This brief paper explores changing conceptions of digital libraries and how they fit into the broader information flows and information landscapes in Higher Education and beyond. It includes a few observations comparing how thinking about these questions has evolved within the very different planning, funding and implementation contexts in the United Kingdom and the United States. The central argument here is that, over the past decade in the Higher Education and research sphere, we have seen several large-scale technological shifts, with all of the accompanying organisational, social and cultural changes, proceeding largely independently - the transformations of scholarly practice, of teaching and learning, of scholarly communication, and of 'traditional' research library services. In particular, the resulting transformation of traditional research library services is not necessarily well matched to the range of needs presented by the other transformations. The independent and parallel nature of these transformations has led to a great deal of focus on questions that may ultimately be proven to be, at best, poorly posed, and at worst, of little importance; notably 'what is a digital library?', or 'what do digital libraries have to do with research libraries?'
I do not have space to detail the enormous changes that are well advanced in the practice of scholarship both in the US and the UK (and indeed worldwide). But it is important to sketch some of these changes at a high level to provide context for the discussion of libraries, digital libraries and new infrastructure for research and education that is to follow. To provide a sense of timeframe, while these developments have roots that go back decades, they really emerged in the mid-1990s and gathered tremendous momentum (and really substantial funding) around the turn of the new century.
These changes include the growing reliance on computational resources for modelling, simulation, analysis, and data collection; advanced observational and data capture instrumentation; the use of advanced networks, to access and share data and computational resources and experimental and observational apparatus; collaboration technologies that allow groups of scholars to work together independent of geography, both synchronously and asynchronously; the increased importance of very large datasets and databases both as evidence to support scholarly inquiry and as a means of documenting, structuring and communicating scholarship. One might think that these are mostly applicable to the sciences and engineering disciplines, and indeed much of the early adoption did come from those sectors -- 'e-science' was an early shorthand term for these changes in scientific practice.(Perversely, in the United States, we refer to 'cyber-infrastructure' as the shorthand term, taking about what is needed to support the changes rather than the disciplinary changes themselves, largely as a result of the ground-breaking report of the Atkins committee . The terminology of grid computing is also often used as a shorthand for many of these developments, and in anthologies like  or  one can find many disciplinary case studies of scientific transformation. But in more recent years, we refer to e-research or e-scholarship as well, as a way of recognising that these same shifts are rapidly sweeping the social sciences, the humanities and even areas of the arts, and that indeed some of the most creative and transformative applications of information technology and digital content are emerging in these sectors .
As the practices of scholarship change, so of course does scholarly communication, which reflects current practices of scholarship. Thus, we see traditional books and journal articles, for example, being supplemented by a deluge of data, multimedia, computer software, simulations, interactive tools, Web sites, databases, and, continually updated online community information resources of various types.
Teaching and learning has also changed, with the emergence of learning management systems (or virtual learning environments, as they are commonly called in the UK); the commonplace use of not only online courses but 'blended' courses that include both online and face-to-face components; and of course the transfer and adaptation of much of the e-science technology (remote access to equipment, virtual lab benches to perform experiments by simulation, etc).
For a wide-reaching look at these developments from a US perspective, see .
Starting in the early 1990s, research libraries in both the US and UK began seriously to ponder what the growing onslaught of digital data, advanced information technology, and high-performance networking meant for their future.
It is striking to me how differently this future unfolded on opposite sides of the Atlantic. In the UK, explorations were characterised largely by the eLib Programme  and considerations of how the overall system of scholarly publishing and communication might evolve, and the place of libraries in this evolution. It was also characterised by a much greater emphasis on the deployment of a national system of library services than we found in the US, which I will discuss in the next section. In the United States, while research libraries also adapted to the modernisation of the traditional scholarly publishing system as it moved to electronic publications, and moved into local digital collection building and digital dissemination roles, they did so much more on an institution by institution basis, sometimes forming statewide or regional consortia, and certainly sharing experiences and evolving best practices, but not as acting as a national system.
In the mid-1990s the US National Science Foundation launched a digital libraries research programme . This involved levels of funding that were very different from the kinds of grants most research libraries were accustomed to; furthermore, these grants were targeted mostly at groups of computer scientists building research prototypes, with libraries (and publishers) typically only peripherally involved, along with social scientists trying to understand how these prototypes were being used. (To be sure, some of the NSF phase I projects did involve libraries or publishers in more than peripheral ways, but this was the exception, and the culture of the projects was firmly disconnected from the research library community. Indeed, how could a research library make sense of an information management and retrieval system that was being deliberately designed to be short-term and experimental in nature as anything other than a technology prototype?)
But digital libraries were fashionable, they were well-funded, they generated great interest during the great 'dot-com' bubble, and they were frankly sometimes threatening (and sometimes deliberately used as a way of threatening) research libraries in the US - if these libraries were not on the road to becoming digital libraries, they were backwaters, obsolete, 'book museums'; they were in danger of being supplanted or overtaken by commercial competitors. Much of this was, to be blunt, complete rubbish, at least in the near term, but the development of these information management and retrieval systems that were called 'digital libraries' and the confusion between these and what actual libraries as organisations do, and the systems that they might use to accomplish those missions, gave rise to a major problem in public perception.
In the US during the latter half of the 1990s there was great obsession within the library (and broader Higher Education) communities about how to define a digital library    and about how with the present and future forms of research libraries. As most of the funding for digital library research - and particularly for massive prototype projects - dried up, and the NSF-lead programmes focused much more on technologies that might be helpful in a wide range of information management and retrieval settings, the character of this debate shifted and became, I think, more constructive. The focus shifted to questions of how research libraries could more effectively support teaching, learning and scholarship in a changing environment.
By about 2000 the collection of network-based information resources and services supporting Higher Education had become extraordinarily complex, diverse and crowded. Another aspect of this shift of focus away from the teleology of digital libraries was an interest in understanding these other information resources and services, determining how library-provided resources and services might best interact and interface, and ultimately often trying to determine which functions and purposes might be best assigned to which class of resource or service. The UK library community was perhaps a bit earlier than the US in explicitly recognising some of these issues, for example with the important 1998 conference on 'Information Landscapes for a Learning Society: Networking and the Future of Libraries 3' , but the issues here have received growing attention on both sides of the Atlantic as this information landscape (or information ecology) has continued to grow and diversify. Today, this includes systems and services such as Learning Management Systems/Virtual Learning Environments, Institutional Repositories, Portals, Digital Libraries, Digital Collections, Digital Asset management system, and various departmental, workgroup and personal systems at the local institutional level, access to similar systems at other Higher Education institutions, plus a range of national and international data archives and more general repositories, specialised research corpora, commercial offerings from publishers, scholarly societies and other, and collaboration environments of all types.
In UK Higher Education planning and funding for libraries and information technology has a strong centralised element; this has facilitated developments as diverse as Super Janet, national site licensing; national authentication strategies such as Athens  ; the implementation of national disciplinary data archives such as the Arts & Humanities Data Service , or support services such as the Digital Curation Centre . It has allowed the development of strategies to address the information landscape problem discussed earlier such as the Resource Discovery Network  and a system of specialised subject gateways. I would not want to describe any of these developments as 'easy' - I know that they were not, and that in fact they required a great deal of thoughtful crafting and consensus building by the UK Higher Education, library and information technology leadership - but from the perspective of an American, it seems such initiatives are 'easily' accomplished in the UK in comparison to the situation in the United States. The key element in the UK system which is lacking in the United States is a well-established way to move from conceptual thinking to programmes and implementation.
In the United States, with relatively little funding or central planning for Higher Education as a system, it is very hard, complex and time-consuming for the library community to translate intellectual consensus into programmes and from there into deployment and operation. The typical pattern is a series of individual institutional choices (within the context of sharing of thinking, plans and best practice through various national organisations); perhaps the development of regional consortia or the use of existing ones; or the construction of not-for-profit organisations that can channel and aggregate funding from libraries or Higher Education institutions, such as Internet 2 (properly the University Consortium for Advanced Internet Development ), JSTOR , ICPSR  or OCLC . Research libraries in the US as well as the UK recognise that they are going to need to follow the path of greater interdependence and interreliance, of institutions specialising and offering specialist expertise broadly throughout the Higher Education community, of more common infrastructure, but it is very hard to do within the US framework. Interestingly, there is considerable organisational and economic innovation going on in this area, much of it involving the design of mediating not-for-profit organisations; both Ithaka  and Educause  are doing some very sophisticated thinking about these issues.
The scholarly literature - to say nothing of the broader cultural record that forms part of the essential evidence that underpins future scholarly inquiry - is enormous; the systems of publishing and cultural production that create it are large, complex, and diverse. Neither change quickly.
It is really just in the last few years that research libraries have begun to take effective action - national deposit agreements; the JSTOR Portico programme; the Stanford LOCKSS system - in order to deal with the transition of the scholarly journal publishing system from printed to electronic editions. As with so many of the changes libraries face in the digital environment, the solutions involve interdependence, specialisation and service provision; and co-ordinated collective actions; they involve the design of solutions for the system of research libraries.
Similarly, it is just in the past few years that advances in information technology price/performance have reached the point that massive digitisation of the historical scholarly and broader cultural record has become possible, although there are still profound non-technical barriers to such digitisation due to intellectual property issues. Here again, we see emerging discussions in the US about how libraries might coordinate the digitisation of public domain materials, and national projects in the UK to address major national collections. Third-party not-for-profit players like JSTOR have emerged to digitise and preserve large swaths of the historical scholarly record on behalf of the higher education community. We are also seeing the emergence of centralised commercial players, notably Google, in the mass digitisation sphere. These mass digitisation efforts are too important to the future of scholarship for any research library to ignore, but also too large for any research library to take on independently.
One issue that I predict will move to center stage soon for the higher education community and its libraries: As the historical scholarly and cultural records shift to digital form, the way scholars, commercial companies, and many other groups will make use of them will also change. We will move from reading first to searching (where technologies and practices are already well established, including internet search engines) and then quickly from searching to much more complex and sophisticated computation - text mining, inference, correlation, analysis. To the extent that this offers results of importance to the research community and particularly to the public - help in scientific or medical breakthroughs, for example, there will be growing demand to facilitate it. (It is no accident that some of the heaviest investment in these technologies for the analysis of literature corpora, both in the public research sector and the private sector, are aimed at the biomedical literature.) The legal, licensing, and technical problems in supporting this kind of computational use of major literature corpora are immense.
Looking to the future, it is also taken some years to begin to understand how changes in scholarly practice as part of e-research create requirements that go far beyond the modernisation of the traditional scholarly system into questions of data curation and stewardship, for example; or strategies for managing very large and complex software collections over time; or the demand for disciplinary informatics specialists to support faculty research. This understanding and the underlying requirements are of course still evolving, and indeed will always continue to evolve, but the broad outlines of the new scholarly communication practices seem to be getting clearer. Perhaps we are at the point now where research libraries, in collaboration with information technologists, scholars and researchers, and others, can move to the next stage in the co-evolution of these institutions and the services they offer.
Perhaps the great mistake that we made during the late 1990s in all of the discussion about digital libraries and the transformation of the research library (presumably into some sort of appropriately defined digital library) was that we were trying to translate artifacts of existing scholarly workflow and social practice into the digital realm rather than starting with an understanding of how the workflows and social practices of research, scholarship, teaching and learning, managing and preserving the scholarly record changed in the digital world, and indeed how we might make them more effective and robust; and only then looking round for the institutions and services necessary to support these new practices. We started with the question of how the library should change, rather than the question of how the world of scholarship that the research library serves was changing, and what that demanded from libraries.
There is an insightful and important arc of research on the connection of library and information services to the changing demands and practices of research - see  for one set of perspectives, and some of Lorcan Dempsey's excellent recent thinking about 'networkflows' (a classic Dempsey punning coinage) (see for example, ) recently refocusing attention on this in a more pragmatic way . Interestingly, Dempsey points not just at the intensively-discussed issue of changing practices of scholarship that I've emphasised here, but also at the emergence in the UK and elsewhere, in response to research assessment and funding practices, of 'research support systems' which help to automate support for the administrative workflows imposed on academic departments by these developments.
And there is much more that is relevant: studies about data curation and preservation needs (for example, ); studies of faculty informatics needs and practices ; studies of disciplinary data sharing, deposit, and reuse norms and practices; studies of changing practices of scholarly authoring (see, for example, the work of Peter Murray-Rust in Chemistry or the work of Ed Ayers and Will Thomas in History); and studies of how institutional repositories are being used (see, for instance    to name but a few).
So what has happened to the digital library? At least as I define digital libraries, what happened was that we realised that they are just tools, a bundle of technologies and engineering techniques - that find applications in a surprisingly wide range of settings beyond higher education and research. They are now providing essential services in many areas of commerce and professional practice - for example, law (a very early adopter with Westlaw and Lexis), medicine, finance, etc. Digital libraries also play an essential role in the service offerings from a huge range of government entities. Interestingly, they do not show up in these settings with all the library context and baggage that made them so confusing in the research libraries setting; there is no expectation of stewardship beyond the point of regulatory requirements or economic viability for content in most of these digital libraries. Even questions about the broad public access that networked information makes possible is contextual: for public government information, this is an advantage, but many of these systems are designed for closed, high-paying user communities and there is no pretence of democratising access.
One can also ask very good questions about what happens to public libraries in a world more and more infused with information technology and digital connectivity - but also, at least in the United States, a world where there is more and more contention for diminishing public funds, where society is getting more competitive and economically stratified and the social safety net is vanishing, leaving a substantial number of people desperately needing access to education, and social services, including information-oriented social services, and relying on the public library to provide these services. In the United States, I do not think that most public libraries have had an identity crisis over their potential future as digital libraries - they do not have the time, being overrun with demand and under-resourced. (I am in no position to make informed comments on what is happening with public libraries in the UK.) Or - and I would argue that it is getting clearer this is probably a very distinct question - one might ask questions about what happens to cultural memory, its creation, its care, and its use in an ever more digital world, and the roles of public libraries and other cultural memory organisations in the future of cultural memory; one might also speculate about the roles that digital library technologies might play in this future. But these would take us far from the focus here on libraries, and specifically research libraries, embedded within the research and higher education sector. I mention them here only to suggest to the reader that the history and future outside the research libraries in higher education follows different paths.
As we look at how developments may continue to unfold in the coming years, the defining differences in the US and UK future are not just how research libraries and other parts of the infrastructure supporting research and higher education are funded, and the structural provisions for translating a conceptual consensus into planning and then into funding and implementation within higher education. Also crucial here are the differences in the dynamics of the strange mix of community, collaboration and competition that characterises relations between and among higher education institutions in a given national context. Just as the shift to a network-based world has mandated that research libraries function as a coherent system, we will see similar pressures on the overall higher education system over time; but the specifics here will vary greatly across the Atlantic.
We are in the middle of a very large-scale shift. The nature of that shift is that we are at last building a real linkage between research libraries and the new processes of scholarly communication and scholarly practice, as opposed to just repackaging existing products and services of the traditional scholarly publishing system and the historic research library. In this shift we have left the debate about digital libraries behind, recognising this now as simply shorthand for just one set of technologies and systems among many that are likely to be important.
Thinking about infrastructure for the practice of scholarship, the conduct of science, teaching and learning, management of the intellectual and cultural record; thinking about workflow - or, to use Dempsey's term, 'networkflow' -- as all of these activities are increasingly facilitated by the networked information environment: these are the ways forward to understanding the future of research libraries in Higher Education over the next decade.
A special thanks to Richard Waller for his patience and steady encouragement as I prepared this article, and for his superb editorial help.