Culture Map: Encompassing the Art of the Search Engine|
At the crossroads of abstract art and scientific visualization,
Culture Map appeals to popular fascination, while at the same time
attempting to provide some useful indications about the character of the Web. It
is an image system that responds to the appearance of new content and the
disappearance of old content from the Web. The work's title refers both to its visual
embodiment of color compositions, and to its representational goal: describing the
composition of the Web.
Mapping the content of the World Wide Web has become a fairly subjective endeavor. Web content is in constant flux. Protocols, browsers, and programming languages are all changing. Creating an up-to-date index for the content of the whole Web is not really possible. Even the largest existing databases, which seem to be those compiled by search engine companies, only reflect the condition of the Web.1
Most search engines use data accumulated by the constant activity of software "spiders" that comb the Web for keywords. The purpose of the spiders is to associate keywords with specific Web pages. Some search engines maintain bigger databases than others, and their algorithms differ in how they rank the pages they encounter. Culture Map is not directly concerned with ranking. It creates visualizations based on the quantity of "hits" reported for particular keyword searches. So, for example, on a given day, a search engine may report that it has indexed 5,000,000 pages that contain the word "Art", whereas one year later it may have 7,000,000. Since February of 2000, Culture Map has been keeping track of several search engines, recording how their responses have changed.2
Likening search engines to maps is a little misleading, because conventional geographical maps are more visual. Search engines require that one formulate a particular question before seeing any results. In making this assertion, I am distinguishing between search engines and the smorgasbord of links that adorn portal sites today. Many of the portals now combine two paradigms for categorizing information: Web directories and search engines. Each represents the Web differently. Web directories foreground grids of topics like "Shopping," "Sports," "News," and "Business." Each of these categories, in turn, leads to a more detailed list of related topics. Because the links featured in Web directories are reviewed, selected, and organized by humans, the information presented corresponds to the values and ideologies of the judges. One motivation for Culture Map was that I found several major portals were beginning to marginalize categories like "Art" and "Science." The changes I began seeing on some of the most influential portal sites seemed motivated mostly by marketing zeal, and I wanted to address this drift toward commercialism.
The starting points offered to Web surfers are valuable virtual real estate, and so it is no surprise that large companies figure in the ownership of the most popular portals. But, what effect does the intense commercialization of Web reference material have on the use and development of the network? Will newcomers to the Internet be able to perceive the ways it is changing?
At this point it should be noted that I am not a scientist, and although Culture Map engages the mechanics of scientific visualization, its concerns extend into aesthetics as well. Before continuing with an explanation of Culture Map's visualizations, I will explain how it happened that, instead of using conventional pie charts or bar graphs, Culture Map turned into an abstract color composition.
Culture Map's resemblance to abstract geometric painting is due in part to a visualization technique developed by Dr. Ben Shneiderman. The diagrams he began making in 1990, which he calls "treemaps," are not unlike geometric abstractions.3 Like a rectangular pie chart, quantity is expressed as a fraction of the whole. Shneiderman conceived of treemaps to visualize the amount of disk space used by various folders on a hard drive. His technique has also been used to depict stocks.4 I decided to apply some of his ideas to depicting the metamorphosis of Web content.
Ironically, Shneiderman's representational images have the look and feel of non-representational paintings. The austere rectilinear paintings that once elicited hostile reactions have become part of the iconography of science as well as modern art. As an artist I like a familiar visual vocabulary that I can put to new uses. Given that many people who see Culture Map are somewhat alienated by art that is dematerialized, systemic, and conceptual, it seems appropriate to provide a familiar point of reference. Culture Map's formal allusion to abstract geometric painting relates that tradition to new artistic practices appearing in network "space."
In order to guarantee itself the completeness and wholeness that defines or better delimits art, abstraction has shunned reality -- that is, it has shunned real, created space in favor of artificial, illustrated space. The answer to this problem is not a return to conventional illusionistic space, to Albertian perspective; but some kind of turning around is in order. Certainly invention and inspiration are called for.5Stella's mention of "real, created space" sounds a bit mystifying to my ear (probably a reference to Caravaggio). Still, he is not alone in resisting a return to perspectival space as he seeks ways to resolve the felt limitations of abstraction. At the time Stella was writing, interactivity was beginning to attract the interest of artists.6 Interactivity is an invention that can add a semblance of depth to abstract space, without perspective. This "answer" raises some basic questions, however. If a Mondrian image is made interactive, does it become representational? What should it represent?
In a series of lectures published as "Working Space", Stella directs attention towards the working space of artists and its correlation to the images on their canvases. At times this seems like a painter's idle curiosity about the methods of his predecessors -- the orientation of canvases and models, for example. But his observations also address the social and institutional framing of subject matter.
Searching for a way to advance abstract art, Stella looks back to Caravaggio with appreciation for a sense of space that was becoming independent of church architecture. This issue may seem remote from the business of Internet art, but parallels are not hard to find and Stella's critical inquiry suggests some pertinent questions. Church architecture is no longer a pressing concern for most artists, but there are comparable structures artists may wish to eschew. Software architectures that contain Internet art leave much to be desired. The working space of the Internet artist encompasses the technical infrastructure and popular software used to convey cyberspace and telepresence. What limits and powers enframe artistic activity within this new space?
Since the 1960s Hans Haacke has made a number of works that apply systemic analysis to art. His "Visitor's Profile", which appeared (though the computer apparently didn't function) in the 1970 show "Software, Information, Technology", curated by Jack Burnham, was a prototypic system-art project that envisioned dynamic tabulation and feedback from visitors to the Jewish Museum. Later works like "MetroMobiltan" (1985) dealt specifically with corporate sponsorship in the domain of culture.7 At the turn of the century, the centrality of the museum in the presentation of art has been challenged by the proliferation of online art. The relationship between Internet artists and the public is mediated as much by portals and search engines as by museums. The importance of the portal warrants more system-analytical work that acknowledges this change of cultural affairs.
Like the earlier move from the walls of the church to the canvases of the collector, the emergence of the Internet as a venue for art has confronted artists with non-traditional issues and problems. While Web directories and search engines may seem like somewhat banal subject matter for art, they form among Internet users a kind of common experience. However artless and conventional the portals may appear, they constitute provocative strategies of representation. They express important assumptions about how enormous quantities of information should be schematically presented, or abstracted.
The two prevailing categorization paradigms that I have identified -- Web directories and search engines -- serve as de facto maps for Web content. Both function like maps in that they guide people toward destinations on the Web. Search engines are not immediately visual like a road map, but they contain enormous amounts of information about Web pages. As Web content changes, many of the changes are discovered by the spiders, which follow links that lead from one page to another.
By contrast, Web directories use comparatively small databases that are not constantly augmented by software spiders. Instead of surveying the Web, they depend upon people to come to them with suggestions. Alterations to the main interfaces of the Web directories (a grid of categories like "Shopping" and "News") do not occur automatically when Web content begins to change; rather, design decisions are made by the specialists who manage the directories. So while Web directories do provide accessible visual starting points for navigation, the degree to which they adapt to changes in Web content is suspect.
Despite the hackneyed navigation metaphor, there are few means of mid- and long-range foresight when moving through the Web. It is a coded, symbolic space and marketing now plays a substantial role in determining which symbols are prominent. Traversing the corporate portal is increasingly like using the free urban maps offered to tourists: the sponsors' establishments are easiest to locate. Old maps produced by explorers were inaccurate too, but they have gradually become more accurate, with the aid of technology. Undoubtedly in the years to come the illusionistic spaces now found in video games will find their way into the portals. But it is not clear that such technical advancements will lead to better, impartial reference material. Instead of surveying the Web, today's mapmakers are, more and more, studying the psychology and behavior of the person who is looking.
In accordance with market research and the example of Yahoo!, the portals have begun to look a lot alike. Culture Map asks whether the categories typically featured by portal companies in their Web directories really reflect the content of the Web. Accurate maps keep their subject matter in proper proportion. But how does one quantify Web content? Culture Map uses the survey data of three prominent search engines.8 Each day Culture Map records the number of hits that are reported for keywords like "Arts," "Romance," and "Shopping" by search engines. This enables a crude estimation of the prevalence of sites pertaining to particular topics. Visualization of these estimates places the conventional logic of the common top-level portal categories in juxtaposition with the more empirical reports of the Web spiders.
Using Culture Map, one begins by selecting keywords from a list of 32 common "top-level categories."9 Next the selected terms are compared graphically. Figure one shows that the term "Sports" (red) is more abundant than "Science" (yellow). It is possible to select which search engine's data is used, or one can view an average of all relevant data.
In addition to showing the relative abundance of different keywords, Culture Map depicts fluctuations and trends. It serves as an aid to memory, keeping a daily record of search engine results and providing several ways to visualize that data. One option is to view the recent "performance" of several keywords as a stock ticker (Fig. 2). In this way Culture Map mimics the urgent statistical reporting about the stock market, and it speculates what such attention might mean in the cultural sphere.
In perhaps its most interesting visualization mode, Culture Map lets one watch changes in the relative abundance of, for example, "Shopping" and "Family." It is easy to suppose that terms associated with electronic commerce are becoming more prevalent, but Culture Map makes such trends visible as time-lapse movies. The first frame shows the terms as they were reported in February of 2000 (Fig. 3). The image then metamorphoses to reveal the proportions most recently reported. The evolving composition depicts the dynamism of Web content. People who produce Web content may see in these tranformations, in some small measure, the effects of their own actions.
Both the artificial lens of the time-lapse movie, and the ironic metaphor of stock market analysis, point to the emerging qualities of the Web. With the Web changing shape as it expands, it is unwise for would-be cultural leaders to ignore the network's emerging characteristics. Considering the important role electronic media now play in the formation of social ideas, values, and expectations, it is critical that artists and intellectuals concern themselves with the representation of this new "working space." The future visibility of lightly-marketed cultural offerings could be at stake.
Consider, for example, that there is no obligation to publicize a Web site with search engines. So many documents that are accessible via the Web are not known to search engine "spiders." Although registry databases exist that identify the names of most servers, it is not a trivial task to determine what Web pages are available from each server.
Only the quantitative report of how many pages contain a given term is used. Any mapping of the Web's content is bound to be imprecise. CultureMap does not pretend to reveal any rigorously objective picture of the content of the Web. That content is seen through the lens of three profit-driven search engine services. As regards the ranking of the search results, I have been assured personally (May 30, 2001) by Ricardo Echevarría, the Director of Multimedia Development for one of the world's largest portal companies, Terra, that page rank is a commodity sold by all the major portals.
Shneiderman, Ben. "Tree visualization with treemaps: a 2-d space-filling approach", ACM Transactions on Graphics, vol. 11, 1 (Jan. 1992) 92-99. See also: http://www.cs.umd.edu/hcil/treemaps/
The "Market Map", and many other visualization projects, are described here: http://www.cybergeography.org/atlas/.
Stella, Frank. Working Space. Cambridge and London: Harvard University Press, 1986.
Notably, Roy Ascott's contribution to Electra, organized by Frank Popper at the Musée d'Art Moderne de la Ville de Paris in 1984.
Haacke, Hans. "MetroMobiltan". Weber Gallery, New York, 1985.
For details see: http://artcontext.org/cultmap/docs/help.html#method
Thirty two categories were selected, in January 2000, after reviewing about ten English language portals. These are "Top Level" categories that were found on several sites.