![]()
The Future of Special Collections in Emerging Information Delivery Programs
David S. Zeidberg
In his last book, Disappearing Through the Skylight: Culture and Technology in the Twentieth Century, the late O. B. Hardison, Jr., writes:
Disappearing through the Skylight is about change in modern culture. … Because the changes have been fundamental, the concepts–and even the vocabularies and images in which the concepts tend to be framed–no longer seem to objectify a real world. It is as though progress were making the real world invisible. …
This book is about the ways culture has changed in the past century, changing the identities of all of those born into it. Its metaphor for the effect of change on culture is “disappearance.”1
Hardison examines nature, history, language, art, and human evolution and how their embodiments are disappearing into information about them. While libraries are not discussed per se, libraries are experiencing the phenomenon of cultural disappearance as well. We long ago embarked on a path using technology, and specifically computers, to improve the access to our collections. The library profession rightly saw that automated access would be a more effective means of processing vast quantities of materials in a laborintensive profession. We also foresaw a tenuous financial future for libraries, which has now come to pass, so automation allowed us to do more with existing staff, and now fewer staff.
As we turn into the 1990s, however, the role of technology in libraries has become increasingly more complicated. At best, we are getting mixed messages from the library leadership about the role of technology in libraries, and I believe the leadership must return to a position of balance between acquiring and providing access to research resources, and developing the methods of access to them.
At worst, technology is becoming an end itself, with information about the research resources replacing the resources themselves as the predominant “product” of the library. This latter view of technology in libraries has been abetted to a certain extent by other trends in librarianship, which are beyond the scope of this paper, but which do not bode well–the closing of major American library schools, and the United States government’s attempt to downgrade the librarian rankings in its federal GS series. Both of these trends suggest, among many other things, the diminished need for the intellectual training of the librarian to facilitate use of research materials, and the increased role of technology as the end rather than the means. These trends have brought into focus some thoughts about changes in research in academic communities and what these changes mean specifically to Special Collections in academic libraries.
To examine the mixed messages, let us begin by returning to Hardison for an example of “disappearing” resources. One of his discussions which comes closest to representing the dilemma facing libraries and the scholars who use them occurs in his chapter “Artificial Reality.” In it, Hardison describes the development of optical disks and hypertexts in education, and specifically as they could be applied to literary studies. He uses Shakespeare’s The Tempest as his example, and onto his optical disk he theoretically piles the text and its variants; essays on the colonization of the New World, shipwrecks, and cannibals; maps of the period; designs of the London stages; drawings of key scenes; a glossary of Shakespeare’s language and explanations of his grammar; explanations of magic; depictions of costuming; actual performances of scenes. Finally he breaks off:
But enough! The possibilities are endless. The limiting factor is storage and optical disks have already created storage capacities for personal computers rivaling those of mainframes of the 1970’s. The hypertext Tempest is a teacher’s dream of what a student reading The Tempest should have.
What does hypertext do for–or to–The Tempest? Unfortunately, the answer is not as simple as it might seem to be in the abstract. The clear implication of hypertext is that The Tempest is not a literary work to be enjoyed but a heap of facts to be memorized or a puzzle to be solved or a mystery to be explained.
If you imagine a reader using hypertext, you have to imagine a constant movement from text to glossary to grammatical comment to classical dictionary to Bermuda map to textual vari-ants to drawings of Ariel to. …
… the items just listed … form a collage. They are not linked because Shakespeare put them together that way in The Tempest but because of the interest or curiosity of the reader or the demands of a teacher. The process of “reading” hypertext is therefore different from what is normally meant by the phrase “I read the play.” With hypertext, the process is interactive and discontinuous–almost the opposite of reading The Tempest continuously from beginning to middle to end or seeing it performed.
The reader must constantly decide whether to read a sentence in the play or invoke the drawing of the Elizabethan theatre to find out how the shipwreck might have been staged or [etc., etc.]. The sum of the fragments chosen becomes the text the reader has actually read. Each “reading” produces a new text–a megatext–made of bits and pieces of Shakespeare interspersed with bits and pieces from other sources.
When “read” in this way, the play tends to disappear into the hypertext like water into a sponge. Admittedly, computers did not create this situation. The disappearance of the literary text began in the nineteenth century with the theory that literature should be studied as science studies nature. However, computers make the disappearance of the text irreversible by plunging the reader into an information swamp from which there is no escape other than turning the computer off and pulling out a paperback copy of the play–or, better yet, attending a performance (as we did last night), since the performance does not have footnotes.
Hypertext makes clear a fact that was often noticed before it appeared on the scene. We are coming to the end of the culture of the book.2
Hardison wrote this in the mid to late 1980’s as a theoretical example. His sentiment was echoed in Terry Belanger’s 1991 Malkin lecture at Columbia University, in which Belanger concluded, “the book is going to go the way of the horse.”3 Today, these hypertexts already exist. The Voyager Company in Santa Monica now has works in the public domain, such as Alice in Wonderland and Pride and Prejudice, on hypertext discs, complete with the capability of note taking in the margins and place marking. IBM has produced two interactive programs: Columbus: Encounter, Discovery and Beyond and Illuminated Books and Manuscripts, whose title is particularly ironic since the work is not about medieval and Renaissance works with illuminations, but rather include the following eclectic titles–Tennyson’s “Ulysses,” Shakespeare’s Hamlet, Dr. Martin Luther King’s Letter from Birmingham Jail, the Declaration of Independence, and Black Elk Speaks all “illuminated” by the hypertext secondary sources of encyclopedic scale attached to each.
Clearly the commercial sector sees a market for these hypertext products. When libraries are involved in their production, how they are being produced and to what purpose take us to the point about mixed messages from the library leadership. The Library of Congress, for example, has developed “The American Memory.” Kenneth Bacon has described it as “an ambitious … program to put some of (the Library of Congress’s) unique collections on optical and compact disks and then make them available to libraries around the country.”4 The program is designed to be a selection of the library’s holdings about three periods of American history: the 19th century, 1890–1920, and 1920–1939.
The key word though is “selection,” and its use can imply helpful or dangerous results, depending upon which Library of Congress administrator’s view one accepts. The Librarian, James Billington, sees the program as “a wonderful tool for school libraries, and it will lure people who don’t normally use libraries by bringing in new technologies.”5 Billington is suggesting a usefulness which might be compared to LC’s Center for the Book’s “Read All About It” program, which links library sources to television programs derived from them and encourages viewers to be readers as well. If Billington means the use of “The American Memory” in this context, then the program’s selectivity should encourage new readers to visit libraries and explore the complete resources.
But that is not how Billington’s director of special projects, Robert Zick, sees “The American Memory”: “Libraries of the future will be a service, not a place,” Zick says. “What you once had to go to the Library of Congress to see can now be viewed at home. That’s the capability.”6 While that may be the project’s capability, it is not achievable now, and that may lead to some danger in the interim. The present selectivity itself can be dangerous, and while we may have the technological capability to be comprehensive, in reality we are limited by funds and thwarted by the sheer size of collections.
Let us look at the practical aspects first. Where collections are of a manageable size, conversion to an optical disk format can have several advantages. The Southwest Museum in Los Angeles, for example, has photographed every one of its artifacts and placed the images on an optical disk using the interactive program Questor. Through a wide variety of access subjects and other descriptors, a researcher can scan through thousands of artifacts to find the particular one or small group for first-hand study. The researcher has complete access to the collection more quickly than any manual file could ever have provided with the added advantage of seeing the items directly. This is a wonderful enhancement to the older practice of the visible file, which placed contact prints on the catalogue cards, and which had its own relative expense. Few libraries or museums with extensive holdings could afford to create visible files, and those that did were usually limited to a single card catalog entry per image.
The optical disk/hypertext approach gives the researcher much broader access through its interactive database than a single visible file entry. A researcher at the Southwest Museum can limit his or her search, for example, to Chumash Indian baskets, further delimit the search to the tribe’s branch located in the Tehachapi mountains, and further define the search by date or period. This is access at its technological best, with an important by-product of preservation. Complete visible access through an interactive database greatly reduces the amount of handling of the original materials, both by staff who retrieve and reshelve them, and by the readers who would otherwise have to handle a larger number of items before finding the subject of their research.
For a collection of a few hundred thousand items, the current technology can enhance access, but what about larger collections? The Department of Special Collections at UCLA houses the photo archive of the Los Angeles Times, which dates from 1893 and includes almost 2,000,000 images. It has been suggested to us by many people–scholars, library administrators, the Times staff itself–that it would be ideal if the collection were on an optical disk, for all the reasons described in the Southwest Museum example above. But at what cost? The cost of the equipment is negligible when compared with the labor. Panasonic, for example, has developed a component system that can photograph an image directly onto the optical disk and provides a parallel database for describing the image into variable fields of the processor’s invention. The cost of the system is approximately $35,000 (and it will probably reduce as the technology develops).
In libraries, labor remains a barrier that the seduction of technology often causes us to forget. Using a 40 hour work week, the standard work year is 2,080 hours. Suppose we had a top-rate processor for the Los Angeles Times optical disk project, and suppose there were minimal problems of identification in the photographs. Under these ideal assumptions, let’s further assume that the processor could record an image onto the disk and provide the database description for subject(s), photographer, and date at the rate of one every ten minutes. Six images per hour times 2,080 hours per year would process 12,480 images per year. At that rate, it would take more than 160 person-years to process this collection. A processor qualified to do this level of work in our library earns approximately $25,000 in today’s dollars. So the cost for this project in salary alone would be $4,000,000. One could argue that at $2 per image, that’s a bargain, but this does not take into account benefits, indirect costs, and inflation. And in reality, where is the money in the first place?
Were this a grant proposal, most granting agencies would question placing equal value on all 2,000,000 photographs in the Los Angeles Times archive. Perhaps only 10% of the photographs in the collection were actually published. Maybe we should select those for the optical disk/hypertext project. Readers most often make requests for the original photographs from research they have already done with the newspaper microfilm, so that process of selectivity makes some sense. But we must be careful not to ignore the remaining 90% of the photographs. Newspaper photographers often shoot an entire roll of film or more to produce one “publishable” image, and the rest of the negatives on the roll may give readers insights to their research which the published story did not. So while the 10% project will only take 16 person-years instead of 160, we would still need to arrange related photographs physically in the collection to those on the optical disk and link them with a note that will direct the reader to those related images not on the disk.
This one example of a visible collection addresses the problems faced in trying to put a collection of that size into an optical disk/hypertext program. Returning to Zick’s suggestion that one day “everything” at the Library of Congress will be available this way, I refer the reader to the 20,000,000 monographs in the Library’s holdings, which probably average two hundred pages per title. Leaving aside Hardison’s interactive model for The Tempest, where is the labor and money to convert these 4 billion pages to optical disk, only linking them to their existing cataloging? What about the many more millions of manuscripts, prints, photographs, drawings, maps, motion pictures, and sound recordings at the Library of Congress? Or should we only put those works into the program which are “important?” Who will decide this?
The approach of selectivity takes us back to the Library of Congress’s “American Memory” project and the dangers inherent in it to which I alluded before. If “The American Memory” stands, as James Billington says, as an attraction for new readers to the library’s complete holdings on those periods of American history, then it serves a good purpose. Robert Zick suggests that “everything” in the Library of Congress will someday be at our electronic fingertips; I hold the foregoing discussion up as the practical reason why this is unfeasible. If one reverts to a selective approach, an immediate danger is that the selectivity will be misconstrued for comprehensiveness. And whether the approach is selective or comprehensive, another immediate concern is whether all in our free society will have access to this technology.
In creating a representation of the American history holdings of the Library of Congress, the very process of selecting the items for the optical disk makes secondary resources out of primary ones. Even the broadest, brightest, and best-educated process of selection remains one person’s or one committee’s perspective. The selectivity, and form the representation takes, parallels what we used to call books. A scholar ideally would research the whole corpus of materials on a subject, formulate a thesis about the material, organize the “relevant” primary resources to support the thesis, and write it up. As readers of the bo...
