited, and see which spaces are available. In the Norton web version, the closest one can come to such information is a Java Applet window available from the "Links" button at the bottom of the screen. The list of spaces in the Java window, however, are not selectable. It is merely a list; one cannot choose to revisit or reload a page from the list. The "Back" option will do it, but since the paths diverge and bifurcate, trying to move "Back" is, or should be, more complicated.

The Joyce hypertext on the Norton site does provide some thoughtful commentary on the use of hypertext, warning readers against the fictionality of closure: "Closure is, as in any fiction, a suspect quality, although here it is made manifest." Would it were so. The web implementation of these hypertexts is far simpler, far more strightforward, than the Storyspace versions. And the technology is hidden; one cannot navigate the space visually by moving around on the map, or by viewing a list of visited spaces and currently available links. The web version is very much like the web itself--geared towards simple consumption, simple delivery of content, simple interfaces and ease of use. Storyspace is complicated, often overly or overtly so, presenting choices one didn't even realize one had to make. And this is, I'm saying, a good thing.


The WWW, that hyperlinked part of the Internet which seems often to be the Internet, is hypertextual, but it is not hypertext. Perhaps it is something more, but in many ways it is something less as well. Joyce's hypertext fictions make regular use of unmarked links and the guard function to create a reading experience that is very much unlike surfing the internet or clicking through indexical lists such as Yahoo. As the WWW increasingly moves towards content-delivery models (Internet Explorer's channels, Netscape's Netcenter, Go Networks, and good old Yahoo), the transparency of the technology increases. A web browser (the person, not the software) wants information, quickly and easily. A poorly implemented search function, for example, will return dozens if not hundreds of irrelevant hits. This is not, of course, a good thing. But in hypertext fictions, such uncertainty may well be an avdantage.

The web model is already familiar, seemingly unmediated. Rarely does one see the "warty little knobs and boogers" of HTML which Ted Nelson complains about. The web isn't really hypertext as Joyce or Douglas uses it; the web is HTML, rendered transparent by the browser. One needn't think about how the connections are made, what codes are operating beneath the visible text on the screen. In the Storyspace versions of Joyce's hypertexts, however, the "map" of the hypertext is always visible, partially obscured by the page one is currently visiting.

The Norton Postmodern American Fiction website's buggy java implementation of Joyce's hypertext, then, provided a means of making the model, the form of the hyperlinked text, unfamiliar again. It made reading unfamiliar, since the outcome of each click was in doubt. The technology becomes visible when it fails; it reminds us of its strangeness; we see that it not only has form but it gives form, and that this form is not something that just happens. Someone made this. We don't always think about formal models when they function as planned. Equally, we should view with suspicion any model of hyperspace/hypermedia/hypertext that seems seamless ("seemless"). Such models offer no choices, no control over the space one finds oneself in. The ability to navigate a virtual text depends on one's awareness of the limits of that textual space. There must be some noise in the system, if only to inform us that it is a system. A perfect virtuality would be a totalizing dream from which we could not awake.