The study began with a survey questionnaire circulated to the academic libraries that were members of the DLF in January 2001. Twenty-one institutions responded.2 The data they supplied were illuminating on several points, including the different approaches that libraries have taken to build their digital library programs and the extent to which the complexion of any program is tied to campus personalities, circumstances, and needs (Greenstein, Thorin, and McKinney 2001). The data compelled a closer look at the softer influences not so readily identified by a survey. Accordingly, extensive interviews were conducted with key staff members at six DLF member libraries: the California Digital Library (CDL), Harvard University, Indiana University, New York University (NYU), the University of Michigan, and the University of Virginia.
These six programs were selected because they represented distinct variations in attributes that the survey identified as being potentially key determinants of a program's distinctiveness. Among those attributes is age. Virginia and Michigan are two of the oldest digital library programs in the United States. NYU was included because it is a relatively new program, begun in earnest only in the last two years. Harvard and Indiana represent the middle-age members of the sample.
Another distinguishing characteristic is a program's orientation—that is, the main focus of its work. Not all digital libraries focus primarily on digitally reformatting analog items in their collections and distributing them online. Harvard and NYU, while possessing digitization capacity, focus principally on providing systems environments and infrastructure capable of managing digital assets as may be acquired or used within their host institution. The CDL is similarly focused but is actively assessing funding and business models that may support more investment in digitization. Michigan has focused both on digitization and on the development of access systems. Indiana stresses digitization in the context of teaching and learning. Virginia, on the other hand, emphasizes innovative use of information technology (IT) in support of research and teaching. Its main focus is consequently user services, with collections and system development playing a supporting role. The range in orientation of DLF digital libraries is illustrated by data from the DLF survey. In the year 2000, digital conversion costs for member libraries ranged between $38,000 and $1,145,000; the average spending on all aspects of digital library programs was $4,341,798 ($2,641,798 if the costs of acquiring access to commercial electronic content are excluded). Fewer than half of the DLF libraries surveyed invest primarily in digital reformatting programs. Most have oriented themselves toward the development of technical infrastructure and of various reference and other end-user services.
A third characteristic used to select programs for case study interviews was organization. Information about how digital library programs are organized is available for 18 of the DLF's members. Two—Harvard and the CDL—are confederal organizations to which a number of libraries contribute at some level. Harvard's program supports the more than 90 libraries that make up the Harvard library system. The California Digital Library is a library in its own right, but it provides services to faculty, students, and libraries at the 10 University of California (UC) campuses. As confederated organizations, Harvard and CDL are unique within the DLF as well as among academic libraries generally. An additional six DLF member digital library programs are managed as separate departments or units within the library. Michigan and Indiana are representative of this approach. The University of Virginia represents another organizational form—one in which digital library effort is found in several library departments but coordinated through a committee. This distributed but coordinated approach is common to 4 of the 18 DLF members for whom organizational details are known. Two further approaches were evident in the survey data but are not represented in the case studies because they are indicative of very small and immature programs. Two DLF member libraries reported in the survey that their programs were too small to have some determinate organizational form, while a further four claimed that their programs comprised a range of uncoordinated activities taking place in different library departments.
A final attribute that distinguishes a digital library program is the library's relationship with surrounding academic departments and information services, such as academic computing and IT. Although not easily quantifiable, closeness may be measured by such factors as the facility and experience of collaboration between the library and these surrounding departments, and the extent to which strategic planning in one department includes representatives from and takes substantive account of planning in other departments. Size may have something to do with closeness. Among the surveyed institutions, Virginia and NYU are the smallest and also have the closest relationship with other departments. They are joined by Indiana, however, which, at more than 33,500 students on the Bloomington campus alone, can hardly be characterized in terms that describe a liberal arts college "feel." Harvard and CDL, because of their confederal character, are perhaps furthest from faculty and IT units. This study, although based on the results of the survey and interviews, has also been informed by other investigations that the DLF has sponsored in its attempt to understand various aspects of the digital library and by the numerous formal and informal discussions that have supported or resulted from those investigations.3
In sum, we encountered digital library programs that were developed
Whether a library director's support is sufficient to launch a successful digital library program remains an open question. The hands of university presidents and provosts are clearly at work in our case studies. Since 1995, UC President Atkinson has been a great champion of the CDL. University of Michigan President Duderstadt (1988–1996) identified issues surrounding an information and IT agenda as being among his priorities. He invested in the transformation that created a School of Information and in the aggressive development of the technology infrastructure. Both ultimately benefited Michigan's digital library interests.
Creative leadership, another important ingredient, comes from a number of sources. At Michigan, it emerged from close collaboration among three people: one in the School of Information and Library Studies (now the School of Information), one in the university's Information Technology Division, and one in the library. Such collaboration is apparent at Harvard, in a triumvirate whose members are drawn from within the university's library system; at NYU, in a combination that is orchestrated and shared by the dean of libraries and the CIO; and at Indiana, in a group modeled closely on Michigan's and comprising members of the library, the IT service, the library school, and the School of Informatics. Creative leadership can also stem from an individual. At Virginia, the digital library program owes a great deal to a single librarian who, in the 1980s, was already thinking deeply about what a twenty-first-century academic library should be.
The need for executive leadership is self-evident. No digital library program, however well supported and envisioned, can develop without people who possess the appropriate technical, tactical, and even diplomatic skills. Programs that otherwise had key elements in place, such as funding and mission orientation, were delayed to some extent until such leadership could be found. This was the case at Indiana and NYU.
Just as digital library leadership is required at several levels, ownership of the digital library program needs to be felt widely across its host institution. This may be less important for the startup phase of a digital library program than for its long-term operation. All six programs represented in the case studies traced their origins to inclusive strategic planning exercises—exercises that involved academic faculty, senior university managers, some library staff, and, in some cases, representatives of other information services. Representatives of Michigan's triumvirate traced their program's origins to a planning process that included a yearlong, campus-wide faculty symposium on electronic information and a report prepared for campus administration on this process. The work included intensive analysis of distributed computing operations and the economics of campus information provision. NYU's mandate stems from library and IT planning vetted by faculty advisory groups and university administration. Harvard's stems from a planning committee that included deans, librarians, and faculty. CDL's grew from a planning process that was initiated by librarians but was ultimately ceded to a committee comprising representatives of all constituencies: faculty senate, campus administration, system-wide administration, information technologists, and libraries. The committee itself was instituted by the chancellors (campus chief executives) and the university provost. Of the 21 digital libraries surveyed, 76.2 percent (16) are guided by digital library development policies created as part of a broader university IT strategy. These data suggest that digital library development is at some level conceived as part of a university-wide planning process.
The model of a digital library program that is relatively autonomous in its early days is evident elsewhere with significant variation. Virginia cut its digital library teeth in experimental units that were outside the normal line of management—in the library's Electronic Text Center and, in the academic unit, the Institute for Advanced Technology in the Humanities (IATH), which is located in the main library. Through a series of digital library projects, staff (and through the staff, the library) gained competence in creating, managing, and distributing electronic information. The Electronic Text Center was located in the library but was initially independent of other library departments and services. Even IATH reported directly to the associate provost for research and was thus removed to some extent from the "academic line." Virginia continued for some time to foster experimental efforts in relatively autonomous organizational units. Thus, as it sought to develop experience and capacity with new digital materials (e.g., images, statistical data), it opened new media centers that focused on the technical and other challenges that these materials introduced without being clouded by traditional library practices or even by practices that were growing up around the Electronic Text Center and IATH.
Indiana's program introduces a variation on the same theme. Here one can point to a number of experimental initiatives, each mounted with a considerable degree of autonomy from mainstream operational departments and from one another. VARIATIONS, originally an audio e-reserve, was a flagship project run out of the Music Library as a personal project of that library's director. The Library Electronic Text Resource Service (LETRS) is an e-text center much like the one at Virginia. It is located in the library but is relatively autonomous of the traditional library departments. Digital Images Delivered Online (DIDO) was an early experiment that served faculty in the School of Fine Arts.
In 1996, Indiana established the Digital Library Program (DLP), which brought together these three projects. Through amalgamation, Indiana achieved economies and cross-fertilization at virtually every level, enabling a more coherent approach to research and development, selection and deployment of digital library technologies, and staff appointments. Amalgamation was not intended to bring the DLP into a more traditional library unit and thus out of its skunk-works environment. Organizationally, the DLP reports directly to the dean of libraries rather than up the normal operational chain. Moreover, the DLP is funded discretely, rather than from operational budgets that support traditional library services. Programmatically, initiatives are still mounted at least in part to build on digital library competencies as they accrue, to add staff to complement skills already in place, and to acquire technology as may be needed across the program.
Trajectories at Harvard and the University of California differ from those at Indiana, reflecting the programs' confederal nature and system-wide orientation. At Harvard, the digital library program grew out of technical services developed on a system-wide basis, notably around shared cataloging needs. The CDL was instituted in parallel with a system-wide technical service (the Department of Library Automation), which it then subsumed. Although evolving from these more established services, digital library initiatives at CDL and Harvard also took on at least some of the attributes of the skunk works. Thus, while CDL made a point of launching at least one new operational service every six months, it fashioned numerous experimental initiatives and even established a small unit under a director for education and strategic innovation. At Harvard, where the Library Digital Initiative was organizationally housed in the same unit that supplied system-wide technical services, the program's emphasis on digital library infrastructure assumed a five-year development path during which time few infrastructural components became operational. The program was seen as a highly practical applied research program that would lead ultimately, but not immediately, to the development of operational systems.
The digital library's initial locus in experimental units a step or more outside the traditional management line was important. It created a space where experimentation could happen in a way that could not disappoint the service expectations of faculty and students, or even of other librarians. In some cases, digital library staff members were spared the normal bureaucratic decision making, hiring, staff development, and procurement procedures that applied to other more traditional library services.4 As such, they could more quickly populate, launch, redirect, or even end experimental initiatives, and could respond to funding, staffing, and other opportunities as they arose. This nimbleness was secured at a cost, however. With few exceptions, institutionally based digital library programs (as opposed to programs such as those at Harvard and CDL, which operated at least in part on a system-wide basis) were challenged later to integrate their staff, work, and thinking about library collections and services into the mainstream of the library.
The need to promote digital library programs was felt acutely by the programs that were launched in the early 1990s. At that time, few of the agencies to which libraries had a natural or historic entrée were making any substantial investment in activities that we would today associate with the digital library. The National Science Foundation's (NSF's) DLI1 initiative changed the profile of digital library activities and the amount of federal dollars that might be available to support them. Libraries, however, did not appear on lists of the first successful NSF digital library grants. The inexperience on both sides of the funding fence only emphasized the importance of marketing, as digital libraries were forced to consider how best to present their research interests and needs to members of a scientific community with which they had hitherto had only limited contact.
The role of foundations (particularly The Andrew W. Mellon Foundation but also the J. Paul Getty Foundation and the W.K. Kellogg Foundation) was also important in kick-starting digital library initiatives. Here too, however, libraries were obliged to get themselves on the agencies' informal bidding lists and to make their case for being there. It would be interesting to know how many failed funding applications a digital library program made on average before achieving its first success.5 Certainly, such failures were earning experiences that helped digital libraries focus their aims and build their technology expertise. These experiences also helped libraries learn to present their funding cases convincingly, in a language that was comprehensible to those outside the library community and in a way that demonstrated the broader significance of their institution's own efforts.
No amount of external grant funding, however, can replace the need for a substantial institutional commitment. So-called core funding—whether obtained from gifts, from additions to the library's permanent budget, or from a reallocation of existing funds—was claimed as essential by all 15 of the survey respondents who answered questions about how their digital library programs are funded. Such commitment takes different forms. At Michigan, Indiana, and NYU, it is in the form of contributions from university departments that are co-investing in the digital library's development. At Harvard and UC, local funding flows directly to the digital library from the office of the university president. At Virginia and Indiana, most startup funding came from money the university librarian had reallocated from other library purposes. Interestingly, as digital library programs mature, reallocated library funding becomes more important, at once reflecting and contributing to the challenges of mainstreaming the digital library.
Greater innovation is apparent from the mid-1990s—for example, in experimentation with digital reformatting. Surrounded with lofty rhetoric about universal access to all human knowledge, innovative scholarship and teaching, and national and international digital libraries, some startup programs focused on providing online access to selected holdings. It is too soon to judge the results of these early initiatives. The risks involved for early adopters may have been so great that they were forced to exaggerate the claims for their work. Consequently, few of the earliest digitally reformatted collections stood a chance of meeting the expectations that grew up around them. Judged in light of initial expectations, the earliest online collections are, with notable exceptions
The competitive dynamic is still apparent in startup digital libraries, and it may be intensified as unsettled space on the digital library frontier becomes more difficult to locate. Interestingly, fewer startup programs appear to be staking their claims on a particular technology or practice. Instead, they are emphasizing service orientation (as at Carnegie Mellon, for example, which emphasizes user support services) and local collections and scholarship (as at NYU, whose digital library program reflects a local orientation toward art and performance across disciplines).
The competitive disposition of young digital libraries may also help explain the relatively slow, sometimes fractious progress of associations of digital libraries and of collaborative digital initiatives. There are, to be sure, interesting examples of fruitful information interchange and shared investigation, particularly in the development of metadata standards, e.g., the Text Encoding Initiative (TEI), Dublin Core, and Encoded Archival Description (EAD).
One can also point to early and successful information exchanges, notably the task forces, which still exist, sponsored by the Coalition for Networked Information. Still, the ubiquitous penetration of networked technologies creates opportunities for deeper forms of association and resource sharing (with collection development, for example) that have been left almost entirely unexplored. Constrained organizationally from pushing technologies to their logical extent, young digital libraries fell back on safer modes of cooperation in talking shops that explored new ideas and in shared cataloging activities. The former resulted in statements of principle. The latter extended the purview of shared cataloging to some nonbibliographic records (e.g., archival finding aids and descriptions of some digitally reformatted materials) and explored mechanisms for constructing virtual, as opposed to union catalog, databases.
The young digital library is, in summary, an immature, experimental organizational form. It explores new opportunities and gathers new competencies. It does this within the safe harbor of soft-money projects and other activities that are organizationally, financially, and even culturally removed from traditional library services.
Jewell (2001) shows the extent to which leading research libraries have routinized the complex processes of identifying, evaluating, negotiating access to, and supporting use of electronic information that is commercially supplied by third parties. Pitschmann (2001) demonstrates an evolving and highly sophisticated understanding of the pitfalls, opportunities, and real costs for libraries that choose to organize access to "free" external Internet resources through various subject gateways, portals, and other linking services. Smith (2001) demonstrates that the digital libraries that were reformatting materials from their collections in the late 1990s (and not all digital libraries were), did so strategically rather than experimentally. Reformatting was being done, for example, to profile selected rare and special collections, to support specific teaching and research needs, or to manage and conserve selected general holdings.
The sea change in attitude and approach is partly economic. As part of the maturation process, digital libraries transform interesting skunk-works projects into essential library infrastructure. At this stage, failures can no longer be written off as learning experiences gained at limited cost and subsidized with external or soft money. Attitudinal shifts also reflect changing understandings of digital collections and digital library architectures. Where online collections are concerned, maturing digital libraries recognize that they are unable, by themselves, to supply end-users with what they really want—enough online information to meet their specific and evolving information needs. Consequently, they set a high premium on interoperability and on definition and adoption of the standards required to achieve it. The drift toward modular systems architecture is also responsible for new attitudes that favor adoption of more established standards, since systems modules (locally and globally arrayed) must interoperate at a fundamental level. Libraries at this stage demonstrate a desire to pool their collective uncertainties and to define, and then frame, a broad suite of practices as benchmarks in which they can all invest and upon which they can more safely and predictably build. Attempts to develop and codify digital library standards and best practices are tapping into this very fruitful seam.
Some of the library associations that take the lead in quantifying traditional aspects of library use have been relatively slow to respond to this new and pressing need. The reasons for this are complicated. To begin with, the metrics are complex and difficult to agree upon. How, for example, should we define what constitutes a "use" of a networked information object? Second, the library associations that are so well suited to developing statistics for traditional library use are typically membership organizations that are driven by consensus, which, in this case, is difficult to engineer. Further, the measures themselves can potentially disrupt the organization by fundamentally altering the criteria by which it admits and excludes new members. Debate about e-metrics is quickly transformed into debate about what institutions should be recognized as leading research libraries and is accordingly difficult to resolve.
Some of the best analyses of user behavior and need take place at the grassroots level in what can only be described as a series of largely uncoordinated guerilla attacks that are mounted at the institutional level and by ad hoc and informal associations. Denise Troll Covey uncovered a wealth of these in a survey of use-assessment methods at numerous digital libraries (Troll Covey 2002). Among the revelations emerging from these fragmented efforts is the extent to which users want to work in highly personalized and malleable online environments, that is, environments that present them with the information and services they actually need at any one time. The operational lessons for the library are twofold: (1) users want seamless presentation of collections and services, irrespective of where, by whom, or in what format they are managed; and (2) libraries should consider deploying user-profiling technologies that enable users to configure a networked information environment that meets their specific needs. Both lessons, if taken seriously and reflected in new operational services, have revolutionary implications for the library. The first would integrate the library into a globally arrayed network of information services in a way that challenges its historic organizational insularity. The second potentially obscures from the user's view the library's importance as a portal to that global network, because chunks of the library's collections and services are removed from the library environment and placed into new contexts.
The maturing digital library takes very seriously its users' needs and interests through its support for a suite of activities that have become known as "e-scholarship." Although the phrase has a frustrating tendency to take on new meaning every time it is used, its definition usually includes initiatives that enable scholars to produce and disseminate "publications" with minimal intervention from third-party commercial publishers. Overall library interest in supporting innovative forms of scholarly communication (or e-scholarship) at this point perhaps has less to do with transforming scholarship than it does with a strategy to increase pressure on publishers, who have increased prices dramatically in the past 10 years, particularly in the sciences.8
Virginia's quest for general technical solutions is more a response to users' needs than to economic forces. The media centers at Virginia have adopted their own range of approaches to the production, management, and distribution of the electronic information for which they were responsible, and each has an independent and highly distinctive online presence. The problem is that, from a user's perspective, the fragmentation of digital collections makes little sense. As Virginia thinks about capitalizing on the mass of digital information it has created or acquired through its media centers, it is beginning to envision an online environment that allows patrons to search seamlessly across collections, regardless of whether they were available in print or digital form, and, if in digital form, regardless of whether they were managed as texts, digital images, geospatial information systems, or other types of data. Virginia's notion of an online environment through which scholars can obtain high-quality information—irrespective of how, where, by whom, or in what format it is managed—required it to think hard about building a technical infrastructure that could be generalized and would be capable of supporting its heterogeneous collections.9
The apparent importance to the maturing digital library of a general core infrastructure is only amplified by investigation into newer digital library programs. NYU's initial investments in its digital library have been made with explicit reference to the experiences of older programs whose project work stumbled temporarily on a proliferation of diverse data objects and delivery systems. Not wanting to repeat a development path that was becoming well worn, NYU focused its earliest investment (as did Harvard, Columbia, and one or two others) on the development of core infrastructure, assuming that content and services could eventually be built on top of that infrastructure.
The maturing digital library's quest for core technologies seems to bolster rather than to undermine its emphasis on applied research. Michigan, Virginia, Harvard, and CDL, for example, have obtained or are seeking permanent, base-budget funding for the research and development efforts that are focusing on the identification of core technology. They recognize that such research and development funding is essential to their programs' successes, and that the programs cannot be sustained exclusively, or even largely, on external funding and soft money. At Michigan, investment flows into the development of DLXS.10 At Virginia, it is reflected in the Digital Library Research and Development Unit, established in 1999. At CDL, it is evident in the establishment of a small unit under a director for education and strategic innovation. Harvard is arguably the most interesting example. As a service to a confederation of largely independent library organizations, the Libraries Digital Initiative (LDI) did not launch as a series of independent projects. It began instead by building a common digital library infrastructure that would support Harvard libraries in their development and exploitation of digital collections. As it reaches the end of its first four years (and the end of the $12-million, one-time funding that launched the program in 1998), the LDI has developed first-generation infrastructure. In its second phase, it will seek a means of ensuring that the infrastructure will be developed and maintained out of whatever base-budget funding is available to the program.
The digital library's organizational integration into mainstream library services is another aspect of its maturation. In its startup phase, the program may be relatively unknown to library staff. Interviewees at more than one institution noted how the programs in their earliest phases were better known nationally than on campus or even within their host libraries. As digital library collections and services develop, their impacts are increasingly felt on local library staff. Reference staff find themselves fielding queries about digital library collections; catalogers consider descriptions for digital objects where once they focused exclusively on bibliographic materials; conservation officers find their work flow and priorities shaped to some extent by the progress of local digital reformatting activities.
The impacts do not flow in one direction only. Library staff members who are responsible for the development of print collections assert their perspective in setting priorities for digital collection development decisions. Bibliographic cataloging practices are considered as the library develops systems that allow users to search across collections, irrespective of their format and location. The hardware, network protocol, and systems choices that technical services make have implications for platform choices in the digital library. If the maturing digital library program emerges out of the skunk works and into the mainstream of library services, it does so in part because the program's work at some stage can neither be safely ignored by nor conducted in isolation from other library services. Rationalization is encouraged by the library's relatively inelastic personnel budget. Simply put, digital library programs are able to develop collections and services beyond their experimental phase only by organizing themselves so they may achieve an appropriate functional division of labor. At the very point when a digital library program is ready to hire its second (possibly even its first) metadata librarian, for example, it may also be ready to think seriously about establishing a stronger liaison with the cataloging department.
Rationalization is most evident at institutions such as Michigan and Virginia, which have older digital library programs. At Michigan, the program that reported directly to its three founding partners from the library, the School of Information, and the campus computing service was incorporated into the library, and the program's executive director was appointed as an associate director with responsibilities for IT services and electronic collections. This initial reorganization may have had more to do with changes in the personnel and priorities of the university's senior management than it did with any intrinsic need to rationalize digital library programs.
Still, once implemented, this structure created opportunities for—and even required—further organizational adjustments. Once the program was embedded within the library line, senior library management took pains to smooth its operation with existing library services. New money allocated to the library by a friendly university administration helped considerably in firming up infrastructure support and providing continuing funds for new program development. The library also introduced a committee structure that maximized opportunities for cross-fertilization among library managers in traditional and nontraditional units. By placing the Digital Library Services Division under the associate director, the digital library obtained representation on the library's senior management group—its executive council. From that perch it participated in thinking strategically and operationally about a broad array of library functions that interrelate with digital content, systems, and infrastructure. At the same time, managers responsible for those traditional functions helped think about and shape the digital library's future. As the program evolved, additional departments (the Scholarly Publishing Office, for example) were added as peer units to the existing operational one—the Digital Library Production Service.
Virginia's experience of reorganization is not altogether different, though it was not quite as fundamental. Virginia's program, unlike that of Michigan, was always represented in the library's senior management group through the deputy university librarian. One consequence is that the library's senior management had early opportunities to encourage the cross-fertilization of library staff in new and existing units. From an early date, the Electronic Text Center provided basic HTML instruction to library staff—offerings that were as popular as they were effective in giving those staff members insight into the opportunities, as well as the pitfalls, inherent in the online environment. Virginia's staff-share program was also effective in developing mutual understanding across library units and in supporting two-way transfers of professional skills. Structural reorganization, which came later, focused primarily on aligning effort at Virginia's media centers and on developing common services, such as digitization facilities and research and development units that they can share.
The DLF survey showed some evidence that consolidation of digital library activities is beginning to take place generally, at least with respect to the development of digital collections. For example, digital library staff members have a limited role in acquiring access to commercial electronic information. At 95 percent of the libraries responding to the DLF survey, that responsibility was vested in the subject bibliographers and other professional staff who were responsible for acquiring print materials. With respect to selecting materials from library holdings for digital reformatting, three-quarters of the survey respondents said that such responsibility was located largely, if not exclusively, in digital library departments and initiatives. Surprisingly, however, in half of those institutions, responsibility was shared with subject bibliographers—that is, with staff in traditional library services.
Reorganization of the maturing digital library has a number of practical implications. As staff members from outside the original skunk works become more involved in developing or supporting the use of digital library collections, the program is at pains to document its practices in various process recommendations, including, for example, guidelines for standards implementation, flow control, and rights clearance. Indeed, the extent of this gray literature provides some measure of a program's maturity. In the skunk works, processes and practices are in flux, and few are responsible for developing and implementing them. As the digital library matures, its collections and services rely much more on professional staff who are distributed across the library and who will require access to such reference materials. Again, the influences flow in two directions. While the maturing digital library typically takes responsibility for practical and process guidelines, the library's senior management, assisted by staff from across the library, supplies the policy framework in which the program's practices are applied. Early in the adolescent stage, for example, collection policies are revised so they can govern the selection of commercial electronic resources as well as bibliographic materials. Later developments include preservation policies that take account of digital as well as bibliographic materials and policies that articulate the library's orientation toward intellectual property and the exercise of the fair-use exclusion with both digital and print materials. Regardless of whether the documentation guides or reflects evolving practices, it is significant. Its very existence recognizes initial thinking about how digital library activities fit into and help fulfill the library's overall collection and service goals. For this reason, it is an important milestone of the maturing digital library.11
Institutions are differently positioned to effect the reorganization that is required to integrate digital initiatives into the fabric of the library. Staff development is a significant issue. However well suited professional librarians are to developing and maintaining high-quality digital libraries, they apply their skills differently in digital and traditional environments. Mastering different techniques requires training opportunities that all institutions are not equally positioned to supply. In addition, reorganization may require libraries to think anew about how, by whom, and in what combinations certain tasks are done. Here, too, there are numerous and substantial differences between institutions. Some lack the nimbleness and facility that is needed to effect such changes, which, at a fundamental level, will determine whether and at what pace a digital library program will mature.
As digital library programs mature, they gain visibility through their online presence, their success in acquiring external funding, or their ability to attract regional, national, or even international acclaim. They rely as much as ever on effective promotions to senior managers and to external funders. They also find themselves having to appeal to new audiences: to their faculty and other patrons and to other library staff whose own work, as we have seen, is influenced by and contributes to the digital library. Faculty members have a substantial stake because the campus library supports their research and teaching. They also have very different needs with respect to library collections and services—needs that turn on discipline, age, and personal preference. Without exception, the programs where interviews were conducted took pains to promote themselves to the faculty. This was true even of Harvard's LDI, which serves the faculties through the libraries.
At the institutions where interviews were conducted, promotional activities had three things in common. First, the programs took pains to demonstrate that digital libraries could be supported in a way that did not threaten funding for traditional library services, in particular for the development of print collections. It is unclear whether this is a realistic assessment of future funding models or merely an anticipation of future battles for the digital library.
Second, the programs capitalized wherever possible on successful services—that is, services that were well received by patrons, and by faculty patrons in particular. Senior managers at CDL, Harvard's LDI, Indiana, and Virginia were well aware of how much support for their programs is built on early successes, even those successes that owed little or nothing to the digital library program. CDL claims Melvyl (UC's union catalog) as a birthright, even though it was developed by the Department of Library Automation some years before the CDL's founding. CDL's continued maintenance and enhancement of Melvyl (e.g., with electronic "request" services) helps garner the public support it needs to run more experimental initiatives such as those associated with its e-scholarship program.
The credibility that Harvard's LDI requires to develop operational services that do not see the light of day for several years grows out of its host unit's successful and timely delivery of core services such as Hollis (the university libraries' shared catalog). Indiana's DLP builds on the visible successes of LETRS, the audio e-reserves VARIATIONS, and IUCAT, its union catalog. Virginia wrapped its first digital library forays in the cloak of earlier successes, notably with VIRGO, the library's online catalog, and with course tools, the instructional technologies that the library supports in tandem with the university's computing service. In addition, the programs take pains to promote new successes as they become available. In its first several years, CDL as a matter of policy launched a new service or a service upgrade every six months with appropriate system-wide publicity. Virginia's Electronic Text Center vigorously promotes the success it has had delivering its online texts as e-books. Michigan promotes an array of digital initiatives for users, such as geographic information systems (GIS), numeric data, humanities texts, and visual resources that have grown out of research projects.
Early digital library experience at NYU provides an interesting counter-example of the importance of service successes in garnering support for programs that might otherwise be perceived as experimental, or even peripheral, to core library activities. At NYU, in an early digital library initiative supported by foundation funding, the library attempted to place records for full-text electronic resources into the OPAC catalog environment. The initiative encountered several insurmountable obstacles. The host library was inappropriately organized to support an initiative that had both research and operational service components. It also lacked the needed technical expertise. Perhaps the largest hurdle had to do with the poor understanding of metadata that existed at that time. The project concluded, but it was never deemed to be successful by library staff and others who were involved in building and supporting use of the campus's scholarly information environment. This negative experience hampered the digital library program when it was begun again in earnest, and on a far more ambitious and comprehensive footing, years later. Professional staff had to be convinced that not all digital library initiatives were so speculative and so unfruitful.
A third common thread exists in the way that digital library programs promote themselves on campus. Interestingly, they portray themselves as fostering innovation while helping the library fulfill traditional roles. At Harvard, for example, the LDI presents itself both as a means of enhancing the university libraries' abilities to acquire, manage, and encourage scholarly exploitation of the digital scholarly and cultural record and as a mechanism for integrating digital and nondigital holdings. The innovative aspect of the LDI grows directly out of its traditional function, i.e., through the development of new collection management techniques, some of which allow faculty to search for the first time across Harvard libraries' disparate collections. Michigan's digital library program is also promoted as a means of developing and extending traditional collections, notably through digitization and robust access mechanisms that complement traditional bibliographic systems. Where Harvard finds innovation in integrated searches that the LDI permits, Michigan finds it in the enhanced functionality that scholars and students find in monograph and serial collections that can be explored as databases. Michigan's digital library program, especially its digital reformatting of monographs and serials, is a vital component of its traditional preservation program. It has recently moved its preservation-reformatting program from paper to digital as the default strategy for preserving brittle materials. NYU's embryonic digital library program, meanwhile, is oriented around the university's strong interdisciplinary interest in aspects of the performing arts and its location in New York City, capital of the performing arts. Virginia promotes its program's support for the library's historic service orientation toward students and faculty. Through the media centers, Virginia's digital library program helps faculty members in their research and teaching. It also supports faculty innovation in the classroom and in scholarly publishing.
The model is not so fantastic. Indeed, it is already beginning to emerge. JSTOR acts as a distribution agent responsible for ensuring online access to many scholarly journals. It is also an archival repository that ensures that these journals persist in their digital form. As libraries develop trust in JSTOR as both a distribution point and a digital archival repository, and as library patrons get used to accessing scholarly journal content in electronic form, libraries are beginning to remove from their shelves copies of the print journals upon which the JSTOR collection is based. "Deaccessioning" print volumes makes economic sense and allows the library to redirect scarce resources once tied up in the management of JSTOR journals—resources that paid for shelving space, print preservation, circulation, and reshelving work, for example (Lawrence, Connaway, and Brighman 2001). The scenario evolving around JSTOR is missing only one ingredient—a small number of print repositories that ensure persistence of the printed journals that make up the JSTOR collection. Still, one hopes and suspects that it is only a matter of time before at least a few such repositories emerge on national and regional bases.14
The model that relies upon distributed provision of interlocking service functions works as well—or better, given the volume of information—for materials that are born digital (i.e., those that exist from the outset in a digital form, whether as databases, geospatial information, images, sound, video files, or mixed media applications). The model is being tried in the delivery of library reference services, but it is too soon to assess the result (McClure and Lankes 2001). There is evidence that the model's development need not be impeded by legal and business considerations such as those that may surround the distribution of copyrighted material. The experience of JSTOR and other services, such as the Art Museum Image Consortium (AMICO) and HighWire Press, suggests that there are models that satisfy those who hold intellectual property rights or copyright to information as well as those who have an interest in using that information for research and teaching.
One might argue that the library itself is the single greatest obstacle to a more distributed and economically rational provision of information services. It is difficult to cede to third parties responsibility for collections and services that have historically been provided in-house and upon which library patrons rely so heavily. It is especially difficult when those who are forced to consider such fundamental reorganization are encumbered with professional, cultural, and organizational baggage that defines a high-quality library as one that supports in a single place a very wide range of collections and services—a range so wide that it may now be beyond the reach of any single library.
The fragmented organizational approach to information provision that exists on most large campuses harkens back to a prenetworked world when discrete information service functions were built on and delivered through distinctive technical platforms. In the early 1970s, campus telecommunications, mainframe computing, audiovisual departments, and libraries delivered discrete information services using very different technologies. In 2002, these discrete functions have migrated into a networked space where they are vastly more difficult to distinguish from one another, in part because they rely upon the same infrastructure. Convergence is evident to some extent in the DLF survey data: 55 percent of the libraries responding noted some formal relation with their IT or academic computing service. Ninety percent claimed either a formal or an informal relationship. Only two respondents (11 percent) indicated they had no relationship, formal or informal, with any other information organization on campus. Convergence has also fostered competition as information organizations assert their claims against that portion of the university's resources that can be devoted to information management generally.
The DLF survey documents not only the problem's complexity but also some interesting opportunities for addressing it. Excluding administrative and financial data from its purview, the survey sought to determine where libraries perceived the loci of responsibility for the production, distribution, and maintenance of electronic information produced on campus resulting from or supporting research and teaching. Responses demonstrated that the production and distribution of online research and teaching materials was taking place outside the library—in academic departments and the units responsible for providing academic staff with technical support.16
At the same time, libraries perceived themselves as one of the few units on campus with an interest in preserving these materials. Therefore, the library may be positioned to stake a credible claim to some custodial role with respect the campus's digital information assets. The library may fulfill this role in cooperation with other units that have better capacity for providing some of the technological functions associated with large-scale data management and preservation. Some progress is clearly being made. The survey demonstrates that 4.8 percent of the DLF member libraries have developed or contributed to digital preservation policies that include electronic information created on campus but outside the library. At the same time, of four DLF member libraries that are developing digital repositories, only one is doing so in collaboration with other campus-based information organizations.
The waste is excessive, even in good economic times. Bibliographic references that may be included with online course materials are not automatically hot-linked to the library's online catalog; in some instances, they cannot be linked. The digitally reformatted materials produced by the library, and the many more such materials produced at other libraries, museums, and archives, and to which most libraries now link through various means, are placed outside the pedagogical purview as it is defined online by instructional technologies. The digital library as a learning resource is therefore put at risk. Meanwhile, faculty and students work in isolation from a wealth of well-organized, high-quality information that is directly relevant to their learning and for which they have already paid, for example, through university and library spending.
Some emphasize the digital library as a collection development arm. CDL, Cornell, Harvard, NYU, and Stanford, for example, are exploring means of archiving selectively those scholarly materials that are uniquely available in electronic form. Michigan, Cornell, and Yale are using or have used digital library techniques to develop and manage fragile printed materials. There, digital surrogates are created for fragile books and used to develop print and online access copies to save wear and tear on the originals.
Some libraries are beginning to realize how digital technologies might help reduce the costly and redundant management of print materials held by numbers of research libraries. Where libraries can reliably deliver access to online editions of scholarly journals (editions that evidence suggests faculty and students prefer anyway [Dagar, Greenstein, and Watson Healy 2001]), they are beginning to be able to reduce redundant print copies while ensuring that at least some print editions are made persistently available.19 Nine campuses of the UC are canceling redundant print editions of selected journal titles while ensuring that sufficient print copies remain in selected campus libraries or shared storage facilities. The 11 Midwest university libraries in the Committee on Institutional Cooperation are beginning to use the Center for Research Libraries in Chicago as an archive for the print journals included in JSTOR. Similar approaches are being explored for other materials.
Some digital library programs emphasize services over collection. But almost half of the DLF members digitize selected rare, special, and other collections as a means of encouraging access to and use of those collections by scholars, students, and a broader public. Library digitization programs were driven initially by curatorial interests. They targeted unique special collections that research libraries owned. Today, an expanding dialog between libraries and scholars is allowing libraries to consider what online resources humanities scholars need. In the humanities, scholars investigate meaningful bodies of information and build ideas and conclusions from a context. Therefore, which collections are digitized will make a difference in how quickly online scholarship will grow. Research using digital resources is beginning to have an impact in the humanities, although it has not replaced the traditional critical mass of books and archives that scholars investigate. Research projects such as the Valley of the Shadow and Perseus are moving disciplines into uncharted and exciting online research territory, and some online journals now link to digital repositories.
Conversion of analog materials to digital is often confused as representing the sum total of what digital libraries do. Clearly, it is not. Experimentation in new forms of scholarly publishing, notably at Virginia, but now increasingly elsewhere, is another kind of service orientation. So is the provision to students of online class resources, such as sound recordings and art history images at Indiana. Other variations can be seen at Carnegie Mellon University, North Carolina State University, and the University of Illinois, where digital library programs explore in depth how users behave in online environments with a view to improving those environments.
The digital library pioneers who led their organizations at an exciting and exploratory time created diverse programs, but with synergies that hold promise through the application of network technologies. It is obvious that network technologies permit users to access information regardless of where information is located, but there are still significant obstacles to deep resource sharing, including the effect that competition among universities has on library cooperation. In addition, the question of whether and to what extent the networked activities of research libraries will affect or inform public libraries, or even those academic libraries that are more oriented toward teaching, remains unanswered.
Historically, libraries fulfilled diverse functions by acquiring and assembling collections, services, and professional staff in a single place. Access to the library and its professionals required proximity to them. The rapid development of network technologies has not eradicated the significance of the library as place and as owner of collections, but it has diminished and will continue to diminish it. Network technologies are forcing a fundamental transformation of the library and the university itself. Against this background, the record of accomplishment recorded in these pages is extraordinary.
In the research library, achievements occurred precisely because these libraries were located in large and complicated environments. First, they existed within the framework of research and development that occurs on large campuses and were, in some instances, able to harness that expertise to their advantage through collaboration. Second, with significant budget, staff, and space, they were able to move discreetly some resources from traditional library activities into digital research and development. To do this, they needed to have a strong vision of the future that enabled them to push a largely process-driven environment into a research and development arena— something that was mostly unknown in the research library in the late twentieth century. This is why so many young programs relied on skunk works or laboratories where library staff worked away from daily operations.
In the traditional environment, the size of the academic library's collection was the single biggest determinant of its national and international status. How the collection was used and how often it was used were matters of smaller concern. In a networked environment, the reverse is true. Print collection size and scope continue to be important in the research library community, but as rich local holdings are supplemented by the ever-broadening array of networked information that includes digital collections, the importance of the physical collection's size is diminishing in favor of instant desktop access. In some disciplines, books from the library's stacks are still used heavily and will continue to be used, but in other disciplines networked resources are eclipsing print.
Because large research libraries are developing and presenting information online, they are grappling with how to present networked information—its look, feel, functionality, and context—and taking on new interpretive and publishing roles. For the first time in decades, research libraries are thinking hard about how to develop and support use of their scholarly and teaching resources. Not surprisingly, the expertise required to plan and create online collections and services that bring together resources for teaching and scholarship is rarely found completely within the library. It occurs instead through the combined thinking of scholars, information professionals, and technologists. The quest might have the beneficial result of integrating the library into the academic community after decades of it being a stand-alone resource. In some cases, the survey and case study data suggest that this trend is already occurring.
The coordinated provision of networked information services is probably the biggest challenge that libraries, along with everyone else in the academy, face today. In fact, the "stovepipe" organization that is common within universities and the tradition of faculty working as individual entrepreneurs are both significant barriers to the inclusion of the library in enterprise-wide planning. The current rush by faculty to create online learning environments is a good example. Without a campus-based technology approach and collaboration with the library to include digital resources already available, course management systems are doomed to be redundant and limited endeavors at a time when budgets are already stretched too thin. In an environment that has long valued individual, department, or school achievements and that mostly markets itself in bits and pieces to its various constituent groups, the notion of a more coordinated approach to the provision of networked information services, e.g., the digital library, course management tools, or preservation of the university's digital assets, is one that is strange to many in the academy. (More than four years ago, Brian Hawkins and Pat Battin [1998] explored the critical interrelationship between the library and the higher education enterprise as well as the need to transform the academy's traditional organization in order to use technology effectively.)
Early digital projects initiated within the research library have now grown into programs. Programs that began in separate organizational units within the library are moving into the fabric of everything the research library does. As research libraries develop new digital collections and access tools, new physical spaces where scholars, information professionals, and technologists can work together on digital challenges, and new ways to assist scholars in their work (e.g., helping to preserve and access research in digital repositories or assisting scholars to publish with minimal intervention from commercial publishers), the next challenges will include the ability of the research library to embed itself ever more deeply in the scholarly environment and in the transformational change that will occur in the academy at some point from the pervasive use of information technology.
2 A list of these libraries appears in Appendix 1.
3 See Jewell 2001, Pitschmann 2001, Smith 2001, and Troll Covey 2002. The DLF also commissioned Outsell, Inc., to conduct a survey of the dimensions and use of the scholarly information landscape. A report is forthcoming.
4 This is also true in the development of the Library of Congress's American Memory program.
5 For example, Indiana tried three times without success to obtain relatively small amounts of funding from the Library of Congress/Ameritech competition. Three years after its first attempt at that competition, the program received a $3-million grant from the National Science Foundation.
6 DLF-endorsed standards are available at: https://www.diglib.org/standards.htm. Criteria for a digital archival repository of electronic journals are available at: https://www.diglib.org/preserve/criteria.htm.
7 This is becoming evident in work conducted by the DLF with Outsell, Inc. The work is described in Dagar, Greenstein, and Watson Healy (2001).
8 Libraries, as collectors of cultural and scholarly heritage, are in a strong position to supply the services that help researchers, teachers, and learners navigate, find, interconnect, interpret, and use information in whatever form it exists. Libraries are also poised to capture those interconnections and interpretations and manage them as new information. Instead of concentrating on introducing new electronic scholarly journals that attempt to compete with expensive and traditional commercial e-journals—a business that libraries know little about—libraries may find that they are better off building and exploiting traditional services to support and nurture new forms of scholarly communication.
9 See details about Virginia's work with FEDORA at: http://fedora.comm.nsdlib.org/.
10 Core or base-budget funding is not the only means by which Michigan supports what it sees as essential technical development work. Having developed a suite of digital library tools in DLXS that can be generalized, Michigan supports their adoption by other programs, thereby supplementing its technical development funds and offsetting some of its development costs.
11 The local policies, standards documents, and implementation guidelines that have grown up around DLF members' digital library programs are listed in a database at: https://www.diglib.org/pubs/techreps.htm.
12 Colleges and other institutes of higher education that emphasize teaching over research also developed academic libraries to physically assemble and collocate information, individuals, and services. Their collection and service purview, however, is more defined by local pedagogical requirements than is that of the research library.
13 This does not suggest that, henceforth, printed books should be read only "on screen" or on some hand-held device. Rather, we should begin to think more creatively about delivering electronic information through a variety of means, including, for example, localized print-on-demand services.
14 The Center for Research Libraries is attempting to acquire a complete file of the print copies of journals available through JSTOR. See http://wwwcrl.uchicago.edu/info/jstor/crljstor.htm.
15 This is evident in the heavy use that scholars make of the few archives that have been established to collect and manage these data, for example, the Inter-University Consortium for Political and Social Research in Ann Arbor, Michigan.
16 The library's role in producing and distributing electronic information was focused almost entirely on digitally reformatted materials drawn from library holdings. Although the library developed the materials in support of research and teaching, they were frequently not developed in consultation with scholars involved in those research and teaching activities.
17 Costs for cataloging collections and making the catalogs accessible online are probably a library's greatest cumulative expenditure. Although bibliographic information for print and electronic holdings is now regularly funneled into the online catalog, much remains to be accomplished in special and archival collections, prints and photographs, microforms, slide libraries, sound recordings, maps, and the numerous other special collections that research libraries own and manage. The task is much larger if extended to Internet resources.
18 At this point, the amount of money spent annually by DLF members on subscriptions to commercial electronic information dwarfs all other digital expenditures. Electronic journals and reference databases are paramount, but libraries also collect or provide access to electronic texts, statistical data, digital images, geographic information systems, and other digital objects.
19 Most large research libraries use off-site warehousing for less-used print and other materials (Association for Research Libraries 1999).