Hypertext invites traversal across a set of texts in an order chosen by the reader. The links between texts may be traversed in either direction, or a link may be reversible. These possibilities change the process of reading from the transit through a so-called linear text, which appears to be a one-way process. In fact, reading a literary text from beginning to end may involve processes that make it non-linear in several important ways: settings recur and are compared across sections; characters develop in unpredictable ways that call for retrospection; the theme of the text may be felt early in reading giving it an anticipatory or emergent quality. But literary reading, although it includes these non-linear processes, in general appears to be non-reversible, involving processes of modification and transformation. Although at the level of plot events may be encountered by the reader within a cause and effect structure, aspects of which may be retraceable (the lost can be found, mistakes can be corrected), most phenomena are clearly non-reversible: the meaning of an event may change in the light of subsequent knowledge, or a concept introduced early in the text may be transformed later on. Literary reading in this respect replicates processes of growth or change that are typical of living systems and that are non-reversible: the chicken does not return to the egg or the flower to the bud.

In coming to understand chemistry it may not matter whether I learn about the properties of nitrous oxide before I learn about the principles of other oxides. If I wish to understand the history of the concept of oxides, however, it is important to know that the theory of phlogiston proposed by Priestley to explain some of the effects of oxidation preceded the new understanding of oxygen put forward by Lavoisier. This acquisition of knowledge is not reversible: the world views made possible by a theory of phlogiston or by a theory of oxygen are substantially different. Once you understand Lavoisier's account you cannot reinstate the theory of Priestley. Developments in knowledge are not merely linear: they modify or transform what was previously known (the more dramatic historical instances were dubbed paradigm shifts by Thomas Kuhn). It is this kind of knowledge that is realized during literary reading.

It is not clear how far hypertext, with its reversible links, could be designed to support such non-reversible reading processes. The chapter of Yellowlees Douglas suggests that when revisiting certain nodes a persevering reader of Joyce's Afternoon finds that different links have become available, which appears to simulate a modifying process; but this example seems a special case: it is not typical of the use of hypertext in general, nor of the typical reading of a literary text. The standard reversible model of hypertext, in brief, seems ill-suited as a vehicle for reading literary texts as we have known them until now. Whether Joyce's narrative signals a shift to a new literary form able to accommodate non-reversibility remains to be seen. We also cannot tell yet whether such narratives would prove attractive enough to readers to displace conventional literary forms.

The argument about linearity is one important theme of the two chapters by Rosenberg and Moulthrop. A second theme which cuts across it is the use of spatial tropes to represent hypertextual reading or thinking, as in Rosenberg's appeal to geometry, or Moulthrop's use of the terms synapse or rhizome. The argument in these two chapters is, at one level, a dispute over topological figures which largely preempt alternative forms of representation. The alternative conception that Rosenberg postulates, with the help of writers from Bergson to Prigogine, is the non-spatial concept of Becoming, difficult to grasp precisely because we have become so used to spatialized modes of representing knowledge. While I will attempt to characterize this version of the concept here as an aid to understanding Rosenberg, my commentary will also derive from my own understanding of Becoming during literary reading in which, as Coleridge put it, we "become that which we understandly [sic] behold & hear."

Rosenberg

Rosenberg situates his critique of hypertext within models drawn from physics, in which time plays two different roles. He contrasts two different conceptions: in the dynamic, which is reversible and deterministic, time is incidental and may proceed in either direction; in the thermodynamic, whose irreversibility is derived from contingency and non-linearity, time only proceeds in a forward direction (p. 269). Hypertext, as it is currently represented by theorists such as Landow, is said to belong to the first conception. Later in his chapter, drawing upon Prigogine, hypertext is also characterized as belonging to the realm of Being (geometric and nonlinear). This is contrasted to the realm of Becoming, seen as the realm of the contingent and probabalistic (p. 276).

Rosenberg argues that hypertext theorists have drawn on these tropes from mathematics and physics to authorize an ideology of resistance and liberation. But he argues that this shows the complicity of hypertext with spatialized, deterministic systems. Although hypertext is said to place the reader in control, structures of nodes and links "create rhetorics entrapped in the necessarily logocentric geometry of regulated time and space" -- a conception that echoes the spatial assumptions of classical rhetoric (p. 274). The vaunted nonlinearity of hypertext can only be realized under conditions of symmetry or reversibility, showing that nonlinearity in practice means multilinearity (p. 275). For the reader, then, the experience of reading is indifferent to the flow of time, with one linear pathway equivalent to any other. In the absence of the irreversible, modifying processes of Becoming, hypertext reading actually represents a form of stasis.

Rosenberg works out the correspondences between the geometrical commitment of physics and hypertext reading in detail (pp. 279-285). He argues that in reading hypertext, although the jump across the link may itself be dislocating, "a normalcy emerges as . . . the hypertext reader, acclimates to the new geometry or new sequence of lexias" (p. 283). It is worth noticing that this claim can be verified with the account of Yellowlees Douglas: in her readings of Afternoon her commitment to plot reasserts itself with each reading, thus restoring some semblance of narrative normalcy. As Rosenberg points out, the aim of avant-garde hypertext is to prolong the dislocation that might liberate us from "the hegemony of geometry" (p. 283), but this is a conception already realized in Moulthrop's notion of "breakdown," which I have suggested attempts to build upon some questionable conceptions of reading. The reading that liberates will be situated within the experience of Becoming. But here, Rosenberg shows how the notion of "becoming" that hypertext theorists have appropriated from Deleuze and Guattari is misleading.

Hypertext theorists, he notes, appeal to the "the nomad and rhizome as concepts articulating the contingent condition of 'becoming' that enables resistance to domination" (p. 272). Or, he adds a few pages later,

Logic is hierarchical; nonlinear association is smooth. Here we may wish to resort to Deleuze and Guattari's distinction between striated, arboreal structures and smooth, rhizomatic structures, suggesting . . . that linear, hierarchical structures in hypertext are logocentric, smooth, nonlinear structures are nomadic, and so on. (p. 277)

But this distinction will not work for hypertext, which is unable to transcend geometric structuring. The becoming that Deleuze and Guattari propose is derived from Bergson's concept of duration, the experience of pure subjectivity that precedes any conceptual model of time and space, such as that of Kant (p. 288). Thus, in the model of Deleuze and Guattari "only tactics of subversion that remain indifferent to geometry can produce a liberatory state, a localized death," or, as Rosenberg adds, the horror aimed at by the avant-garde or the postmodern sublime (p. 291), which explicitly repudiates modern, mechanical conceptions of time as well as the inauthentic avant-garde of Cubism or Futurism (p. 290). In this context, Rosenberg suggests, "hypertext, as an avant-garde medium, remains an inadequate vehicle for that transgressive and liberating moment" (pp. 285-6).

Futurism: cf. Lanham

But how are we to understand the possibilities of Becoming in Rosenberg's system? One well-known aspect of Becoming is described by the second law of thermodynamics, that is, the concept of entropy, which states that "given any isolated system, that system will move in the direction of greater entropy or disorder, ultimately reaching stasis, or equilibrium" (p. 284). While Rosenberg notes the extension of entropy theory by Prigogine to unstable systems to show how order can be generated out of chaos (i.e., chaos theory), he does not pursue the biological implications of this concept. Life can be seen as a local exception to the entropic principle which governs the universe: the increasing order and complexity of life forms has often been conceptualized as negative entropy. Thus we can conceive of reading as a negative entropic process, one in which during interaction with a literary text we experience an increase in order, or a process of becoming. However, chaos theory, another physicalist model, is probably not an appropriate framework for understanding reading. Borrowing tropes from physics, Rosenberg notes, indicates that the arts are a poor relation (p. 271): literary reading, which obeys laws distinctive to its own domain, calls for understanding on its own terms. For example, the self-organizing process described by Prigogine and Stengers (quoted, p. 289), in which bifurcation is preceded by a maximum of fluctuations and randomness, seems remote from the transformative processes that occur during reading: here, a new perspective during reading is anticipated by emergent properties at the level of feeling or kinaesthetic imagery, which direct and order the conscious processes of response. In other words, the aesthetic law we seek in the case of reading appears to reverse the physicalist law of chaos theory: such a law must be a realization of our organization as biological systems, an outcome of our neuropsychological evolution.

The disappearance of the intuitive consciousness, described by Bergson as duration, is particularly apparent in hypertext reading. Its loss, in this account by Bergson, translates readily to hypertext -- an application I emphasise by the insertion of additional words in brackets:

In vain, therefore, does life evolve [for the hypertext reader] before our eyes as a continuous creation of unforseeable form: the idea always persists in that form, unforseeability and continuity are mere appearance [beyond the next link], the outward reflection of our own ignorance. What is presented to the senses as a continuous history would break up, we are told, into a series of states [or nodes]" (cited, pp. 286-7).

Becoming, then, is only possible outside the geometrical uniformity of spatio-temporal existence that hypertext has enforced. This, finally, is the lesson of Rosenberg's challenge to the underlying premises of hypertext theory. Whether Rosenberg's neglect of the biological context in favour of physics is the most appropriate way of understanding hypertext is a topic for another occasion.

Moulthrop

Meanwhile, Becoming continues to be the dream of Moulthrop. His theoretical position on hypertext is demolished in advance by Rosenberg, but he seems to acknowledge this only briefly: it doesn't prevent an extended set of claims about the power of hypertext to install a new culture. This would be due not to the communicative possibilities of hypertext but to its "rhizomatic" form. The terms in which Moulthrop frames his encounter with Rosenberg makes this clear. He poses two forms of "resistance" to hypertext, adopting with this term a rhetorical stance able to incorporate all conceivable objections to hypertext, current and to come.

The first phase of resistance, he suggests, is "operational": facing the vastness and randomness of hypertext, users will seek to tame it by imposing structure on it -- either the rhetoric of linking that Landow (1991) proposed, or the personal appropriations foreseen in Ulmer's "mystory." Moulthrop (writing here in 1994) foresees quite accurately the maneuvers of the internet community to tame their vast sea of hypertext: the emergence of structured search engines, such as Yahoo's hierarchical tables of contents, or the personal list of "hot sites" that nearly every author of a home page offers. The second phase of resistance will be ideological: here Moulthrop frames Rosenberg's critique of hypertext theory as "resistance" at the same moment in which he acknowledges the "bad faith" within which hypertext proponents such as himself "have been known to operate" (pp. 309-10). How does Moulthrop respond to Rosenberg's charges? There are two parts to the response, acceptance followed by evasion.

First, he accepts Rosenberg's charge that the supposed nonlinearity of hypertext is really only multilinearity. He agrees that "Such matrices are always edifices, never autonomous zones; they are structures that do not allow for deterritorialization" (p. 310). Implicated in this response, however, are two failures of understanding: Moulthrop does not note the central thrust of Rosenberg's argument, which is that the dynamic systems on which hypertext is based are reversible; nor has he realized the consequences of "deterritorialization" for the topological perspective of hypertext theories such as his own. The mode of becoming that would realize the dream of a new culture requires representations that are neither reversible nor topological, but in Moulthrop's topographical world "deterritorialization" is literally unthinkable.

Moulthrop's second move is evasive: to forward his dream of a new culture he puts in place the "New Dope," a fable of Pynchon: "the minute you take it you are rendered incapable of ever telling anybody what it's like, or worse, where to get any" (cited, p. 311, from Gravity's Rainbow). The dope is perfectly self-defeating: it represents self-enclosure, a failure of community, the ultimate narcissistic imprisonment. It certainly cannot represent either enlightenment or death. In other words, it negates precisely the new culture that Moulthrop would bring into being. In the light of history it offers a secular replication of the absolutism of protestant grace, since "It is the dope that finds you, apparently" (cited, p. 311). For the Plymouth Brethren, for example, the granting of elect status was due neither to good works nor any moral qualities: it was simply endowed on the individual by the act of God. For Moulthrop the appeal of the new dope is that it represents "a true alternative to the capitalist order." This remains a fantasy, however; he cannot offer a route towards implementation. His efforts to imagine the new order of hypertext falter at this point, except for a second evasive move under the heading "Resistance is Futile," in which he retells the story of Karl Crary. But this move too is doomed, since (as I have shown elsewhere) in this conception of hypertext all discourse is at an end.

Moulthrop thus falls back on a provisional, halfway notion, that of transition. Like the political revolutionaries whose strategies he tries to adopt (Jacobins or Bolsheviks), Moulthrop cannot see how the promised land is to be realized except dimly, as a dream, thus progress towards it must be installed permanently (cf. the "permanent revolution" of the Bolsheviks after 1917): "The transition seems likely to be both permanent and perpetual," he announces (p. 317). Yet the only image of that transition state that he can offer is a return to the image of Japan with which he began his chapter (p. 299), a navigation of the streets of Tokyo: "Head east till you come to the Ono-Sendai Building, hang a left at the statue of Colonel Sanders," etc. Such traversal is tied to the geometric as firmly as any of the hypertext thinking that he had already admitted to be "bad faith."

So what function has Moulthrop's insistently repeated "dream of a new culture"? In his final paragraph he suggests that "We may find ourselves one day arriving as the first nomads of cyberspace, voyaging smoothly across the grids of consensual hallucination" (p. 317). This, of course, is consistent with his appeal to the "New Dope" of Pynchon: in a word, a fantasy drawn from hallucinogenic drug culture is deployed to oppose all current and previous forms of culture or communication. It is worth examining a little further the strange positions this allows Moulthrop to inhabit.

First, he repudiates prior forms of identity, suggesting that a new one can be dreamed into existence. In the opening of his chapter he cites Joyce's suggestion that we seem to "be undergoing a change of identity, weaving a fresh social fabric" through a "marketplace of semiotic exchange" which is now pervasive. This points to "a reformulation of the subject, a truly radical revision of identity and social relations" (pp. 299-300) which attempts to "stand outside any stable order, old or new." While Moulthrop calls on Lyotard's suggestion that we can no longer appeal to any metanarratives (master narratives or myths), it is clear that his own fantasy is itself a master narrative, since it takes upon itself to place within a totalizing perspective all earlier forms of identity and culture. "The dream of a new culture is a fantasy of immanent change" (p. 300), he says. The scope of the "fantasy" is shown by his later comments that tie the hypertextual revolution to "opposition to the rightward political drift of the West and the demise of state socialism in the East" (p. 301). While Moulthrop runs into difficulty in being specific about the new culture, as I have suggested, his eschatological claims could hardly be any wider if hypertext is to produce something "outside any stable order, old or new."

Second, the stance of Moulthrop would effectively jettison any of the instruments by which his dream could be realized. The production of a new culture can only be based on a transforming process applied to the old. That is, an existing organ takes on a new role, or a new role is derived from a new relationship between organs already in place. Only an evolutionary model of this kind is adequate to understanding how, for example, the new Greek city state of ancient Athens came into being, or how the American Revolution put in place a resilient new republic during the 1770s. In Moulthrop's anti-evolutionary world we can only dream another culture into existence by a total rejection of the past, but history shows that to do that is to evoke one of the nightmares of history (the state terror of the Jacobins, the Bolsheviks, the Khmer Rouge). A rhetoric dedicated solely to "the designation of places or occasions" (note the topographical figure once again) leads to such instruments of terror as Robespierre's Festival of the Supreme Being, symbol of an order imposed on a society which had ceased to possess any instruments for thinking about cultural transformation.

I offer these historical examples not because hypertext theory of any kind has the power to recreate culture (Moulthrop's fantasy is largely wish-fulfilment), but to draw out the inner logic of Moulthrop's claims: the historical analogies help to show why such arguments are self-defeating. No attempt to create a new culture on the ruins of the past has ever succeeded: history provides not a single example. Lasting changes in culture come about through changes in relationships, existing entities taking on new roles, inversions in valuation, and similar modifying processes. Moulthrop's inability to conceive of such change is shown in several other ways. Here are several further striking examples.

Moulthrop, despite Rosenberg's critique, attempts to appropriate Deleuze and Guattari's model of Becoming for hypertext theory. Becoming, he tells us, is offered by smooth space, in which "the points are subordinated to the trajectory." Paradoxically, however, "smooth social space is mediated by discontinuities. It propagates in a matrix of breaks, jumps, and implied or contingent connections which are enacted . . . by the viewer or receiver." And as one example, Moulthrop puts forward the "parataxis and bricolage of broadcasting": in other words, the advertising culture of television with its "Now . . . this" (p. 303). As a model of hypertext functioning, this seems confused: television is precisely that medium in which breaks or jumps cannot be said to propagate contingently. No advertisement has any contingent relationship with the advertisements that precede or follow it, nor with the program that it interrupts. The rare occasion when advertisers complain about the context in which their products are shown, such as the recent dispute over Nothing Sacred which was said to offend Catholics, is the exception that proves the rule.

Moulthrop then further confuses the issue by suggesting that "interactive media," said to share the characteristics of smooth space, "exhibit the same phenomenological structure as cinema and video" (p. 303). While hypertext is reversible and multilinear, these characteristics certainly don't belong to film or video, which offers only one, linear order of viewing. But the confusion continues with Moulthrop's inability to pin down what exactly a hypertext link or jump really is, as shown in this next comment with its proliferation of terms:

Hypertexts are composed of nodes and links, local coherences and linearities broken across the gap or synapse of transition, a space which the receiver must somehow fill with meaning. In describing the rhizome as a model of discourse, Deleuze and Guattari invoke the 'principle of asignifying rupture' (9), a fundamental tendency toward unpredictability and discontinuity. (p. 304)

The three terms for a link used here, gap, synapse, and rupture, are far from synonymous. A television advertisement could be described as a rupture: its placement within a program (a film, for example) is, in relation to the meaning of the film, asignifying, if you attend both to the film and the advertisement. A gap, on the other hand, would occur from merely disregarding the advertising break (as I usually do by muting the sound and looking elsewhere); normal viewing resumes after an interval. A synapse is a quite different type of space: placed between a nerve axon and an adjacent nerve cell body, a synapse facilitates the transfer of (chemical) neurotransmitters from one nerve cell to another unless the receiving nerve has placed inhibitors to block the exchange of chemicals across the synapse. This mixture of terms points to Moulthrop's inadequate conceptualization of the hypertext link: it cannot be all of these things, and it may be none of them. Only the synapse could be proffered as a genuine model of propagation, but Moulthrop seems not to intend this term seriously since it doesn't recur in the chapter, and he fails to show how it could be considered a model for hypertext linking.

The third (and last) issue I will mention arising from Moulthrop's account of hypertext relates to the nature of knowledge. He typically conflates different kinds of knowledge, a problem that underlines once again the indeterminacy of the dream that hypertext is supposed to realize. Accounts of the formal properties of hypertext are produced as evidence for the kind of knowledge that hypertext makes available, a confusion of form and content that Moulthrop inherits from other writers. For example, according to Nelson, "interactive media will encourage 'populitism,' the dissemination of specialized knowledge within unconventional or unofficial networks" (p. 304). But what is disseminated here is information, not knowledge. Similarly, Moulthrop imagines readers who are free to construct their own networks of links "not sanctioned by the present divisions of culture and discipline -- free to construct idiosyncratic networks of knowledge, or 'mystories'" (p. 305). Both these claims, examined closely, are a justification for the dissemination of misconceived knowledge, with resultant loss of power and contact with reality, and the marginalization of the receiver. Knowledge without an appropriate context for understanding, such as a university degree should provide, cannot but be superficial, disempowering, or dangerous. Again, the large scale hypertext system at Boeing for aircraft maintenance is an information processing system: it cannot perform the transformative role that Moulthrop attributes to it.

Moulthrop cites Pynchon, once again, to sharpen the distinction between hierarchy and hypertext. Leni Pökler tries to justify her model of knowledge as a form of parallelism, represented by "Metaphor. Signs and symptoms," etc., to which the systems man Franz replies "Try to design anything that way and have it work" (cited, p. 307). What Moulthrop overlooks in his discussion of this passage is that the aesthetic realm has realized Leni's way of thinking for centuries. The distinction proposed here is not between nomos and logos (p. 306), as Moulthrop seems to think, but logos and mythos. Myths, as encoded in literary texts from The Iliad through The Rime of the Ancient Mariner to Gravity's Rainbow, are effective because they engage us at a level where identity is genuinely in question and transformation possible. The culture that Moulthrop seeks has been with us all along, and continues to be created. Moulthrop, however, has already discarded it since he believes it to be enslaved to "the monology of print" (p. 308).