The introduction of the first Web browsers in 1994 and 1995 was a pivotal moment in establishing an intransigent panoptic state on the Web, but a subtle proprietorial architecture had previously been encoded into the Web’s particular form of hypertext. As one of the Web’s chief inventors, Tim Berners-Lee (1999), acknowledges, hypertext markup language l) for the Web sought to maintain authorial and editorial control over Web content and the characteristics of the networking of Web documents themselves through the insertion of hyperlinks. However, although such practices are the norm today on the Web, early hypertext systems, particularly Ted Nelson’s Xanadu system (1960), allowed any user to add, delete, modify, or edit text and hyperlinks. Nelson thus envisioned a hypertext program that enabled universal access and promoted a populist ideology, generalist knowledge, pluralist politics, and a consequent engagement with controversial and radical ideas.[2] At the center of Nelson’s system was the hyperlink itself, typically a word or phrase that when chosen would automatically take the user to the referred document. In Nelson’s vision, the hyperlink was perhaps the single unifying element of the medium of hypertext: it laid bare the very conditions and practices of computerized connectivity, thematic associations, lines of argument, and potential ‘‘paths’’ of reading that a user could follow.
As has been argued elsewhere, hyperlinks serve distinct indexical goals, suggesting—and indeed providing the means to make— associative connections (Elmer 1997a; Shields 2000). With the advent of graphical Web browsers (Netscape’s Navigator and Microsoft’s Explorer), the indexical characteristics of hyperlinks were given added prominence through the deployment of a graphic user interface (GUI)—in particular, the choice of an iconic index finger as the cursor that enables the activation of a hyperlink. Thus, once a user positions a cursor on a hyperlink, the cursor is transformed into an index finger, signifying not only the link (as a keyword, an image, or a phrase) but also the actualization of Web connectivity. Bennington (1994) refers to this indexical characteristic as the process of designation, or the enabling of a particular course of events—in this instance, the connection to another Web document or site. Web hyperlinks are therefore not merely a sign or metaphor, as semioticians like Charles Peirce conceive them. Rather as a deictic index, they actually enable the connectivity of the Web-as-index.
After such decentered gateways to interdocument connectivity, association, and networking (via hyperlinks) were redesigned and formatted by Tim Berners-Lee (1999) and others for the Web (as hypertext markup language), though, they still raised serious questions about usability. How can a decentered document system with theoretically multiple entry and exit points (hyperlinks) be made workable? How can such a system be structured to enable efficient storage, search, and cataloguing functions? In short, where might users effectively begin to search this electronic, networked system?
The browser, through its default, start-up Web page (typically a Netscape-AOL or Microsoft ‘‘portal’’—in essence an online categorized index), attempted to provide a searchable and universal point of departure for Web browsers. The popularity of such portals is unchallenged: AOL, MSN, Yahoo!, and Terra-Lycos, continue to consistently rank in the top five most visited sites on the Web.[3] Yet as Introna and Nissenbaum (2000) remind us, such portals (or to be more precise, the databases that are offered for these search engines) encompass fewer than half of all the sites on the Web. Search engines and portals, in other words, do not provide access to everything on the Web but rely on a host of strategies to construct selective representations of the Web. In addition to asking users to submit their sites for registration (as Yahoo!, for instance, recommends), search engines such as Google rely on automated cataloguing agents or ‘‘spiders’’ to construct their databases. According to Netcarta Corporation (1996, 1), the main producer of Internet mapping technologies,
Spiders—also known as intelligent, user, or Internet agents —are commonplace on the World Wide Web, and perform a variety of onerous tasks automatically so users can be more productive. The most common spiders simply roam and mine the Internet for data.
Search engines like Google are not constructed by simply following hyperlinks from document to document (or from Web page to Web page). They also confer relevance and authority to Web sites based on the number of incoming links. Sites with more links, in other words, are more apt to rate highly on Google’s lists of search ‘‘hits.’’ The Google database thus constructs a hierarchical metaindex from a topography of hyperlinks. With the help of the browser, the search engine—as a database of Web pages and sites—renders an indexical form of Web connectivity. As a consequence, Web sites and pages are no longer atomized or decentralized.
Surveillance on the Web is a similar means of sorting and categorizing users and is enabled by a similar logic of connectivity, not unlike how hyperlinks connect documents. Web surveillance is largely concerned with facilitating reliable and continuous connections between users and (largely) commercial servers. In fact, one might argue that on the seemingly discontinuous Net, making connections (and keeping them) have become the primary commercial strategy. Continuous computer-mediated communication via instant messaging programs, for example, has become a hotly competitive technology that Microsoft, Yahoo!, and America On- line have used to acquire new subscribers, users, and consumers.
E-tailers and other consumer-driven online Web sites initially viewed Internet users as a decidedly disconnected class and made periodic requests to large servers for content, services, and commodities. In Internet parlance, users and e-tailers were mired in a ‘‘stateless’’ environment, with no overarching mechanism for continuous, two-way interaction. With the advent of the first Netscape browsers, however, the Internet became both increasingly more visual and decidedly more connected, especially between individual users and Web site proprietors. Netscape’s cookies provided Web sites and servers with continuous and often automatic access to users’ hyperlinked browses. They also challenged users’ exclusively personal use of their personal computers. The PC hard drive has become for most purposes the end node of the Net—the point at which personal information is continuously integrated into the medium of the Web.
[2]Refer to <http://xanadu.com.au> for more on Nelson’s ideas and the Xanadu hypertext system.
[3]See the Jupiter Media Metrix top 50 Web site rating reports at <http://www.jmm.com/xp/jmm/press/mediametrixtop50.xml> (accessed June 1, 2001).