By Way of Introduction

In the following essay, I will attempt, among other things, to approach the ways in which writing on the Internet represents a marked change from writing as stable, anchored to context and authorial intent. Whether written specifically for use on the Internet or written for print and later placed online, Web-based text has built into it or grafted onto it, certain “meta,” or peripheral, information designed to take into account the users who might search for it or eventually stumble across it. This meta information, including hyperlinks, keywords, images, headlines, titles, and meta descriptions, works on the periphery of the text itself to make its parts accessible to an audience by making them visible and relevant to search engines and Web editors.

Readers within this system are thus empowered to search among and choose between texts and Web artifacts, ultimately writing their “own” narratives based on those choices. This essay will draw parallels between web-based writing and contemporary theory, particularly Jacques Derrida’s Signature Event Context, in order to show how web content demonstrates that essay’s suggestions regarding the ultimate un-anchor-ability of language. It will also seek to contest the standards of print-based writing in order to demonstrate, structurally, the ways in which web-based writing can defy the kind of authority-reinforcing structure by which previous iterations of language were defined. This text, therefore, as should become evident, will contain meta-information designed to facilitate searching within, outside, and around it, which will hopefully open it to interaction with potential readers and within new potential contexts.

One of my main goals in this essay will be to examine the way language constitutes itself on the web by removing any illusion of anchoring in author, intent, stable context, or truth value. I will also attempt to show the way that search is itself a kind of utterance, performative in structure, which changes what it seeks to find in the mere desire to find something. The goal here is not to discourage searching but to suggest that searching using community-generated content, rather than top-down, editorialized content, may help a reader to overcome the limits of categories and space that affect non-digital language, ultimately allowing users to access information that they may not even know to ask for. Searching based on intuition, then jumping off the structures that facilitate searching, moving even temporarily without the aid of meaning-imparting institutions, is a way to find information users never knew they were looking for.

On Patterns, Codes, and Visibility

Marshall McLuhan suggests in The Global Village that “data overload equals pattern recognition.” In A Thousand Plateaus, Gilles Deleuze and Felix Guattari suggest that the rhizome, their metaphor for modes of production, including data proliferation, “is an acentered, nonhierarchical, nonsignifying system without a General and without an organizing memory or central automaton, defined solely by a circulation of states,” but they go on to state that a “rhizome is made up of plateaus,” which are “always in the middle. . .” Daniel Dennett, discussing evolution and how it tempers an infinite realm of possibilities, asserts,

A system has a degree of freedom when there is an ensemble of possibilities of one kind or another and which of these possibilities is actual at any time depends on whatever function or switch controls this degree of freedom. Switches (either on/off or multiple choice) can be linked to each other in series, in parallel, forming larger switching networks, the degrees of freedom multiply dizzyingly, and the issues of control become complex and non-linear. Any lineage equipped with such an array confronts a problem: What information ought to modulate passage through this array of forking paths in a multi-dimensional space of possibilities?

Although they come from very different theoretical traditions, these authors approach the same problem: that for any given uncontrollable, acentered, dizzying array of information, some—at least temporary—focus, control, or entrance must be present to make pattern recognition, and therefore, visibility (iterability and utterability), possible. With regards to the masses of information found within, on, by, and around the Internet, the portal website, whether it works as a search engine or as a site of recommendation, offers the entrance necessary to make those networks of information visible to users.

What, then, to explore the alternative, would happen were there not some method of accessing, more than once, any given mass of information, specifically writing as it appears on the Internet? As Derrida suggests, “A writing that is not structurally readable—iterable—beyond the death of the addressee would not be writing.” He notes conversely, ‘there is no such thing as. . .[an] organon of iterability—which could be structurally secret” (Derrida 7-8). The implication here is that the only way for a mark to remain secret or invisible would be if it were not iterable, and therefore not a mark. On the Internet, existence as an organon of iterability qualifies the language of a website as a mark, but mere existence doesn’t guarantee that the site will ever be seen. In order for a mark to effectively exist on the web, it must make itself visible to users. So, to answer the question initially set out, without some method of access, any information on the web would be forgotten as soon as it occurred, which would ensure that it wouldn’t ever pick up momentum, let alone create force or effect among its audience.

The earliest problem of the Internet was that although there was a wealth of knowledge and information being stored on servers around the world, there was no method of accessing that information; the information may as well have been invisible because it was not indexed, quantified, or related to anything else on the Internet, which made it not only impossible to search for, but also impossible to stumble across accidentally. In the mid-nineties and before, to even learn of a website one had to have a personal recommendation, invitation, or some other initiation to be admitted into the discourse of “the Web.” What was missing at that time was a tool that could not only itself find and categorize the vast array of information but could also provide entrance to users, so that they could search for the information they felt might be out there, but were otherwise ignorant of.

Repeatable patterns were missing from the early Internet—collections of information had no relation, loose or otherwise, to any other, making navigation a plodding, start-and-stop process that could never be guaranteed a repeat. Without being findable in a reliably repeatable way, information languished unseen by the users for whom it was constructed. Users made queries in attempts to find information but often got no response, indicating that the documents were unable to interact with users—to even let users know “I’m here!” What was missing was a way for information to compose itself such that it could respond to queries posed by searchers.

Search Is the Destination

In order to access or retrieve any computer file or website, one must know the name, address, or general location of the text. However, the issue of access is complicated. There is the question of finding what one knows is there but isn’t sure how to find, and there is the deeper question of finding what one doesn’t know is there. In the pre- or non-digital world, in order to access a text one doesn’t know exists, a recommendation is necessary—by another text, a friend, a teacher, a snippet from a film, maybe something like the Dewey Decimal System’s categories. Then, the recommendation must be compelling enough for one to find the reference, borrow it, or go out and purchase it. In other words, the recommendation must be memorable enough to sustain the protracted search that will ensue. What about in the digital world though: does accessing the unknown-unknown involve a different process? This is where, as the Internet and its problems of access grew, the search engine came in.

In order, though, to talk about search engines, one must first address the “portal website.” The portal website is, as the name implies, a go-to site that serves as an entranceway to the rest of the Web. Portals such as Yahoo! employ teams of editors and ontologists to group information by determined categories, as well as to create value, in so-called “editors’ picks,” “best bets on the Web,” or simply “Headlines.” Here, the authority of the portal operator is implicitly trusted. But, as Clay Shirky points out in his Web-focused lecture Ontology Is Overrated, there’s a problem with this kind of organization:

Look for the word “queer” in almost any top-level categorization. You will not find it, even though, as an organizing principle for a large group of people, that word matters enormously. Users don’t get to participate in those kinds of discussions [in] traditional categorization schemes.

Even though, as this passage implies, there may be some substitute category for “queer,” the problem is that anything indexed under that politically correct heading was probably also found using politically correct search terms, which means anything written outside that language wouldn’t be included. The point that Shirky makes here is that if a user doesn’t have the ability to search using his or her own personal categorization scheme, the nuances of his or her desired terms of search will be lost.

Portal sites thus contain a built-in exclusionary possibility, and thus inevitability, even as they attempt to provide access to the masses. Portal websites were not constructed as means of hiding or diverting access to information; on the contrary, they were constructed to make it possible for users to find anything. Unfortunately, the categories failed, mainly because categories devised by a select few just didn’t translate to the masses who went searching online every day. Just as language, as a part of its structure, can lose track of its author’s intentions, so too, these portal websites’ intentions of access were diverted. As Derrida asserts, “a possible risk—is always possible, and is in some sense a necessary possibility,” meaning for portal sites that part of what made their existence possible was their exclusion of lots of confusing and contradictory data (“Signature Event Context” 15). The result? Portal Websites mean users are only able to find information that adheres to a small group of editors’ concepts of category and value. If the popular adoption of Internet as information central was something of a Big Bang, an explosion and proliferation of strings of potential energy and data, the nearly immediate reaction by those who wanted to create meaning of all that information was the opposite: a sharp contraction of all that potential into little dense spheres of category, light years apart and with no real means of meaningful interaction or overlap. Search on the Web, therefore, started as a top-down process of adding meaning to a mass of information by imposing on it arbitrary and essentially univocal values.

It must be acknowledged here, though, that a need for some means of access, whether top-down or not, was necessary at the time portal websites were created because “early on you couldn’t put a search box in front of people and expect that they would know what to do. . .users were new to the experience; there were no preset habits attendant to [web] surfing” (Battelle). For this reason, portal sites websites included search capabilities, although content was still mainly composed of listings of categories, which were meant to make it easier for users to find what they were looking for—a particular book could be found under the heading “Books,” for example. “A hierarchical approach simply made sense for a public trying to understand the wild and rather disorganized chaos of the early Web. As surfers moved from a stance of exploration (‘What’s out there’) to expectation (I want to find something that I know is out there’), search as a navigational metaphor began to make more sense,” which is where those little search boxes found their genesis (Battelle).

As users began to comprehend that there was plenty available on the Web and that they only needed to find the right path to that information, the search box began to be used much more than before. Here, a blinking cursor and blank box transferred responsibility to the user, asking, “what do you want to see?” Still, search engines remained somewhat unreliable. Unless a user knew the perfect query with which to call up a piece of information, that information still lay hidden. The reason for this has to do with the structure of the search engines and the documents themselves. Each relied heavily on its authors to generate certain “meta” (that is, self-referential) information by which it could be categorized, identified, and thus found. The algorithms created by the search engines’ programmers made assumptions about types of categories and the ways information could be relevant to an individual. In short, search engines undermined the success of their endeavors by using the ideas and work of a select few to create value and relevance for many. Additionally, search terms were at first near-impossible to come up with, because as John Battelle points out in The Search, “in the initial stages of the Internet, [. . .] no one knew what was out there.” Because both user and search engine had to practice asking and answering relevant questions (finding an iterable structure) before they got any good at it, searchers and search engines all had to consider—“what is relevant?”

At that relatively early stage of search, there was no way for users or search engines to know what was out there, let alone what was relevant—because search engines had not yet indexed everything on the Web, or the ways they had categorized their indexed material was impossible for a user to anticipate, which made coming up with a relevant search term nearly impossible. Further, as many early search engines found out, selecting from a “dizzying array of choices” was hard for users, as well as for the search engines themselves. Search engines found that their methods of ranking information, based on all kinds of meta-text was unreliable, either because its authors didn’t realize exactly how the information they used as meta text contradicted the purpose of their site or because they were simply trying to manipulate the system in order to achieve a higher ranking. In the same way that in language, according to Derrida, “the intention animating the utterance will never be through and through present to itself and to its content,” web utterances were often contradictory and authorial intentions diverted. Because of these difficulties, early in the game it was very nearly impossible to get a result relevant to what one might be searching for. So, although the searcher intuited that the information must be out there, there was no way to access it—it may as well not have existed.

What Google innovated was a technology called PageRank that revolutionized the way search works. Its creators increased the relevancy of websites returned in a search by incorporating the age-old idea that “the process of citing others confers their rank and authority upon you” (Battelle). In order to measure rank and authority, Google set up an algorithm that managed to take into account both the number of links leading into a particular site, and the number of links exiting into each of the linking sites, which “mirror[s] the rough approach of academic citation counting. . .” (Battelle). With this innovation, Google placed more importance on what others said about a site rather than what the site’s author may have said about him or herself. In this way, the search began to expand the network without author, without source, without center, ranking everything in relation to every other piece of information—rather than ranking it for “truth” value or authorial intent, a text is valued on its ability to spread, to network, and to be referred to by others.

As Google and its complicated algorithms evolved toward increasing relevance in search, individual users also began to learn what kinds of search queries returned relevant results, and as they did so, modified their search behavior. In fact, Battelle points out that users generally enter five search terms each time they’re looking for something, suggesting that users refine and revise their method of questioning until it turns up something useful to their needs. Another component of Google-style search is the information itself—the bank of websites out there competing to be seen on that coveted first page of search results in Google. These websites also take into consideration how users might search for their particular brand of information, and successful website design and SEO (search engine optimization) incorporates keywords and meta-information designed to be the most relevant result to a user’s search within that category. Knowing that websites will try to find ways of being relevant, whether they have the user’s best interest or search intent in mind or not, Google uses secret, complex algorithms based on a combination of keywords, images, headlines, and citation found within a site in order to prevent website designers from pretending to be more relevant than they really are.

But there’s a limit to these search engines, and that limit is the algorithm that determines relevancy. There is a trade-off. Although users learn better to intuit how others might categorize information, and learn to ask all kinds of questions to find what they want, search engine responses can only be as good as the temporary and secret algorithms the company programmers use to separate ‘”real” relevance from “false” relevance.” The effect of the Google-style search engine is that while it provides far more relevant results than portal websites and earlier search engines, it always contains in its results the dangers of irrelevant results, that is, positive search engine placement that was ill-gotten because the designer of a website was trying to “game” the system in his or her favor to appear at the top of the list of results. In much the way Derrida lays out the idea that “the structure of possibility of this utterance” intelligible, although mistaken, misleading, or badly formed. . .”includes the capability to be formed and to function as a reference that is empty or cut off from its referent,” every supposedly relevant search engine result contains within it the possibility that it may simultaneously be the same and different from what it cites, because relevance only becomes a goal when irrelevance threatens to overtake it.

Nevertheless, search, especially Google, continues to be a destination for most designers of content for the web, first of all, because it works when you want reliable, repeatable, relevant information, and second of all, because businesses or artists or politicians attempting to sell themselves on the web have no better way of reaching the public than by appearing in the number one slot on the first page of results in Google. “Visit us on the web” is just the new way of sending a customer to the best possible source of marketing and sales, no longer a brick-and-mortar establishment, but a virtual marketplace.

Making It All Happen

Search engines, in their freedom to count any information as “relevant” regardless of truth value or authorial intent, echo Derrida’s discussion of communication in Signature Event Context, in terms of “freeing the performative from the authority of the truth value, from the true/false opposition, at least in its classical form, and [substituting] for it at times the value of force” (Derrida 13). This concept of the performative in the context of Internet search is especially relevant because of its identity in force, rather than its meaning in an author’s intent or its truthfulness. Derrida defines the performative as an utterance which “produces or transforms a situation” and serves as “a ‘communication,’ which is not limited strictly to the transference of a semantic content that is already constituted and dominated by an orientation toward truth” (14). Here, the term “search query” is a “performative,” for any series of terms entered into a search engine with the intent of creating a list of results and possible choices is an utterance that “produces or transforms a situation.” Additionally, that a search query entered into the blank box on Google’s home page could be any combination of words adhering to grammar or not, with regard for “truth” or not, makes it performative in its existence as a “communication” not “limited strictly to the transference of semantic content.”

To return to the idea that a search query is performative with the built-in ability to transform a situation, we must acknowledge that this Internet version of the performative utterance comes in the form of a question. A search term or phrase, whether fragment, “mashed potatoes” or “mashed potato recipes” or full interrogative “how do I make mashed potatoes?” always asks a question. This question, when used to activate the search algorithms of a search engine, then pulls up page after page of results, effectively materializing that information for the user. Thus, the retrieval of a web-based artifact through some imperative or some interrogative, that is, some performative statement or action, as described by Derrida, “produces or transforms a situation, it effects” (Derrida 13). These artifacts, therefore, must be written to be read; they thus become rewritten into a new context: the context of the question to which they serve as answer.

The search query thus constitutes a performative utterance in its calling up of information. It describes the thing which it intuits is out there but for which no location is known, while it also calls that thing into being by entering a search query that is answered by the text’s materialization. The text is then transformed as the result of being called forth in search because it becomes a new answer and thus a new text each time it appears in response to a different string of search terms.

Readers as Writers, Writers as Readers

Every sign [. . .] can be cited, put between quotation marks; in doing so it can break with every given context, engendering an infinity of new contexts in a manner which is absolutely illimitable. This does not imply that the mark is valid outside of a context, but on the contrary that there are only contexts without any center or absolute anchorage.” (Derrida 12)

What does it mean to enter a word or phrase into a search engine, expecting to get some result? Search terms represent an acknowledgement, in the act of writing, that these terms are citations, that they have already existed in writing somewhere. As Derrida suggests in the above passage, such citations produce a myriad of possibilities supported by a myriad of incalculable contexts. A searcher, therefore, in making a language-based query, pulls matching, fragmented terms from whole documents, recontextualizing them in terms of their own, personal search results. In effect, the act of search is an act of rewriting that which one is searching for, making the potential reader the writer before he or she ever becomes the reader.

To explore readerwriter roles, we will attempt to lay out the properties of search and find. Searching for a term from a bank of others’ knowledge and information means learning to describe it in ways other than you yourself might initially intuit to write—that is, constructing a search term that searches other people’s materials means constructing search terms in their words. To capitalize successfully on the knowledge that is potentially available, therefore, one must learn to think, not collectively, and not individually, but from many different individual perspectives: as many as possible, in fact, for the more ways of searching, the more likely you’ll find a relevant result.

Searching is pattern recognition, because in order to search for something one knows exists or intuits exists, one must identify a meta-description of it. In searching, one isn’t recreating the exact text of what one hopes to find—because the searcher doesn’t know exactly what he or she is looking for. The searcher enters terms that represent some pattern, whether in category, author, group, etc. . . The terms are entered with an expectation that they will return an array of choices; indeed, users of search engines say that the most important thing to searching is getting a wide range of information, not just one piece of information. What this means for the searcher is that the searcher enters somewhat generalized information that could apply to a number of artifacts.

The successful or relevant items returned in a search correspond to the patterns laid out by the searcher; they capitalize on self- and community-generated meta-descriptions—that is, on the references made to them, which confer authority. Therefore, the key to successful searching is learning to write and read patterns in a variety of ways—not only through website content, but also through meta-textual content—that is, through the structure, the captions, the headlines, the keywords, the metaphors, the references, the images, etc.

Just as the act of searching online represents simultaneous reading and writing, perusal of web-based documents is also a reading and writing act. The hyper-textual aspects of text on the Internet mean that a user has a series of entrances and exits on each page, each representing a choice of path. Just as a page of search results in Google represent choice, and therefore, the imposition of the searcher’s intentions on that information, so too, hypertext links offer the reader a chance to write. George Landow, in Hypertext, says this: “All hypertext systems permit the reader to choose his or her own center of investigation and experience. What this principle means in practice is that the reader is not locked into any kind of particular organization or hierarchy” (13). Choosing a “center of investigation” is complicated by the fact that a user does not start from the document—the user must first get there somehow, which means that with multiple ways to access a document, the “center” is as much outside as inside the document.

The ability of a user to choose what pages to read and what hyperlinks to click on “blurs the boundaries between reader and writer,” allowing a reader to, in essence, write his or her own text whose narrative traces its way along the “click-stream” of the user (Landow 13). For there is no rule on the Internet that anyone must read to the end of a work, watch to the end of a video, or even remain on a page for more than the time it takes to click the next hyperlink. This lack of central control or authorship creates a text that is, in the words of Mikhail Bakhtin, “constructed not as a whole of a single consciousness, absorbing other consciousness as objects into itself, but as a whole formed by the interaction of several consciousnesses, none of which entirely becomes the object for the other” (qtd in Landow 11). Information constructed with hyperlinks, therefore, represents a system of communication in which the page a user is reading is not the end of the line, the point where “truth” is transmitted; rather, the transmission of value or force is determined by the movements of a user through that text. Here, again, Derrida’s thinking on language can be applied to the contents of the Internet in his assertion that “writing is read; it is not the site, ‘in the last instance,’ of a hermeneutic deciphering, the decoding of a meaning or truth” (Derrida 21).

So, what does hypertext do in this context? It knives open a text, surgically removing its center and eviscerating it, widening the entrance point, and in fact, flattening the body such that every point is an entrance point. With hypertext, no longer is a text going to remain stable; it will be open to endless recontextualization. Landow theorizes instead “one experiences hypertext as an infinitely de-centerable and re-centerable system, in part because hypertext transforms any document that has more than one link into a transient center, a directory document that one can employ to orient oneself and to decide where to go next” (Landow 12). Here, it is evident that a reader writes his or her own text by choosing where to go among a network of choices.

Bottoms Up – The Giddy World of User-Generated Content

As we get used to the lack of physical constraints, as we internalize the fact that there is no shelf and there is no disk, we’re moving towards market logic, where you deal with individual motivation, but group value. (Shirky Ontology Is Overrated)

To return to the earlier discussion of search and its limitations, we should ask whether there could be some means other than using editorialized content or algorithm-determined relevance, by which users could find useful information on the Internet. The above quote from Clay Shirky hints at the possibilities: the web is not constrained by space, time, or top-down logic. What Shirky is referring to is user-generated content, that is, content that has been created and tagged by many, many users, rather than by a few authorities. The idea of market logic he refers to is that even though the individuals tagging information may have contradictory or competing points of view, they provide, in sheer numbers, a more accurate depiction of how a piece of information could be found than a single, really consistent, really knowledgeable person. An example is the site del.icio.us, a social bookmarking site that allows users to create tags (identifying descriptions) for any website they view with whatever keywords occur to them. As del.icio.us describes it,

del.icio.us is a social bookmarking website, which means it is designed to allow you to store and share bookmarks on the web, instead of inside your browser. This has several advantages. . .you can share your bookmarks publicly. . .you can find other people on del.icio.us who have interesting bookmarks and add their links to your own collection. . .You have access to the links that everyone wants to remember. You can see whether two people have chosen to remember a link, or whether it was useful enough for a thousand people to remember — which may help you find things that are useful for you, too. (del.icio.us)

Here, the individual motivation of a user comes into play, but group value is gleaned because that individual tagging scheme is placed with all the other individual users’ tags, creating a network of search terms created by and for users intimate with that information. As Ontology Is Overrated states it, “Each individual categorization scheme is worth less than a professional categorization scheme. But there are many, many more of them” (Schacter qtd).

To understand where the ideas of user-generated content can be linked with Derrida’s previously referenced discussions of language, we can compare a quote from Derrida and one from Shirky:

an opposition of metaphysical concepts (e.g. speech/writing, presence/absence, etc.) is never the confrontation of two terms, but a hierarchy and the order of subordination. Deconstruction cannot be restricted or immediately pass to a neutralization: it must, through a double gesture, a double science, a double writing—put into practice a reversal of the classical opposition and a general displacement of the system. It is on that condition alone that deconstruction will provide the means of intervening in the field of oppositions it criticizes. (Derrida 21)

Merges are Probabilistic, not Binary – Merges create partial overlap between tags, rather than defining tags as synonyms. Instead of saying that any given tag “is” or “is not” the same as another tag, del.icio.us is able to recommend related tags by saying “A lot of people who tagged this ‘Mac’ also tagged it ‘OSX’.” We move from a binary choice between saying two tags are the same or different to the Venn diagram option of “kind of is/somewhat is/sort of is/overlaps to this degree”. That is a really profound change. (Shirky)

What we see in Derrida’s quote is that in order for deconstruction to provide a “means of intervening in the field it criticizes,” it must resist binary oppositions and reactionary swings from one extreme to another. With Shirky’s quote, we actually see this movement in play; rather than rejecting traditional top-down categorization by rejecting categorization altogether, user-generated content uses the structure of category, but subverts it, changes it into categorization “where individual differences don’t have to be homogenized,” and where overlap and difference, rather than mutual exclusivity, represent the norm—best of all, user-generated categorization schemes work from the bottom up to create value. User-generated content, in the end, strengthens its position in the canon of useful web information not by reinforcing existing categorization schemes, but by being responsive to users’ changing views of categorization.

What You Didn’t Know You Didn’t Know

Taking a cue from Shirky, we can suggest that since units of content on the Web don’t have to adhere to any of the particular constraints of categorization or space that served as the condition of their creation, readers and writers on the web are free to reenvision whatever model of organization makes sense or is useful, even temporarily. This is happening in all kinds of sites, from del.icio.us to stumpleupon to youtube to many other user-generated websites designed to remain as much in flux as the users who access them daily, hourly, and on a constant streaming basis.

Here, the past and present of web search can be pulled together to demonstrate effect on the user. Whereas in the beginning of the Internet familiar categorization schemes were necessary to let users know what existed, the now assumed presence of all meaningful information makes it possible for those former schemes to be jettisoned. What this means for users is that previously held notions of language being tied to authorial intent, a single context, or some purported truth value are being dispelled in favor of a realization that information is only relevant in its ability to interact with users, not in its static value independent of them. With texts thus open to constant recontextualization, readers, writers, and web users have the ability to see language from a variety of perspectives and contexts, based on the language not just of the author, but based also on their own language in tension with the many users who have also tagged that information.

Readers on the web, then, have the ability to see information from others’ perspectives, allowing them to find things they themselves may not have had the language to know how to ask for. This is one of the main powers of the web—that is, the possibilities laid open by searching for something we cannot name, and finding things we never knew we were looking for. Therefore, as we sit down to a terminal, it is anything but.