The strong theme of these two essays is that we have not yet done electronic scholarly editions right because we do not have an adequate environment, tools or vision for the digital future of textual scholarship. Nothing I say in this essay is intended to disparage the fine and complex electronic editions underway or recently accomplished on HYPER platforms (Nietzsche, Woolf, Wittgenstein), or at IATH (Rossetti, Piers Plowman, Blake), MITH (Dickinson), or projects like Beowulf, Boethius, the Dead Sea Scrolls, the Codex Sinaiticus, Archimedes Palimpsest, or the Canterbury Tales, etc. These each represent in one or more ways the leading edges in electronic editing, but none yet offers much more than local solutions to local problems of literary representation.[2] They share too little in common to be called a united front or even a community of scholars. Electronic publishing is still too new. At this stage there are more questions than answers. But my tentative conclusions are that we need to develop a collaborative electronic workspace for the construction of textual “knowledge sites” that will be dynamic, interactive scholarly environments.[3] To be collaborative, the contributions of each scholar should be made up of modular components, connectable, and extendible, such that the parts can be enhanced, repaired or replaced without damaging the network that comprises the whole — whatever it is that the whole turns out to be. The products of our scholarly activities should not aim to be finished mega-wholes to be looked at but not touched; instead, we should aim to contribute component parts to be worked with and enhanced. We should contribute to a growing work site, added to by many different scholars to create wide-ranging knowledge, where users can take control of their copies of the archives and editions to do with as they see fit, and not be restricted to uses and materials the developers saw fit to foresee. The payoff will be that instead of having electronic editions conceived, constructed, and completed by an editor and published like books to sit neglected on the shelf, electronic textual scholarship will occupy a communal space where the intellectual work of the future will be built on the textual foundations laid down electronically by the present generation.

To begin, however, I pay homage to a different approach to humanities computing and particularly to the development of electronic scholarly editions. Without this approach many of the astonishing developments in computing would not have occurred. This approach begins with a local problem and seeks a local solution to it. To begin with the sublime: given the works of Dante Gabriel Rossetti and given the capabilities of our technical assistants and our hardware, firmware, and software, what solutions can we imagine and develop? Could we not put a marvellously conceived and developed body of electronic images, texts, links and analytical tools into the hands of scholars, at least for those at the University of Virginia? To move to the ridiculous: Given the conditions of most people’s access to the World Wide Web, asked a Mr. Hart, would not Project Gutenberg put something, we don’t much care what, into the hands of many people who otherwise would have nothing? Or, less ridiculous but bordering on the obscene: Given that most libraries in the world do not have the resources or space to obtain and shelve the scientific and scholarly journals of the learned world, would it not be good to digitise or otherwise make electronically accessible full runs of these materials and sell subscriptions (i.e., rent them out) at exorbitant prices, which the libraries can pay for year after year and still have to pay again next year or by cancelling the subscription lose everything paid for so far?[4]

Identifying a problem and coming up with a viable solution to it in terms that are specific to the problem and to the focused vision of the proprietors of the problem and its solution has led not only to some marvellous innovations in humanities computing, it has almost always produced a solution that is local, limited, proprietary, and/or inapplicable to the problems of others. This approach has been fruitful and practical, and has frequently led to products that satisfy not only those who discovered and solved a problem but also the funding agencies that made the work possible. But it is not good enough.

If we want to develop a significantly different model for constructing electronic editions, it would be good to begin with a description of the problem — a description that is not driven primarily by a deadline or a desire to get one project finished and posted to the web. The problem of how to represent literary texts electronically begins with an attempt to understand how literary works exist.[5]

And then we need an environment for electronic texts and their analytical tagging that is open to fellow scholars.[6] And we need a tagging tool that will allow textual scholars (not highly trained technical assistants) to associate their analytical tagging as stand-off mark-up to already existing texts. Such a tool will also allow others to add tagging without affecting the files already on offer. And of course we will need tools to allow users to choose the tagging they wish to have affect their views of the text. Text files and variants files and annotations should no longer be mixed in a single file for which every attempt to add tagging or correct errors render the whole vulnerable to accident.

Both because verification of textual knowledge requires direct examination of sources and because the materiality of source texts is an important aspect of their cultural frame of reference, knowledge sites should begin with images, not just transcripts, of texts, and should extend to other products of analytical and interpretive scholarship. For safety’s sake knowledge sites should be distributed (and mirrored) widely. One essential tool that should accompany every file and every data set that links files is an authenticator that warns users when the content of a file has been contaminated or tampered with, but that tool should not prevent contamination or tampering. It should simply label the altered version as no longer authenticated.

And in imagining this collaborative environment of a growing body of data and of scholarly enhancement, we need always remember that the individual components must remain simple, though the aggregate may be complex, because individual, connectable - and therefore unconnectable - files are easier to maintain, transport, replace, and repair than are massive complex megalithic finished products aging and deteriorating as the community of scholars moves on to new things.

The component structure approach is beginning to show up in the DISCOVER project (growing out of the HYPER-Nietzsche Project); in the Pico della Mirandola project at Brown University; in the NINES projects, in NORA, and in JustInTimeMarkup (JITM) developed for Australian literature.[7] It is also beginning to affect the prototype pioneering projects like the Whitman Archive, the Blake archive, the Piers Plowman Archive, and the Cervantes project at Texas A&M. I do not think that the systems that we need for this kind of editing have been well developed yet, but we are getting closer. The big first job, of course, is to get scholarly editors to give up being control freaks and to begin thinking about working in collaborative communities. This is harder than it might seem because the print world did not require it; in fact discouraged it. We have to learn how to do it.

The history of scholarly editing is not famous for its attempts to make the fruits of textual research easily accessible to readers. Textual scholars are perhaps the only creatures on earth capable of being enthusiastic about textual apparatus. Jo Ann Boydston, addressing the Society of Textual Scholarship in the early 1990s, spoke “In Praise of Apparatus.” [8] She had just finished the monumental task of editing the works of the American philosopher John Dewey and was impressed by the importance and usefulness of the apparatus. Her paean to apparatus would have struck responsive chords in the hearts of German Historical/Critical editors, for many of whom the editor’s primary task is to construct an historical textual apparatus. The text itself, according to Gunter Martens, writing in 1975, could be dispensed with or published separately, since the real business of textual criticism was contained in the apparatus [Marten 1995]. The Germans might not, however, be happy with Professor Keith Elliott’s promotion of eclectic editing for the New Testament, though the editorial goals he proposes and calls “thorough-going editing” could not proceed except eclectically nor rendered scholarly without a carefully constructed textual apparatus [Elliott 2005, 47–58].

Textual scholars do not need to have the importance of the history of texts explained to them, though their commitments to textual criticism and textual histories are explainable in a variety of ways: some have an inherent interest in textual histories as part of the dynamics of composition, revision, and publication; others care more for the complex of social interactions that affect the textual results; others believe such histories help us to distinguish right from wrong readings; and still others care because they see in textual differences interpretive consequences unavailable to us when we read a single text in isolation.

Regardless of the reason, textual scholars are committed to textual histories as records and as aids to scholarship and to interpretation. Unfortunately, such commitments are often compromised by lack of will to do the work, lack of funds to publish the results, or lack of knowledge to implement a feasible electronic design. Most of all they are prevented from influencing the literate world at large by failures of imagination on how to convey the information and the significance of it to non-textual critics.

What follows resulted in part from frustration over the fact that so much effort goes into the construction of apparatus and, often, so little effort is put into using the apparatus during the writing of interpretive essays. But these arguments are also driven by the realization that electronic representation introduces changes of a radical kind, such that, one’s hope for an electronic surrogate for material texts is forever doomed, while one’s hope for a thorough understanding and enriched and dynamic interaction with literary works may best be accomplished electronically.

All literary critics acknowledge that for the study of literature we need texts. Some, though not all, also acknowledge that for the study of texts we need not just any old text but specific texts. But even then, some, God bless them, think that if they just use the text of a scholarly edition, it will save them having to actually know anything about the history of the text — that is, having to come to terms with the specifics of the text in the form of drafts, manuscripts, typescripts, proofs, magazine publications, books, revised editions, and the ways these are reported in scholarly editions. If we are studying the history of Shakespeare in the 18th century, however, it should, but often does not, go without saying that we will need 18th century editions to work from.

And we need other things that provide evidence of the social, political, economic, and daily life surrounding the production and reproduction of the texts we wish to study.

To conduct this business we need libraries and travel funds. We do not need computers. Computers are just a convenience, as my colleague Tony Edwards suggested at a conference in Leicester. But as Willard McCarty said to me afterward, yes, of course, one can walk from Leicester to London, and in that sense automobiles and trains are a convenience.

I have thought about that notion of convenience, because it seems not worth our while in electronic editions just to make things more convenient. We feel we have to provide something that could not otherwise be done — something that shows that the electronic world supersedes the print world. But actually, now, being able to get from Leicester to London in 1 hour and 15 minutes is something that cannot be done on foot. If, however, trains at the end of your journey dumped you out onto green park grass with straw in your hair, we might object that the convenience had its down sides, but it doesn’t. You get to St. Pancras station and walk out as if nothing unusual had taken place.

So, too, if the scholarly electronic edition, as a convenience, dumped a text in front of you smudged and rendered unrecognisable because its source was not acknowledged, its fonts and formats had all been changed, its covers and dust jackets had been torn off and discarded, its contexts of origin had been completely disregarded, and its words and punctuation damaged in both palpable and impalpable ways, we would say the convenience had its down sides. These are in fact the conditions of most electronic texts now available.

Convenience in texts has another very important role. In the age of print, scholarly editions have not been famous for their convenience. They are so expensive that most people do not own them, and there are rules and perhaps laws against taking a book from the library and dog-earing the pages, or writing in the margins, or, having done that, failing to return the book to the library. Hence, most literary criticism is conducted in the presence of convenient, cheap, often paperback, editions — texts with absent or false acknowledgements of sources, texts with no traces of original fonts, formats, or forms or even descriptions of those forms — texts, in short, with dubious claims to textual accuracy. And why are these the tools of choice for literary criticism? They are convenient.

Electronic formats have a chance of providing literary critics with convenience that is at least as good as getting on a train in Leicester and arriving in London with hair in place and creases intact. If one is inclined to complain about the trains, imagine how one feels about scholarly electronic editions, none of which has figured out how to be as convenient as a paperback —as cheap, as personalizable, as portable, or as dynamically interactive as a personally owned, printed book.

One of the problems is that scholarly electronic editions are the results of attempts to be as good as the scholarly print editions so few people use. If only we can be more complex, more innovative, more comprehensive — and one hopes more accurate and easier to use than a print scholarly edition — we would then, we think, have the electronic edition of the future. And of course we all know better than to aspire to be as good as a discount store paperback. We accomplished that abomination almost as soon as computers became available: its name is Project Gutenberg and its sisters, fan websites.[9]

But let’s not go too fast here. What do literary critics and students use for access to texts when they are not using their own cheap paperback copy with the turned down page corners, underlining, and marginal notes? They are using texts from Project Gutenberg — texts of sources unknown or misidentified, texts proofed cursorily, if at all, texts in random fonts and formats. And why do they do that? Because it is convenient — which includes the notion that they may have no viable choice because they don't have access to a good library and the Internet provides inferior goods, even when it is the only place to look (conveniently). It is not because they do not care about the quality of texts. No one, given an equal choice, would say, I prefer an inaccurate text of unknown origin. No one, or very few, I hope, given easy access to contextual information about the texts being studied, would say, I really do not want to know. But they proceed without knowing because it is convenient — or they rely on the most convenient source of information, the Internet, known generally as “The Source of All Truth” — and, unfortunately, “The Source of Much Else, as well.”

I was once told that the likelihood that a scholar or student will check the accuracy of a supposed fact is in inverse proportion to the distance that has to be travelled to do the checking. If it can be checked without getting up, high likelihood; across the room, probably but maybe not; out the door across the campus to the library, only if highly motivated. Why? Convenience.

I am not saying that no one uses scholarly editions; but, for example, fifteen years after the publication of Grindle and Gatrell’s magnificent edition of Thomas Hardy’s Tess of the d’Urbervilles, ninety per cent of new critical essays cited other editions.[10] This is not a unique case. I’m not saying that scholarly editions are always better than other editions; if the early volumes of the Ohio Browning edition live up to the denigrations of its scholarship found in the reviews[11] and in the explanation one of its own board members gave for his resignation, there must occasionally be good reasons not to use the scholarly edition.[12] And not all literary critics have a choice: when there is no scholarly edition, something else must do. It is not always the case that doing the convenient thing is bad, but it is natural, and scholarly editions have not had the effect they should have had on criticism primarily because they are not convenient either in price or structure.

Why aren't scholarly editions convenient? Because textual criticism is complex; because the print medium has required that textual scholarship’s complexity be compressed and made economical; because the textual condition of a 400-page novel entails not only the 150,000 to 300,000 thousand words on those pages, and not only the 1.5 to 3 million characters and spaces that represent those words, but also on the relation between the textual elements in the manuscript, the proofs, the magazine version, the first book appearance, and the revised edition or editions and perhaps more besides. Not only that, but understanding textual history entails knowledge of the contexts of textual origination — knowledge of what went without saying in the texts. It is not easy to deal with complex textual histories and contexts economically and compactly. That is why scholarly editions are not convenient. And that is why the convenience of a cheap paperback is “purchased” at such a great loss. It seems that many literary critics have no idea how much they are paying for their convenience. But let’s not try to reform them. Let’s admit that before scholarly editions begin to influence literary criticism, they must be made convenient.

Convenience for readers — for students and critics — is not the only goal. Convenience is a necessary goal for editors as well. Editing entails a variety of tasks, not all of which have to be intellectually challenging or have to entail reinvention of editorial architecture every time. Further, the quality of the textual criticism and editing need not depend on one person taking charge of all the details. Editors have traditionally thought themselves to be in charge of the quality of their editions, and they have been very protective of the fruits of their labour. With print editions, this has amounted to battles between the editor and the publisher and with what scholarly editors tend unfairly, perhaps, to refer to as the publisher’s minions. In electronic editions this protectiveness of editorial work reveals itself in the pains taken to make sure that users of the edition cannot appropriate the text by changing it. If you have a paperback edition and find an error, you cross it out and write in the correction. If you want to add a note, you do it in the margin. With an electronic edition, you can’t do either. Electronic editions are not easily appropriated, personalized or augmented by “mere users.” The convenient scholarly electronic edition may need to have a new structure to facilitate open-ended collaborative action that might lead to editions growing and changing even when the original editors have lost interest or died — even when they disapprove.

Furthermore, convenience for readers and editors is not good enough. Scholarly electronic editions need to be convenient for the technical experts who not only have to build them but who will have to maintain them, repair them, and transport them to future platforms and software. With more and more libraries creating digital repositories, we need editions that can easily be stored and maintained by digital librarians who simply cannot devote their lives to mastering the intricacies of each individual project — because if we don’t, the shelf life of electronic editions is too short to justify the investments of time, money, and intellectual labour. Print books on acid free paper last centuries; when we have figured out how to do that electronically, we will have something to crow about.

Before the goal of the convenient edition has been reached it will have to address readers, editors and technicians, both now and in the future. Falling short is not good enough. Giving a megalomaniac editor full control over the edition is not good enough.

The scholarly electronic edition of the future — the one that will actually be used and therefore influence literary study and criticism — will be convenient: it will be as cheap as a paperback book, with a user-friendly interface (adaptable by the user to suit his or her condition, whether the user is a scholar, a student, or a tourist), and will be treated as the user’s own, with bookmarks, highlighting, space for marginal notes, and the ability to annotate or even change the materials that appear on the screen in what must truly feel like the user’s very own private copy. It will be convenient both for the editors that build them and the scholars who augment them. It will be maintainable, with component parts that are replaceable and amenable to being supplemented as new data and new uses for textual data develop. It will be convenient, moreover for technicians — both now and in the future — adhering to standards except where the standards impose intolerable limitations.

In order to avoid the down sides of paperback books, the electronic edition must give accurate access to representations (images) of specific historical forms of the text and specific critical editions of the text and to the ancillary materials that contextualised the texts at the time of origin and the times of reception that we care about. It would be even more convenient if the accumulation of scholarship related to the work were also at hand. And to encourage that for the future, the site of the edition should be the knowledge site for the work and have a place for new scholarship to be attached to it. For such convenience to exist for readers, convenience must be built in for editors and technicians as well.

Parenthetically, Elsevier and his merchant brothers oppose these ideas. Any company that can make money from the ownership of knowledge as a commodity in whose making it had no hand and to whose growth it offers no support is a member of an evil empire that is our enemy. It would be convenient if the electronic knowledge site of the future — with its accurate and enhanced texts, dynamic work spaces, and potential for sharing in both development and production — put such companies out of business or transformed them from the exploitative sycophantic parasites they now are into useful and contributing citizens in the intellectual community.

Visions, particularly visions of the future, may be made of buzzwords. We are suspicious of these creatures, but perhaps they cannot be or even should not always be avoided — perhaps some “have legs.” [13]

We need a Collaborative Literary Research Electronic Environment. And for it to work as shared space, not in the absolute control of a general editor or other owner but available to scholars in the field, scholarly contributions to it must have modular structures in which the components contributed by one scholar can sit usefully and comfortably cheek by jowl next to the components contributed by a colleague, a rival, and an enemy. Each component should be, in this modular space, in some sense a completed whole, self-contained, capable of being moved, removed, replaced and repaired without damaging the integrity or dynamics of other components. Content data may need to be separated from its related analytical data and both may need to be separated from their currently enabling software.

Each module in the structure should connect to other components so that in the aggregate they act as a network of related parts and not only as a sum of independent elements.

Each connectible module should be extendible so that it can be the basis for enhancements of its own special contribution to the work site or knowledge site. If the component module is basically the representation of a specific text, for example, it should be extendible by allowing extra and different tagging to be associated with it and extra and related texts, and extra and related contextual materials, etc. If the component is a bibliography of reviews, it should be possible to associate supplements to that bibliography. Supplements must be connectible without disturbing the integrity of the component already there, running the risk of introducing corruption while trying to enhance the component. It is not good enough to say that in a data file, tagging can be added without affecting text. Accidents happen; tagging a text-file corrupts it.

As well as being available to users who simply want to do research on the materials in the knowledge site, each component and the network of the growing, non-proprietary, shared work environment network that is the knowledge site should be openly accessible to qualified scholars, with a mechanism for offering contributions of new scholarship. As scholarly editors in the electronic arena, we cannot focus all our attention on rearranging the furniture in the existing rail cars. We need to know that different arrangements, valued by other scholars, must be possible. We need to know that when a component is damaged or rendered obsolete by new discoveries or rendered inadequate by the emergence of new questions, the damaged or inadequate component can be repaired, enhanced, or supplemented without damage to the whole knowledge site. We need to know that the electronic repositories of our literary scholarship will not, through complex or eccentric design, seem like too much trouble to port into developing software on developing platforms.

I end with the story of the English Poetry Research Carrel, which is one idea of how a collaborative Literary Research Electronic Environment would make a literary work of art look.

In Timbuktu (or Leicester) J. Q. Scholar sits, bent upon learning what can be known of John Keats’s “Ode on a Grecian Urn.” There is no research library nearby. Curiosity is stimulated but not satiated by the books on Keats that are at hand.

Scholar opens a web browser search tool and types John Keats’s “Ode on a Grecian Urn” and clicks on the first item: English Poetry Research Carrel, which contains a list of poets. Scholar clicks on John Keats. A web page appears, looking like a desk surrounded by books with a number of options for finding Keats’s poem and materials about it. Available items include four manuscripts in facsimile and transcribed, none in Keats’s own hand (none exist), a magazine appearance, two book appearances in Keats’s lifetime, a scholarly edition, introduced, edited, and annotated by Jack Stillinger. In addition, there are biographies, bibliographies, and commentaries in “full text.” J. Q. Scholar can with one click find, open, and begin reading the poem. Any curiosity about the particular edition is satisfied in a small window. A desire to know if any editions differ from the one open is satisfied with a click, producing either textual variants or a variant text — either as image or transcript. A bibliography of commentary on the poem is a click away with links to full texts more often than not. Reader curiosity is what drives access. A search engine across all parts of the carrel, but capable also of going into the Internet at large, will help give the reader a sense of control over the array of options. And on the virtual desk will be the tools for extracting, quoting, highlighting, and selecting texts and for personalising the Keats Knowledge Site, for adding things from outside the Carrel, for storing Scholar’s own contributions to his personalized site, and for building the scholar’s own new scholarly work on Keats’s poem.

Scholar, who is of no certain age, located anywhere — Timbuktu or OtherSimilarPlace — could be fifteen or fifty-five, a seasoned or a beginning searcher. The ways into the carrel should be already familiar and intuitive, the depth and sophistication of entry should be the user’s choice.

This is fiction because, to my knowledge, no such carrel and no such site exist. Our goal is to try to construct the electronic infrastructure that will support the literary and scholarly content and the navigational and analytical tools that will provide a place for investigation and dynamic interaction with the Carrel’s English poetry materials.