Page 96 →Chapter 5 Production Collectives
Page and Screen
The curatorial skills required to transform a reference text that contains narratively arranged words and images into a useable, accessible, unified compendium with searchable content are frequently unnoticed by casual readers. This invisibility is partly due to the simple fact that solid organizational schemas and usable navigation elements such as cross-references and strategic links are largely unnoticeable when done well. Many genres are meant to be read as linear narratives that are read front to back. This is not the case for reference texts, which the reader often picks up in search of specific information. Consequently a multivolume reference work like an encyclopedia must include strategic, manageable interface elements that make the information findable and, ideally, make it easy for the reader to locate related information. These strategic links, recomposed texts, descriptive metadata, and information architectures are suasive elements that contribute heavily to the ethos of arguments, reinforce worldviews, and sanction knowledge, as Chambers’s interlinking of Newtonian topics demonstrates.
Jodie Nicotra has powerfully addressed elements of this skill set, arguing that understanding findability and information architecture as compositional skills is fundamental to teaching forms of digital literacy that account for the social, networked nature of contemporary writing. The discipline of writing studies has long been interested in rhetorical contexts of literacy, but as she has argued, familiar ways of teaching essayistic writing no longer fully account for the ways that writing is developed or functions on the web: “Now more than ever the focus is much more on the organization of the total network than on the individual producer of texts. . . . The sheer amount of information with which we’re dealing now and the medium in which it primarily occurs has perhaps given the importance Page 97 →of organization over individual authorship a heightened intensity. Thus, the issue of findability [and information structures] is an important one for contemporary rhetoric and composition.”1
Nicotra focuses on folksonomic tagging in social sites like Flickr as an example of small, collaboratively produced texts that negotiate and facilitate persuasive findability. This distributed development of information taxonomies is both rhetorical and indisputably categorized as writing, She contends, creating a space for students to develop grounded understandings of audience awareness and a metacritical awareness of network participation. The labor and small texts necessary for this work are not less important or diminished writing simply because they are less visible writing. Rather the converse is true: these elements are essential for functional, networked texts and carry with them the value attached to writing-as-skilled-craft.
These skills apply equally to print and digital encyclopedias, but digital affordances render them more dynamic and more easily managed. These elements also help both writers and readers realize the full potential of digital environments. Form is a particularly powerful element of textual agency, and hyperlinks are a vital textual form that enables lively, traceable textual connections and paths for readers as well. They are also important tools for curators, since category links enable taxonomic sorting with a single click that pulls all pages with the same tag into a unified query result. The same collection of articles can be sorted in myriad ways, resulting in a variable text that can recontextualize data through juxtaposition. The potential for such serendipitous juxtaposition is as important and intentional an element of the modern encyclopedia as careful arrangement has become.
Nonhuman actors are essential for developing curation at this level, from the typographic conventions that distinguished cross-indexes in print encyclopedias to the complex, dynamic infrastructure of digital reference works. These actors enable the development of sustainable interfaces and shape the performances of readers. Two levels of interfaces are important in curatorial processes: the page or screen that the reader uses, of course, but also interfaces that allow the material production community to perform the work of preparing the curated text for use. Production workers are another crucial part of curatorial collectives, and yet, like the structural elements that they frequently directly influence, they remain largely invisible to the reader. Without their skill, labor, and coordination, these texts would largely not exist.
An exploration of the ways that human and nonhuman actors influence development of both levels of interfaces must begin with an understanding of the problems of arranging knowledge in information-rich environments and building in textual features that make the resulting information structure usable for Page 98 →readers. This work is undertaken not just by the curatorial communities, but in printing houses and labs, both of which make heavy contributions. Furthermore the interfaces shape not only the text but also the content contributors and the development community that supports those interfaces. All these members of the collective play important roles in the process of building encyclopedic texts and pushing them into successful circulation.
Chambers wrestled with the problems of information infrastructure during a period that was itself confronting the realities of information overload. At the same time the intensity of natural philosophy experimentation and worldwide explorations sponsored by the Royal Society produced continuous impetus for a widespread effort to taxonomize the natural world. The process of arranging the encyclopedia was inevitably intertwined with this encompassing cultural invention process, since a usable encyclopedia had to effectively and efficiently order knowledge that was being contemporaneously produced by experts in multiple fields. The organizational challenges were monumental: “Chambers had to confront the reality that Bacon himself acknowledged,” Yeo has written. “Knowledge, however acquired, had to be recorded in print in order to be available for comparison and cross-checking. . . . The progress of knowledge depended on reliable storage, retrieval, and communication.”2 These problems of findability and taxonomization placed the encyclopedic author firmly within the realm of church and state, since curating this text required negotiating tensions concerning religious respect and the borders of religion and science. A taxonomy is rarely just an taxonomy; the problem of ordering knowledge bridges the questions of what is knowable and permitted to be known.
Until the early eighteenth century, reference texts were typically arranged thematically.3 Comenius’s Orbis Pictus Sensualium, for instance, contains 150 chapters with categorical divisions such as “inanimate nature” or “humans and their activities.”4 Other lexicons and grammars, which were often used as reference texts, followed suit, as did dictionaries. Such schemas respected the divine order of the natural world but were also necessarily vague, presenting a challenge for new readers of a text. Bacon had planned 130 sections for his own failed reference text that was sectioned in imitation of the Divine Work: Partitions of the Sciences, New Method, Natural History, Ladder of the Intellect, Anticipations of the Second Philosophy, and Active Science. While other encyclopedic projects did not mimic these divisions exactly, topical arrangement remained the standard. Chambers himself offered a map of topical divisions of knowledge in the preface.
His map shows forty-seven major areas of the sciences and arts. However, his primary goal was not classification in the sense of natural philosophy but rather “a reduction of the vast bulk of knowledge into a lesser compass.” He also demonstrated the ways that the Arts formed an important division of the Universal Page 99 →Knowledge that was essential to educated readers. By incorporating the arts as essential knowledge, he reinforced the title page’s claim that this work was indeed a universal dictionary.
Chambers’s taxonomy of knowledge. Image courtesy of University of Wisconsin Digital Collections Center
However, Chambers did not order his text according to this map. One of his central innovations was popularizing alphabetization by individual topic as a common means of ordering encyclopedic information. Alphabetization allowed for easier rearrangement in subsequent editions that would need to incorporate new articles and account for linguistic shifts. He also became enamored of it for its arbitrary placement of subjects.5 The topical juxtapositions imposed by alphabetization exploded any potential linear readings of his text. Chambers cheerfully extolled the heuristic virtues of this system: “When numbers of things are thrown precariously together, we sometimes discover relations among them which we could never have thought of looking for.” This novel means of arrangement represented a significant innovation that could be read as an imposition of man’s order on God’s creation, and Chambers went to great lengths in the preface to deflect potential charges of heresy. He argued that it provided a clearer means of order for a broad reference text, especially given what he called “the present wild state” of the English language.6
Page 100 →He was the first to introduce extensive cross-indexing in an encyclopedia, thus creating a traceable information network within his own text.7 His primary objectives were to help guide readers to other topics in the same basic area of knowledge and in the process promote a Newtonian worldview, but he also rendered nonlinear reading an overtly encouraged aspect of working with an encyclopedia. This expectation is made clear in the instructions that appear at the end of the first volume’s front matter, which explains to the reader how to navigate the unfamiliar cross-indexing between the tipped-in images and the central text: “The figures relating to each art are placed fronting the name of the respective art, in the body of the book, and are referred to under that title as Tab.Architecture, Tab.Geography, etc. To each figure is also annexed the word for whose exemplification it serves, so that the reader may either go from the word to the figure, which exemplifies it, or backwards from the figure, to the word which explains it.”8
Here we see Chambers abandoning an element of authority that we commonly accord to authors of narrative texts: the ability to dictate the ways in which a reader moves through the text. The readers of the Cyclopædia may follow their interests forward or backward between figures and texts—or anywhere at all within the book, tracing the network of cross-indexes. In the second edition these pointers to other relevant topics were typographically distinguished from the text of articles by being set in capital letters. This abiding visual convention was not developed by Chambers, though: it was an important innovation by the printing house.
Printed Page as Interface
The bound book has been with us so long now that it seems natural in our everyday interactions, but it is as unnatural as any other technology, a handheld nonhuman collective. It, like all other technologies, is a situated, integrated part of human life rather than a neutral object, and an important element of authorial agency.9 The book is a socio-technical system, comprising not just the author and publisher but the full range of humans and nonhumans whose collaborative work is essential to the finished product and its circulation. In the case of the texts under consideration in this study, humans’ development of the bookmaking craft and, later, moveable type enabled them to reproduce large-format, multivolume texts efficiently enough that they could be distributed to the literate public through paid subscriptions—as with the Cyclopædia—rather than limiting editions to handwritten manuscripts commissioned by the truly wealthy.10
The compositional contributions of the printing house illuminate another aspect of curation that accounts for audience concerns from a somewhat different angle. In this cultural moment when the Romantic Author was not yet reified or pervasive, the composition process was understood as far more distributed than Page 101 →the author-focused angle that is naturalized today. When a manuscript came to the printing house, the production process was not one of “slavish reproduction,” as Adrian Johns has noted.11 In her study of authorship and the early modern English text trades, Lisa Maruca has argued that “those who worked within many professions of the print trade (from printers to publishers to writers) understood books and other print products to be the result of a collaboration of many hands and the process of textual production to include not only writing but also the work—and workers of—technology.”12
While the text’s composer may have kept marketability and economic factors in mind entirely or not at all, the printer worked with these issues as a foremost concern. Having been commissioned by the publisher, the printer was bound in a financial agreement that focused on cost, labor, and marketability. (After all, successful sales might well mean additional printings.) The compositor, printer, and binder all worked in the service of their master, not for the writer. With the scope of their performance constrained by the economic factors of the project, they focused on creating a usable reference text with content features that could be worked with sustainably across multiple editions. The quality of the materials that made up the book were also shaped by the printer’s agreement with the bookseller and, further, by the bookseller’s success with subscribers. The visual ethos of the text would have been closely accounted for, since leather bindings or gilt titling would have contributed to a distinguished final product that was suitable for the private libraries of wealthy or professional subscribers, while the loose serialized pages were more casually presented to the lower classes. All these factors in turn shaped the prestige and reputation of both the printer and the bookseller.
The printer’s “arts,” Maruca wrote, render the text legible13—and, in the case of reference texts, usable. Practitioners of the trade considered themselves to be significant creative contributors to texts, sometimes to the extent that they superseded the messier contributions of the writer through their corrections. Printers shaped the central ethos of the final product in a variety of ways that are now invisible to the twenty-first-century reader. Decisions about typography and style would have been made by the publishing house and printer rather than the writer, thus demonstrating an extension of curatorial agency that includes development of the material form of the text in the curation process.
In the case of the Cyclopædia, the most evident contributions are found within technical decisions about the typeset text. While Chambers was intensely interested in the state of the English language and wrote about philological issues in the 1728 preface, there is no evidence that he directed typographical changes in his own text. These hand-set typographical alterations, which were contributed by the compositor at the printing house, were labor intensive and contributed Page 102 →to the project’s most basic ethos. Two primary typographic changes were made between the 1728 and 1738 editions: a shift away from Germanic capitalization and the introduction of fully capitalized cross-reference terms. The capitalization shift is evident from the first sentence in the table: “Natural History” becomes “natural history,” “Sometimes” becomes “sometimes,” and so forth. These changes alter the fundamental feel of the text, moving it from the then-typical style of Germanic capitalization to a more modern typographical style that was quite progressive for its time. In a genre that depends on cultivating an ethos of not just reliability but also timeliness, this change contributed to the text’s kairotic presentation.14 With these alterations the typesetter and printer make significant contributions to this text in ways that are quite apart from the central composition process. This labor functions as an important but largely invisible aspect of curation, occurring at a midpoint between the time the text leaves the original composers’ hands and before it reaches the binders, the bookseller, and the reader. As with practically any other stage, nonhumans—in this case the lead type, the type drawer, the composing stick, the galley, the forme, the paper, the ink, the press—play an important role alongside the human craftsman.
In the example shown in table 2, few stylistic alterations have been imposed. In the first paragraph “Mine” changes to “subterraneous place or mine.” The third paragraph has been split. What is more important, three additional indicators of cross-indexing have been added in all capitals: METAL, OAR, and MARCASITE. These changes enhance the interface that is the page by developing a distinct typographical convention that signals to the reader that there is more to be found, a path to follow, in much the same way that readers of digital texts understand that a blue or underlined word is a link that leads to another page. The move to fully capitalize cross-indexed terms provided a similar visual differentiation, setting these words apart from other proper nouns that were capitalized through Germanic capitalization conventions. This important innovation marks the visual emergence of a network within the text, explicitly inviting readers to invent individual readings.
The widespread adoption of typographically distinct cross-references as a navigational convention in the reference genre means that readers of encyclopedic texts nearly always face a different range of possibilities than do readers of prose or poetry. Many genres are designed for linear reading that begins at the start of a narrative, goes on to the end, and then stops. Any reader may determine an individual path through any sort of text: skipping the introduction, flipping ahead to find out the ending, or reading strategically for strictly informative purposes. Still our primary conceptualization of the reader is of one who reads novels, essays, articles, or poems from the beginning on till the end. However, the encyclopedic reader encounters a text specifically built to suit individual performances. One Page 103 →rarely reads an encyclopedia front to back; rather the usual point of entry is a topic of interest, from which the reader may continue to compose an individual, temporal experience of the text. Links in Wikipedia perform much the same function and are a fundamental feature. Similarly someone searching Wikipedia primarily for contemporary bands and movies experiences it primarily as a compendium of popular culture, following series of links that organize informational paths in what Joshua Reeves has called “flowing” media experience.15
Table 2
Parallel comparison of the iterations of the Minerals article
1728 | 1738 |
MINERAL, in Natural History, is sometimes used in the general for Fossil, and applied to any Body, simple, or compound, dug out of a Mine; from which it takes the Denomination. See MINE. | MINERAL, in natural history, is sometimes used in the general for fossil; and applied to any body, simple or compound, dug out of a subterraneous place or mine; from which it takes the denomination. See MINE. |
In this sense, the Metals, Sulphurs, fossil Salts, Semi-metals, &c. are Minerals. See FOSSIL. | In this sense, metals, sulphurs, fossil salts, semi-metals &c. are minerals. See FOSSIL. |
On this Principle, they divide Minerals into two Classes; the one fusible, and malleable; i.e. which melt with Fire, and stretch on the Anvil; which are what we properly call Metals. The others want those two Properties; and are what in the strict sense we call Minerals. | On this principle, they divide minerals into two classes the one fusible, and malleable; i.e. which melt with fire and stretch on the anvil; which are what we properly call metals. See METAL—The others want those two properties; and are what in the strict sense we call minerals. See OAR and MARCASITE. |
This performance is much like that of the traveler choosing a route, as Eco, Magli, and Otis have suggested: “Just as a railway system is structured in a rigid manner while authorizing different routes, an encyclopedic knowledge can be structured and nonetheless oblige the one who consults it to elaborate conjectures about the best route.”16 These individualized interests also influence readers’ experiences of becoming more active curators of encyclopedic texts should the opportunity and inclination present themselves. The encyclopedic form shapes Page 104 →their performances: by providing a series of textual links that can be arranged to form a path; by dictating the nature and tone of acceptable contributions; and by providing an authoritative, informative ethos that some forms of vandalism (such as misinformation) push specifically against.
Dreams of Mechanization
Cross-references quickly became the standard in printed reference texts, but the obvious potential offered by a networked system of links between related topics and the problems of print led to dreams of developing a mechanized encyclopedia. The material problems of curating ever-expanding and changing information texts suggested that developing a mechanical system that could facilitate users easily dropping in new information would clearly be worthwhile, and the incredible potential of developing personal information ecologies was certainly not lost on either scholars or the military-industrial complex. Forward-thinking individuals kept searching for ways to develop a mechanized reference system, rhetorically positioning the idea as a liberatory expansion of educational access. Another common selling point was the ways a mechanized encyclopedia would help humans move beyond the temporal limits of printed texts. It was not until the release of Encarta in 1993 that a fully technologically networked encyclopedia would be successfully launched, but a number of individuals made proposals for such texts in the intervening 265 years that elapsed since the first edition of the Cyclopædia. The first modern proposals for such an encyclopedia came from a man well known for futuristic technological dreams: H. G. Wells.
In 1936 and 1937 Wells delivered a series of lectures on what he called “The New Encyclopedism,” a centralized, curated collection of the world’s knowledge that he claimed would enable mankind to enter a new age of peace. Later collected in a volume entitled The World Brain, these lectures describe in detail what such an encyclopedia might look like:
A World Encyclopædia no longer presents itself to the modern imagination as a row of volumes printed and published once for all, but as a sort of mental clearing house for the mind, a depot where knowledge and ideas are received, sorted, summarized, digested, clarified, and compared. It would be in continual correspondence with every competent discussion, every survey, every statistical bureau in the world. . . . This Encyclopædia organization need not be concentrated now in one place; it might have the form of a network. Quite possibly it might to a large extent be duplicated. It is its files and its conference rooms which would be the Page 105 →core of its being, the essential Encyclopædia. It would constitute the material beginning of a real World Brain. . . . If such a thing is to really live it should grow rather than be made. It should never be something cut and dried. It should be the survivor of a series of trials and fresh beginnings—and it should always be amenable to further amendment.17
He pointed to microfilm, which the American Library Association had just endorsed in 1936, as the mostly likely physical material for such a project.18 While he cast the project as proletarian in both scope and access, he nevertheless recommended retention of a “directorate and staff of its own type, specialized editors and summarists.”19 Still the World Brain would differ from previous encyclopedias written “by gentlemen for gentlemen” and be distributed as widely as possible all over the world. The volumes would be in constant cycles of revision and replacement, continuously redeveloped and redistributed as textbooks, dictionaries, and shorter reference works for “individual and casual use.”20
It was not long before another inventor offered a somewhat more compact model for a networked encyclopedia. By the time Vannevar Bush published his 1945 proposal for the Memex as a solution to information overload, he had been considering the problems of managing substantial amounts of military information for some time. He began to formulate plans for a semiautomated system a few years before World War II, when he served as head of the National Defense Research Council (NDRC) and later the Office of Scientific Research and Development (OSRD).21
His Memex would be an infinitely expandable, interlinked series of microfilm cards projected onto slanted, translucent screens. The economics of production had finally advanced enough, Bush noted, that long-hoped-for machines such as calculators and computers could finally be produced on some marketable scale in the near future. A cheap, reliable machine should be able to provide a solution to the problem of directly linking a primary text with the reader’s ancillary notes. A simple algorithmic indexing system would connect the two and be controlled via a keyboard, buttons, and levers.
The resulting system would be an “enlarged intimate supplement to . . . memory.” It would also readily lend itself to hyperlinked cross-indexing: “Whole new forms of encyclopedias will appear,” predicted Bush, “ready made with a mesh of associative trails running through them, ready to be dropped into the Memex and there amplified.”22 Users would be able to purchase some content—books, periodicals, and newspapers—from companies and drop them into the system. Business correspondence could be inserted as well, and “direct entries” would also be possible. The Memex was widely discussed but never went into production. Still many Page 106 →historians have suggested that it clearly influenced early developers of computers and the Internet, particularly J. C. R. Licklider and the ARPANET research team, who continued military research that led to the first digital transmissions.23
One of those early adopters, Ted Nelson, launched what would become the most extensive (but ultimately unsuccessful) attempt at a digital encyclopedia prior to Wikipedia. Nelson has been actively developing Xanadu since 1960, when he envisioned it as “a simple-to-understand electronic publishing system for the world, and a new technical way of simplifying and improving the world’s storage data.”24 Nelson aspired to develop nothing less than a hypertextual system that would store, index, and facilitate publication of the entire world’s literature, and his concept is alternately cited as an early draft for both the World Wide Web and Wikipedia. He was insistent that the project would also facilitate a system of automatic micropayments to authors/creators in exchange for reuse or quotation of their work. While his stance was vehemently anticopyright—he felt that anyone should be able to publish information—it was just as vehemently pro–intellectual property in this requirement for payment. Tim Berners-Lee, in his account of meeting Nelson, suggests that this insistence on a pricing structure that was consistent worldwide was one of the reasons that Xanadu never succeeded.25 Nelson put together a project team in 1979, made the preliminary design public in 1981, and pressed ahead with the database structuring. The project was acquired by Autodesk, which invested an estimated US$5 million before discontinuing its sponsorship in 1992.26 It has since largely languished, although Nelson continues as the project evangelist and launched a 2014 version called OpenXanadu.27
Each of these unsuccessful projects represents a situated desire for a networked encyclopedia that is articulated within the potential and needs of its time. Whether they sought to manage government documents or literature, each of the proposals addressed the exigencies of information overload and explored potential collaborations with nonhuman actors that could facilitate a means of distributing and managing information for as many users as possible. In every case interfaces have been a crucial factor in the success of networked encyclopedic projects.
Wikipedian Interfaces, Development, and Curation
Building a digital encyclopedia that is sustained on multiple levels by digital networks demands an even more extensive collective that includes hardware, coders, and robots. These actors mark significant, obvious differences in the Cyclopædia and Wikipedia while also connecting the latter to the precedents set in the eighteenth century. Rather than relying on publishing houses, printers, bookbinders, and typesetters for material production and ethos, this digital encyclopedia Page 107 →depends on strict server maintenance, standards-compliant code, well-developed visual templates, and other central aspects of site management and architecture. It also takes advantage of the potential for automated management by deploying a vast array of bots to handle textual production and management, including surveillance and correction of vandalism. All these actors are essential and interdependent, just as important as the human editors who receive most of the credit for creating Wikipedia.
Wikis have been in active use since the Portland Pattern Repository’s companion website, WikiWikiWeb, was established by Ward Cunningham in 1995.28 Cunningham named his platform after the Hawaiian term for “quick,” “wiki wiki,” and built the original wiki interface around design principles that facilitate what has become known as the wiki philosophy: open, organic, mundane (simple), universal, overt, tolerant, observable, and convergent.29 In keeping with these principles, the markup of wiki text does not require the user to know standard HTML or other markup languages. Instead wikis have historically relied on a simplified system of wiki code, which the backend system converts to HTML and uploads. This low interface barrier is intended to invite updates by a new and casual users who do not necessarily know or wish to learn standard coding languages.
Wiki platforms are fundamentally networked on several levels: physical, intertextual, and social. Since they rely on a database running on a central server that supports access by multiple users, a wiki can exist only in a digital environment.30 Users move within a loose, organically changing network of intertextual and interpersonal connections, particularly in larger wikis like Wikipedia.31 The looseness built into these designs enables radical collaboration, “because of [their] total freedom, ease of access and use, simple and uniform navigational conventions, and apparent lack of formal structure.”32 Digital rhetoricians have noted that individual and community ethos in digital environments are also influenced by technology itself, and that this ethos has the potential to shape human-computer interaction33 and form a barrier for users.34 This contention applies equally to wikis and the confluence of technologies that operate within them. The low interface barrier and basic wiki design principles influence and are influenced by the community ethos and recursively shape the labor at hand.
As with the printing houses that produced the Cyclopædia, the coders of Wikipedia represent multiple communities who make contributions to the form of the text as well as particular aspects of its content. Since the development of digital tools does not rely on the time-consuming and costly process of material manufacturing, which demands additional craft knowledge and skill sets, it has also been possible for discrete groups within the coding community to devote their efforts to developing tools that better facilitate workflow, limit server downtime, and even write the encyclopedia itself. The Wikipedian coding community has Page 108 →from the beginning been both broad and collaborative. It is composed of limited staff working with many volunteers in yet another example of commons-based peer production.35
In the same way that the printers of the Cyclopædia were responsible for a number of genre conventions and for the ethos of the typesetting, Wikipedia’s coders are also responsible for many of the features that have come to be associated with the genre of the Wikipedia page template: the easily navigable static sidebar, content boxes with section links, and the information summary box that appears at the top right of each article.36 The development community rolled out the platform’s categorization system in 2004, bringing site-wide, hyperlinked taxonomy within reach.37 The use of automatically implemented templates standardizes the visual ethos of the encyclopedia, enhancing perceptions of reliability and authority.
The front page of a Wikipedia article, with page tabs at the top.
Image courtesy Wikipedia under CC BY-SA license.
Wikipedia’s interface presents a powerful example of the social, articulated nature of interfaces. MediaWiki, the open-source wiki platform that powers the encyclopedia, is a multilayered interface with tabs for reading, editing, and discussing the topic of each page, as well as for automatically archiving all activity on the article since its inception. Over time the interface itself has been rearticulated to better facilitate the labor it enables, illustrating Johndan Johnson-Eilola’s contention that “the interface is not simply a tool, but a structure for work—a set of forces articulating a specific form of work, a set of forces articulating work.”38 In this case the work at hand is textual curation. The Wikipedian development community has customized the MediaWiki interface to better facilitate distributed curatorial work.
The design and deployment of the wiki backend and interface necessarily shape both contributions and community. Any registered user—human or Page 109 →nonhuman—can perform direct action on the text, adding new information, removing contributions, inserting images, video, or sound, linking to other texts, or contributing a dizzying array of other potential changes. Users can communicate directly on the Talk page for each article or leave notes explaining changes. The richness of this environment demands strong curatorial skills that are specific to the potential of digital environments. Curatorial skills reach beyond basic editing to address the scope and infrastructure of large texts in specific ways: project management, strategic linking, and creating metadata. Many of these tasks entail what the community calls wikification, or “formatting according to Wikipedia style.”39 Wikifying can include converting plain text to hypertext, adding interwiki links, or restructuring the articles into the usual article format of an overview followed by subsections, as well as the specific tasks mentioned above. Done correctly these curatorial elements create a text that is not just well written but well structured, composed of multiple potential paths that are both findable and usable.
The Edit page for for an article, showing simplified wiki code. Image courtesy Wikipedia under CC BY-SA license.
The interface features that are visible when readers first click through to an article have changed little. The Article, Talk, Read, Edit, and View History tabs are typical across wiki platforms. The Edit Source tab was added after the development of the Visual Editor in 2012. Article and Talk are the two primary layers of any article, allowing the reader to move between two interactive experiences: consumption of multimodal text and social consideration of the text. Users click on any of the four tabs on the right side of the tab bar to read, edit, or view the history of either the article or the related editorial discussion.
For many casual users of the encyclopedia, the Article view is the only view they ever notice. Many never click on any other tab. Barry Newstead, Wikimedia’s Page 110 →Chief Global Development Officer, has estimated that 90 percent of users outside Wikipedia’s “core community” never realize that the pages are editable.40 Attracting new editors is a constant challenge owing to the fact that many simply never realize the potential for individual participation exists. Those who do choose to contribute are asked to work with a bare-bones interface that presents them with simplified wiki code.
While the interface presents familiar, simple icons that most editors have encountered in other interfaces, it still serves as a barrier to developing a diverse collective. As Adam Banks has argued at length, interfaces are inevitably rhetorical, evidencing cultural norms and literacies in ways that invite participation from those who identify with these norms and whose prior literacies lead them to intuitively navigate these interfaces.41 Wikipedia’s long-standing core community demographic of wealthy white males with considerable backgrounds in gaming, coding, and digital communities had proved unsustainable, making diversification necessary for not only ethical but pragmatic reasons.
Wikipedia’s Visual Editor interface, which offered a more welcoming WYSIWYG option for casual editors. Image courtesy Wikipedia under CC BY-SA license.
A primary difficulty presented by the editing interface is that it requires editors to work directly with code, however simplified. Wiki code is considerably simpler than other coding languages, but it still requires contributors to cross the barrier of reading and using markup within their written text. The learning curve is low, but learning—that links require double brackets, to give just one example—is still a sufficient barrier for many busy or distracted users. It requires time to figure it out, and many casual contributors with very limited leisure time simply do not have that time. This requirement limits the contributing community in ways Page 111 →that only strengthen the exclusionary aspects of the community, since women and other underrepresented demographics are the potential contributors most likely to have limited time and, possibly, limited digital literacy or access. The substantial heaviness of the platform also limits contributions by editors who use mobile phones as their primary computing device, as is frequently the case in African nations. The interface is intended for editors working from desktop setups or laptops with more screen real estate and, optimally, a mouse.
The Wikimedia Foundation recognized the rhetorical effects of the editing interface a number of years ago and prioritized replacing it with a visual editor that would offer a default WYSIWYG option for editors. “We must adapt our editing technology to the changing web, improve the social experience for new users and grow participation in the Global South,” announced the 2009–10 foundation report.42 The interface update was part of a multifaceted strategy for attracting fresh editors to the project.
With one-click formatting options, new users no longer encountered the initial barrier of integrating wiki syntax with their text. Developers thought that users with basic digital writing experience involving blogs would find the interface particularly comfortable, as it mimicked popular blogging platforms such as WordPress. The older, syntax-dependent editing system was retained under the Edit Source tab as a way to appeal to more experienced users who considered themselves to be experienced Wikipedians or core users. A number of other changes also furthered the goal of making the psychological and practical transitions to editing more appealing to readers: adding a multimedia uploader that mimicked common conventions of other familiar uploader utilities, ongoing live usability testing that resulted in minor tweaks to interface aspects throughout the entire workflow, and rhetorical revisions of single-sourced review messages that experienced editors working with the Page Patrol initiative send to those working on newer pages.
The Visual Editor implementation was met with dismay by the core community, who found that it deleted or mangled their edits during the posting process. Their demands to remove the interface were initially rejected by Wales, who was adamant that the Visual Editor was central to a new, more inclusive ethos that Wikipedia would have to embrace in order to continue to grow. A few weeks after the launch, German Wikipedia editors rebelled, and communities from other nations followed. By September 2013 the Visual Editor had been moved to opt-in status—and in order to opt in, users had to know it was available. Newcomers who were unfamiliar with the options were effectively limited to the older, code-based interface that had proved to be prohibitive to participation, and the hoped-for change in the ethos of the editing interface was tabled.43 This design element is one of many reasons that the editing community has not expanded since its peak year, 2007.
Page 112 →However, the communities that remain continue to be productive. This productivity is localized in smaller communities formed around curatorial tasks, as, for example, the Military History group has.44 This group is typical of the thousands of interest-based groups that have sprung up within Wikipedia, most of which are dedicated to mundane, necessary topics that individuals find some passion for. I focus on this sort of group because it exhibits aspects of the incremental activity that drives iterative expansion in distributed authorship rather than the swarming behavior that is more typical of topical areas that are currently in the news or subject to popular controversy. This group’s page announces that its members are “dedicated to improving Wikipedia’s coverage of topics related to military history” and openly invites participation in improving any of the more than one hundred thousand articles within the scope of the project. This sort of loosely directed curatorial effort is the foundation of growth within the system.
For subject-area communities, the job of producing relevant texts serves as a central impetus. This group’s community page announces that its members are “dedicated to improving Wikipedia’s coverage of topics related to military history” and openly invites participation in reaching any of its current four major milestones: having 750 of its articles approved for featured article status (77.7 percent complete), achieving five hundred items of featured content other than articles (80.2 percent complete), improving two thousand of its articles to “good” status (90.8 percent complete) and having at least 10 percent of its articles rated B-Class or better (79 percent complete).45 Each of the classifications is linked in order to better direct members; for instance by clicking on “featured articles” the reader is taken to an alphabetized list of articles on military history that need improvement.
The community has organized departments devoted to assessment, motivational contests, coordination, and incubation of new groups and initiatives within the project. The hyperlinked task list directs community members to lists of articles needing specific types of curatorial work: references/citations, accuracy, structure, integration of photographs, tagging, and assessment. An extensive list of “Articles to Be Created” provides starting points for editors who prefer to start from scratch. As is typical of most groups, the group maintains an Open Tasks page listing articles that need general or specific improvements. Many of these improvements focus on wikification: 35,972 articles need work on reference or citation, for instance. Others need work on “coverage or accuracy,” “structure,” and “photographs.” At this writing, the assessment list was backlogged, as was the list of articles that needed project tags fixed. The page also lists content currently under various sorts of review: peer reviews, “good” article candidates, featured article candidates, and the like. These task lists highlight the community’s focus on curatorial tasks as essential to community standards of good composition and Page 113 →point community editors toward these tasks as priority items. This sort of broad yet detailed project management functions as the topmost level of curatorial work: defining the parameters of a project and identifying the work to be done.
Interested editors self-select to contribute to these pages, making additions that they feel qualified to offer. A significant proportion of the work involves wikification, particularly in the form of link management. In my limited sample, link management was the most common task performed. As is the nature of wikification, this work was not limited to concerns about the written text of the articles, but rather to linking those articles within the variable network of the encyclopedia.
Insertion, deletion, and maintenance of interwiki links made up the highest proportion of tasks in the sample. Links to other pages within Wikipedia serve as digital forms of the cross-indexing, encouraging readers to linger in the text and create their own contingent experience by writing paths within the network. For example at this writing the Fortification article has been edited forty-eight times to add links to other article pages, and only one link was removed. It is a typically densely interwoven hypertext with multiple links in nearly every paragraph, offering the reader a quick leap to topics as diverse as Sumer, Oppida, the Yongle Emperor, and Concrete. Additionally it offers an extensive list of other relevant pages in the “See also” section, as is typical in many established pages. Aside from offering pathways to readers, this dense interlinking serves an internal rhetorical purpose: hailing the editors of related pages to encourage collaboration and return linking, both of which attract increased attention to the article. Algorithms facilitate this identification: the “what links here” function on each page automatically compiles incoming links for review.
The addition of category links further strengthens ties within the network on both textual and social levels. These links appear at the bottom of each article, providing users with a means of navigating by topical area instead of searching for individual pages. These taxonomic links are a core element of the system’s information architecture, so important that a substantial WikiProject focuses solely on their maintenance. On the Fortification page, heavy editors of the topic did not insert the category links, which were instead added by members of the Categorization WikiProject.46 In the process of working through that WikiProject’s task list, these editors often by necessity deal with article topics that they are unfamiliar with, and occasionally this leads to conflict. For example the Falconry article is currently categorized as “Falconry” and “Intangible Cultural Heritage of Humanity,” but at one time it was also categorized under “Blood Sports.” The core page editors, who included several experienced falconers, took issue with this categorization and removed it, leaving the note “Blood sport? wtf?” There was no further conflict, and the edit has stood for some years.
Page 114 →Implicit moral or ethical judgments are not uncommon in taxonomic systems, as a number of scholars have noted. The same was also true of this exchange in the Gardens article, which features a list of gardens mentioned in famous literary works, including religious texts. The title of this list underwent some revision early on, beginning its life as “Gardens in fiction.” On April 2, 2004, the editor Fennec changed it to “Gardens in fiction, religion and myth” because of concerns about alienating audience members. This editor noted, “Watch what you call ‘fiction’, people are liable to get upset:).” While the edit imposes change on a previous writer’s work and the tone of the note is rather terse, the note also explicitly works to avoid conflict through the use of a concluding smiley. The edit also moves the title toward a more typically encyclopedic description. There was no overt conflict in the history or discussions over this edit, but at some point in the life of the article the list title changed again. It currently stands as “In religion, art, and literature.”
Another prominent category of interwiki links is links to the same article in Wikipedia translation projects. This link form builds a digital network between the various language versions of Wikipedia, increasing the chances of non-English speakers locating the article in their own language after being directed to the English version by, say, a U.S. Google search. These links offer an invitation to underrepresented editors and readers, rhetorically positioning Wikipedia as an international community that does not wall off the prominent, largest version—the English Wikipedia—from the other 288 official Wikipedias.47 Building paths that encourage international contributions also encourages translation work as well as potential diversification of viewpoints.
Finally external links in the Reference section of each article provide pathways to cited resources. These links enhance the logos of pages, providing verifiability for claims made within articles and discouraging contributions of original research. These links require several sorts of curatorial labor: verifiability assessments ensure that citations are in place and linked; link bot patrols flag broken links for replacement; and both humans and bots monitor links for spam or link farming. This work, along with the link maintenance previously described, tightens the connections of a monumental network to a plethora of texts and entities both within the project itself and on outside web pages. Taken as a whole, these connections coalesce into an articulation of Wikipedia’s place as a public resource that exists firmly within the open culture of the web rather than in any specialized educational environment.
Encouraging and facilitating rigorous curatorial practices is so important to the maintenance and quality of Wikipedia that the Foundation supported the development of the Page Curation tool suite in 2012. This interface provides experienced editors with a feed of new pages and a curation toolbar that facilitates Page 115 →more efficient review of articles. For the first decade of the project, page patrollers relied on algorithmically generated feeds of new pages and used third-party applications for more efficient curation of content. The new interface provided a centralized dashboard for flagging curatorial issues on pages or offering Wiki-Love as a reward for carefully curated work. Sending WikiLove in the form of a personal message or by awarding community recognition in the form of Barnstars, Kittens, or other award badges that are posted on the recipient’s User page is a way to offer positive reinforcement to newer editors for their contributions. Wikipedians hope that these sorts of rhetorically situated welcoming moves will encourage retention of new editors and, consequently, extension of the community.
Wikipedia’s Page Curation toolbar with single-sourced message tags for a broad variety of curatorial situations. Image courtesy Wikipedia under CC BY-SA license.
The Javascript panel “enables editors to retrieve basic information on the page’s creation and status, mark a page as reviewed, tag it, mark it for deletion, send WikiLove to page creators—or jump to the next page on the list.”48 Curatorial management of links is sufficiently central and important that tags concerning link issues constitute four of the seven most common practices: correct URL usage, additional footnotes with attendant links, additional references with related links, and category linking. By clicking the box to the left of each problem description, experienced editors can efficiently leave tags to instruct page contributors regarding improvements that will bring the page up to project guidelines and thus Page 116 →invite continued contribution, encourage additional readership, or, depending on the editor’s philosophical leanings and the state of the page, avoid deletion. The ease of this interface encourages curatorial work within the system by lowering the learning curve, lowering frustration with third-party workarounds, and improving the amount of curatorial work that can be accomplished in one sitting.
The Curation Toolbar also formalized curation as a core term in Wikipedian nomenclature. Until at least 2010 curation was not a common descriptor of the labor that occurred within Wikipedia; most tasks related to the construction or maintenance of articles fell under the umbrella heading of “editing.” Many of the tasks that the Curation Toolbar renders explicit are those that have been prevalent in the system for many years: basic editing, metadata development, sourcing, interwiki linking, categorization, monitoring compliance with the neutrality policy, and monitoring vandalism. By making the work of maintaining not only the text but also the structure explicit, it acknowledges the centrality of this compositional work. Integrating the terminology also rewards editors who have historically invested extensive time and labor in curatorial work, especially during years when the expansion of the text received considerable media attention while structural work was rarely remarked on at all.
Curation badge available to editors who review newly created articles with the Page Curation tool. Image courtesy Wikipedia under CC BY-SA license.
At the same time, curator badges became available for inclusion in Userboxes, offering a way to signal affiliation in the same way that other badges signal WikiPhilosophies. These badges explicitly articulate these tasks as central to the labor of maintaining a rigorously constructed encyclopedia. By publicly identifying themselves as Curators, self-selected Wikipedians lay claim to a broader skill set that includes editing but also extends beyond it to the tasks of wikification and quality control. Displaying one of these badges also advertises the potential for doing this work, allowing a point of entry for the new or the curious who may inquire about what it means to be a curator or who simply look it up themselves.
These curatorial skills and tasks demonstrate the relationality of human and nonhuman friction within these projects. The text, the tools, and the human all come together in order to compose a useable reference work that simply would not be possible without these distributed performances. The contributions of Page 117 →nonhumans are apparent at each level of this process, as they are necessary for the construction and sustainability of any interface, whether it be print or digital. Nonhumans assist in building and maintaining. They shape paths and barriers for human movement through a text and into a community through small and often unnoticeable textual elements such as links and cross-references. As vital as those contributions are, these actors still have more to offer: the autonomous construction, maintenance, and surveillance of texts.