Digital text

### 1. The notion of digital text      It may seem a moot question to inquire into the nature of a digital text – the obvious definition seemingly being any combination of words and images displayed on a computer screen. Thus any digital composition may be called a "text," however much sui generis: a website as a whole, or an individual page, or a data base, or even a graphic file with plots or images. The structuring and the scope are different in each case: a page has much narrower confines than a website, and so does a single sheet within a larger relational data base. But they share the electronic medium in which they are communicated.
     If we start, however, from a different assessment of what digital thought properly is, particularly in its differentiation from the electronic dimension, then we will reserve the term digital for a conceptual structure that differs substantially from the standard text structure as rooted in the tradition of writing sequentially on fixed display surfaces (from clay to paper to the screen).
     A database is digital in the sense I am advocating, but it does not in and of itself develop any argument. It is true, and significant, that the very design of the database and the choice of data respond to criteria that must be formulated explicitly, outside the confines of the database itself. But the primary intent of the author is to hand over the key to someone who may then develop a proper argument, within the limitations of the selection and categorization criteria established by the author.
     A website has the greatest potential to be a proper digital text, but in current practice it is so only to the extent that it includes a database. The discursive component of a website, on the other hand, is no different from a printed text,with a regular prose flow, embedded graphics, cross references (in the form of hyperlinks).

Browser edition

The term "browser edition" (which is directly pertinent to a discussion of the global record) may be used to refer to the proper digital nature of a website, seen in its logical and conceptual organization, independently from the (accidental) fact that it happens to be available electronically on the World Wide Web. A browser edition, whether local to a given computer or available on the Web, has a digital potential unlike that of any other medium. And this potential is properly digital. This is what I am seeking to define in the present context.

Hyperlinks - informational and thematic

     Hyperlinks are the hallmark of all browser texts (the acronym HTML embeds the term in its first letter). But there are important gradations in their use.
     The most common type is informational: a given term is linked to an entry that provides additional information about a given topic. A text so construed presents a multitude of pointers which do not specifically serve the purpose of construing an argument. If an argument can be derived from them, this is accidental, it is not proposed by a structural concatenation of the links which remain, in and of themselves, like shots in the dark.
     Different, and more subtle in its implications, is the type whereby an argument is constructed intentionally through hyperlinks. I will call this type thematic in the sense that a given theme is intentionally pursued across the hyperlinks. While an information hyperlink is invoked as a name or a keyword, a thematic hyperlink is invoked as a concept that unfolds a theme and wants to be linked back to its original anchor, where the theme continues to unfold.

A test

A simple test may help to clarify the difference between the two types of hyperlinks. When you print on paper a text that includes only informational hyperlinks, neither the narrative nor the argument are impacted by the lack of the interactive ability to open a link. There is a an informational gap, to be sure, but the argument flows unimpaired. The omission of the hyperlinks' target does not detract from the coherence of the argument.
Try, on the other hand, to print on paper a text with thematic hyperlinks. The narrative flows because it is construed with regular syntax, but the argument is incomplete. It is originally construed as a multi-block or polyhedral argument where all the pieces are interlaced and intrinsically cross-referenced. A proper digital text with thematic hyperlinks has a prismatic nature whereby it needs all the targets to which the hyperlinks point. The printed version of a single page not only robs readers of the ease with which additional information can be obtained, they are in fact unable to follow fully the argument as originally intended.

Structure of a digital text

     Structure is the formal counterpart of an argument. An argument develops in a multi-linear direction, and is structured accordingly. To the extent that a digital text develops an argument, it is also multi-linear. More properly, it is "polyhedral," in the specific sense that multiple lines do not lead only to points, but to planes that are indicative of a larger structure.
     This concept is basic to the articulation of the text as much as it is to the flow of thought. A digital text is fluid but not unstructured. In fact, the farther do the tendrils of multi-linearity reach, the more complex should be the underlying structure that oversees the direction and the interlacing of the filaments. While in a linear text one is oriented in a single vertical direction (as expressed by the notions of "above" and "below"), in a digital, multi-linear text one has instant connectivity with all points of the compass and, more importantly, with the full array of which each of those points is but a part.
     This is made possible by a careful and structured use of hyperlinks. While informational hyperlinks are entered in an ad hoc fashion, thematic hyperlinks are in keeping with an overriding and organic structural system. Thus each hyperlink remains within the purview of the intended structure of the argument, it is born out of the argument itself, it is a pointer to something which is not outside, but is, rather, within the argument itself. Rather than exiting the argument, the hyperlink of a structured digital text brings the "reader" (hence, not a "user") back to the larger structure of the intended argument, to the thematic, multi-linear development of the author's thought. Rather than shots in the dark, thematic hyperlinks of a proper digital text ricochet inexorably back to the start point, illumined in the process by the exposure not of single bits of information, but of multiple parallel argument structures.

Structure vs. function

     One of the most visible aspects of an electronic text is the search function. Conceptually, this is the major distinction between a printed text and its mirror image in .PDF format. It may well be the only conceptual difference – and I deal with it more at length in connection with the notion of digital reading. But it should be noted that it does not, in and of itself, affect the structure of the text, with which we are concerned here.
     In an important way, however, the search function is of relevance, if indirectly, to the digital nature of an argument. Search presupposes structure. At its most cursory and ephemeral, it is the structure of momentary curiosity: a hyperlink elicits an interest in something which is generically within the sphere of interest of the user. As such, it does not aim for a finite end, and what develops is a potentially endless hopping. One may of course venture on unexpected connections that stimulate great thoughts. But this is accidental. Like all intuitions and inspirations, it is of course most welcome, but it is not inscribed within the framework of a reasoned argument. It is not, a function of any given "text."
     At its most thoughtful, on the other hand, a search is oriented towards a specific goal, it discards what is not pertinent, and it responds to the structure of the query as posited in the mind of the inquirer. A digital text, then, channels the attention of the reader in the direction of multiple searches that are all in correlation with each other, as part of the wider structure of the digital text itself. In this case, a search is a function not only of the reader's query, but of the author's as well.

A comparison

     In their introduction to a classic anthology of Max Weber's writings in translation, Gerth and Mills distinguish two distinct styles of writing. "One tradition corresponds to the drift of English towards brief and grammatically lucid sentences... Authors...believe in addressing themselves to the ear; they wish to write as if they were speaking" (p. v).
     "The second group address themselves to the eye of the silent reader," they believe that "a long period bespeaks of greater deference for the reader than do twenty short sentences. In the end the reader must make them over into one by rereading and recapitulation." (p. v, the last sentence being a quote from a German writer). The two authors further characterize this style by saying that the writers of the second group "use parentheses, qualifying clauses, inversions, and complex rhythmic devices in their polyphonous sentences. Ideas are synchronized rather than serialized. At their best, they erect a grammatical artifice in which mental balconies and watch towers, as well as bridges and recesses, decorate the main structure. Their sentences are gothic castles" (p. vi italics mine).
     It may be said that a digital text combines the two styles into one. While the complex architecture of the website recalls the castle, the self-standing and autonomous identity of each segment endows the text with an agility that is comparable to the effect of the "brief and lucid sentences" of standard narrative.

### 2. Fluidity

Structure and deconstruction

     If an argument is to have structure, it is essentially "constructed," i.e., it proceeds along a path that can be inspected, verified and, precisely, "argued." But how can a structured construction be the model when we are, it seems, called to deconstruct it after all?
     "Deconstruction" is a treacherous term, but treachery is intentionally built into its very referent. For the term is, and is not, a noun of action: while one is seemingly called to deconstruct a construct, one is in fact called not to construct a construct in the first place. The core of the paradox is precisely that the act of deconstructing ends up being, in practice, a full exercise in structural analysis. For to "deconstruct" is generally taken to mean to break apart a structure in order to better understand it. But what is more "constructive" than to "deconstruct" in such manner, what more structural than to articulate a structure?
     The deeper, and more valuable, meaning of deconstruction is in pointing to the need of producing a construct, yes, but one that instead of being frozen is fluid, one that retains the full dynamics of creative thought rather than suffocating it in all too rigid a mesh, one that is imbued with life at the same time that it opens itself for dissection. The real goal, then, is not to deconstruct in such a way as to produce an alternative frozen construct, but rather to create a construct that is intrinsically self-deconstructing.
     That is, precisely, what a proper digital text can offer. An argument is indeed constructed, but in such a way that its unfolding is self-declaratory. It is not the dissection of a cadaver, but a self-displaying vivisection. Indeed, structure rules supreme, and its parameters are as sharply defined as any – but they are not static. The argument is shown in flux, and flux itself is arguable, controllable.

Updates

     The most obvious implementation of this status of fluidity is the immense potential for constant updating offered by web publishing.
     Updates offer the most tangible demonstration of fluidity. A text is never definitive, and yet is always final. But it is a different type of fluidity from the one that affects the structure of the text. It is the extrinsic fluidity that reflects the flux of the author's thought, whereby additional information is added, or an opinion is changed. The very rapidity with which updates can be effected entails a difficulty in keeping track of the structural dimension of the argument, of the text. Hence a case can be made for implementing alternative ways to introduce updates in the digital text, so as to channel fluidity in a more productive direction.

### 3. Mechanisms      The concrete embodiment of a digital text occurs through a number of mechanisms. Some should be seen as logical extensions of traditional means of constructing a text: the first three discussed below belong in this category. Other mechanisms have become possible only with the electronic medium, and present problems which must be addressed to ensure an effective use of the digital text.

### Statement of purpose: preface      notion of preface - see Harvard Classics volume

### The frame: site map and table of contents      The frame of a website is generally conveyed by the site map, which provides with a tree-like structure the logical concatenation of the pages within the site. It is, for the most part, a stenographic notation that helps the author in keeping track of where the various elements belong.
     I privilege the notion conveyed by the traditional term "table of contents," which I develop in the sense of an outline. Here, emphasis is placed on articulating, at the same time, the substance of the components and the coherence of the whole. One can then develop a perceptual relationship to the "long argument" that is developed as a central thesis or complex of theses, within the website.

### Titled segmentation      In its digital version, a running text should be broken down into segments each of which has a title, all segment titles appearing at the top of the page. This yields and autonomous status to the individual segments while, at the same time, identifying the coherence of the page through the list of titles given as heading.
     The first advantage of this procedure is that it allows the reader to access the individual strands of the argument through a progression of links that go from the higher to the lower nodes. It offers a type of perception of the fragmented continuity of an argument that replaces the visual perception one has of a printed argument where one can leaf through the pages of a book or an article and retain the sense of the whole while focusing on the segments.
     The second advantage is that it is possible to construe hyperlinked argument across the full spectrum of the website and of course across the entire span the web as well.

### Side bars and pop-up windows      side bars - page from Erasmus

### The evidentiary base: links and footnotes
### Tagging Vivisimo. Alexandria project.

### Portability      It seems fair to say that digital publishing has not yet gained full acceptance in the academic environment. And this in spite of the enormous outlay of funds and the incalculable proliferation of websites. The problems that I have stressed above (and in relationship to digital reading) pertain to the intellectual level. But there are also important practical consequences to be considered.
     The major one is portability, i.e., the degree to which data can maintain permanence by being imported into progressively newer formats and support systems. These two aspects may present unexpected difficulties.
     Format conversions are far from satisfactory, especially when the base structure is complex. The best solution seems to be to have the basic data in the simplest possible format (we have chosen ASCII for texts, and JPEG or TIFF for graphic files). An extra measure of safety is to have the data available in a non-digital medium as well.
     Support systems should serve two parallel needs. The first is to guarantee current access on any platform. In this respect, a web based online server, with one or more mirrors, provides the best assurance. Second – for the sake of future upgrades, it is important to rely on strong institutional support, one that would guarantee regular maintenance and migration of data.

### Intrinsic vs. extrinsic integration      Integration is an overriding goal of all computer projects, particularly at the level of the interface between programs and data entry. But, even when successful, it is often limited because it remains essentially extrinsic. By this I mean that different components are indeed assembled, often with sophisticated equipment – but the correlation happens through a process of juxtaposition. Thus photographs are taken, points in space are geo-referenced through total stations, 3-D models are produced, detailed typological inventories are drawn up. Each set of data is inserted in a database system that is rich of attributes. At best, these are interconnected in a relational mode. But each remains essentially a universe of its own, juxtaposed to the others. This is what I call extrinsic integration.
     The goal of intrinsic integration, on the other hand, is to achieve the correlation of (1) all the pieces, (2) automatically, (3) from within and from the start, and (4) at the highest level of the pertinent universe. As for (3), the correlation among the components is invoked from within each component at the very moment that the component is first produced, rather than being superimposed after the component's data base structure has been achieved. As for (4), all components, however heterogenous in either substance or format, must end up in a structured whole where the coordination happens at the level of the whole, not of the parts (which would result in mere juxtaposition). The Browser Edition of the type proposed here is the concrete approach I follow in order to meet these requirements.

### 4. Four types of digital texts      We may distinguish four different ways in which a digital "composition" or "text" may be construed, depending on grammatical coding (tagging), degrees of complexity, display format envisaged.
     Each of the four approaches is briefly described below. The following chart gives a synoptic overview of the criteria used to differentiate the various types of "digital reading," and adds simple examples taken from everyday experience.

	grammatical coding	model	example
1. pre-configured narrative	absent	linear	normal web page
2. tally	simple	sequential list	telephone directory
3. matrix	complex	non linear matrix tabulation	online bank statement
4. self-generated narrative	complex	hyperlinked	-

1. The pre-configured narrative

Reading a linear text channels attention in a preset direction, the one proposed by the writer. Obviously, this privileges the writer's point of view, which is, after all, what writers intend to do in the first place – proposing a given argument. In a printed version, footnotes and cross-references help readers redirect their attention along parallel registers, and thus to initiate and pursue alternative paths.
We may distinguish two types of electronic format, PDF and browser. The former is a perfect mirror of a printed text with added advantages such as search capabilities. The latter is conceived digitally, and exp A drawback of an electronic version of a linear text is that it weakens the perception of the whole, except within a single page. In other words, one does not have a sense of the larger scope within which the individual pages fit. (A detailed table of contents in outline format helps to obviate this drawback.) On the other hand, there are the major practical advantages, in particular minimal costs, ease of access, practically unlimited possibilities of graphic documentation, and the power of search functions.
This important intellectual function (which started only relatively recently, in the nineteenth century) is addressed in the electronic format by hyperlinks.

2. The tally

inventory/list approach
The simple example given above is that of a telephone directory. The level of organization is at its simplest: a large amount of data is sequenced through a single sorting key, the alphabetical order. The reader can randomly checking data for an argument that is being developed outside the digital medium itself

3. The matrix

data base: systematic search on the basis of organized data

4. The self-generated narrative

the argument is developed from within, through a systemic following up of links
related to inventory approach, in that one seeks evidence for argument, but from within (DEFINE THIS)
INTERACTIVE, more developed than footnotes and cross-ref