Welcome to "the late age of print" (David Bolter). But this age is in no hurry to end. The mixed company, the intermediate existence of hypertext amid print and books, is likely "permanent," not a passing stage. Our reading and writing skills need to become more complex.
In the work many of you at KMI have done, you've given a lot of attention to presenting argument and knowledge structure and to extracting structure from collections of data and documents.
I have been more concerned with intentionally authored hypertexts with structures that are complex or contested enough that no straightforward map or hierarchical structure will present them adequately. In "Scholarly Hypertext" (ACM, Hypertext '97) I talked about these as jointly composed complexes of texts where no one owned the whole and no single author structured it. But in this talk I am more concerned with structures that result from intention and construction by a single 'author,' although the author could be a group and the text could be a part of some still larger net.
For examples showing argument structures of different complexities: compare the Storyspace maps of the simple "Hypertext as Resistance" (an unpublished hypertext essay) with the more complex maps in Socrates in the Labyrinth (Eastgate Systems) where the argument structure cannot be captured on any single screen.
Hypertext can make argument structure evident; for example, stretches of argument can be separated yet linked, spaces can be provided for backing and sub-arguments, maps and outlines can clarify argumentative lines, typed links can indicate rhetorical "moves," annotation tools can offer directed questioning, and so on. There are many problems to be worked out, but, presuming we have worked through those problems and developed good tools for presenting argument structures, what do we do with them?
Argument is supposed to have persuasive effects on the reader. For that purpose is there any reason to use hypertext rather than printed books? There might be presentational advantages to hypertext as a mode to store knowledge and present it. These advantages might come from linking as a means of laying out structures, from "live" outlines and summaries, and other similar techniques.
But suppose we want to be "more fully hypertextual"? Can we still have persuasive effects? By "more fully hypertextual" I intend no essentialism. Hypertext is a technology, not a literary genre; it can be used in many different ways. Still, any medium makes some things easier and some things harder, thus tempting us to exploit features of the medium. The feature of hypertext I want to highlight here might be indicated by saying that hypertexts have no edge of the page. We are accustomed to the idea that books have no fixed length; the book can grow indefinitely. In hypertext the margins are endlessly wide. There is always more space for sideways moves and remarks, space for new kinds of moves and relations. Any intended sequence can get lost when the proliferating marginalia become the text.
"Is rhetorical interchange possible in a form that works against the linear-sequential argumentative structures that have traditionally underpinned rhetoric? If an author fully deploys the resources of hypertext, can she still present a point of view for critique and analysis, or is she limited to posing questions and raising issues without asking the reader to try on an answer for size?"
(This quotation is from Rhetorics of the Web: Implications for Teachers of Literacy by Doug Brent at the University of Calgary. My emphasis.)
Elsewhere in the same text Brent says: "According to my reading [David Kolb] is not convinced that hypertext is a very good medium for argument as we have come to know it, and neither am I." I'm more ambivalent than he says, but I do agree that the more "fully hypertextual" the text becomes the harder it is for the text to have a straight argumentative effect. Is that a good or bad thing?
A popular model for hypertext is: Exploration! Find out new information and test variant concepts! But what if you aren't exactly searching for data and you already have definite concepts and propositions that you want to urge on the reader?
Take as an example of intermediate existence a current work in progress of mine. This is a long text containing several types of writing (description, narration, polemic, constructive arguments of which some are fairly concrete and some very abstract background. The work concerns "places" today (where a place is more than an area; a place is correlated with human actions and norms). I want to deny certain accusations against modern and postmodern places, claiming these rest on bad presuppositions about the nature of place. I will urge new concepts while admitting contemporary problems and suggesting some guidelines for remediation. The topic has strong divergencies, polemical debates, and conflicting and complimentary partial descriptions that don't integrate easily into one final view of the topic. It is intended, also, for different audiences: philosophers, architectural theorists, planners, writers about place, debaters about mod/postmodern, and others. Because of this variety the project seems natural for linked hypertext presentation.
But I want effects on the thinking and doing of specific audiences who have not encountered hypertext, and may have little desire to do so. Nor are they literate in hypertext. Even worse, as Brent says, hypertext may actually undermine my desired effect.
What are those effects? I want the audience to accept some conclusions, to assent to certain propositions, to adopt some concepts -- more concretely: to give up certain complaints, to propose certain policies.
What then do I want from the reader? I want critical evaluation and agreement. I'm not just setting out structure to be strolled through and admired -- not: "ooh, how complex and interesting!" Nor am I building a structure to be explored and discovered as in a mystery narrative -- not: "now I see the whole structure that was presented only in glimpses!" That's a helpful result, but not enough. Nor am I just creating a world that readers try on -- not: "so this is what it feels like to have this point of view." That too is useful but not enough. Contemplation of structure can leave the reader unaffected or perhaps changed unconsciously by a kind of osmosis, uncritically. I want my arguments to offer conclusions that ask for assent, not osmosis. There are specific assertions and propositions to be judged and assented to. The reader should exit from the text changed, but in a self-critically evaluative way.
So I want to make sure that the reader encounters conclusions and arguments as items to be judged. The conclusions should not be accepted based on their source or textual prominence. They are only as good as their arguments -- their own merits. A lot of ingenious structuring is needed to make sure that the propositions and arguments do get encountered and evaluated.
To assure the textual prominence of the argument I may need to provide textually privileged concepts, and a clear unified voice with privileged textual gestures. At the very least I will need privileged textual locations: summaries, tables, diagrams, concluding chapters, overviews, maps, etc. But these have problems in hypertext: because such textual elements can be multiplied, or commented on, or undermined. In the edgeless text where there is always more space and more kinds of spaces available for moves and meta-discussions; the argument may not dominate the text. The node that asserts its own authority becomes at most one element within the net or web.
This is not unique to hypertext; it is true in the Library where there can always be more books commenting on your book. But in hypertext the single work may have no edges. There is that tempting space as you write. The other discourses, and maybe the other of discourse, that surrounds the argument becomes more visible.
Now, these problems can be "handled" by ingenious structure. There are ways to structure a single-author hypertext so as to force textual prominence and authority and a clear view of the argument and curb excessive non-linearity. But the more you use such devices to structure the reader's experience the less reason you may have to use hypertext at all.
Consider again my work-in-progress example. I try to rebut attacks on today's mode of living in space (real or virtual). These attacks argue that we live today in no-place, that true dwelling in a thick environment has been destroyed, that a sense of locality is replaced by simulacra, that formless sprawl forces one-dimensional uniformity of place dotted with theme parks that pretend to difference, and that all this is caused by some global process (capitalism, technology, semiotics, or whatever).
There is much truth to these attacks, but we should beware the pre-supposition that true dwelling requires rooted organic centered places, or the opposite pre-supposition that our true existence belongs nowhere, that our dispersed inhabitation can only be ironic.
In the work-in-progress I try to study the general conditions for places to exist. This involves examining the nature of the grammar of place, as well as some abstract considerations about being and change in cultural objects. I argue that we find new kinds of unities and places and dwelling, for instance, some types of distinction and discontinuity have always been a condition for an area to be a place. But now, more extreme discontinuities are emphasized within new kinds of place-unity.
This means that we can use some qualities of contemporary places (discontinuity and linkage) against others (simplification and flattening of identity). And so, in dealing with current simplified places (such as suburban sprawl) do not idealize the little village, nor the imperial city; rather, complexify, de-serialize or non-linearize, find or make links.
I remark in passing that there are obvious connections here to discussions about the relation of hypertext to more familiar texts.
This work-in-progress text, with that core argument, is aimed at several audiences and it is surrounded by some other texts (narratives, polemics, applications, etc.). Given the varied audiences, a mixed strategy seems called for, print and hypertext.
Currently I have an idea for a mixed format. Imagine a large archive hypertext and a book which offers a slice of the archive (or several smaller books offering different slices). The archive is on a CD tucked into the book (or it could be on a web site, or both -- a CD would bring closure but perhaps encourage more careful reading and attention; a Web site would bring more access and updatability, and a kind of serendipity for readers using search engines).
Readers would have both the archive and the book that had explicit links into the archive. The printed book would provide accessibility, "solidity," convenience, influence, ubiquity and durability. The hypertext would provide linkage, clarity-complexity, freedom of invention and structure.
But the real issue is the mixed content. What would be in the archive other than the book differently presented? Certainly it would include the text redone into separate nodes, and multiply linked with its structures strongly mapped by outlines, summaries, and navigational aids. It would have more explicit structure and links than the book.
But why bother? The archive must be more than an alternative re-presentation of the book. What else might it be? We could add more stuff: more pictures, perhaps other essays that were extensions of the ideas or more detailed concrete examples and applications. There could be critical dialogues, and further explorations of the bases and background of the conclusions. This is all fine, and it would be a way of offering valuable context, especially if there were several smaller books aimed at different audiences. But is it enough to justify making a hypertext?
Could we do more? Be more "fully hypertextual"? Why try? What would we gain for the effort?
There could be explorations of alternative ways of posing questions, divagations, distractions, tangles and multiple voices that feel their way around the topics but don't lead to conclusions. That feels more hypertextual, but what then happens to the desired effect? What happens to those propositions and recommendations? Do they get surrounded by this cloud of stuff? Have we split the text into The Structure and The Cloud? Into The Argument and The Play?
To answer these questions we need to look more closely at a presupposition that has remained unexamined so far. How does structure fit into being "more fully hypertextual"? Indeed, what is structure?
"The fact is buildings don't fly. They don't crawl; they don't whirl; they don't move an inch. They exist as structure. . . . Structure is how architects impose their will on chaos. They make things that stand and are ordered in a specific way, with a sequence of rooms that mean something or dictate the way human beings move through them. This is exactly what a novelist does when he or she structures a narrative. You dictate the way in which somebody experiences a story. It's a way of taking the flux we're working from and paring it down to something significant.
I like Decon. It's architects experimenting with the unmoving and the built, seeing how far they can push it before it ceases to make sense, how close to chaos they can make a building and still render it meaningful. That is what I do with a narrative. My books always have a story, but they're about as far from traditional storytelling as you can get without lapsing into meaninglessness.
Delirium is serialized on the Internet, where people can decide how to read it. I am asked, "Doesn't it appall you to give so much artistic will to the audience by allowing them to navigate your text as they choose? Doesn't this diminish your powers?" It doesn't. I consider the works of architects analogous. The architect designs a floor plan; he doesn't dictate the order in which the rooms are to be experienced. He gives over the options of navigating that building to its occupant. That doesn't make the architect any less of an architect, any less the author of a building. The walls are set in place. The plan is the plan. Similarly, my book on the Web has an unvarying plan. You can navigate it anyway you like, but I wrote it."
(This quotation is from an interview with Douglas Cooper, "The Plot Thickens,"Architecture, July 1998, 43-51. My emphasis. Cooper's novel, serialized on the Web, is Delirium (Hyperion, 1998).)
So far I've been equating hypertext structure with mappable link relations just as Cooper equates architectural structure with the plan. But this is inadequate. Cooper's room example is incomplete. Besides the plan there are the norms. Besides the nodes and links (rooms, doors, walls, passages) there are the norms that define what the rooms are. This normative structure is the grammar of a place, a set of norms for appropriate use. Dining room, study -- these involve norms as to what one should do, not what one will do -- you can dine in the bedroom, but it's still normatively a bedroom. Cooper is not the author of those norms. They do change but not easily by individual action.
These norms and the plan may be at odds. Two architecturally distinguished spaces in the plan might be part of the same normatively functional space. For what is appropriate to do there the spatial distinction might be irrelevant. Or it could be the reverse, one architectural space on the plan might be normatively divided.
So too in hypertext (as in any text) there's a normative structure, not just links. There are sets of norms defining appropriateness. This bit is this sort of text, to be used in this way, this link or pattern of links means we should judge that node as a conclusion. There are speech acts, rhetorical moves, textual conventions.
So there is link structure and there is normative structure. Often the link structure is designed to express the normative structure, but an author could set the two against each other, or use link structure to set normative structure ajar.
This is an old game in literature and visual art, which have been busy for a century twisting formal structures and playing them off against the norms and conventions of "painting" or "gallery" or "museum."
Architecture can do this too: plan against the plan. Consider these examples taken from the experimental houses of Peter Eisenman:
In Eisenman's House 3 we have an intersection of plan structures with one another.
A yet more radical twist can be found in his House X.
At first this plan looks more normal, though somewhat fractured. Notice, though, how the plan breaks into four quadrants. This is more obvious in a three-dimensional view of the house.
What Eisenman then does is to pit this formal division in the plan against the functional division of the house.
Eisenman remarks that:
"Traditionally, the initial demands made on any form are always, how does it conform to its prescribed function and what does it mean? Thus a system of spaces which is articulated in four parts as in House X would traditionally be expected to house four different functions (or four similar functions requiring separation) or four different meanings. Moreover, the viewer would expect to be able to understand the particular articulation of each in functional terms. But as long as his primary expectancy is functional the user cannot begin to structure his experience into any 'other'-based order. One way to subvert this kind of expectation is to set the functional order against the formal order. In House X, this was achieved by a deliberate separation of proximate functions. Thus, spaces with the same or related functions are not placed within the same quadrant but instead are divided between quadrants -- it is thus necessary to move from unit to unit rather than within a single unit to accomplish a single purpose -- and vice versa. In other words, functions are made to straddle forms rather than to fit neatly into them. For example, movement from kitchen to dining room or from public space to another similar public space is from unit to unit, while conversely movement from public space to private space (for example, living room to bedroom, private study to family room) is within the same unit."
(From Peter Eisenman, House X (New York: Rizzoli, 1982), p. 90. My emphasis.)
Literary hypertexts sometimes create this kind of collision of form and norm within narrative structures (see, for example, Stuart Moulthrop's Victory Garden). So, being "more fully hypertextual" can mean adding 'more structure', but also playing with this doubling of structures.
But then the question becomes: Can this kind of play be of any use in argument and rhetorical structure? Or is it useful only in narrative and literary experimentation? What effect might be sought by doing this?
Think of structures of appropriateness in exposition and argument, norms of what follows, what counts as backing. Core argument structure -- the logic and math side -- is more a priori, but the logical structure must be combined with speech act norms in order to be useful. Could one play with those norms, turning link structures against them? This is different from simply multiplying arguments or rhetorical moves.
Here are some hesitant examples -- hesitant because we don't really know how to do this -- deliberate multiplication of arguments and paths that puzzle the rhetorical norm, reuse that forces the same nodes into different rhetorical roles at once, breaking the flow of expectation with interruptive meta-voices doing unexpected or inappropriate things, statements that don't fit the usual norms but suggest new ones, stretching implications, showing how other things might follow from odd readings or odd uses . . .
These are akin to the standard deconstructive gestures of giving multiple readings and creating or finding textual undecidables. The deconstructive parallel reminds us that this play can be and is done in print too. Maybe it works even better there because there is no explicit map, so even the formal structure can be left undecided. On the other hand, in hypertext the explicit link structures provide a new tool, a new space for such maneuvers. This may mean fighting against the atomism of hypertext -- its boxes and arrows -- even as that atomism gives us new tools. We can turn them against themselves, refuse to take them as uncontested background, make them explicit, play them off against themselves and against the normative structure.
Supposing such play is possible, why might one want to do it? Is there any worthwhile effect for argumentative/expositional hypertext?
We might do this to make plain the roles things play, to show the difference between the two kinds of structure, to show how they depend on each other and stimulate each other. Or we might do it to criticize the norms, to make norms self-aware so that they can be judged or adjusted. We might try to catch a nascent new function (as Frank Lloyd Wright's open plan houses at the beginning of the century assisted at the birth of a new mode of domestic life). Or we might make explicit by twisting it how we inhabit the text. We might make that inhabitation self-aware and possibly self-critical, or assist at the birth of a new kind of textual inhabitation.
For argumentative texts, in my field of philosophy, there are some examples that play with the usual conventions and norms. Think of Kierkegaard, Nietzsche, Wittgenstein, Cavell. Nietzsche is usually cited in this connection, and he does deal with these issues, but his texts are so self-consciously fragmentary and depart so from the usual norms that they put less explicit pressure on them because they don't use them.
In some ways Kierkegaard is a more promising example. He does argue explicitly, but his terminology is not consistent and he lets unexpected things follow on one another in puzzling ways. His authorial personae vary and comment on one another. "His" propositions don't go easily together, but neither do they just contradict each other. They remain in a curious tension that is not quite that of opposing arguments. He shows the reader by example how to occupy different positions. He generates multiple conceptual structures as a way of doubting the adequacy of conceptual structures. He wants us to learn to inhabit conceptual structures in new ways.
So here is a possibly desirable effect for a more "fully hypertextual" argumentative hypertext: to judge and assent to conclusions, really hold them, but also to be self-consciously critical about the act of judging and assenting. And this might be -- the Kierkegaardian moment -- without this new inhabitation itself being based on allegiance to a still larger structure.
It takes a lot of structure to question the role of structure. The "more fully hypertextual" presentation is about getting the reader to inhabit the text and the argument differently. This is behavior in mixed company that hypertext might be proud of.
Of course, all this puts great demands on the skills of the reader and the writer. They have to deal with larger interwoven nets and shorter nodes, overviews, typed links, lots of structure and navigation, as well as all this textual play. The reader and writer need to be aware of structure of multiple levels and kinds, and of the process of norm creation; they have to understand and change and judge and relate and hold it open -- to be there in place and to be aware of the processes of being in place. They are not just contemplating the structure but participating self-critically in the structuring act.
There's a familiar danger. The reader may skim. The influence of the web is pernicious here (look at Nielsen's studies). With a page of Kant's or Aristotle, the argument just sits there looking impenetrable and challenges you to read it again and again. With a hypertext node you feel that if it is difficult, you can move along and something else helpful or interesting will show up soon. There need to be strategies for slowing down the reader, strategies for navigation and location, and contemplation, as well as readers who know how to respond to those strategies with the right kind of active attention
However, there is also a less familiar danger. Currently, the hypertext writer can get off too easily. It's too easy in hypertext to make a lack of discipline or insufficient thought look superficially profound. It's too easy to use shocking transitions as a substitute for questioning and to substitute easy association for complex relationships. It's too easy to substitute facile meta-moves for complex plays with structure against itself. In other words, and again parallel with places today, it's easy to make a temporal or spatial series of simpler intensities and structures instead of more complex interwoven identities and structures.
Hypertext could help us avoid this, because the habit of hypertextual active attention to linking and structure could enhance our inhabitation of place as well as text. But the audience and authors must learn together to develop both the new textual objects and the new literacy skills. We need to experiment with writing and reading in the more open space provided by mixed company, seeing what happens, and who we become.
Date created: August 19, 1998 Last modified: August 21, 1998 Copyright © 1998, David Kolb