In the usually-rarefied world of textual scholarship, where decades used to pass without any cause for undue excitement, few words have aroused as much enthusiasm and hope as "social." As my title implies, it is one of those happily elastic terms which you can apply to objects, to the processes of making the objects, and to the abstract concepts underlying both objects and processes. So we might speak of "social editions" as realized and actual physical entities; of "social editing" as some kind of collaborative process which creates editions; "social texts" as the ghost in the machine inspiring our endeavours.

The problem with elastic terms is exactly that they are elastic: the same word might have quite different, and even opposing meanings, in different contexts. It would be a beautiful thing if there was a perfect harmony between the three uses of "social" in my title. Thus, one might work in a collaborative way ("social editing"), and by definition create a "social edition" which embodies, again by definition, a "social text." But a moment's thought exposes this as wishful thinking. One could work collaboratively (there are many kinds of collaboration), and create something which is a traditional editor-created edition with no claim to be "social," let alone contain anything which could be described as a "social text." It can be, too, that the word "social" is so redolent with appealing associations that one is tempted to use it just for those associations. A "social edition" puts us in mind of "social networks," of Web 2.0, of the mass access to knowledge and power offered by the Internet, and hence of the better age for all which the digital revolution might bring in its train. Faced with a choice between making a traditional, print-based, editor-driven scholarly edition, or making a ground-breaking, democratically-empowered, digital edition for and of the people, who would not choose the latter? Indeed, even if what you are doing is the former, the advantages of calling it "social" might be so compelling that you give it the name "social," although it is really no such thing.

In this chapter, I will review some of the ways the term "social" has been applied in recent years in the overlapping fields of textual scholarship and the digital humanities. As a starting point, one should distinguish between the uses of "social" when used for a concept ("text"), a process ("editing"), or an object ("edition"), and one should be aware of the tensions between these uses. As textual scholars, we should be particularly sensitive to how words can be misused and the dangers of this misuse: labelling something "social" when it is not "social" is likely to lead to cynicism and disillusion. On the other hand, the promise of the "social" is real. If we can identify exactly what social editions, editing, and texts are, we might be able to make our work as textual scholars far more accessible to many people, open what we do to many more people, and create editions far richer in their uses. These are prizes worth pursuing.

Social texts

The third term in my title, "social texts," has the longest history in textual scholarship: all of thirty years or so. The origins of the concept of the "social text" lie in the writings of Jerome McGann and Donald McKenzie from the mid-1970s on. McGann and McKenzie, working separately and initially unaware of each other, both came to the concept of the "social text" from different directions: McGann (1983), from his experience in editing Byron and in reaction to the Anglo-American "copy-text" editing theory then dominant; McKenzie (1996), from his experience of researching the history of Cambridge University Press and his work on Congreve. Both concluded that the texts we read are shaped by many factors besides their original author and are transformed on their way to the reader, by publishers, critics, teachers, and generations of previous readers. Both concluded that the material forms the texts take are critical to understanding of the social nature of texts (hence McGann's "bibliographic codes"), and both argued that any approach to texts—and hence, textual editing—must be grounded in this double awareness of the social and material nature of texts. The two are vitally linked: the physical expression of any text is the result of the social factors underlying and determining this instance of the text and is itself a key to understanding of those factors. For McGann (1983), as an editor, this means that an edition must reflect, or express, or invoke, this "social text." McKenzie (1996), whose early work was more in literary and publishing history than in textual editing, does not speak of the "social text" as the goal of an edition. As his writings testify, for him the social and material aspects of a text are to be explored through discursive analysis: essays, not editions.

As several critics have remarked, and the succession of verbs "reflect, express, invoke" suggest, it is not at all clear what an edition of a "social text" might look like, or even whether such an edition is possible (Howard-Hill 1991). At a minimum, one might expect a social text edition to include a record of all the different material versions of a text. It happens that McGann's interest in "social text editions" coincided with the advent of the digital revolution. From 1990 on, just as McGann and others were theorizing "social text editions," it became possible to gather all the material versions of a text, through digital images and transcriptions, and present them together. McGann himself was a pioneer in this enterprise, both in his theoretical writings and in his practical work with the Rosetti digital archive. In a review article on Jim Mays' edition of Coleridge, in which he contrasts May's print edition with his own work on Rosetti, he declares that "[a] central purpose of The Rossetti Archive project was to prove the correctness of a social-text approach to editing" (McGann 2006, 32; see also McGann and Buzzetti (2006)). Citing McKenzie, he asserts, "McKenzie's central idea, that bibliographical objects are social objects, begs to be realized in digital terms and tools. The Rossetti Archive proves that it can be done" (McGann 2006, par. 38). So, what exactly does the Rosetti Archive do? And is it a "social text" edition?

Over the nearly two decades now of the Rosetti Archive's existence, it has changed considerably. As it first appeared from 1997 on (following the "research report" of 1993), the Archive consisted almost exclusively of images and texts of Rossetti materials published during Rossetti's lifetime. The 1993 "research report," also described as a "research demo," is available via the Wayback machine. It is now not possible to recover the first versions of the archive (which McGann (2002) and other writings describe). The description given here derives from the traces apparent in the Wayback Machine. The 1993 report declares its aim: "In an ideal imagining the Archive will hold a digital image of every textual and pictorial document relevant to the study of Rossetti." Accordingly, the first versions contain materials relating to Rossetti's The blessed damozel—as images of manuscript, print editions and painting—, joined from 2003 on by resources giving text and images of other Rossetti works (see "Visit Rossetti's studio," referenced from McGann's Rossetti Archive "Research report," in the 22 December 1996 Wayback Machine snapshot). At first, the Archive contained a substantial bibliographic record of Rossetti materials, with a very few of them appearing online. Over time, more materials have been added online, accompanied (in many cases) by substantial commentary. Interestingly, the Archive itself makes no claim to present a "social text edition," or indeed to present an "edition" at all: these claims are made by McGann elsewhere (e.g. 2006). This is also true of the Blake Archive, strongly influenced by McGann's ideas, created at the same time (and by many of the same people) as the Rossetti Archive, and containing a very similar mix of texts and images of primary materials with scholarly commentary. Nowhere that I can find do the Blake Archive editors refer to this as a "social text edition."

Several observations are pertinent. First, while McGann (2006) proclaims that the archive explores the possibility of a "social text edition," his description of it stresses its computer possibilities (the ability to gather multiple texts and images and view them in many ways), without any further discussion of exactly what a "social text edition" might be. One feels that a "social text edition" should be more than a collection of resources relating to different forms of the text presented in digital form with an array of computer tools: but what more, McGann does not say. Broadly, an "edition" incorporates a series of editorial decisions concerning documents; an "archive" simply holds the documents. The affordances of the digital era act to blur this distinction, as online archives do not just "hold" the documents (as a library might): they present images, transcriptions and detailed metadata relating to the documents, as editions might do (for the distinction between "archive" and "edition" see Price (2009)). One might infer from the archive itself that the "something more" might be the scholarly commentaries which the Rosetti Archive provides (similar to those provided by the Blake Archive editors). These are substantial, informative, and scholarship of the highest order. Yet, the effect is that the edition looks very familiar indeed: like a traditional work of textual scholarship, carefully gathering all the materials and presenting them with discursive commentary. One is reminded of Tanselle's forceful observations on McGann's and McKenzie's writings about the "social text," that good editors have always attended to the production and cultural factors shaping texts (Tanselle 1991). There is, it seems, nothing new here. Second, the materials included in the Rossetti Archive stop at the end of Rossetti's life, as does the Blake Archive. Why? Removing this limitation, and incorporating all the texts made by everyone right up to the present, would indeed make this a very different object. Surely a "social text edition" would be interested in all the forms of the text, right up to the present. Indeed, one might argue that in this respect a traditional variorum edition is more of a "social text edition" than is the Rossetti Archive. Third, this is very much McGann's edition: an edition of something he describes as a "social text," but his edition. Even if one concedes the Rossetti Archive presents a "social text," it is not a "social edition"—or if it is, so is every edition ever made.

I am not aware that any editors since McGann have declared that they have made or are making a "social text edition." One might take this (and indeed the self-description of the Rossetti Archive as an "Archive," not as a "Social Text Edition") as indicating that McGann's notion of social text editions is still-born. However, one may see the influence of McGann's ideas in several areas. Scholars in the burgeoning area of book history have absorbed McGann's and McKenzie's perceptions in their study, but without making scholarly editions. Within scholarly editing, one may ascribe the prominence given to the documents and to their analysis, at least in part, to McGann's work (for example see Gabler (2002), Gabler (2007), or Gabler (2010)). One may also trace it in Paul Eggert's (2013) writings and editorial work, focussing on linking the histories of works with their changing receptions and expressions over time: thus his Biography of a book: Henry Lawson's while the billy boils. Most significantly for this study, the essay "Underpinnings of the social edition" (Siemens et al. 2010) references McGann's and McKenzie's "social theory of text" as a precursor to the authors' own work on "social editions" and refers frequently to "social theories of text" in the model of editions developed by the paper (Siemens et al. 2010). It seems that Siemens and his collaborators see a "social edition," made by many people, as by definition a "social text."

Social editing

I observe above that McGann's Rossetti Archive is in one respect a very traditional edition: the product of the work and vision of a single editor. This editor had many collaborators, but there is no doubt that the impetus and direction is McGann's. The Archive is described (on the "Home" and "About" pages of its post-2008 incarnation) as "The complete writings and pictures of Dante Gabriel Rossetti, edited by Jerome J. McGann," and it appears that there is nothing in the edition which is not made directly by McGann himself or by people working in direct collaboration with him as his students, his technical advisers, or his chosen academic partners. However, the development of the Web as a medium for collaborative work, and the nature of much editorial work, has given rise to "social editing": the possibility of editions made by many people, not working in direct collaboration with one another, often not even known to each other. The possibility of opening up scholarly editing to wider group collaboration through the Web came late to textual scholarship. Even before the Web, indeed, one may find instances of what Yin Liu terms "scholarly open collaboration"in her article in this collection on the Oxford English Dictionary. Models of Web-based open collaboration were fully developed across the Web in the Wikipedia and the Zooniverse projects and in the term "crowdsourcing" coined years before organized attempts to create open collaborative editing projects were initiated. No doubt the entrenched model of scholarly editing, where an edition is made by a single editor or a group of editors working very closely together, had much to do with this tardiness. However, in the years from 2010 on there has been a rush towards different forms of "social editing," following this model. The "Transcribe Bentham" project is the most prominent of these, but it is far from alone. At least two other sophisticated systems exist for the making of full-featured scholarly editions involving open collaboration: the Huygens Institute's eLaborate project and the "Virtual manuscript room" developed by the Institute for New Testament Research at the University of Münster. There are many other "social editing" tools in development by scholarly projects: however, these are typically either closed, undeveloped, or inaccessible (or all three).

In addition, there are many online transcription tools that were not developed specifically for scholarly editing projects but are available for group use in transcription projects: thus, the list of thirty tools compiled by Ben Brumfield (http://tinyurl.com/TranscriptionToolGDoc). Most of these systems produce output either as plain-text transcriptions, which is not suitable for scholarly editing purposes, or, as in those based on Wiki systems, markup inadequate for scholarly editing. Their main use appears to be for the making of materials for personal research (thus, online transcriptions of letter collections and the like: for example, the collection of William Wenham letters at FromThePage). Even when the materials to be transcribed are of scholarly interest (for example, the medieval manuscripts in the T-Pen project), the simplified transcription protocols supported by the system would not enable meaningful scholarly use. Further, these are transcription projects, not editing projects: editing is more than transcription, in the range of editorial decisions which have to be made (including just what has to be transcribed and what principles should guide the transcription). However, the three projects named above all enable scholarly editing, creating fully marked-up transcripts within full scholarly editing environments. All three are "open," to various degrees. "Transcribe Bentham" is the most open and is deliberately designed to allow beginners to start transcription very quickly. The Münster system is less inviting and presumes a higher level of knowledge and commitment, but also permits new users to be contributing within minutes. eLaborate differs from these two in that it is not a single editorial project, but rather a system for hosting collaborative projects. It differs too in that a typical eLaborate project is not open to anyone who wants to join (as are the other two projects), but one must go through some form of joining process (typically, by emailing the project leader).

It is notable that these three projects, like the Rossetti Archive, all adhere to the same editor-driven model. Crowds are invited to join and contribute, but the shape, direction, and policies of the edition are completely under the control of a core editorial team. Indeed, the control goes further than this: once the reader has made and contributed a transcript to the system, he or she hands it over completely and has no further control (or perhaps even access) to it. This is most clearly articulated in the "Transcribe Bentham" project, where once a transcription is contributed, it is reviewed by the project editors and "locked." Once it is locked, even the original transcriber cannot change it. It appears that the transcripts, so contributed and "locked," might be used ultimately either in the ongoing print edition of Bentham's works, or in some form of Bentham digital repository. Beyond transcription, it appears that contributors will have no involvement with the editorial work of the project. In contrast, the Münster New Testament project offers contributors a wider range of activities: reporting the existence of online images of manuscripts, identifying the text in them, transcribing the text. However, it too presumes that the role of the contributors will be, indeed, to contribute: the editors will take the contribution and then use them (or not) in the various editions being made by the Münster and related teams. Strictly, these are "crowd-sourcing" projects, in which the editors determine exactly what work is to be done and how it is to be done, invite others to do some of the work, and then use the contributed work as they see fit.

One can see the attractions of this model. It allows well-established editorial projects to maintain their policies and practices, while opening up some of their work to others—so allowing (a project may hope) for a great deal of work to be done at less cost than might otherwise be necessary. Particularly, this approach meets the objection often voiced by editors, that the quality of the edition might be compromised. The editors remain in complete control; nothing which does not meet their standards will be published. Further, the few years of experience we have had in crowd-sourcing projects has shown us that there is a second benefit in the involvement of "citizen scholars" in the making of scholarly materials. It is this: as well as there being value in the materials the volunteers contribute, the volunteers themselves become energized by their involvement. Project after project reports that volunteers did not just contribute materials, and do not just become fascinated by the individual tasks, but they become involved in the whole area of scholarship of which their task is a part. As Ben Brumfield puts it, they become enthusiasts, patrons, and advocates (Brumfield 2013). Brumfield points to a 2013 article by Trevor Owens explaining that library and archive collections which have engaged in crowdsourcing of transcription and other tasks have discovered this added value in crowdsourcing. If nothing else, this suggests that volunteers might be motivated to do more than the tasks project leaders might assign to them. Owens (2013) puts this eloquently:

At its best, crowdsourcing is not about getting someone to do work for you, it is about offering your users the opportunity to participate in public memory.

Far from being an instrument which enables us to ultimately better deliver content to end users, crowdsourcing is the best way to actually engage our users in the fundamental reason that these digital collections exist in the first place.

One can see that what is true of transcriptions of Civil War letters for a public archive could also be true of transcriptions of Greek New Testament manuscripts as part of a large scholarly edition. Indeed, the bulletin boards and comments sites of the Münster and "Transcribe Bentham" projects (and of many other similar crowdsourcing sites) are replete with evidence of not just commitment, but joyous engagement (for example, this comment by a contributor to the Australian Newspapers Online Project: "you lot are so cool," Holley (2009)). People who become involved in this way become better readers, supporters of the project,—and, even, editors themselves. It is notable that contributor engagement in all three of these projects fits Haythornthwaite's (2009) model of a "community" (or, "heavyweight peer production") rather than a "crowd" ("lightweight peer production"). As Causer and Wallace observe, following Haythornthwaite, contributors who are part of a "community" rather than a "crowd""tend to be smaller in number, to be less anonymous, and to respond to more complex tasks and detailed guidelines" (Causer and Wallace 2012): exactly what a leader of a scholarly edition project would want of his or her contributors. Many scholarly editions already have a ready-made community close to hand in the form of circles of scholars, students, and readers already dedicated to the study of a particular author or text, or group of texts. The Münster Greek New Testament project has quarried this community to great effect; many more may do so.

Another feature of these three "social editing" projects is that they enable a very clear division of tasks. The lead editors can decide what tasks are suitable for what levels of skill and assign those accordingly while reserving complex editorial tasks for themselves. One can imagine that as social editing projects develop, as contributors show themselves capable of more demanding tasks and willing to undertake them, and as editors learn to trust their contributors, that the range of responsibilities undertaken by contributors will expand. In theory, it might expand to the point where the distinction between "contributors" and "editors" dissolves. In practice, at least with existing editions which have (in many cases) a long history of established practice and hierarchical control, this seems unlikely.

Mirroring this gap between "contributors" and "editors" in these instances of social editing is another gap. This gap is between a very few projects which aim to achieve scholarly editions created to a high scholarly standard and which have established an online presence to allow others to contribute, and a large number of other projects which exist, on a more-or-less ad hoc basis, to create transcripts and other materials typically related to a particular resource: a collection of letters from or concerning one person or family, or in one archive, or with a single thematic connection. Most of the tools listed by Brumfield are used in this way. There is no question of the utility of these tools and of the resources created within them: essentially, they are destined to be used alongside online images of the originals which they represent. These tools are typically free and take just a few minutes time to set up for a new transcription project. The very popular Omeka system, used widely for creating an online presence for archival and other cultural memory artefacts, now offers a plug-in, Scripto, to allow users to "crowdsource the transcription" of Omeka content. Like many other systems listed by Blumfield, Scripto is based on MediaWiki software. Accordingly, it lacks fundamental features essential to a scholarly edition in digital form: it can handle only a limited range of characters (which rules out almost any editor working outside modern Western European materials) and supports basic Wiki-style markup only (which rules out even simple structural encoding, let alone representation of more complex markup).

This gap, between projects which are content with what can be done for free and those which have a demanding agenda and are able to create a tool to achieve it, is more like a chasm than a gap. All three of the projects I highlight here—"Transcribe Bentham," the Münster Greek New Testament Project, the Huygens eLaborate project—were developed at considerable expense. "Transcribe Bentham" has won at least two major grants: £262,673 in 2010 from the Arts and Humanities Research Council and $538,000 from the Mellon Foundation in 2012 (Causer and Terras 2014). To the average scholar, used to grants of a few hundred dollars for some travel or materials, these sums are astounding. We can assume that substantial sums were necessary for the Münster and Huygens systems. One notes, wryly, that if one had paid instead for the transcription so far achieved by the "free" volunteer labour contributed to the "Transcribe Bentham" project—5729 "folios" as of 3 July 2012, each with some 500 words, or 2500 characters, thus around 3 million words or 15 million characters—the total cost would have been in the order of $50,000 (based on the $0.89 cents per 1000 characters cited at http://accesstei.apexcovantage.com/Home/PriceMatrix, hence around $14,000 and adding a generous multiplier for the difficulty of the material and for supervision costs). As well as the cost of developing the software, a project would also need to reckon with the additional costs of supervision of contributors of varying skill levels.

Despite these caveats, the advantages of "social editing" as deployed in these and in many other less complex projects are compelling. While the financial advantages of this approach are probably less significant than one might suppose, one could justify the effort by the building of a community around each edition and the consequent benefits to both editors and readers (see Causer and Wallace (2012)). We can expect that many other editorial projects will find ways to involve their communities in their work. For this to happen, the cost and difficulty of creating an online editing framework for a scholarly edition project will have to reduce substantially. One can pick up and configure an "off-the-shelf" system which allows many people to carry out a very limited range of editing functions in a rather simple format, but it should be possible for editing projects (and indeed, single editors) to set up a system to support a wider range of editing tasks, creating data fully adequate for scholarly use, without prodigies of programming and massive grant funding (indeed, without any programming and no grant funding).

Finally, it is factually inaccurate to use the term "editing" to describe the work done by contributors to these projects. They are engaged in a scholarly project, carrying out highly-specific tasks as transcribers, indexers, annotators, but they are not "editors" in the sense of making decisions concerning the aims, strategy and details of the edition, or in effecting the complex processes entailed by these decisions. One might better describe these activities as "community engagement in scholarly editing projects” rather than “social editing."

Social editions

In the foregoing discussion of "social editing" I referred to the possibility that the narrow range of tasks currently entrusted to on-line contributors in projects such as "Transcribe Bentham" and the Münster Greek New Testament Project might become much broader. In theory, one could imagine that every editorial activity, without restriction, might be open to every contributor. It is this vision which underpins the concept of the "social edition" emergent in the last five years and closely (if not completely) identified with the work of Ray Siemens and his collaborators. Two articles by this group (2010 and 2012) present a definition of a "social edition," and an actual edition, by Siemens and his collaborators, of the Devonshire Manuscript is claimed to be an instance of a "social edition." Two characteristics are presented as key to "social editions." First, online collaborative "social" tools are used in their creation:

A definitively "social" edition employs new and emerging tools for interaction around such activities as transcription, bookmarking and bibliography-building, flagging and tagging, commenting and annotating, linking to contextual material (especially for names and integration of bibliographic information), glossary and other analytical functions, and all other pertinent activities that sit at the evolving intersection of social media and the electronic scholarly edition (Siemens et al. 2012, 453).

Second, the "social edition" is made jointly by a "community of practice": by anyone and all who contribute to its making, using social media tools. Within these communities, all activities are shared by all members:

In a "social" edition, textual interpretation and interrelation are almost wholly created and managed by a community of users participating in collective and collaborative knowledge building using social technologies (Siemens et al. 2012, 453).

In this account, the relations between editors, contributors, and readers within the community are altered:

[T]he "social" edition privileges a new kind of scholarly discourse network that eschews traditional, institutionally reinforced, hierarchical structures and relies, instead, upon those that are community-generated (Siemens et al. 2012, 453).

In this shift of power from a single editor to a whole community the role of an editor is redefined. The editor is no longer "a single authority," determining and shaping what is important to the reader, exerting "immense control over what the reader can engage" (Siemens et al. 2012, 453). He or she is now (if anything, one supposes) a "facilitator," one who is participating in a "process" of collaborative "textual knowledge creation," and not the "progenitor" of a product, a scholarly edition.

Thus presented, as a series of general statements about what editing has been and what it might be, this appears an irresistible set of propositions. Editors in the past, we are told, controlled just what the reader read, determining with unilateral authority—in effect, dictating—what the reader should think. In the social edition of the future, we will be freed from this traditional, hierarchical, top-down strait-jacket of knowledge. Instead, we the people, the community, will make knowledge ourselves. Who could prefer the old-fashioned, hierarchical and authoritarian to the modern, egalitarian and communitarian? The rhetoric suggests that we are on the verge of a Textual Spring, with social editions as the expression of a new unbounded world of community-based knowledge creation.

However, on closer analysis, doubts emerge. Consider the first assertion: that a social edition uses Web 2.0 tools. That may be true, but many editions which are most certainly not social editions also use Web 2.0 tools (notably the Münster Greek New Testament editions). The use of social tools says nothing about the "social" status of the edition. This argument is so weak that one wonders why the authors make it. This throws the whole weight of definition of the "social edition" on the second assertion: that a "social edition" is characterized by the dissolution of the authority of a single editor, now replaced by the wisdom of the crowd. In the first place, the description presented in this paper of the "traditional institutionally-reinforced hierarchical" editions of the past is a caricature. In textual scholarship, the notion of "definitive editions," complete with the seal of approval of the MLA Committee for Scholarly Editing and with any pretence to the kind of authority here denounced, was comprehensively demolished through the 1970s and 1980s, following foundational work by three scholars in North America—Peter Shillingsburg (1996), Jerome McGann (1983) and David Greetham (1999)—all prominent in the foundation of the Society for Textual Scholarship in that period. I doubt that one could find now a single textual scholar in North America or Europe who works to the model presented by this paper, as the survey of scholarly editing around the world in Greetham (1992) amply illustrates. Indeed, the writings of Fredson Bowers and Thomas Tanselle, and many others, show that so-called "definitive editions" were never as "definitive" as this characterization suggests. Rather, as Prue Shaw cites her mentor the great Italian philologist Gianfranco Contini as repeatedly saying, an edition has always been a working hypothesis ("un' ipotesi di lavoro"). Shaw remarks, "[T]he text reflects or embodies the best hypothesis the editor is able to construct to explain the inter-relationships among the individual extant copies, and the relationship of all of them to the author's original … Only an unwise or presumptuous editor would claim that an edition was definitive" (Shaw 2006). Siemens and his collaborators are surely aware that no scholarly edition made in the last decades has pretended to the singular authority they denounce: thus their frequent and approving citations of McGann. So why do they assert this? Or are they asserting that in this democracy of knowledge, all opinions are equal, so that the views of an editor who has spent a lifetime studying a particular text, as Shaw has done with Dante, is worth no more than that of the casual reader with minimal knowledge of Italian or Dante? Is an editor like Shaw, presenting elaborate and detailed arguments about how the textual traditions of Dante's Monarchia and Commedia evolved, exerting "immense control" over her readers; is the "editorial and analytic lens" she offers an infringement upon the liberties of her readers, as her editions privilege what she knows over what others know?

Some statements in the Siemens et al. (2012) article ("immense control"; "traditional institutionally-reinforced hierarchical") suggest that the authors are gesturing towards some concept of a complete democracy of knowledge, free from control and hierarchies. But other statements suggest that they are aware that in the communities of knowledge creation, not all opinions are equal. It appears that, after all, the editor is to retain some kind of role distinct from that of others in the community: he or she is to be a "facilitator." The text of the 2012 article offers no definition of what this "facilitator" is to do, beyond that he or she will not be a "traditional" editor. However, a footnote in the final published version, written one can suppose in response to reviewer comments on earlier drafts (and perhaps to my own comments on a pre-publication version, kindly sent me by Ray Siemens) does try to spell out something of what this "facilitator" might be. He or she is a "facilitator for community enrichment," undertaking "interpretive curation": "The work of the editor of the social edition is to make this kind of curation possible for members of the community of practice to undertake" (Siemens et al. 2012, 460). What is being "curated" here, it seems, is not the edition itself (which the previous footnote asserts might be made by traditional means), but "interpretations" made by the community after the initial edition is made. With no stretch at all, one could describe Prue Shaw (2006) as exactly such a "facilitator," in the sense that her editions enrich and enable the interpretations of others (as good editions always have). It appears that, after all, this "facilitator" might be an editor, as we have long understood the term.

The uncertainty as to the role of this "facilitator" (who might be, indeed, an "editor"), stems from a deeper flaw in these arguments. It is quite clear what Siemens and his collaborators mean by "social" in the phrase "social edition": pages in the 2012 article are spent describing how social media can create edition components, and a whole separately published annotated bibliography (originally intended as an appendix to the article) describes many instances of social networking applied to digital knowledge creation (Siemens et al. 2011). It is not at all clear what they mean by "edition." The 2012 article cites John Unsworth's categorization of the seven "primitives" of humanities scholarship: "discovering, annotating, comparing, referring, sampling, illustrating, and representing," and observes their centrality to the making of editions (Unsworth 2000). The article details five additional activities as "primitives" in the "new modes of engagement with digital objects" they see engendered within social editions—collaborative annotation, user-derived content, folksonomy tagging, community bibliography, and text-analysis—and notes with approval that they are "user- rather than creator-driven, evolving rather than fixed, collective rather than individual, expansive rather than inclusive, and open source rather than proprietary and closed" (Siemens et al. 2012, 452). Indeed, a scholarly edition might contain all of these elements, but this is not all a scholarly edition contains. This description omits all the parts of an edition which require decision: the choice of who the edition should be aimed at; whether it should present an edited text; if so, on what principles should that text be edited; what conventions of spelling, punctuation and presentation should be followed, and many more, right down to decisions on single characters. For editors, these questions revolve around the terms "document,""work," and "text," questions which each edition must determine for itself, and questions which have been made more complex by the digital turn (for discussion of the terms document, text and work, and their significance in textual scholarship, see Eggert (2009), Greetham (1999), Robinson (2013a), Robinson (2013b), Shillingsburg (1996), and Tanselle (1989)). These decisions may be far from easy, and editions can be, and often are, controversial: one thinks of Gabler's Ulysses, or the Wells and Taylor Shakespeare, not to mention the centuries-long struggles over the text of the Greek New Testament. One finds in the text of this article no trace of awareness that these issues might be problematic, and no hint as to how such decisions are to be made in the "social edition." It is not credible that some kind of crowd consensus could achieve decisions to everyone's satisfaction, where generations of scholars and readers have not been able to do this, and Siemens and others do not even attempt to make this case. In editions, the making of decisions is linked to authority: we trust an editor, and the edition he or she makes, because we understand the decisions upon which the edition is based, even if we do not agree with them. We especially trust the editor to situate, on our behalf, the edition in the landscape of document, work and text. But what is the authority of the "social edition?" What is the authority of the crowd? Any account of editing which does not deal with these issues is radically incomplete.

As well as the two articles, the collaborators have produced an edition, The Devonshire manuscript social edition, published in 2011 in Wikibooks (no date of publication is given in the publication; this date is derived from the Wiki "Revision history," which dates the first version to 14 December 2011). The title proclaims this is a "Social edition," and so one might reasonably expect that the practice of this edition would clarify what is not clear, or not discussed, in the articles. This edition is a superb work of scholarship. The transcriptions are done to the highest standard according to an exceptionally finely-detailed set of guidelines; the introductions and annotations to each poem are thorough and comprehensively researched. One would not hesitate to recommend this edition as a proof that a digital edition can achieve the same high standards of excellence, and hence authority, as can a print edition. But there is no sense in which this edition, as published in 2011 and as now instanced as I finalize this chapter in August 2015, is a "social edition," as described in the text of the 2012 LLC article. It appears that the plan and policies of the edition, their expression in the transcription and annotation principles, and the final execution of the edition were the products of a small group of scholars led by Ray Siemens, whose name appears first on the edition, and who clearly guided every aspect of the making of the edition over more than a decade from its first conception through to its actual publication. It is no different in this respect from (say) my own editions of the Canterbury Tales, or indeed almost any edition ever made (including all those "traditional institutionally-reinforced hierarchical" editions decried in the 2012 article). One might argue that it has now become a "social edition" because of its publication on a "social" platform, Wikibooks, in theory permitting unbounded involvement and alteration by anyone who cares to contribute. But in practice, following an attempt to vandalize the online edition, the Wikibooks administrators have enforced a review policy on all contributions: hardly "social" indeed (personal communication, Ray Siemens). Even without this restriction, the core XML files of the manuscript itself cannot be altered, nor can the alterations be displayed, within the online edition.

The careful reader will observe that in the last paragraphs, I qualify my references to the 2012 article as being to "the text" of the article. Several footnotes, apparently written in response to reviewer and reader criticisms, tell a different story. The first footnote (p. 458) does refer to the problematic concepts of text, document, work and authority (acknowledging that I had drawn the authors' attention to these in my own comments on a draft version of the article); footnote 21 attempts a fuller explanation of "authority" in the social edition; footnote 22 offers to explain just what a "facilitator" might do. At the end of footnote 21 (p. 460) one finds this sentence:

To construct a social edition, we must rely on earlier theories of editorial practice and disciplinary conventions to determine our source text and ultimately the digital representation of that text.

One seems to hear the sound of air escaping from an over-inflated tire as one reads this sentence. After all the article's rhetoric, the text of the social edition—the core of the edition—is to be made as editions always have been: following the same "theories of editorial practice" and "disciplinary conventions." This same footnote shifts the focus from the "social edition" to a "social archive," citing Dean Irvine's advocacy of editions as instances of "socialized text" (Irvine 2006, 202-3). It appears that what is to be "social" is not now the initial creation of the edition, but its afterlife, as it becomes surrounded by, and embedded within, a continuing unrolling of conversation, commentary and discussion, enabled by and promoted through social media. This is an attractive formulation, as the edition becomes the ground for enriched reception of texts by their readers, reshaping itself as readers change. The Wikibooks publication of the Devonshire manuscript is a gesture towards this formulation, though not a very effective one. Indeed, the decision by the editors to release the XML source for the core manuscript transcriptions free of all restriction makes possible what would indeed be a realization of what one might call the "socialized edition," where others take up and build on the work done by the initial editors, creating new editions and new readings. One can see this process in (for example) the Whitman Archive, which began (as the title suggests) as a repository for Whitman materials but is increasingly a forum for, and access point to, multiple readings of Whitman. This view of the "social edition" may be less revolutionary than the parts of the 2012 article which advocate a complete remaking of how we create editions, but it may have no less profound effects on how editions communicate.

Conclusion

The arguments of this chapter—that neither "social text editions" nor "social editions" exist and that the phrase "social editing" is misleading—may disappoint those who would like to see textual scholarship share the up-to-the-minute excitements of the social media. There is no need for this disappointment. If we strip away the unrealistic expectations invoked by mis-application of the term "social," the possibilities offered by the digital media in all three areas are real and substantial. We may not have "social text editions," but we can have editions which represent the original documents, and all the processes which shaped the text throughout its history, far more fully than ever before possible. We may not have "social editing," but we can have many more people involved in the making of editions (both enriching the edition, and their own experience of the text) than ever before possible (as envisaged in the description of the Shared Canvas data model in the Smith and Viglianti chapter later in this volume). We may not have "social editions," but our editions can become the locus of a continuing conversation between our readers and ourselves. These are worthwhile goals.