[Pages 23-26 in print version. © the Johns Hopkins University Press 1992].
Discussions and designs of hypertext share with contemporary critical theory an emphasis upon the model or paradigm of the network. At least four meanings of network appear in descriptions of actual hypertext systems and plans for future ones. First, individual print works when transferred to hypertext take the form of blocks, nodes, or lexias joined by a network of links and paths. Network , in this sense, refers to one kind of electronically linked electronic equivalent to a printed text. Second, any gathering of lexias, whether assembled by the original author of the verbal text, or by some else gathering together texts created by multiple authors, also takes the form of a network; thus document sets, whose shifting borders make them in some senses the hypertextual equivalent of a work, are called in some present systems a web.
Third, the term network also refers to an electronic system involving additional computers as well as cables or wire connections that permit individual machines, workstations, and reading-and-writing-sites to share information. These networks can take the form of contemporary Local Area Networks (LANs), such as Ethernet, that join sets of machines within an institution or a part of one, such as a department or administrative unit. Networks also take the form of Wide Area Networks (WANs) that join multiple organizations in widely separated geographical locations. Early versions of such wide-area national and international networks include JANET (in the U.K.), ARPANET (in the U.S.A.), the proposed National Research and Education Network (NREN), and BITNET, which links universities, research centers, and laboratories in North America, Europe, Israel and Japan. The fourth meaning of network in relation to hypertext comes close to matching the use of the term in critical theory. Network in this fullest sense refers to the entirety of all those terms for which there is no term and for which other terms stand until something better comes along, or until one of them gathers fuller meanings and fuller acceptance to itself: "literature," "infoworld," "docuverse," in fact, "all writing" in the alphanumeric as well as Derridean senses. The future wide-area networks necessary for large scale, interinstitutional and intersite hypertext systems will instantiate and reify the current information worlds, including that of literature. To gain access to information, in other words, will require access to some portion of the network. To publish in a hypertextual world requires gaining access, however limited, to the network. The analogy, model, or paradigm of the network so central to hypertext appears throughout structuralist and poststructuralist theoretical writings. Related to the model of the network and its components is a rejection of linearity in form and explanation, often in unexpected applications. One example of such anti-linear thought will suffice. Although narratologists have almost always emphasized the essential linearity of narrative, critics have recently begun to find it to be nonlinear. Barbara Herrnstein Smith, for example, argues that, "by virtue of the very nature of discourse, nonlinearity is the rule rather than the exception in narrative accounts" ("Narrative Versions, Narrative Theories," 223. Since I shall return to the question of linear and nonlinear narrative in a later chapter, I wish here only to remark that nonlinearity has become so important in contemporary critical thought, so fashionable, one might say, that Smith's observation, whether accurate or not, has become almost inevitable. The general importance of non- or antilinear thought appears in the frequency and centrality with which Barthes and other critics employ the terms link, network, web, and path. More than almost any other contemporary theorist, Derrida uses the terms link, web, network, matrix, and interweaving associated with hypertextuality; and Bakhtin similarly employs links (Problems, 9, 25), linkage (9), interconnectedness (19), and interwoven (72). Like Barthes, Bakhtin, and Derrida, Foucault conceives of text in terms of the network, and he relies precisely upon this model to describe his project, "the archaeological analysis of knowledge itself." Arguing in The Order of Things that his project requires rejecting the "celebrated controversies" that occupied contemporaries, he claims that "one must reconstitute the general system of thought whose network, in its positivity, renders an interplay of simultaneous and apparently contradictory opinions possible. It is this network that defines the conditions that make a controversy or problem possible, and that bears the historicity of knowledge" ( 75). Order, for Foucault, is in part "the inner law, the hidden network" (xx); and according to him a "network" is the phenomenon "that is able to link together" (127) a wide range of often contradictory taxonomies, observations, interpretations, categories, and rules of observation. Heinz Pagels's description of a network in The Dreams of Reason suggests why it has such appeal to those leery of hierarchical or linear models. According to Pagels, "A network has no 'top' or `bottom.' Rather it has a plurality of connections that increase the possible interactions between the components of the network. There is no central executive authority that oversees the system" (20). Furthermore, as Pagels also explains, the network functions in various physical sciences as a powerful theoretical model capable of describing -- and hence offering research agenda for -- a range of phenomena at enormously different temporal and spatial scales. The model of the network has captured the imaginations of those working on subjects as apparently diverse as immunology, evolution, and the brain. The immune system, like the evolutionary system, is thus a powerful pattern-recognition system, with capabilities of learning and memory. This feature of the immune system has suggested to a number of people that a dynamical computer model, simulating the immune system, could also learn and have memory. . . . The evolutionary system works on the time scale of hundreds of thousands of years, the immune system in a matter of days, and the brain in milliseconds. Hence if we understand how the immune system recognizes and kills antigens, perhaps it will teach us about how neural nets recognize and can kill ideas. After all, both the immune system and the neural network consist of billions of highly specialized cells that excite and inhibit one another, and they both learn and have memory. (134-135) The immune system, like the evolutionary system, is thus a powerful pattern-recognition system, with capabilities of learning and memory. This feature of the immune system has suggested to a number of people that a dynamical computer model, simulating the immune system, could also learn and have memory. . . . The evolutionary system works on the time scale of hundreds of thousands of years, the immune system in a matter of days, and the brain in milliseconds. Hence if we understand how the immune system recognizes and kills antigens, perhaps it will teach us about how neural nets recognize and can kill ideas. After all, both the immune system and the neural network consist of billions of highly specialized cells that excite and inhibit one another, and they both learn and have memory (134-35).
