William Y. Arms
The first federally funded research program in digital libraries was DARPA's Computer Science Technical Reports project . The opening meeting, in 1991, was consumed by a heated discussion about whether the field of study should be called digital libraries or the digital library. This was more than an academic argument. Many of the computer scientists at the meeting had been leaders in the development of modern computing. They had seen networked computing begin as isolated, incompatible islands that merged into the shared framework of the Internet. Should digital libraries be encouraged to develop independently or together?
Questions about the name of the field were laid aside when the NSF/DARPA/NASA program officers selected the name "Digital Libraries Initiative" for their joint program that began in 1994 . Agreement on the name, however, does not answer the underlying question: should digital libraries be self-sufficient islands or should we strive for a single global digital library?
This question can be studied using viewpoint analysis, a technique from software development. The idea is to identify the various stakeholders in a system and view the system from each of their viewpoints. For example, there is a famous New Yorker cover that shows the view of the world from 9th Street, Manhattan . A few blocks of New York City dominate the scene; China and Japan are vague bumps on the horizon. The cartoon is amusing because it represents a universal truth: the world looks very different depending on your viewpoint.
This article looks at digital libraries or the digital library from three viewpoints: an organizational view, a technical view, and the view of the user. From an organizational viewpoint, the world clearly consists of many separate digital libraries. From the user's viewpoint, this distinction is less clear.
Why is this important? Digital libraries research has a mixed record in recognizing major innovations: computer scientists resisted the simple technology of the Web; librarians disparaged the value of Web search engines. Greater emphasis on the user viewpoint, and less on the technical and organizational, may reduce such mistakes in the future.
The Organizational Viewpoint
Figure 1 provides an organizational viewpoint. It shows how the Library of Congress or any other major library might view the library world.
Most of the boxes in Figures 1 correspond to discrete organizations with distinct identities. Many of them have long histories from the time when libraries were defined by their physical buildings. The awareness of their identity leads organizations to create digital library services in which differences between organizations are emphasized explicitly.
Before computer networks, an emphasis on the organizational viewpoint was natural. When libraries were defined by their buildings, an individual patron used a very small number of libraries, perhaps the local public library or a university library. A researcher could spend a career within the bounds of a single library. In a few cities such as Boston or London, several libraries were grouped together, but most libraries felt obliged to provide their patrons with all the necessities of intellectual life.
However, an organizational focus can be annoying for users: early publishers of CD-ROMs promoted their materials by stressing the distinctive aspects of their user interfaces, thus forcing researchers to learn many different interfaces; university libraries have developed Web portals that bring together the resources that they offer, but not necessarily all the resources that a faculty or student uses; the original ACM digital library gave an integrated view of all ACM publications, failing to recognize that a reader uses resources from many publishers; Google Scholar shows somewhat the same myopic viewpoint.
The Technical Viewpoint
The DARPA program officer for the Digital Libraries Initiative once observed that the only reason DARPA funded digital libraries was to stimulate research in interoperability . In this context, the term interoperability describes technical methods to combine services from discrete libraries, i.e., it takes a technical or system viewpoint of digital libraries. This viewpoint has much to say about data structures and metadata formats, the relationships among them, and how they are exchanged. It has little to say about the actual content and it is agnostic about users.
Interoperability research assumes that there are many digital libraries: the challenge is how to encourage collaboration among independent digital libraries with differing missions and resources. (See, for example, .) Early on, the broad acceptance of the Internet protocols and the core Web technology, augmented by library standards such as MARC and Z39.50, provided a base level of interoperability, on which computer scientists have been steadily expanding in areas such as XML, RDF, and Web Services. Ten years later, significant progress has been made in many technical areas, such as mark-up languages, metadata standards, harvesting protocols, and identifiers. Moreover, we can be pleased with the progress in understanding what characteristics of technical standards lead to widespread adoption by independent libraries. However, digital library researchers have largely ignored the efforts of the World Wide Web Consortium to develop technology for a single digital library.
The User Viewpoint
While good progress has been made in interoperability from a technical viewpoint, less progress has been made in turning a plethora of digital libraries into a single digital library from the users' viewpoint. Figure 2 shows how an academic user might view the digital library world. Notice that the Library of Congress appears in the corner of this figure, just as China and Japan had appeared as distant islands in the New Yorker cartoon.
From the user's viewpoint, technology is irrelevant and organizations are of secondary importance. Separate organizations, each with their own identity, can easily become an obstacle. For instance, at Cornell University the university library supports faculty research, as do the central computing organization, the computer science department, and the supercomputing center. Each service is excellent, but they were developed separately and can be awkward to use together.At a user interface level, the almost universal use of Web browsers has created an appearance of uniformity. Many stylistic conventions have emerged in the layout of Web sites and navigation within them, and in details such as standard buttons and the terms used in menus. Such conventions are very important, but they are only the surface. A user who wishes to do serious work using online information will find that superficially similar services have deep semantic differences. Web search engines emphasize precision of the highly ranked hits, while scientific information services emphasize recall. Library gateways attempt to give coherence to the collections and services offered, but the underlying systems are so different that the gateway is only a veneer.
The Digital Library: a Research Agenda
For the past decade, many people have carried out research and development on separate digital libraries and technical interoperability among them. As the early work matures, it would be easy for digital libraries research to become inbred, focusing on detailed refinement of the same agenda. Alternatively, we can think of the digital library from the user's viewpoint.
As a first step, we need to rethink evaluation. The standard way to evaluate a digital library is to give a group of users a set of tasks to carry out within that library. This is evaluation from a system or organizational viewpoint. User testing rarely takes a holistic viewpoint, beginning with the user. For instance, in evaluating the National Science Digital Library for science education , a holistic evaluation would center on a user, perhaps a science teacher preparing a course, and observe all the tools the person used not just those that one library provides and how effective they are in combination.
In software development, viewpoint analysis is part of the process of requirements analysis, understanding the functions to be carried out by a system. Requirements that are developed from a technical or organizational viewpoint, may fail to recognize the user's viewpoint. For instance, technical experts resisted HTML because it mixed structural and formatting instructions. Users loved it because it provided attractive displays and brought color images to their desktops. The Internet is truly disruptive technology, yet requirements developed from an organizational viewpoint tend to assume continuity of existing organizations, not disruption.
About twenty years ago, independent computer networks began to merge into the single unified Internet that we take for granted today. Perhaps now is the time for digital libraries to strive for the same transition, to a single Digital Library.
Notes and References
 The Computer Science Technical Reports project was led by the Corporation for National Research Initiatives on behalf of DARPA.
 The program officers were Stephen Griffin (NSF), Barry Leiner (DARPA), and Paul Hunter (NASA).
 Saul Steinberg, View of the World from 9th Avenue, New Yorker, March 29, 1976.
 Another contribution by DARPA was initial support for D-Lib Magazine.
Copyright © 2005 William Y. Arms