Hypertext narrative

Michael Joyce

Hypertext is, of course, young by almost any measure-- 47 years since Vannevar Bush's Memex, 29 since Engelbart's Augment, 27 since Nelson's hypertext, five years since a number of us gathered for the first of the ACM hypertext meetings in 1987.

Narrative is old. Historically in the Sumer, narrative poems joined warehouse inventories as the first texts known. This is quite fitting since, as Ezra Pound has argued, "The artist is one of the few producers. He, the farmer, and the artisan create wealth; the rest shift and consume it." (The Renaissance, Literary Essays)

Narrative creates wealth. Yet hypertext, thusfar at least, mimics most computational systems in shifting and consuming rather than creating wealth. This is because hypertext thinks itself to be structural rather than serial thought-- thought in space rather than thought for space. "Serial thought," says Umberto Eco, "aims at the production of history and not at the rediscovery-- beneath history-- of the atemporal abscissae of all possible communication." ("Series and Structure," The Open Work, 221)

Hypertext, at least when it is seen as constructive rather than exploratory, is serial thought. Its "mode of spatialization," Deleuze and Guattari's term, is being for space, what I call the constructive, a form for what does not yet exist; rather than being in space, or the exploratory and colonizing. Like AI, its elder sister in this science, hypertext would like to claim its creative powers, but to do so must become the art it claims it is.

"Look," we want to say of the hypertext, "see this traversal, this web, this trail, this graph... No one has ever thought this before. Look how, like pearls on a string, these nodes glitter on their path..."

But always some emperor-baiting child within us will out. "Pardon," she says, 'But why, if this is new, is it there already? Surely you mean no one has gotten to this point in this way before; surely that is what is new."

"No, child," we say, "You don't understand. This substance is an object of value unto itself..." Then she asks the terrifying questions: "Who is the author of this new thing? What will you call it? And where will you put it?"

For the reader of a technical communication, theoretically at least, there is some nethermost node, a gleaming target which represents the meaning of a text. (In reality, of course, this node likely encompasses the span of a series of nodes.) This theoretical terminal node can be systematically described by both its location and its links. Even if the meaning is potentiated-- as an inference or discovery, for instance-- the node is thought to be present and reachable. That is, every reading by every reader is thought to be anticipated by the system of the exploratory text. We might call this belief the myth of emerging order.

Within the mythic system of emerging order, the reader's task is to make meaning by perceiving order in space. This characteristic Eco identifies with structural thought: "The aim of structural thought is to discover, whereas that of serial thought is to produce." (Ibid, 221) In the exploratory hypertext, interaction is recognition; the system of the text is the author of its additions.

Should the reader wish to add her perceived meaning to the document, there is theoretically no impediment, only a question of privilege and location. We determine the reader's privilege and judge the utility of the addition according to the match we perceive between the reader's order and what we take to be inherent order. We ask if a reader can add to the document: Is the summary unique? the inference supported? the discovery consonant? That is, is this addition authorized.

We likewise determine where a reader can add to the document. Once you are privileged, you can write into a technical document if its structure accounts for what you add. The document is seen as the atomic shell within which the valency of each addition finds its weighted place. The addition is a terminal node; the text may be seen as leading to it. Yet the addition is perpetually marginal. Interaction does not reorder the text, but rather conserves authority. Reordering requires a new text, new authority.

The myth of emerging order seems especially instrumental in maintaining technical hypertexts. Because technical hypertexts are thusfar treated like centralized rather than distributed knowledge (manuscripts rather than books), individuated meaning must be marginalized and inherent order privileged. Authority, because it is given, is ungained. Interaction remains a utility for the individual reader, i. e., annotation. Hypertext programming - in the double sense of instructed machine behavior and information content, the thing which, like situation comedy, weather map, or docudrama, is shown on the screen - is privileged, centralized, and self-sufficient.

Hypertext narrative transposes this mythic system. If a technical reading is marginal to its text, the text of a narrative is marginal to its reading. Technical communication is governed by the myth of emergent order, i.e., a belief that the structure of meaning emerges from the text. Hypertext narrative produces the present-tense contour of meaningful structures. Meaning in narrative is an orderly but continually replacement of meaningful structures throughout the text. "Although conventional reading habits apply within each lexia," says George P. Landow, borrowing this term for the nodes-&-links from Barthes, "once one leaves the shadowy bounds of any text unit, new rules and new experiences apply." ("Hypertext and Critical Theory" in HYPERTEXT..."). We judge an addition to narrative by asking if the story continues to "read right," a reader's question, instead of how in technical communication one asks if it fits.

Self-sufficient meaning is not generally characteristic of non-trivial narratives. Only riddles or games (and few of those) offer a target node which truly holds their meaning. In most cases, meaning literally has no place in narrative; the meaning of narrative is not in its space but rather exists for the space of its unfolding. The moral of a story which has one is part of its story, not its meaning. Meaning in narrative is always strictly potentiated; outside and in some sense prior to the text. It is unreachable and thus not yet linked.

"The fable unwinds from sentence to sentence, and where is it leading?" writes Italo Calvino, "To the point at which something not yet said, something as yet only darkly felt by presentiment, suddenly appears and seizes us and tears us to pieces." ("Cybernetics and Ghosts," The Uses of Literature, 18) Every reading by every reader becomes privileged and authorized. Reading becomes (and therefore alters) the system of the text.

The reader cannot avoid adding her own meaning to the document. Addition is not merely privileged but required in order to span to the nethermost, target node outside the text. Interaction is perception of the principles of ordering. Because hypertexts are read where they are written and written as they are read, interaction is the assumption of authority over the replacement of one writing by another. The reader's orderly perception leads to the placement of an apparent meaning.

Under such circumstances, each addition is an initial node to the replaced story she tells in her reading. The target node becomes instead a source node; the text may be seen as leading from it. To accommodate the location of the addition, the text itself becomes marginal; meaning reorders the text. The meaning of a story which had none becomes part of its story. Reordering requires a new text; every reading thus becomes a new text.

Despite these heroic claims, readers are thusfar not particularly apt to participate in maintaining, i.e., co-authoring, narrative hypertexts. Co-authorship (rather than the Disneyite laurels of perceived first-personhood) is the true index of interaction. It is not the case that a reader can only interact if she is convinced she is in her own idea space or story world. Millennia of human experience of empathy (not to mention the most recent 100 years of psycho-biology) not only suggest that I can be lost in your story or your idea, but also that I am lost if I do not make it my own. It is in this sense that the four principles of the TINAC Dryden Statement (1988) suggested a future text:

  1. No interruptions: Reading should be a seamless and uninterrupted experience. Its choices proceed from the expression of possibilities as a narrative medium and depend upon the complicity of the reader in the creation of a narrative. Reading is design enacted.
  2. Any person: Interaction manifests itself through recognition, sympathy, and witness as much as through impersonation, perception, and exploration. Apprehension of character is participatory design.
  3. Every ending: Closure, as it has been described by Andrew Sussman, is "the completion of self by the reader. " It is, in this sense, design determined.
  4. A read-write revolution: Interactive narratives are Óevolving narratives,Ô written, whether by reader or writer. Authorship is an invitation to active design.
Hypertext narratives become virtual storytellers and narrative is no longer disseminated irreversibly from singer to listener or writer to reader. It exists instead as a cycle in which readers become co-authors and artificially intelligent systems "read" their responses. At the moment, most interest centers on the machine side of this cycle, focusing on hypertext or virtual reality systems and intelligent agent/navigators that can structure or generate narratives. However, even in a system which generates narrative, interaction is the assumption of authority over the replacement of one writing by another. Thus the human side of the cycle demands primary attention.

To do so, the primacy of the text must be marginalized. The reader declares independence from the software agent and the contingent structure of the virtual reality alike. If the reader is programming, the reader is programmed. Hypertext narrative asserts the authority of the individual reader, no longer privileged and centralized by the system of the text but rather by her reading itself. "Once we have dismantled and reassembled the process of literary composition, the decisive moment will be that of reading" says Calvino, "even though entrusted to machines, literature will continue to be a 'place' of privilege within human consciousness, a way of exercising the potentialities contained in the system of signs belonging to all societies at all times." (Ibid, p. 16)

There is a danger in asking too much of a technology in its early years (look at AI); but there is a similar danger in allowing a technology to mature unchecked. Change is incremental but also instrumental: it defines as it refines. Hypertext is young; narrative is old. In its pre-adolescence hypertext attempts the impossible: a preservation of hierarchy through managed individuality. It is consumerist at base. Desire given the form of floor plan, the shopping mall or living room as loci of being in space.

In the marketplace at least, hypertext for now remains unsurprisingly a process of shifting and consuming value within a known map and body of material. Whether cloaked as the Multimedia Solution or as part of the jingly joy of "everybody's doin' it, PS2ing it!," our age is one where, as Hakim Bey characterizes it, "speed and 'commodity festishism' have created a tyrannical false unity which tends to blur all cultural diversity and individuality so that 'one place is as good as another'." ("The Temporary Autonomous Zone" in T.A.Z....) Our image of hypertext as multimedia is something like a shopping mall, itself the Platonic form of that form of a graph known as a hierarchical tree. The reader is a consumer on a shopping spree, a squirrel leaping branches, while on a high branch Peter Jennings plays the part of that wise nocturnal bird, Scully's Navigator embodied as the software agent called The Reporter (MacWeek 6:9, 3/2/92).

Our image of hypertext as virtual reality is the wire-frame living room of cyberspace decorated with overstuffed tulip-back wire-frame chairs. Its ripstop, crayola-colored upholstery is dithered and fractal-nubbed. Squatting in air, one is seated in the stretched simulated-man-made fiber, thereafter it is possible to copulate with a lobster, examine a cancer cell, joust with a medieval man, or fly up to the ceiling fixture by moving an imaginary stick. The reader is furniture mover in the carefully modelled three dimensional space of a fractal furniture warehouse, swatting at mosquito vectors in a grid of 3D sound.

The warehouse is the familiar locale of most avant-garde theatre, although this time the props and flats are rendered in light. A virtual universe of attention has been given to modelling what in old-fashioned narrative one called setting and mood. Tediously earnest rule-bases attempt to log precepts of Aristotle, Brecht, and Propp in order to generate lovable Fujisney software-agent action figures, which in old-fashioned narrative one called characters. Yet not enough attention has been given to understanding narrative at the interstices, i.e., beneath the virtual seat cushion where a jumble of spilled tokens, sesame seeds, abracadabra apparatus, and salt changes lobsters to princes, transfigures light fixtures to nebulae, and makes medieval jousters weep in sun bleached parlours.

Jay Bolter suggests that AI is nothing more or less than a machine instantiation of drama. "Artificial intelligence programmers are themselves engaged in a perpetual search for the author," he writes, "they will admit to being the authors of their programs but they refuse to take credit for the output... If the program writes a story..., they look for the author in the program itself... which for them constitutes intelligence." Yet as such AI is potentially an art which women and men and their machines (but not machines themselves) can use to create value. "At all times the computer, like all previous technologies of writing," says Bolter, " is intelligent only in collaboration with human readers and writers." ("Artificial Intelligence," in Writing Space, 191 & 193).

We make a like claim for hypertext narrative, whether as Eastgate reader, expanded book, multimedia, or virtual reality. In doing so, however, we must own up to our responsibility - even pledge our fidelity - to the forlorn, misunderstood and lonesome machines who increasingly live both for and with us. We are so dazzled by the arrival of the cyborg at the eternal costume ball that we begin to think it can dance barefoot and without us. We form ourselves in instamatic memory and camcorder consciousness, situate ourselves upon the hypertextual subconscious beneath the interstitial seat cushions. Snapshots don't watch themselves, nor stories tell themselves. Narrative is not an application but the ur-system of hypertext. Our emperor-baiting child asks, "Who is the author of this new thing? What will you call it? And where will you put it?" Doing so, she uncovers her disguise, disclosing herself in her three aspects as our old friend and goddess, Mousa ex machina. The muse child is revealed in her aspect as the old woman; her question is one of consciousness; it is asked and answered in the asking by mother Narrative herself: You are the author, she murmurs, you call the new work history and you place it in your mind, your mind in it.

Narrative is the series of individual questions which marginalize accepted order and thus enact history. Hypertext links are no less than the trace of such questions, a serial conversation with structure. "If structure is [thought to be identical] with the mechanisms of the mind, " says Eco, "then historical knowledge is no longer possible." (Ibid, 232) We redeem history when we put structure under question as narrative and hypertext do in their essence. The true history is under question and thus, in some sense, outside History. "Is it possible to imagine an aesthetics that does not engage, that removes itself from History and even from the Market?" asks Bey, "or at least [one that] intends to do so? which wants to replace representation with presence? How does presence make itself felt even in (or through) representation?"

Hypertext is authentically concerned with the serial rather than the structural; with being for space rather than in space; with consciousness rather than information; with creating knowledge rather than the mere ordering or preservation of the known; with presence rather than representation. The value produced by its readers is constrained by systems which refuse them the centrality of their authorship. What is at risk is both mind and history.

Center for Narrative and Technology Jackson Community College Jackson, MI 49201

An earlier version of this essay was presented in the panel "Hypertext, Narrative and Consciousness" at the ACM Hypertext Õ89 meeting, November 6, 1989, Pittsburgh, PA.

Texts cited

Bey, Hakim. (1991) T.A.Z. The Temporary Autonomous Zone, Ontological Anarchy, Poetic Terrorism. Autonomedia, Brooklyn, New York.

Bolter, J. D. Writing Space: The computer, hypertext, and the history of writing. Hillsdale, N.J.: Lawrence Erlbaum and Associates, 1991.

Calvino, Italo. The Uses of Literature. translated Patrick Creigh, Harcourt, Brace, Jovanovich, 1982.

Deleuze, Gilles and F'lix Guattari (1983). The Smooth and the Striated. in A Thousand Plateaus: Capitalism and Schizophrenia, University of Minnesota Press, Minneapolis.

"The Dryden Statement," Broadsheet and electronic minifesto, TINAC Report, Dryden, New York and elsewhere 1988 and after.

Eco, Umberto. The Open Work, translated Anna Cancogni, Harvard University Press, 1989.

Landow, George P (1992). HYPERTEXT: The convergence of Contemporary Critical Theory and Technology. John's Hopkins Press, Baltimore.

Pound, Ezra. Literary Essays of Ezra Pound, New Directions, 1954.