1998 Joseph Leiter Lecture

The virtual library: an oxymoron?*

By Jean-Claude Guédon, Département de Littérature comparée, Université de Montréal, CP 6128, Succursale "Centreville", Montréal, Quebec H3C 3J7 Canada

 

Introduction What Do "Digital" and "Virtual" Mean? The Political Economy of Print What Political Economy for Digitized Documents? The Virtual Library and the Dynamics of Digital Documents Conclusion

Introduction
Large cities generate ever-fluctuating landscapes and Paris is certainly no exception to the rule. Recently, the southeastern side of the city saw a mammoth piece of architecture rise over a portion of land that, until then, had been distinguished mainly by railroad yards, warehouses, and other eye sores of the industrial age. Tourists had never paid much attention to the Tolbiac and Bercy sector of the City of Lights. However, and quite suddenly, a new structure was transforming the landscape. The Bibliothèque Nationale de France, the last monument built in honor of French President Mitterrand, was seeing the light of day. While 1998 is slowly drifting by, the library is receiving truckload after truckload of books. Between ten and twelve million volumes will thus move across the city and find an entirely new environment. For how long? That question is best left to the next century, especially if we begin to think about another trend that is not quite as exciting: libraries, hospital libraries in particular, are being phased out on the strict basis of narrow economic arguments.

Are libraries still multiplying or are they on the way out? On the one hand, Paris, but also San Francisco, London, and Vancouver lead to optimistic predictions. Even Montreal is discussing a large project of this kind, although the financial context of Quebec is anything but favorable. Amidst all the discussions about the possible disappearance of the printed book, libraries keep on multiplying and growing as if none of this really mattered. On the other hand, the cost of documentation, the way digitized documents are being commercialized, and the general feeling of helplessness in front of poorly understood phenomena lead to understandable fears.

What does it all mean? Should we compare the several large projects mentioned earlier to the final flare of a star going nova before collapsing into oblivion? How do they relate to all the heady talk about virtual libraries? How do the latter relate to the parallel closing down of various categories of libraries? Where is the hype and where is the reality?

Behind its monumental façade, the answer Paris provides to these questions is not totally obvious. Although quite striking, the architecture of the new building harbors ambiguities. Shaped like a parallelepiped delineated by four angular towers—four open books perhaps—the bibliothèque sketches the outline of an enormous, yet virtual, box against the sky. More subtly, the eye, in striving to actualize the suggested structure, cannot decide which way this incompleteness works. Do the potential dimensions of the new national library point toward further growth? On the contrary, do they pull our gaze downward, all the way to a gradual dissolving of the library structure? Could this be the consequence of the digital onslaught of zeroes and ones? Wittingly or not, Perrault, the architect, injected ambiguity into his latest achievement, as if to express a degree of ambivalence and even anxiety.

What Do "Digital" and "Virtual" Mean?
Digital libraries, virtual libraries: these terms are often used interchangeably although they should not be. The former refers to a mode of coding, while the latter points to a process, an evolution. In its present binary form, the word "digital" refers to the ultimate attainable form of encoding because it relies on the absolute minimum of two signs. As such, it appears as the limit of the process that led humanity from the largely emblematic cave paintings found, for example, in Alta Mira or Lascaux, through the first ideograms that later emerged, to syllabaries and full alphabets [1–2]. With the growing consciousness that signs were arbitrary entities, came the realization that entirely artificial systems could be designed for special purposes: hand signals for deaf people, for example, or the Braille system for blind people. The latter, incidentally, really corresponds to a six-bit system of writing. As Braille libraries began to appear in the nineteenth century, digital libraries have actually been around for over a century.

The word "virtual," as in the phrase "virtual library," refers to an order of issues altogether different from that of digitization. In particular, we must not confuse virtual with "unreal". Despite appearances, virtual reality is not an oxymoron because virtual is opposed to actual, not to real.† The virtual is nothing but potential and, as such, it is reality (possibly) in the making [3]. Great medieval thinkers in dealing with the problem of the virtual (or potential were not probing the static nature of reality; instead, they were raising a profound question: how does being come into being? How can the seed give a full plant? How could God create? The virtual refers to a dynamic situation, to a passage from some state to another, to a flux. By contrast, the opposition between real and unreal remains static, like a tightly guarded border.

While attempting to avoid sounding like a scholastic of the worst kind and speculating on the number of angels virtually dancing on pins, it might be useful, however, to reflect on the number and kinds of libraries that can hang on to a string of bits, and on the implications of this apparently strange question for the future of these precious institutions.

To be meaningful, the phrase "virtual library" must point in the direction of some kind of activity or transition. A virtual library thus carries the tense promise of its actualization. For example, when printed books came upon the scene, book collections quickly appeared, but these did not yet qualify as libraries as we understand them. At best, these collections were but the virtual promise of what was to develop quite a bit later. The historical trajectory leading from (largely private) book collections to their public counterparts recounts the actualization process leading to the modern library. Interestingly, the visions and values attached to libraries were deeply transformed, but not the form. By contrast, present-day libraries attempt to build collections or provide access to some digital materials, but they have not found the right recipe or recipes to reconstitute library functions and roles around the new materials. In other words, they do not yet know how to actualize themselves as digital libraries around these new materials, and, in particular, they do not know how to change form without, at the same time, changing values. Preserving values requires putting material change in its place, so to speak.

Electrons are no better at the virtual stance than paper is, but as they are the "new kids on the block," and as we are all impressed by their novelty, we become confused. Consequently, we often treat digital encoding as if it incarnated the virtual dimension in toto, while, in reality, it has simply, but deeply, transformed what we could virtually expect of the future. In other words, digitization and networks change the rules of the documentary game. What remains to be seen is whether the former objectives can still be maintained within, or even despite, the new framework. And if not, what translation rules should apply if a few fundamental values are to be preserved?

While trying to actualize within the new context of digital documents, libraries must remain conscious of the wide spectrum of material forms upon which their documentary collections rest. This consciousness is what I meant when I mentioned that we needed to keep material change in its place. To exist, a message needs some material substratum, be it clay tablets, wood bark, stone, papyrus, parchment, paper, magnetic tape, etc. It also needs some organizational scheme, be it columns, lines, or cells.‡ Libraries must take advantage of these forms to make good use of all their potential in such a way as to preserve and even enhance their vision, their objectives, and their fundamental institutional values.

In the case of printed books, libraries actualize their service potential in various ways, for example, through lending and through direct browsing access to the stacks. With digital materials, libraries must learn to actualize the basic functions of their libraries through other kinds of developments that will take into account the peculiar material characteristics of digitized documents. Librarians of the future will thus need to distinguish signs from matter and yet not treat the latter as a pure and neutral instrument with regard to the former. Indeed, changing the nature of the information vector leads to profound modifications in what might be termed the social and institutional life of documents. When, in a celebrated passage, Hugo has one of his characters declare that the book will kill the cathedral, he strikingly underscores the fact that a cathedral does not behave like a book! For one thing, it is far less portable. He also underscored the fact that new vector forms could strongly push older forms aside to the point of making them largely irrelevant to most people's concerns.

Some illustrations help clarify the import of Hugo's idea. For one thing, a library of scrolls does not work like a library filled with printed codex books. Reading a scroll requires both hands, so that the simple task of taking notes requires two people. This fact implies different logistics, a certain level of noise, etc. A codex, by contrast, can be used and copied alone and the first seeds of silent reading probably coincided with these developments. Various forms of text segmentations that ultimately led to word separation, paragraph arrangement, and punctuation also began to emerge, probably as a way to make the interface more convivial. In actuality, the historical situation was a good deal messier and certainly less "progressive" than space allows here. Martin offers a nice balance between synthesis and details in his excellent and fascinating Histoire et pouvoirs de l'écrit, first published in 1988 [4]. Elements of texts can also be retrieved more easily in a codex, because flipping through pages is faster than scrolling. As a result, bits of texts came to be associated with facts. Life would have been more difficult for Comte and all the positivists, had they worked with scrolls. In effect, the advent of the codex opened the door to early forms of information surfing and the ability to browse clearly distinguished the codex from the linear form of reading that the roll of papyrus strictly enforced.

In contributing to the transformation of reading habits, codices also affected the functions of reading. For example, imagining dictionaries without the codex book would have been very hard indeed. Collections of codices, however clumsily, began to behave like a data bank because of their peculiar characteristics. Also, because the codex was cheaper, it probably favored a wider diffusion of information. In fact, the groups that first made an effective use of the codex must have enjoyed some kind of informational advantage leading to more prominence and power.§ In the process, texts became more external to the mental processes of the reader. Starting as memory prompters, in particular for the poets, texts ended up being the augmentation tool (to use the vocabulary of Englebart) of the philosopher [5]. In any case, texts placed within codices certainly served the cause of critical and analytical attitudes and we all know how central they are to our educational values.** Therefore, the codex opened vistas and libraries had to learn how to accommodate them. In fact, libraries gradually invented themselves around these developments. Digitized documents have raised the same kind of challenge. In short, whole cultures tend to mutate when messages sport a new material form because messages take on a different meaning and are used differently when their material appearance changes. There is more to cheap pocketbooks than meets the eye and there will be more to digital documents than can be anticipated, particularly when we think about the future of libraries.

Libraries have always harbored a virtual side and it has largely to do with their ways of dealing and even negotiating with peculiar kinds of documents. Let us remember that the parchment manuscript, because it was so rare, was chained to the wall as if it were a unique piece of art. On the other hand, the cheap, glued paperback probably eased the transition to open stacks wherever that right had not yet been acquired. With digital documents, libraries must redefine themselves. Librarians must design new modes of access, new forms of storage, new indexing tools, and new ways to reach constituencies that transcend geographical and even national boundaries. They must think differently about retrieval and usage. They may even, as we shall see shortly, take on new functions and new roles. In short, in thinking anew about most of the fundamental mechanics of their institutions, as well as their functions, librarians are pushing them from a virtual to an actual stage. This change is what going digital really means. It is no mean job!

The Political Economy of Print
We have had printed codices around for so long now that, like good friends' faces, we do not see their shape any longer. We just recognize them. Yet, the history of the printed volume, as the likes of Febvre, Martin, Eisenstein and, more recently, Chartier [6–8] have documented, is accompanied by major cultural and political changes. These changes would have evolved quite differently, had the historical rendezvous with print been missed.

No need here to fall into the traps of technological determinism or, conversely, to treat technology as essentially neutral because of some phenomenal instrumental adaptability. We now know that some degree of intention lies at the heart of technical design, but that this virtual stage-the term "virtual" applies well here-never enjoys perfect actualization. Technical objects always achieve a little more and/or a little less than planned; also, they are always used differently from what had been anticipated. In fact, the history of the telephone and that of radio demonstrate that the gap between intended and actual usage can be quite wide. The career of the Internet provides another good example of these complex systemic transformations that are about as deterministic as the weather system, and just as difficult to predict. Digitized documents will not fault these commonsensical observations.

The familiarity of printed books leads us to treat them almost as if they were natural objects, but the best historians manage to restore a sense of strangeness and novelty to these old friends. They remind us that the printing press ultimately rested on the mastery of hard steel to punch the matrices. These, in turn, mold fonts made of a complex alloy based on tin, lead, and antimony. The metallurgical skills needed to manage a three-tier system of material hardness were anything but casual in the fifteenth century. In other words, printing with moveable types required access to a kind of high tech that only princes could reliably manage. They were part of the very expertise lying at the heart of weapon building and money making. Yet, despite the exotic skills on which it rested, printing did not require the complex vertical integration of resources that characterized the production of manuscripts, from the production of sheepskin to the organization of scriptoria. Neither did it require a huge capital outlay. Setting up a press was not simple, but it was easier than setting up a Cistercian monastery. In fact, printing took power away from the Church and, to some extent, from princes and kings.

Although the earliest printed books were as good at mimicking manuscripts as our present-day electronic journals are at imitating print journals, they could not avoid incorporating a number of innovations that ultimately affected their career in a big way. Multiplying texts by the hundreds or by the thousands cannot be planned and justified in the same fashion as scriptorium-style copying. Manuscripts were produced largely in response to direct institutional demands and needs so that financing was almost never in doubt, despite the high unit cost. This system allowed for a great deal of control of publishing, but when the production of printed books began, they faced an uncertain economic viability because no adequate economic model existed. The price of printed documents, although still high, was lowered so dramatically that they became accessible to many more people. Therefore, the printer could no longer tally his customers ahead of time. In response to this increased uncertainty, printers developed an abstract market model that eventually provided economic viability to the fledgling printing industry. The need to advertise the new wares urbi et orbi led to the creation of catalogs and indexing tools, as well as reviews and evaluations that gradually transformed the whole social meaning of text production. Literature as we now know it today, that is to say with its appendage of bibliographies and critical reviews, is simply unthinkable without print. More generally, the literature of a field or a domain was reshaped. New kinds of collections appeared, as well as new kinds of collective memories. Libraries, first conceived as princely collections, began to emerge. The relationship between knowledge and power became far more visible and Bacon, in any case, did what he had to do to explain it to his contemporaries.

The intervention of market mechanisms also created a truly novel situation for the social meaning of documents: piety, knowledge, and beauty found themselves strapped with a price tag. Documents could no longer be viewed purely as documents; they also began to appear as commodities. In effect, print placed all documents in a situation somewhat analogous to that of a play. A written play remains virtual theater until performance actualizes it. However, the actualization of the play through a performance requires some form of economic intervention, be it the support of a patron of the arts or the entrance fee to some theater, such as the Globe in London. Likewise, a text remains in a virtual state until it is actualized through reading. However, in the case of print, actualization generally requires buying, or sometimes renting, the text. Ultimately, libraries managed to offer free access to some printed materials, but this resulted from revolutions, be they political or industrial, and not from a sense of equity or justice suddenly expressed by publishers. On the contrary, and to this day, librarians and publishers often find themselves at odds as their respective agendas regularly clash. The clash is intensifying with the advent of digitization because publishers hope to take advantage of the changing context to negotiate better conditions. Digitization and the Internet do not solve social problems; they simply redefine battlefields.

Commodities, controls: all these terms quickly lead to situations where law and courts of justice are rapidly involved. When texts were condemned, yet managed to circulate, who was responsible? Were the writers? However, much of what they wrote was probably culled and adapted from other texts and their responsibility was ambiguous at best. Brodman†† recounts that the first indexes were organized by Christian names and not family names, a telling comment on the author's uncertain status [9]. Could the printers be responsible? Possibly so, but were they always fully aware of the content and the import of the texts they printed? Could the bookstore owners or the stationers be held accountable? However, what applies to the printer fits the situation of the bookstore owners even better. What of the readers who read forbidden documents? Were they the owners? Did they seek to transgress the law or did they come upon a forbidden text unwittingly? A close examination of European history between the fifteenth and the eighteenth centuries would reveal that all solutions were used at one time or another, often with fatal consequences for the individual defined as the culprit. Many, like Dolet (who was burnt in 1546) [10], went up in sinister smoke.

Today, laws tend to be less dire in their consequences; yet, they face similar dilemmas. If creators of documents are to be held accountable, what does creating a digital document mean, given the amazing ease with which anything can be digitized, cut, and pasted into another document? Should the law aim at network owners? Yet, they have little or no way to know what is passing through their pipes. Should service providers be eyed suspiciously? Indeed, in several countries, law enforcement officers often target them, but this may reflect their greater visibility more than their real guilt. Colporteurs in the 1550s were also burnt with regularity for the same kind of reason. Finally, what of end users? Are they stumbling upon unacceptable documents by serendipity, or were they actively seeking to break the law? Now as then, legal systems everywhere have difficulties identifying the nature of the beast. Moreover, their territorial limitations do not allow them to deal adequately with a truly global phenomenon. The flow of printed books appears simple by comparison.

In the case of print, writers ultimately had to accept the largest share of the responsibility. However, this situation could have killed the nascent industry, had it not been quickly compensated for by the benefits of royal privileges. These gradually evolved into copyright laws, or authors' rights (both property and moral), as they are known in several countries. Thus emerged the "author" in the modern sense of the word, in other words, an individual who owned "intellectual property." This new legal and economic status appeared in the seventeenth century, but it reached maturity only with the Enlightenment. In terms of human history, the author is a recent and complex legal construction that subtly harmonizes intellectual requisites with commercial concerns. In effect, the author is a kind of Janus Bifrons that can alternatively sport the appearance of the pure intellect or that of the entrepreneur—a form of mild schizophrenia providentially cushioned by the editor, for a fee.

Successful market models were slowly refined and improved as various business plans were tried and tested. These ranged from subscriptions to royalties and, as a rule, texts were printed only if economic viability could be reasonably predicted. The economic dilemmas of the early print years have been largely forgotten; yet, they closely parallel those we are presently witnessing in the digital area and they are quite instructive to consider. What is remarkable in the case of print is that a market model gradually invaded all dimensions of publishing, irrespective of the nature of the documents being exchanged. In effect, all of printing eventually succumbed to a commodity model so that, currently, many people experience difficulties in conceiving of texts dissociated from a money-driven market.

The case of scholarly journals is exemplary in this regard. We know that they emerged as a way to relieve the pressure felt by the epistolary giants of the so-called "Republic of Letters." To make a long story short, the Republic of Letters suffered from insufficient bandwidth despite the heroic feats of a few letter writers like Peiresc and Mersenne in France. The printed periodical was invented as a way to improve the distribution of news.‡‡

From the very beginning, journal publishing adopted either one of two economic forms: while de Sallo immediately favored a commercial approach for the Journal des sçavans, the Royal Society in London launched a barter system for its Transactions. In this fashion, one could build a library of publications for the cost of a single journal. No market other than a market for ideas was involved. The concept was brilliant and people involved with scholarly electronic publishing nowadays would do well to revisit it, but it ultimately faced various difficulties ranging from unacceptable delays to censorship [11]. In the nineteenth century, commercial publishers took advantage of those difficulties and they proposed their services to various segments of the research community. In this fashion, they managed to gain a foothold into scholarly publishing and they contributed to generalizing the pricing model that treated the research text like any other printed text, in other words, as a commodity. Nowadays, they essentially control the pricing trends of scholarly publications, with grievous results for the libraries of the world.

One of the long-range consequences of printing was thus to homogenize all documents into commodities, irrespective of the domains covered by the text. To be sure, upon finer analysis, one obviously does not sell a novel like an encyclopedia or a learned journal. But neither does one sell a Mercedes like a Chevrolet or a Rolls Royce and the fact remains that selling motives, with few exceptions, have tended to override all other considerations, irrespective of the nature of the printed document. Libraries, moreover, have been limited to reacting and adapting to a situation they do not at all control.

Interestingly, digitizing challenges the universal claims of the commercial model, particularly in the case of scholarly publications. Digitized birds are notoriously difficult to keep behind bars. Likewise, digital documents can be duplicated and transmitted extremely easily. In effect, the technology no longer contributes to the strict control of scarcity that stands at the heart of any pricing system within a liberal economic scheme. With digital materials, the engineering of scarcity becomes tricky at best, impossible at worst. Consequently, justifying its existence becomes that much more difficult, especially in those categories of texts that have been submitted to it through what amounts to a historical fluke. Why should research scientists or doctors accept artificial restrictions on the dissemination of their research if their primary concern is not making money, but attaining maximum exposure and visibility? As a rule, scholars are more concerned with symbolic capital, as Bourdieu would put it, than with capital proper. Moreover, research is supported mostly by public money and researchers know that publicly supported libraries buy most of the copies of the journals where researchers publish. If governments end up buying from commercial publishers what they often help finance in the first place, then Martin Luther King was right when he declared that capitalism was socialism for the rich and capitalism for the poor! Let us remember that Reed Elsevier boasted a 35% profit rate on scholarly journals in 1995 [12]. Actually, many governments could save a fair amount of money if they bypassed the likes of Reed Elsevier. However, this would also require that research scientists and scholars, on the one hand, and libraries, on the other hand, strike a new pact to redesign scholarly communication to their mutual advantage.

What Political Economy for Digitized Documents
Scholarly electronic publishing is still very much in its infancy. The first electronic scholarly journals such as Harnad's Psycoloquy are less than ten years old. Most other individual projects are even younger while commercial endeavors only began three or four years ago. As set against the backdrop of the evolution of the print age, the present phase of electronic publishing belongs to "incunabula 2.0." As was the case with print in its early days, we are floundering about with various production schemes and economic models even as we try to set up a viable alternative to print. At the same time, financial pressures on many noncommercial editors, particularly university presses, public research laboratories, and learned societies, are steadily increasing. Together, these difficulties and trends clearly point to the need for an alternative economic and technological model and computer technology now appears mature and rich enough to entice us with a credible promise.

Behind the turbulent and effervescent landscape of electronic publishing—more than 6,000 journals are presently listed in the Newjour database maintained by O'Donnell and Okerson§§—a latent unspoken consensus has been emerging about the nature of digitized document. What is fascinating is that this consensus leads to starkly contrasting attitudes to access philosophies. In effect, all players now recognize that digitized documents are too "volatile" to obey traditional marketing rules. They also silently agree that we really do not know very well how to market and sell digital wares. Consequently, some, and I belong to this group, argue that wherever public money is involved, the recourse to market should be eliminated or, at the very least, placed at the lowest possible level. Others, intent on preserving the profits derived from the captive market made up by the research libraries of the world, are trying to revolutionize marketing techniques and thus to ensure a wonderful future to their stockholders.

The only success story that is readily available in "Digitown" is the software industry, and even it is not devoid of many ambiguities. We now know that protecting software is largely ineffective, but only after that solution was unsuccessfully tried a decade ago; on the other hand, planned obsolescence works. However, this may be temporary too as the free source code movement may ultimately ruin such tactics. Linux, and not Windows NT,*** may well become the operating system of choice within the next five years. But for all of its problems, the dilemmas facing the "digital birds" of the software industry appear slight compared to those faced by the publishing industry, be they scientific or literary. Imagine trying to engineer the planned obsolescence of a novel by Dickens or the special relativity papers by Einstein! Imagine promoting Underworld 2.0 by De Lillo in 1999… This would certainly set an entirely new meaning to the stock phrase "Second edition revised and augmented." And someone may find the way to make it acceptable, thereby bringing an entirely new turn to the literary enterprise.

These dilemmas run very, very deep. They largely explain why commercial vendors, normally a traditional and somewhat conservative lot, find themselves in a somewhat surprising position. They are now actively subverting traditional sales practices in ways that only the word "revolution" begins to capture. Even as great efforts are being expended at the World Intellectual Property Organization (WIPO) to tighten and harmonize national copyright laws, big publishers such as Reed Elsevier are effectively moving the battlefield elsewhere. Part of the revolution lies in the fact that copyright laws may become secondary or even largely irrelevant to the future of electronic publishing. Commercial publishers simply consider regulatory laws largely unenforceable and, silently, they have decided to change the basic rules of their trade. Okerson, assistant librarian at Yale University, masterfully depicted the new landscape and some of its implications at a conference held in Caen, France, in April 1998.

Giving up on regulatory laws amounts to moving that much closer to some analog of the Wild West where contracts enforced by six-shooters were the rule. Apparently, many commercial vendors have tacitly decided to do just that, as their growing fondness for individualized contracts testifies. Often, in this new and radically transformed context, libraries no longer own what they pay for. They only buy a right of access to various kinds of materials for a limited time. With digital materials, libraries are reduced to haggling specific terms of use for documents that are reduced to the status of a databank. Who owns the information? Who can use it? How may it be used? What is a site? The contractual terms offered by large commercial publishers are complex to the point of being arcane. At the same time, they effectively define an entirely new political economy for scientific information.

Alas, this brave new world of electronic publishing bears more resemblance to cattle wrangling than to an orderly bookstore. Actually, it is even worse than cattle wrangling because we are stuck with shrink-wrapped cows, accompanied to market by legions of lawyers. Libraries, singly or united within consortia, meet the vendors, similarly equipped. Lawyers, in effect, are the great equalizers of today and thus unexpectedly appear as the modern (but drab) equivalent of the six-shooter. Joking aside, these startling comparisons help us plumb the deepening gap separating the traditional forms of book sales from those shaping up for the digital age.

If you feel jolted by these unlikely metaphors, console yourself with the following thought: consider the bewilderment of your standard scriptorium leader when the new print boys first galloped into the local saloon. But also remember that these shifts correspond to a strategy designed to put librarians in their place, that is to reduce them to a subservient status, equivalent to that of a gas pump. From the vantage of Reed Elsevier, the ideal library pays the right to pump knowledge. Good librarians do so without batting an eyelash and they do not even try to own a local tank: all they need is a nozzle, a pipe, and a cash register!

For noncommercial publishers, the irrepressible character of the famed digital bird can lead to a position that stands entirely opposite to that of commercial publishers. If Reed Elsevier symbolizes the new, incredibly tight, commercial context in a sinister, yet exemplary, manner, Ginsparg, at Los Alamos National Laboratory, offers an alternative implementation of electronic publishing that is just as radical in its implications.††† Ginsparg correctly reasoned that, because actual communication between scientists takes place at the pre-print phase, a free database of such texts would be most welcome. He quickly set one up in high-energy physics, the field he knew best, and proceeded to demonstrate that it would work. And he succeeded. Physicists very quickly recognized its great usefulness and started using it in large numbers. The quantity of pre-prints also began to grow very rapidly. At first based on volunteer work, the Los Alamos database eventually secured significant financial support from the National Science Foundation (NSF) and it continues to play an important, perhaps a determinant, role in defining the future of electronic publishing.

The success of the Ginsparg project demonstrated the real importance, or rather lack thereof, of the traditional commercial vendors. It led to some very interesting discussions about the basic functions of scholarly publishing [13]. Most people quickly agreed that the Ginsparg server certainly fulfilled the communication and the archival needs of physicists. However, the questions of legitimacy, authority, and prestige turned out to be thornier. In my opinion, it is the crucial point that has to be solved by electronic scholarly publishing if it is to succeed, and commercial vendors are doing all they can to corner that facet of scholarly transactions. In the case of Ginsparg's server, legitimacy was provided in a somewhat unorthodox manner. Rather than resting on the label of a particular journal, reputation was established more by direct measurement of interest and commentaries. Paper journals, on the other hand, are capable of providing prestige and visibility, although they contribute little to actual communication, being too slow. Also, libraries, and not vendors, have been tending the archiving of these commercial journals, most of the time at public expense. In general, what could easily be labeled as the Ginsparg challenge contributed to revealing an important point: commercial publishing was not at all indispensable. In fact, the foundation for the role played by large commercial vendors seemed to reflect various forms of institutional power rather than to respond to real needs. Those who own and/or control prestigious titles can influence the pecking orders of research establishments and even whole countries. However, Ginsparg's move opened new vistas and liberated imaginations. Nothing stands in the way of imagining alternate paths to legitimacy, authority, visibility, and prestige.

Somewhere between the large commercial publishers feeling their way in the new digital environment and the strategies symbolized here by Ginsparg, university presses and learned societies have tried to define some sort of middle ground. Alas, their attitude is not dictated by a vision that can be called theirs; instead, they are merely reacting pragmatically to very difficult financial situations. In most cases, their covering institutions tell them in no uncertain terms that they should not lose any money. In fact, if they could earn some, it would be quite welcome. This situation has led them to practice commercialism with restraints, so to speak. In most cases, subscriptions are retained but cost recovery, rather than profit, stands as the central preoccupation. University presses do not seek profit rates as outrageous as some commercial publishers and they generally do not object to libraries actually owning what they buy. The Project Muse at Johns Hopkins University and, more recently, the SPARC project at the Association of Research Libraries (ARL) provide good examples of these attitudes, as does the electronic division launched last year at the Presses de l'Université de Montréal.‡‡‡ What is most striking in these developments is that they all involve innovative alliances where the library appears as a somewhat unexpected or even surprising partner. Yet, these developments may point to a most important point-namely that the advent of digital publishing inexorably causes libraries to become part of the publishing process. This idea alone opens up exciting new vistas. More than that, it is the most important thesis of this paper.

The Virtual Library and the Dynamics of Digital Documents
Between the relentless quest for profits pursued by callous commercial interests and the libertarian aspirations expressed and even implemented by a vocal, but limited, segment of the research community, what can libraries do? What role should they play? Indeed, they face a very difficult task. While document costs grow at an alarming rate that far outstrips inflation, libraries must answer to a somewhat schizophrenic constituency of scholars. It is schizophrenic because the same individuals who complain about local subscription cuts will nonetheless submit their latest articles to the most prestigious journal possible, with no concern whatsoever for its subscription price. As users, scholars demand the best collection possible. As authors, they insist on the greatest visibility as well as the highest prestige possible. Smartness does not preclude shortsightedness so that, out of selfishness or plain laziness, they rarely devote any time or effort toward understanding the connection between these two phases of their working life. Thus, they completely miss the point that their concern for personal visibility indirectly costs libraries a pretty penny. They also miss the point that this system exists only because it helps maintain a highly profitable confusion between a form of economic elitism and true excellence. Not only does it accentuate the difference between poorer and richer institutions within rich countries, but it also deepens the gap between poor and rich countries. Some people wax lyrical when it comes to the "knowledge economy," but if it rests on mechanisms that ensure unequal access to research results, there is little reason to rejoice.

Because libraries are too often regarded as a mere service and not as an essential part of research institutions, librarians tend to remain segregated from the research community in a fundamental way. There lies the main obstacle preventing libraries from playing an important role in redefining the political economy of knowledge. This situation hampers their involvement in various electronic publishing experiments, despite the presence of the few promising exceptions mentioned earlier. Research administrators must become conscious of this obstacle and work toward removing it. If they think this work is too arduous, they should remember how difficult it was to make a humanist work alongside a metallurgist in the fifteenth century. Luckily, the social distance separating researchers from librarians is not nearly as great nowadays. In any case, electronic publishing requires new types of collaborations. It will rest on divisions of labor that are not yet invented.

Librarians must be deeply involved in the designing of scholarly electronic publications for a number of important reasons. First, they have at their disposal sorely needed ammunition. As buyers of the research documentation, they collectively hold large amounts of cash that the greedy eyes of the commercial publishers have accurately targeted. By gradually diverting more of these funds into collaborative electronic publishing and by fostering the sharing of the newly created resources, libraries can usefully support the gradual growth of a communication system that will truly serve the whole research community. They can help liberate the scholarly publishing enterprise from the deadly grip of the large commercial vendors. However, to reach this goal, librarians need the active support of scholars and they need to devote some time and energy to educate and build a favorable constituency. Together, scholars and librarians must find ways to produce forms of prestige and visibility that truly respond to the legitimate urge for excellence and that are equal to the highest forms of prestige that commercial journals currently provide. Work remains to be done on that front and the new stable recipe has not yet fully emerged. It is probable that the so-called Harvard system of evaluation is part of the equation. In part because it relaxes the quantitative pressure in favor of a qualitative pressure in such a way that candidates under review are asked to list only their best publications (often five best) rather than all publications. It may also be that evaluations for career and prestige purposes should be treated independently of publishing evaluations. The point, in any case, is to design communication systems that stand in better harmony with the aims and practices of various research communities, and scholars should recognize the crucial role libraries play in this regard.

Librarians should also be involved in electronic publishing because they have stakes in the preservation of documents. At present, the archiving of many digitized documents accessed through contractual licensing remains an open issue; in many cases, publishers themselves have the sole responsibility of licensing and that state is as new as it is worrisome. Are we going to see publishing houses spawning archival companies, perhaps modeled after UMI's dissertation archives? However, can we trust private companies to ensure continued access to certain types of documents for decades, or even centuries in some instances? They may go belly up at any time or they may simply change their internal policies and suddenly decide to drop the preservation of this or that class of documents simply because it no longer generates sufficient profits. By helping develop a communication system of research that is largely independent of commercial considerations, librarians will be in a better position to influence the preservation of digital materials. They will also stand a better chance to influence the choice of norms and the design of procedures to take good care of our collective memory.

Finally, libraries should also get involved in electronic publishing because, in the case of digital materials, publishing naturally merges with providing access—the central function of libraries. One may argue that libraries have no particular expertise or tradition in the area of editing, but this points to the need of collaborating with university presses and with scholars. Libraries need not edit; but they can publish and they can provide access. After all, libraries have been unwittingly, and somewhat covertly, publishing for decades. Note taking by hand is a form of publishing, however inefficient it may be. With photocopying machines, libraries began to publish in earnest. Incidentally, this equipment became very popular because it did to note copying what the facsimile did to the memo. In any case, this distinction between publishing and access, on the one hand, and selecting and editing, on the other hand, points to a good and fruitful division of labor within the new publishing context. That division too is a hopeful sign for the future.

Conclusion
When we try to understand how technologies affect the institutions where we work, we should never lose sight of our own goals. We also affect our own institutions and the ways in which we translate our agendas within the new technological settings may be of the essence. It is also important to learn how to listen to weak signals. Often, potentially important developments are difficult to spot because they do not fit the vision of powerful interests. I mentioned the Linux phenomenon earlier but you will rarely see it mentioned in the commercial press devoted to computer news. In the late 'eighties, few people paid attention to the Internet. Nowadays, many people, particularly within the telephone companies, dearly regret their former deafness. In 1990, the idea of an electronic journal still looked slightly crazy and best left to starry-eyed dreamers.

Concerning objectives, librarians hold a strong hand in the form of positive values and well-defined roles, but they must never forget them. In fact, with digitization, they must learn to fit their modes of working and some of their functions within these values, rather than the reverse. In particular, they must not panic under financial pressure and accept any deal that seems to save the day at the expense of tomorrow. I often hear librarians tell me that they will worry about the licensing of digital documents only after they have secured the money to license them. This attitude sounds tough minded and very realistic, and it may be the kind of talk to use in front of reluctant deans and university presidents, but it is not a good way to approach the problem in its full generality. Libraries remain our very best hope to prevent basic human knowledge from being completely privatized, monopolized, and ultimately locked up by venal interests. Librarians must never forget that point and scholars should quickly come to assist as they recognize that librarians are their best ramparts against a commercial ethos that may undermine their intellectual and individual integrity.

As for weak signals, they are no longer so weak any more and librarians must learn to hear them because they will work to their advantage. A wonderful trend is unfolding, but few so far seem to recognize it fully for what it is. Yet, examples abound. The development of the Internet, of the Web and now of Linux and the free source code movement all show that, somehow, humanity is discovering the immense power of its intelligence when it is distributed and shared. Raymond's recent paper, "The Cathedral and the Bazaar" [14], published on the Web and already translated into a few languages, points to the fact that when people accept playing nonzero sum games and when they accept sharing and collaborating, they collectively gain a great deal more than by retaining tightly proprietary attitudes on things. Amusingly, this kind of message should have been detected a good deal earlier because, after all, a very famous human enterprise has been thriving on it for the last four centuries: I mean scientific research itself. The scientist gives his results away to obtain visibility, prestige. Only indirectly does this symbolic capital translate into real cash and important institutional positions. The weak signal, which is not so weak any more, is that distributed intelligence beats any other known combination, in particular hierarchical forms of planning coupled with proprietary attitudes. Another weak signal is gathering momentum: it is that the world is finally moving beyond the tired (and tiresome) opposition between radical individualism and just as radical collectivism. The self-reliance of human beings can only go so far, and the regressive dream to fuse into some sort of sect-like community is just as elusive. The world is moving toward a networked individual whose identity bears greater analogies with linguistic phonemes§§§ than with material atoms.

All of this is important for libraries because they will assume a truly interesting role if they participate in the building of this distributed intelligence. In taking on this role, they will give themselves the opportunity to actualize the best out of their present virtual state. Participating in the elaboration and support of electronic publishing constitutes a wonderful first step in that direction. Earlier, I noted that virtual referred to an active transition. Should libraries accept to take their full place in the development of scholarly electronic publishing, they will find themselves involved in a major transition in our civilization. Beyond the publisher's role, libraries will also find that they can (and should) facilitate all possible synergies among members of their constituency. They can help them link into a worldwide network that would have the whole of humanity and its well being as its cause. For that is what fundamental research ultimately aims to do, beyond the narrow horizons of individual careers.

Where groups appear fragmented, dislocated by institutional barriers or warped by ruthless competition or by isolation, exclusion, and segregation, libraries can heal, weave, and mesh. Beyond owning information and providing access to it, libraries can repair the very fabric of their constituencies into communities; they can open new collaborative pathways and thus overcome old barriers. By fully and positively actualizing all the greatness of their present, digitally informed, virtual state, libraries can point the way to a new form of enlightenment. They can point, in effect, to a neo-modernity that could carry us beyond the sterile despair of present post-modern cant. They can bring us back, along a spiral path, to the values of the Enlightenment, only at a higher pitch, and without some of the defects.

Notes

* Guédon, J. The digital library: an oxymoron? Bull Med Libr Assoc 1999 Jan;87(1):9–19.

† The confused use of virtual is quite common. Recently, Guillou, the head of the network of French-speaking universities, AUPELF-UREF, proudly laid the first symbolic stone of a virtual university in Dakar (sic). It is true that with just one stone, or indeed several, the university remains quite virtual. However, Guillou should realize that when the job is completed, i.e., the institution is actualized, his virtual university will have lived.

‡ In French, the word "vector" is sometimes used to express this material dimension of a document. The vector is what carries the signs.

§ The fact that the rise of the codex coincides with that of early Christianity raises some very fascinating issues in this regard. Studying them would lead us too far astray, but simply hinting at a possible link between Christianity and the codex shows how much we may miss by neglecting such apparently trivial matters as the material form of messages.

** A variety of factors ranging from the full alphabet, as Havelock argues in his Preface to Plato (see reference 5) to the solitary exercise of silent reading, also played an important role in this process. However, we should not neglect or belittle the part attributed to the codex.

†† I would like to thank Ms. Frances Groen, Head of Libraries at McGill University, for drawing my attention to Brodman's book.

‡‡ The bandwidth metaphor was used first in Okerson AS, ed. Why are electronic publications difficult to classify? the orthogonality of print and digital media. In: Directory of electronic journals, newsletters and academic discussion lists, 4th ed. Washington, D.C.: Association of Research Libraries, 1994:17–21. Also available online at various locations, for example: http://poe.acc.virginia.edu/~pm9k/libsci/guedon.html.

§§ The Newjour database may be viewed at http://gort.ucsd.edu/newjour/NewJourWel.html.

*** Linux is a POSIX-compliant, Unix clone that was initially developed by a young Finnish student, Linus Torvalds. It was then taken up by hundreds of programmers around the world. Windows NT, of course, is an operating system developed by Microsoft. For information on Linux, see http://www.linux.org. On the free source code movement, see http://www.gnu.org.

††† The e-print archive served at Los Alamos may be viewed at http://xxx.lanl.gov.

‡‡‡ The Project Muse site is available at http://muse.jhu.edu/muse.html.
The Presses de l'Université de Montréal is available at http://www.pum.umontreal.ca. For a description of the SPARC project, see http://www.cni.org/tfms/1998a.spring/handout/Mcase98Stf.html.

§§§ The linguist de Saussure who invented the phoneme concept used to compare it to men on the chessboard. They indeed have some intrinsic value, but it can vary enormously, depending on position and relation to other men on the board.

References

1. Gelb IJ. The study of writing. the foundations of grammatology. Chicago, IL: University of Chicago Press, 1952.
2. Calvet LJ. Histoire de l'écriture. Paris, France: Plon, 1996.
3. Lévy P. Qu'est-ce que le virtuel? Paris, France: La Découverte, 1995:13–19.
4. Martin HJ. Histoire et pouvoirs de l'écrit. Paris, France:Albin Michel, 1996.
5. Havelock E. Preface to Plato. Cambridge, MA: The Belknap Press of Harvard University Press, 1963.
6. Febvre L, Martin HJ. L'apparition du livre. Paris, France: Albin Michel, 1971.
7. Eisenstein EL. The printing as an agent of change. London, U.K.: Cambridge University Press, 1980.
8. Chartier R. The order of books. Cambridge, U.K.: Polity Press, 1994. (Translated from the French by Cochrane LG. L'ordre des livres. Paris, France: Éditions Alinéa, 1992.)
9. Brodman E. The development of medical bibliography. Baltimore, MD: Medical Library Association and Waverly Press, 1954:26.
10. Gilmont JF, ed. La Réforme et le livre. L'Europe de l'imprimé (1517-v. 1570). Paris, France: Les éditions du Cerf, 1990.
11. Meadows AJ. Access to the results of scientific research: developments in Victorian Britain. In: Meadows AJ, ed. Development of science publishing in Europe. Amsterdam, Netherlands: Elsevier Science Publishers, 1980:43–62.
12. Franc A. Reed Elsevier met en vente sa presse et son edition grand public. Le Monde (Paris), 1995 July 20;51(15701):15.
13. Okerson AS, O'Donnell, eds. Scholarly journals at the crossroads: a subversive proposal for electronic publishing. Washington, D.C.: Association of Research Libraries, 1995.
14. Raymond E. The cathedral and the bazaar. [Web document]. [rev. 11 Aug 1998; cited 19 Aug 1998]. Available from Internet: <http://sagan.earthspace.net/esr/writings/cathedral-bazaar/cathedral-bazaar.html>.

Received June 1998; accepted July 1998