"Social media" are now commonly considered Internet phenomena: that is, they are technologies used by people to engage in acts of representation, relationship-building, exchange, and collaboration in networked digital environments. The Oxford English Dictionary (OED) dates the first attested instance of social media to 2004 (see OED, 3rd ed., s.v. "social," S2: "social media"). But, of course, people do not need digital media in order to be social, and "social media" tools depend upon, enable, and enhance existing patterns of human interaction in complex ways. In the previous chapter, Peter Robinson questions our often vague use of the term "social" in the context of making digital scholarly editions. Discussions of this sort often benefit from a longer historical view, especially insofar as scholars may wish to engage members of the general public in academic projects without simple-mindedly jumping onto a "crowdsourcing" bandwagon. Indeed, collaborative scholarship has pre-Internet precedents that provide valuable insights into ways in which human participation in such projects can be managed. I offer the history of the first edition of the Oxford English Dictionary (OED1 for short) as a case study of a successful scholarly open collaboration (or "academic crowdsourcing") project that is instructive both for its early failures as well as for its eventual success. This historical analysis complements and supports the sort of analysis provided by recent academic crowdsourcing projects in reviewing their methods and results.

The term crowdsourcing was introduced to the English language by Jeff Howe in a 2006 article in Wired (see Howe (2006)). Howe subsequently expanded the article into a book, published in 2009 (see Howe (2009)). He defines crowdsourcing as the use of "distributed labor networks""to perform tasks, usually for little or no money" (Howe 2009, 8). The term is now common, although there has been some argument over its meaning or its applicability (Estellés-Arolas and González-Ladrón-de-Guevera 2012). The OED itself did not publish, in its online 3rd edition, an entry for crowdsourcing until June 2013. This is ironic, because the OED is often mentioned as a historical precedent for academic crowdsourcing (see, for example, Wikipedia's article on "Crowdsourcing"; Dunning (2011); Lanxon (2011); Jenkins (2010); and, perhaps more thoughtfully, Rockwell (2010)). In the sciences, where the method has achieved some notable successes, it is often called "citizen science": respected examples include the Galaxy Zoo project for classifying galaxy images and the online protein modelling game FoldIt, whose players, in 2011, successfully modelled the crystal structure of a specific retrovirus protein, a problem that had resisted conventional methods for over a decade (Khatib et al. 2011; Praetorius 2011; see also Eiben et al. (2012)). In the humanities, this approach is often used for transcription projects (Dunn and Hedges 2012; Gibbs 2011), such as Transcribe Bentham (handwritten papers of the philosopher Jeremy Bentham), Old Weather (meteorological data from ships' logs), or Ancient Lives (the Oxyrhynchus papyri). Open collaboration approaches are also used for OCR correction, as in Trove (the Australian newspapers digitization project) and Dickens Journals Online. A better term, further from "outsourcing" and less closely tied to business models, might be "scholarly open collaboration." A large scholarly project involves numerous smaller tasks that humans can do more efficiently or effectively than any machine or algorithm so far developed: comparing images, for example, or proofreading texts. Rather than depending on paid labour to accomplish these tasks, the project invites contributions from a wider community. Anyone is invited to participate, and contributors do so on a voluntary basis, usually working from their own homes and in their spare time.

One product of such a process was the OED or, as the first edition was called, A New English Dictionary on Historical Principles, published in fascicles from 1884–1928: the most extensive and thorough lexicographical work in English, indeed in any language, at the time, and with its succeeding editions surely one of the great dictionaries of any time. We can consider the OED a crowdsourcing project in that, as Knowles (2000, 23) has put it, "the typical contributor was an unpaid volunteer who was not only working 'out-of-house,' but was in fact often living at a considerable geographical distance, with instruction and comment being supplied largely through the written word."

It is particularly important to consider such pre-Internet cases, as Geoffrey Rockwell has pointed out, because, although most examples of crowdsourcing today depend on Internet technologies, the methodology of crowdsourcing is not intrinsically technology-dependent: open collaboration depends rather on human motivation, human behaviour, human relationships, human thought, and human creativity (Howe 2009, 11; Rockwell 2010). The technology of OED1 was basic: innumerable slips of paper, recording examples of word usage sent in from volunteer readers, and a set of wooden pigeonholes to sort the slips, covering the walls of a corrugated iron shed in the editor's garden. The early history of the OED shows that the eventual success of the first edition depended not on the technology used or the participatory method itself, but on the ways in which open collaboration was managed.

The most eminent British dictionary before the OED was that of Samuel Johnson, published in 1755. Johnson's Dictionary of the English Language was the product of an older approach to lexicography: Johnson wrote most of the Dictionary himself, illustrating his definitions with quotations chosen from his own reading. He had the help of about six paid assistants, but their role was downplayed; they were considered technicians of a sort, not contributors to the intellectual achievement of the Dictionary (Reddick 1990, 37-45). A famous story told by Boswell may or may not be true, but it communicates well the legend of the solitary genius that came to be attached to Johnson's Dictionary. When Johnson was told that the Académie Française had finished its French dictionary after forty years of labour by forty lexicographers, and was asked how he could expect to finish his English dictionary singlehanded in the three years he had confidently predicted, ^[1] Johnson supposedly retorted: "This is the proportion. Let me see; forty times forty is sixteen hundred. As three to sixteen hundred, so is the proportion of an Englishman to a Frenchman" (Boswell as cited in Reddick 1990, 1; for more on Johnson's Dictionary, see also Korshin (2005)).

But the lexicographical project of the Académie Française showed that Johnson's single-scholar model of lexicography was already being superseded by a more collaborative approach. Johnson's Dictionary, while admittedly a work of genius, was also notoriously idiosyncratic, reflecting the lexicographer's literary tastes, his politics, and his moral and social anxieties (but see also Weinbrot (2005)). Furthermore, Johnson's generous use of quotations to illustrate his definitions suggested the idea that the purpose of quotations was not merely to illustrate definitions but to serve as the evidence, the basis, for definition. A later lexicographical project, Charles Richardson's A New Dictionary of the English Language, published 1837, also used a vast collection of quotations as the basis for etymologies and definitions, although Richardson's work was unfortunately rather eccentric and often misleadingly opinionated (Landau 2001, 77-78; Reddick 2008). Indeed, the shortcomings of single-authored dictionaries such as Johnson's and Richardson's showed that this data-driven method of lexicography put the methods of dictionary-making beyond the effective capacity of any one man.

It was this problem in method that Richard Chenevix Trench noted in his address to the Philological Society of London, "On Some Deficiencies in Our English Dictionaries," which in 1857 laid out before the Society the principles of what would eventually become the OED:

The literature of our language is so vast, so far exceeding the compass of any one man's power to embrace it all, that innumerable precious quotations must escape the single-handed student; even when he inherits the labours of others, who, single-handed as himself, have wrought before him in this almost boundless field (Trench 1860, 50).

"But how, it may naturally be asked, shall all books be read?" Trench answers this question by pointing to the German dictionary then under way, the Deutsches Wörterbuch, compiled by Jakob and Wilhelm Grimm, which began publication in 1854 and was completed in 1961 (Osselton 2000, 61-64). The Grimms' dictionary acknowledged in its first instalment the labour of 83 volunteers. Trench imagines an even more extensive collective enterprise, involving an "army" of volunteers, beginning with the members of the Philological Society but extending beyond them to the public at large:

it is only by such combined action, by such a joining of hand in hand on the part of as many as are willing to take their share in this toil, that we can hope the innumerable words which have escaped us hitherto, which are lurking unnoticed in every corner of our literature, will ever be brought within our net (Trench 1860, 69-70).

Inspired by this vision, the Philological Society published in 1859 its plan for "a new and more Scientific Dictionary than any at present," invoking the example of the Grimms in appealing to "Englishmen to come forward and write their own Dictionary for themselves" (Philological Society 1859, 1 and 8).

It was a grand plan, but its implementation proved to be more difficult than expected. Herbert Coleridge (grandson of the poet), who had taken on the editorship of the new dictionary, wrote in 1860, in an open letter to Trench, that 147 volunteers had been recruited, exclusive of any Americans that a special appeal across the Atlantic might later bring in. However, of that number, 43 had terminated their involvement, having volunteered for only temporary duties and having successfully discharged them; a further 15 or so Coleridge called "hopeless," individuals who had volunteered but subsequently refused to be involved. Of the remaining 89, only 30 were classed by Coleridge as "first-rate contributors." Still, Coleridge suggested optimistically, these statistics were good compared to those of the Grimms, who had confessed that out of their 83 contributors only six had proven satisfactory and, of those, only one came close to ideal (Coleridge 1860). It was clear that in any randomly selected sample of volunteers, only a minority would be of significant use.

When Coleridge died of consumption the following year, at the age of 31, the editorship fell to the energetic and controversial Frederick James Furnivall, who tackled the dictionary project with characteristic enthusiasm and carelessness. Although constantly distracted from the dictionary by the text societies he founded—including the Early English Text Society, the Chaucer Society, the Ballad Society, and the New Shakspere Society (sic)—by the end of his editorship he had squeezed out of his volunteers and his own reading "over one and three-quarter tons of material" (Mugglestone 2005, 10). Unfortunately, much of it was practically useless. Entire sackfuls of quotation slips were missing and some had been inadvertently destroyed (the remains of the Pa section were finally discovered in an Irish stable, after most of it had been used as firestarter); what remained was, in the words of James Murray, the editor who inherited it, in a state of "primitive chaos" and "utter confusion," and the material itself "rarely to be trusted" (as cited in Mugglestone 2005, 14-15). Murray, who edited most of OED1, was forced to start almost all over again; one of his first actions as editor was to issue a new public Appeal for volunteer readers.

Thus the first twenty-two years of the OED project—from Trench's foundational lectures in 1857 to Murray's first Appeal for Readers in 1879—appear to be marked by failure and frustration more than by progress, let alone success. The collaborative method, conceptually sound as it seemed to be, was not working, and nothing had been published; volunteer readers were losing interest, and publishers shying away (Mugglestone 2005, 10-11). Clearly, if you build it, they won't necessarily come, especially if you don't seem to be building it as much as collecting scrap lumber. One problem, as we have seen, was the quality of the contributors: although Coleridge was a conscientious and sometimes ingenious manager of the early dictionary work—it was he who devised the system of pigeonholes for sorting quotation slips that later editors, including Murray, continued to use (Mugglestone 2005, 9)—his contributors, mainly members of the Philological Society to whom he had appealed for volunteers, did not provide usable material quickly enough to make real progress. The other problem was management. The socialist Furnivall attempted to extend the network of contributors beyond the Philological Society to include more members of the general public, academic or otherwise, so that the Dictionary would represent the English nation completely and not just a privileged section of it (Mugglestone 2005, 10). However, this wider group of volunteers lost motivation as slips they submitted disappeared into Furnivall's chaos and as they received no feedback or further instruction about the value of their contributions (Knowles 2000, 23): Charlotte Yonge, who served as sub-editor of N, expressed frustration but no surprise when she sent sacks of material for her section to Furnivall and, four or five years later, received a letter from him asking whether she had sent him anything (Brewer 2008).

Thus one of the achievements of James Murray, the editor who succeeded in bringing out the first edition of the OED, was to solve, with more success, these two critical challenges of the OED's methodology. He did so first by recruiting more volunteers, beginning with a new Appeal to the Public in 1879. Secondly, and more importantly, he communicated with those volunteers strategically and persuasively, giving the best of them a personal stake in the project. Murray's strategies for managing the OED1 contributors, especially when compared to similar strategies used by successful academic crowdsourcing projects today, amount to a set of best practices for scholarly open collaboration.

There are significant parallels between the knowledge economies of Victorian England and those of industrialized countries today; like Victorian England, we produce far more educated people than we employ to use the education they have received. Scholarly open collaboration depends on a robust population of people not employed as academics, but often highly educated, seeking outlets for their intellectual energies. This pattern is evident when we examine the most prolific contributors to OED1 (Gilliver 2013). Those known to have supplied more than 10,000 quotations include a few philologists; a number of scholars who were not philologists; and quite a few schoolmasters, physicians, and well-read women. A typical reader might be Richard Prior, who contributed some 11,700 quotations by 1888; a botanist by profession, he had also translated a collection of Danish ballads and written a book on croquet. The most prolific reader for OED1 was a Thomas Austin Jr., about whom I have been able to discover remarkably little, except that he may be the man who edited Two fifteenth-century cookery-books for the Early English Text Society, that he suffered from mental illness, and that he had supplied over 165,000 quotations to the OED by the publication of the first instalment. Another significant but lesser-known contributor was James Platt Jr., who reportedly knew every European language and many African, Asian, and North American Aboriginal languages as well; Murray would send him obscure words and Platt would reply within a couple of days with information about the languages from which the words came. But Platt was not a professional scholar; he worked for his father's woolen business in London and had very few books of his own. During the hour he was allowed for lunch every day, he would go to the British Museum library and, in the scant ten minutes available to him there, would select and take out one book which he would read and return the following day, jotting down some information in his notebooks but retaining most of it in his extraordinary memory (Murray 1977, 307-308). The success of Murray's volunteer recruitment was due in large part to his ability to attract and retain contributors such as these, men and women who performed these tasks not as professional lexicographers or even as traditional academics, but out of personal interest and frequently as an exercise of unusually high skill or knowledge.

It was true of OED1, as of today's crowdsourcing projects, that a small minority of volunteers contributed the vast majority of the usable work. Scholarly open collaboration needs a large volunteer base not because more people produce better work—they do not—but because a larger number of volunteers increases the chances of finding good contributors, those who will produce usable work. These "supercontributors" are crucial to the success of an open collaboration project, and thus the recruitment, motivation, and management of those volunteers is also crucial (Brumfield 2012; Causer, Tonra, and Wallace 2012; Holley 2010;.Wallace and Causer 2011). Jeff Howe claims that an important motivator for crowdsourcing participants is social relationships (Howe 2009, 14), the sense of belonging to a community, but the Transcribe Bentham team found, on the contrary, that their "supertranscribers" were more likely to be task-oriented rather than relationship-oriented, somewhat territorial about their manuscripts, and more likely to seek communication with the project leaders than with other volunteers (Causer, Tonra, and Wallace 2012; Wallace and Causer 2011; similar findings have been reported by Galaxy Zoo; see Smith (2011)). A similar pattern emerges for the top contributors to OED1. They included not only William Chester Minor, the American physician incarcerated in the Broadmoor Criminal Lunatic Asylum after he shot and killed a brewery worker, but also Fitzedward Hall, a former professor of Sanskrit who, after a bitter dispute with a fellow scholar, retreated into seclusion under the conviction that he was being systematically persecuted (Murray 1977, 305–307; Stevens 2013, 50-63; Winchester 1998). Murray, although he never met Hall in person, corresponded with him frequently and the two men enjoyed a cordial relationship (Murray 1977, 304-305). It seems reasonable to suppose that a man like Hall found in the OED project a means of participating in scholarship without having to talk to any scholars. The volunteers who contributed to the OED did not form a social network: they appear to have participated because they found intellectual work satisfying and because they found the project itself meaningful.

In three important ways, Murray fostered that sense of being involved in meaningful intellectual work. The first was that he created a narrative in which contributors could participate. His 1879 Appeal—its full title was "An appeal to the English-speaking and English-reading public to read books and make extracts for the Philological Society's New English Dictionary"—set out clearly the scope and intent of the programme. The goal of the project was "a new Dictionary worthy of the English Language and of the present state of Philological Science,""a lasting monument to our language" (Murray 1879, 1, 3–4). The Appeal was surprisingly forthright about the history of the project thus far, frankly admitting that work had come almost to a standstill—Murray uses the phrase "practically dead"—but also assuring readers of a publication contract with Oxford University Press and promising to publish a first instalment of A in 1882 (the first fascicle, A–ANT, was published in 1884).

The second method that Murray used also contributed to the making of a narrative: he made sure that progress was measurable and that volunteers' contributions were acknowledged. OED1 was published in a series of fascicles that subscribers would collect and later bind into volumes; this kind of serial publication maintained interest in a project that would not, as it turned out, be completed until 1928, after Murray's death. There was thus a greater sense of immediacy about the results of the work; volunteers could be given a stake in an ongoing (if unfinished) project, rather than in a finished product whose final form was too far in the future to imagine. Furthermore, each fascicle of OED1 included a Preface, written by Murray, that acknowledged the chief contributors, provided copious statistics to show how much had been accomplished, and discussed the most significant lexicographical issues raised in that instalment (see Raymond (2010)).

Thirdly, Murray's Prefaces, being essentially lectures in lexicography, also served to educate the readers of the OED. They demonstrate one of the most important factors in the success of Murray's management of the OED volunteers: his care to communicate the parameters, procedures, and implications of their tasks in a manner exemplary for clarity and detail. The 1879 Appeal included a set of "Directions to Readers for the Dictionary" that laid out, in numbered points, such instructions as "carefully preserve the spelling, capitals, etc. of the original," and then supplied examples of quotation slips filled out in the proper way (Brewer 2011). But OED1 depended not only on volunteer readers but also on volunteer sub-editors, proofreaders, and consultants. Thus a similarly detailed set of "Directions to SubEditors" was sent to the volunteers who ordered the quotation slips and organised the entries (Knowles 2000, 24-25). OED1's contributors were also learning lexicography. The task given to the greatest number of contributors, the volunteer readers, was itself not very intellectually demanding: to notice a word of interest and to copy it out, along with pertinent information about its immediate context, onto a slip of paper and mail it in. But Murray in his Prefaces showed his readers how their small contributions, taken cumulatively, enabled lexicographers to observe patterns in the English language. This scholarly context doubtless helped many volunteer readers, not themselves professional lexicographers, to submit more useful quotations. Dedicated and productive volunteers who showed a greater understanding of the methods, needs, and challenges of a dictionary project could thus be recruited to roles of greater responsibility.

In conclusion, then, here are some observations about the ways in which OED1 applied the principles of open collaboration. First, lexicography lends itself to this type of approach, because many of its most basic tasks, such as collecting quotations, require human labour that is too large for the conventional human resources available, but can be broken down into small, simple modules that a very large number of people can complete successfully. Furthermore, these are tasks that are not easily automated and for which satisfactory algorithms do not—and may never—exist. Crowdsourcing approaches tend to work best for tasks that involve trading or selling, collecting data, classifying, or problem-solving. The corollary of this observation is that there are many academic tasks that are not amenable to such an approach; for example, crowdsourcing is not necessarily a good way of arriving at authoritative conclusions. One might accept a quotation from a nonexpert reader for the word door, let us say, but one would be better served by consulting a Fitzedward Hall for its Sanskrit cognate or letting a trained philologist write the etymology.

Thus open collaboration is not a way to do work cheaply or at no cost (Causer, Tonra, and Wallace 2012). It is not a substitute for funded research, but a supplement to it—sometimes an essential one. OED1 was not problem-free after Murray became editor; it tended to bottleneck at his place in the process, where he alone could make final editorial decisions, and funds eventually had to be found to hire more staff, including additional editors. Paradoxically, therefore, the successful use of casual volunteer labour will usually require the commitment of staff who work on the project full-time and who will ordinarily have to be paid. Crowdsourcing is sometimes thought of as a kind of spontaneous generation: an amorphous mass of people, the "crowd," forms communities around common interests and produces truth and beauty democratically, selflessly, and organically. This does not happen. Successful scholarly open collaboration in the tradition of the OED requires deliberate and strategic management. The OED could never have been spontaneously generated by a mere crowd; it needed the long-term financial support of the Oxford University Press, the intellectual support of the Philological Society, and, most importantly, the management and communication skills, and the vision, of Murray as editor—a Scottish tailor's son who sincerely believed that he was called by God to edit the Dictionary, and who therefore made it his life's work (Murray 1977, 340-341).

It is reassuring to find that scholarly open collaboration, in the case of OED1 and successful academic crowdsourcing projects today, works best when it is managed as the reverse of exploitative labour. Where contributors are amateurs in the original sense of the word—people who do something for the love of it, not for financial gain—the onus is on the project to attract, train, and support committed and skilled volunteers. Whether submitting data on slips of paper or online, whether finding quotations or transcribing manuscripts or classifying images or modelling protein folding in a game, these volunteers contribute to work that they find important, meaningful, and narratively compelling, the results of which are immediate and visible, and for which they are promptly acknowledged. Not all academic work can be done well by volunteers in this way, but the example of the OED shows that this sort of scholarly open collaboration has a long and successful history; a project does not have to be self-consciously or ostentatiously "digital" to appeal to, or to involve, the public.