The Semiotic Art of Web Sitemaps

George L. Dillon
University of Washington

(On-line version; print version in Discourse Studies in Composition, edited by Ellen Barton and Gail Stygall, Hampton Press, 2002:331-52.)

INTRODUCTION: The dialectic of critique and design

In his recent books and articles, Gunther Kress has been pointing to a turn in our culture away from the densely printed page toward the visual. In an article published in Ilana Snyder's collection Page to Screen and in Gail E. Hawisher and Cynthia L. Selfe's Passions, Pedagogies and 21st Century Technologies, he calls on those who teach the art of written communication to include visual design in the curriculum.1The era of writing texts conforming to stable codes and expectations has given way to one that prizes innovatively visual "texts." And anyone who has taught web-writing will confirm that the lore of composition (split off from technical and business writing) has virtually nothing to say about the meaning and uses of images, and in fact does not encourage or reward their use in students' papers.. Kress and van Leeuwen's Reading Images: the Grammar of Visual Design2began to sketch a semiotics of visual language, but did not go far enough, Kress would now say. Heretofore, he says, semiotics has assumed that we communicate by means of shared codes of signification: analysis of a text would review the set of options that the language code makes available to the writer and discuss the signification of the particular choices made in context. In turning away from code-and-use semiotics, he is turning away from Reading Images, which does still rely on the notion of stable, learned, shared codes. It is not clear from the article what exactly he is turning toward, for books on visual language and information design are proliferating rapidly and they are almost all driven by a lust for practicality, efficiency, and clarity that would make a Benthamite blush. Their terms of analysis are accordingly ad hoc and untheorized.

Here, for example, is an overview from Peter Wildbur and Michael Burke's Information Graphics: Innovative Solutions in Contemporary Design (Thames and Hudson, 1998):

Most of the work illustrated in this book falls into one of three general categories. The first involves information presented as an organized arrangement of facts or data, such as a timetable, a signing system and most maps, from which users are free to extract only that information which they need for a given purpose. The second involves information presented as a means of understanding a situation or process, such as a guide-book, a bar-chart of a stage-by-stage description of how to get a machine to operate. The third category involves the design of control systems, such as that of input and feedback controls of a product or vehicle.(p.7)

Wildbur and Burke continue by quoting Edward R. Tufte, "Excellence consists of complex ideas communicated with clarity, precision, and efficiency and that is just as true of the new media as of the old." (p.7) Even when their language seems to offer some interesting scope ("information offered as a means of understanding"), it quickly returns you to earth with its list of practical texts. We're not talking about verstehen here.

These books, moreover, are long on show and short on tell: they are more like pattern books or catalogs than academic analyses and arguments. They often embrace the "graphics" half of "graphics vs. art" and are markedly hostile to non-evident intentions or the experience of a graphic for its own sake. Even when they are beautifully printed (as Edward Tufte's books, for example) and we are clearly meant to relish their visual sumptuousness, they still officially toe the good-design line. And exactly the same things can be said of Web design books all the way from David Siegel's Creating Killer Websites to Patrick Lynch and Susan Horton's Web Style Guide, and Ken Coupland's Web Graphics Now!3

Embracing a "turn to the visual" does not necessarily mean abandoning the semiotics of code and use. Robert E. Horn, for example, promotes the new amalgam of text and image by showing (!) that it has all the parts and properties of natural language (morphology, syntax, semantics, even metaphor and rhetorical figures such as erotesis, incrementum, apodiosis, etc.).4Horn uses clip-art for his visuals and frequently depicts himself presenting this new view of the visual-textual to groups of business people. He does not concern himself with the positioning of discursive subjects vis-a-vis the new code. Kress, however, is very much concerned with the freedom and creativity of the subject, including the position of the analyst or scholar toward the code, which naturally tends, he says, to be that of critique.

Kress offers design as an alternative to critique. To do so, he constructs a "critique of critique" (87). Critique in his view employs "distanced, analytic scrutiny" to reveal the roots of the code underlying some previous discourses and the processes of the code's users. Even the act of analysis tends to be debunking because it deprives some discourse of its "naturalness;" in addition, critique is critique of ideology, which is to say how the representations of the text serve the interests of those with power to sponsor and promote the text (corporations, agencies, institutions).5Its objects are always the result of past practices and always the discourses of the other, usually the masters. Creative uses of the code are rare, and most discourses are repetitions conforming to the code. Kress has nothing more to say about creativity under code-and-use, and he speaks of dynamism being injected into the system through critique, not through creative or artistic uses.

Subjecting critique to this critique, Kress offers design as future oriented and as shaping the social world instead of serving and conforming to it:

Design takes for granted competence in the use of resources, but beyond that it requires the orchestration and remaking of these resources in the service of frameworks and models expressive of the maker's intentions in shaping the social and cultural environment. ("English," p.87)

This goes even beyond Joseph Beuys' proclamation of "every human being is an artist" to impute remarkable efficacy and transformative power to designers. One wants to tap Kress on the shoulder to remind him that while people with such competence and a flair for innovation fetch a pretty price on the employment market, they do so only if their "intentions" are to execute their employer's project. We're talking graphic (and web) design here, not art.

So is there--could there be--art on the Net? And where would it stand vis-a-vis the clear and direct style for conveying information? The answer to the first question is yes, some of it even self-identified as net.art. The answer to the second is more complicated; the rest of this chapter will sketch an answer along the lines of a stylistics of net.art. The case to be studied, however, is not net.art itself (a burgeoning area) but specifically the use of sitemaps in some representative sites. I will focus on sitemaps because they would seem to meet all three kinds of practicality outlined by Wildbur and Burke: they display the organization of information on the site, they provide a means of understanding the information, and they provide a control mechanism for direct navigation. And--this is the main point--they contribute to the esthetic design and function of the site.


A website is a group of files connected by hypertext (HREF) links. This structure of connection is a network, not a tree, because paths can reconverge and hence there is more than one way to get to a given point, but with a tree there is only one path to each point; its freedom of motion is limited, however, because HTML links are unidirectional and hence you cannot reach every point from any point (hence it is a "directed network"). This topology is the same as that typically displayed in a flow diagram. Sites commonly have one page (the "top" page or "Table of Contents" page) from which the maximal number of points can be reached (i.e., where the paths start) and when graphing sites, that page is often placed at the top of a downward branching network, at the left, as an initial entry point, or at the center of a circle (or concentric circles, a radial network). Each of these figures has certain associations which handbooks describe: Jacques Bertin speaks of the top-down structure as suggesting gravity and "significant order," which other writers such as Kress and van Leeuwen have not hesitated to call "hierarchy;" indeed, Robert L. Harris discusses how top-down organizational charts may seem offensively insistent on hierarchy, and show how circle diagrams can avoid this connotation.6Robert E. Horn discusses "the multiple associations of topologies"(pp. 85-6). These "meanings" are rather diffuse after the fashion of syntactic structures and rhetorical figures in poetry, but if you redesign your organizational charts they must say something to somebody.

Jacques Bertin points out that the connected point topologies of networks does not use position, extension in space, or distance to carry information about the site: the only thing that matters is the network of connections. When extension does signify, we have moved from a network to a map, and from a topology to a topography. This is straightforward enough in the case of a network whose connected points are understood to be spatially located, as in the case of the railroad or telephone network, or of the major nodes of the World Wide Web. Using a map to represent site structure, however, is to commit oneself if only tentatively to spatial metaphor and thus to begin a slide into "cyber" space and virtual worlds. Even the so-familiar-as-to-seem-literal step down to subdirectory of the Windows Explorer filemanager (often carried over to hypertext networks) marks distance from the root measured in the number of clicks you would have to make going one level down each time. This space is measured with machine precision and mindless detail. With other, less standardized maps, rather more suggestive meanings can attach to space and its representation.

Early, slightly pre-web, maps of hypertext sites from a decade ago often mixed text and image and were clearly intended to reproduce the effect of an outline or table of contents: their purpose was to represent the "place" ofeach item in the overall structure, to assist in "going" to an item directly, and to display the full range of contents, so that you can see whether you had visited all of the main heads. These hierarchical trees did not represent cross links between pages, or represent them very well, and this is a limitation that many file-and-folder based website mappers continue to exhibit. They display contents, not paths to contents.

These trees are not properly site maps so much as they are decorated Tables of Contents with direct links to each of the component pages. True sitemaps (two-dimensional, visual representations of site structure) were a novel and prominent feature of the StorySpace hypertext authoring environment developed by Mark Bernstein for the Macintosh in the later 1980s. Bernstein has documented a very general concern during this period with navigation and disorientation of users in the emerging medium.7The experience of distress or medium shock is nicely evoked by this page from Shelley Jackson's "Patchwork Girl" (Eastgate Systems)

Assembling these patched words in an electronic space, I feel half-blind, as if the entire text is within reach, but because of some myopic condition I am only familiar with from dreams, I can see only that part most immediately before me, and have no sense of how that part relates to the rest. When I open a book I know where I am, which is restful. My reading is spatial and even volumetric. I tell myself, I am a third of the way down through a rectangular solid, I am a quarter of the way down the page, I am here on the page, here on this line, here, here, here. But where am I now? I am in a here and a present moment that has no history and no expectations for the future.

Here the experience of disorientation is traced to the electronic text's lack of material extension. In some very practical and untheorized way, the book does not merely represent or even realize the text: it is the text.

Several developments and practices, however, have converged in the last decade to render this new environment less strange and readers less eager for navigational aids. For one thing, Netscape developed the history object to keep complete records of browsing sessions, so that serious retracing became an always available option. For another, increased experience with games, from Doom to Myst and even Microsoft's little Hover game bundled with Windows95 both rehearsed and rewarded hypertext orientation skills. Third, as native net.art and hyperfiction developed, they began to make use of uncertainty and the struggle to orient oneself as reliable parts of the reader's response, and even to treat them reflexively as part of the theme of the work. Bernstein noted that hypertext authors seemed to be inviting the reader's at least occasional confusion, and he cited a similar claim by George Landow that a certain amount of anxiety may be good for readers. Fourth, Windows operating systems now allow more than one window to be open at a time, so that hypertext jumps can open new windows rather than replacing the current one and to "go" somewhere does not require you to leave "here" completely. Regardless of the exact weighting of these various factors, readers do seem to have become less bewildered by Hypertext, or more equipped to deal with its special demands, and the importance of sitemaps for navigation, at least for the net.art end of things, has fallen to secondary status.

This does not mean, however, that the art of mapping sites has withered away; rather, it has freed the maps to offer interpretive takes or perspectives on their sites. Net.art site maps tend to be custom made and site particular: navigation, however, is better served by standard icons and conventions which can work automatically. In truth, on first sight, unique sitemaps are not automatic; they require some testing, learning, and interpretation in most cases. Nonetheless, there are rules that map images to objects that we can draw on and it is the purpose of this study to work out some of these semiotic rules. Some of these rules and principles have been outlined for printed magazines and textbooks by Kress and van Leeuwen, and I will link to their discussion as we go.


OVERVIEW: Kinds of maps

In general, sitemaps are what Kress and van Leeuwen call representations of analytic processes (a map of parts) or they are flow diagrams. Rarely do we find a full network representation (unless we use a special sitemapper and make one), partly because networks, complete with return links, become very jumbled and headless visually (see Figure 5 below). Maps vary along several dimensions, among which that extending from diagram to image is fundamental. A map is diagrammatic ("abstract") insofar as it uses shapes (e.g., circles, rectangles, wedges, stars, and arrows) to represent files in a non-iconic way . That is, the map as a whole could not be taken to be a representation of anything else. The opposite extreme is an image from another domain that is used as a metaphorical overlay, as when database entries are represented as stars in a star field (e.g. the Starrynight database viewer at Rhizome8).

A second dimension of interest is that of spatiality of the site: is the map 2-dimensional (and orthogonal to the reader's gaze) or is it some sort of projection into three dimensions? Sitemaps with three dimensions begin to give visual reality to "cyberspace" and to imply some sort of topography and landscape. Do distant objects appear hazy, smaller, more compressed, or do they exhibit the hyperclarity of early 3D ray tracings before "fog" was added?

Degree of precision and completeness is the third dimension. A map which calls itself "a rough sketch" obviously makes modest claims on this dimension. A map that appears machine drawn is perhaps even machine-mapped (and hence a more "objective" view). Kress and van Leeuwen identify a contrast between boxes and straight lines on the one hand (the "mechanical-technological") and curves, circles and ovals on the other (the "natural-organic")(84).

A fourth dimension concerns the placement of the viewer inside or outside the represented site. This includes images that update to show the viewer's position as she moves through the site. A site with viewer inside is representing immersion in the site. The extreme of immersion is the 3D flythrough using some sort of 3D applet or plugin. These were very trendy a couple of years ago--Apple made one called HotSauce, but ceased development as a cost cutting measure. (It is still available: www.xspace.net). Here for example is a screen capture of the entrance to site I maintain: (grayscaled and inverted)9


FIGURE 1. HotSauce Site Map

I will now illustrate the use of these terms to analyze sitemaps of net.art, grouping the examples into sketched, free-hand maps and computer generated ones.

Looking first at sketches, consider the following image, which is sometimes the opening (splash) page for "jodi" (www.010010110101101.org/jodi.org/map/x.html)10. It is not exactly a sitemap, as it is more a graph of what they consider core, like-minded sites of net.art and hacktivism.(This is 80% scale):


FIGURE 2. www.jodi.org: backbone


This looks like a fairly direct diagrammatic mapping with each box representing a site-URL. The lines connecting the boxes most probably are hypertext links (though why some are heavier than others is not evident); the ellipses would seem to represent a grouping, but the basis for the groupings is not easy to guess (common style, technique, politics?), nor is it clear what sort of ordering the numbers signify. This "map" is itself a challenge or mystery to be solved by clicking on the sites and then puzzling out what the basis for the diagram is. That is, the map does not represent a network of paths and groups that a robot could trace and draw. Rather, it is a claim or set of claims about the way these sites could be or should be viewed.

Here is a true sitemap from Vera Frenkel's Transit Bar / Body Missing site (www.yorku.ca/bodymissing, 1995, 1997)


FIGURE 3. Sitemap for BodyMissing

Here the map calls itself "rough," which may mean drawn freehand or with details omitted or with various inaccuracies. We assume the words are links to pages (which they are) and that the arrows represent links to the page (which they are). The arrows are headed, reflecting the unidirectionality of HTML anchor links, and curved, suggesting "organic" flow from the top. All and only major units and organization are indicated (if you know the site). Presumably some links (return links, for example) have been omitted and the reference to "detail maps" at various places suggests that some nodes have more internal structure. These major nodes are also available from pictures of the interior of the Transit Bar, which is the meeting place for the various threads that make up the site. So these pictures constitute a second set of sitemaps for the site.

"BodyMissing" is a huge site with over 250 pages. The general setup has you imagining yourself as a bartender in the Transit Bar in Linz, Austria, where you begin to overhear narrative fragments many of which seem to bear on a project of remaking lost art works that Hitler had assembled in a nearby salt mine. Numerous artists become involved and do in fact recreate new versions of the lost old masterworks (not copy, not restore, recreate). There is also considerable talk about the other bartenders and piano players, most of it in English, but with bits of French, German, and Swedish. You can spend quite a lot of time there, making all sorts of connections both internal to the Transit Bar world and to war history and art history. The sketched sitemap thus models a sense of the overall frame in which the reader's more detailed and unpredictable experiences will occur.

Both of these apparently handdrawn maps are highly diagrammatic; for one which employs an image as a metaphor, we can compare Shelley Jackson's imagemap for her own on-line site, The Body (www.altx.com/thebody/body.html): (the image is reduced 50% and inverted B/W for clarity; the original looks more like a chalk drawing than a charcoal one)


FIGURE 4. Shelley Jackson: The Body

Though the body-sketch is strictly speaking a metaphor for the concatenation of a bunch of files, the reach across domains is not great, since the site is about her experiences and feeling about the parts of her body. The body is sketched, and so are the hot areas that link to the relevant body part file (though these too are not very accurate). Her refusal to fit the boxes to the figure very closely is a deliberate crudity of craft that says "rough approximation." We should not be surprised if we find that there are other body parts with links that do not have boxes(chiefly around the face), or that there are a couple of boxes with no links. The "rough sketch" as it were explicitly abjures completeness or perfect accuracy. It sacrifices those things in order to convey the overall sense of the work and indeed through its posture and as it were full frontal gaze (great granddaughter that she is of Manet's Olympia) conveys the writer's attitude toward her own self-exposure.

To see what the image does that a straight network map does not, compare this map of thebody site made by the IBM site mapping robot Mapuccino:

FIGURE 5: Network for TheBody

Mapuccino follows links and draws directed arrows corresponding to them, plotting them radially from a "home" center which is the initial file.11(Here, however, the sitemap file is an oldstyle "ismap" that sitemappers can't expand, so Mapuccino has drawn this map from an initial entry at "arms".) The site is a heavily interlinked set of about 40 pages, and comparable maps would be drawn starting at almost any of the hot body parts. Obviously such maps do not afford much in the way of perspective on the site; indeed, perspective is swallowed up in rhizome, even when the rhizome is drawn with machine assistance and displayed with a machine look.

Contrasting to the sketched sitemaps are machine-precise- looking 3D projections of sites. A major source of these is the MAPA program from DynamicDiagrams (www.dynamicdiagrams.com; Paul Kuhn, chief engineer) which displays pages as upright notecards in axonometric projection receding from the viewer who views from above and at an angle. Here is a MAPA sitemap for the Encyclopedia Britannica online site:12


Figure 6: MAPA map of Britannica Online


This map, taken from the DynamicDiagrams site, is considerably newer than the MAPA display of George Landow's "Arts in Victorian Britain" site shown in Hypertext 2.0 (Johns Hopkins University Press, 1997, p. 129). Although both maps are axonometric projections with no perspective convergence of parallels, this one uses some devices of atmospheric perspective and reduction in size, placing the viewer in a more naturalistic relation to the landscape of cards. That is, though still abstract, it is a move away from the pure diagram style of "Arts in Victorian Britain" overview.13

At somewhere near the opposite pole of representational imagery we find this sitemap for Julia Scher's Securityland suite archived at adaweb.walkerart.org: (image has colored lines and hands against a black ground.)


FIGURE 7: SecurityLand map


FIGURE 8--Alternative Sitemap

Here the various subsites having to do with surveillance, regulation and intrusion into privacy are depicted as features of a single structure--the nondescript modern office or public building--represented here in a kind of architect's drawing with major features labelled. If you try "Security Card Access" (and the hand seems to point you that way), you are given a telephone touch pad which also gives links to the four main subsites with the instruction "Please Make Your Selection Now." However, given just a set of numbered buttons, you can only stab hopefully at them to see what happens, as if you have forgotten the security access code to the building. Not only is this useless as a navigator, it also places you at the mercy of the Securityland system--which,I suppose, is exactly where Scher wants you to begin experiencing her site.

My last and most complex example of an axonometric array is the Storyboard page for Doug Aitken and Dean Kuipers' "Loaded 5x,"(1997) a highly graphic slice from the lives of a man, Eddie, and a woman, Annie, along with some supporting characters. (This, like 'Securityland" is archived at the Walker Art site: adaweb.walkerart.org/project/aitken). Aitken and Kuipers modify the MAPA layout of cards by making a thumbnail card from the main image on each page and laying them out some flat and some on edge in fairly exact ranks and files:


FIGURE 9: Storyboard for Loaded 5x

This sitemap generally becomes available each time you come to a page which offers a choice of links. If you summon it, one grayscale version comes up with a red circle indicating where you are. If you click on that circle, the colored version, of which Figure 9 is reproduction, comes up and you can click on any card-image you wish and go to that page. The placement of the ca cards tells us something about their relations: for one thing, cards in a sequence are laid out in serial order, but the longest run, that featuring Anne, takes a right angle turn, and Eddie's pages are not set out in an exactly straight row either. The structure of the site is rather webby, with two main runs and several shorter ones and with considerable crossover between them.

Here is the opening page of the Eddie sequence:


FIGURE 10: Central Eddie

In this frame are all the persons who become centers of attention and whom we see interacting with each other, even the desperate Chrissy in the far distance. The site is made up of many such tableaus with bits and snatches of dialogue from which we eagerly try to synthesize a story or stories. Similarly, we can make considerable headway organizing the the thumbnail pages as a way of mapping the connections between pages and persons, but the rows aren't straight, and there are repeated cards (16 of them), though the pages occur only once in the file directory. (And there are two pages which have no image and hence have no image to represent them in the set of cards.) So the reason for the site's insisting that the name of this sitemap is Storyboard begins to come into focus: it is not solely or even primarily a flow diagram for the site but rather a map of character and narrative thread. That is, being "here" on the Storyboard does not correspond exactly to being "here" in a network of hypertext-linked pages. How far and how exactly Aitken and Kuipers expected us to get in their structure of narrative space is a little hard to guess. "Storyboard" proved very attractive as design, however, and was made the background of a two-page spread in Ken Coupland's recent Web Design Now.14

CONCLUSION:


If you were to set out to design a navigator for a site from which, as Wildbur and Burke say, "users are free to extract only that information which they need for a given purpose," you would do well to consider using basic icons in widespread use with arrows, lines and indentations working just as they do in print documents. That is an excellent example of what Kress means by "code" and writing in conformity with the code: because it is relatively public and standard, it does not need to be learned and tested just for the purpose of navigating a particular site. We can use it fairly automatically and unthinkingly. Compared to a Windows Explorer style file-and-folder map, the ones we have examined are relatively idiosyncratic. Though not entirely private sign systems, they do require a little testing and learning and we are never quite sure how much to rely on them for completeness and accuracy. They call to mind other instances of graphic arrays, such as sketched outlines, architect's drawings, dominoes, or even Scrabble games. They highlight certain features of a site and obscure others which would perhaps all be graphed equally by a file-and-folder display.

There are then three functions that sitemaps may perform on net.art sites (and presumably elsewhere). First, they constitute one take or view of the world of the site. Second they focus and highlight certain parts and connections. Third, they engage the reader as active and conscious of her meaning-making activity: figuring out the map is part of figuring out the site (and there is no doubt net.art has a strong component of puzzle-solving and hermeneutic seeking). Sitemaps are of course interpretive, and that quality comes to the fore when the maps available are more suggestions and pointers than complete and "objective" representations. One might call this a more "creative" orientation toward indicating structure and flow through a site than one that scolded and nattered at writers about users' needs for clear, standard, plentiful indications of the website's structure after the fashion of composition handbooks on the subjects of signposting of structure and indications of transition. Maps of site structure are valuable, but they need not be codified for all purposes. For sites whose value lies as much or more in the experience it gives its user than in the information the user can extract, this sort of open, allusive mapping adds value and may not be wholly unlike what Kress meant to mean by design.

In one sense, the turn to the visual is an obvious and not very interesting claim, namely, that the technology for making and disseminating images has been booming since the invention of photography in the middle of the 19th century and places unprecedented powers in the hands of everyone with a computer. This tecnology will not necessarily alter or expand the functions of images in texts. Images may continue to be illustrative and supportive of meanings more fully realized in the text. In that case, writing handbooks can add sections on Finding and Including Illustrative Images and Clipart, along with others on Typography and Layout and on Use of Color. We could put a couple of the several design books on reserve.

By means of my case analysis, however, I have tried to suggest how the turn to the visual can involve it as another way of conveying meanings and, via offering models, to making claims about the matters under discussion. Finding the ways to see and discuss these meanings is the greater and more interesting challenges now facing writing teachers. It is to understand and teach the visual as a way of writing.

Works Cited

Bertin, Jacques. Graphics and Graphic Information Processing, Trans. William J. Berg and Paul Scott. Berlin: Walter de Gruyter, 1982.

Bernstein, Mark. "Hypertext Gardens: Delightful Vistas." www.eastgate.com/garden.

Coupland,Ken, and Robert Appleton. Web Design Now. New York: Graphis, 1998.

Douglas, Noel, and Geert J. Strengholt. Web Graphics Now! Amsterdam:BIS Publishers, 1999.

Harris. Robert L. Information Graphics: A Comprehensive, Illustrated Reference. Baltimore, MD: Management Graphics, 1996.

Horn, Robert E. Visible Language. Bainbridge, WA:MacroVU Press, 1998.

Kress, Gunther and Theo van Leeuwen. Readings Images: The Grammar of Visual Design, New York: Routledge, 1996.

Kress, Gunther."Visual and Verbal Modes of Representation in Electronically Mediated Communication: The Potentials of New Forms of Text," In Page to Screen: Taking Literacy into the Electronic Era. Ed. Ilana Snyder. New York: Routledge, 1998. 53-79.

________."'English' at the Crossroads: Rethinking Curricula of Communication in the Context of the Turn to the Visual," In Passions, Pedagogies, and 21st Century Technologies. Eds. Gail E. Hawisher and Cynthia L. Selfe. Logan, UT:Utah State University Press/NCTE, 1999. 66-88.

Landow, George P. Hypertext 2.0. Baltimore, MD: Johns Hopkins University Press, 1996.

______.landow.stg.brown.edu/victorian/art/artov.html

Lynch, Patrick, and Susan Horton. Web Style Guide. New Haven, CT: Yale University Press, 1999.

Siegel, David. Creating Killer Websites. 2nd ed. New York: Hayden Books, 1997.

Wildbur, Peter, and Michael Burke. Information Graphics: Innovative Solutions in Contemporary Design. London: Thames and Hudson, 1998.

www.rhizome.org/Starrynight

End Notes

1Gunther Kress, "Visual and verbal modes of representation in electronically mediated communication: the potentials of new forms of text," in Page to Screen: Taking Literacy into the Electronic Era. Ed. Ilana Snyder. Routledge, 1998, pp. 53-79; "'English' at the Crossroads: Rethinking Curricula of Communication in the Context of the Turn to the Visual," in Passions, Pedagogies, and 21st Century Technologies. ed. Gail E. Hawisher nad Cynthia L. Selfe. Utah State University Press/NCTE, 1999: pp. 66-88.

2Gunther Kress and Theo van Leeuwen, Readings Images: The Grammar of Visual Design, Routledge, 1996.

3Dave Siegel, Creating Killer Websites, 2nd.edition, Hayden Book,1997; Patrick Lynch and Susan Horton, Web Style Guide. Yale University Press 1999; Web Graphics Now! ed. Noel Douglas and Geert J. Strengholt, BIS publishers, 1999

4Robert E. Horn, Visible Language. MacroVU Press, 1998.

5Readers of Habermas may think of the problem of the "negative dialectic" of Horkheimer and Adorno: exposing the operation of ideology does not of itself create a valued discourse that is beyond critique.

6Jacques Bertin, Graphics and Graphic Information Processing, trans. William J. Berg and Paul Scott, Walter de Gruyter, 1982, p. 31;

Kress and van Leeuwen, p. 87;

Robert L. Harris, Information Graphics: A Comprehensive, Illustrated Reference. Management Graphics, 1996, pp. 186, 263.

7Mark Bernstein, "Hypertext Gardens: Delightful Vistas," www.eastgate.com/garden/NavTech.html

8www.rhizome.org/Starrynight

9Technically, the viewer is not represented within this scene; rather the viewer perceives it from a viewpoint within the scene. Some 3D browsers offer an optional "avatar" to represent the viewer in the scene (e.g.Blaxxun plugin for Windows viewers).

10To be exact, this one opening page that "jodi" uses, here archived inside another site.

11Mapuccino actually offers several views including a standard down-stepping directory-tree view, a top down branching tree and fish-eye magnification of parts of the network. Unfortunately, IBM no longer makes this available online (www.ibm.com/java/mapuccino).

12This site is also displayed in Wildbur and Burke, p. 132-3.

13This Overview is no longer the one used at the site: landow.stg.brown.edu/victorian/art/artov.html, which reverts to a text-based TOC of lists and sublists.

14Ken Coupland and Robert Appleton, Web Design Now. Graphis, 1998. pp. 8-9

[For further reflections on representing site structure, see Writing with Images, c. 6: Maps of Abstractions]