Washington, D.C.
October 15-17, 1997

Preservation of Digital Information

Toward a System of Digital Archives: Some Technological, Political and Economic Considerations

The Rearranging Effect

Donald J. Waters, Director of the Digital Library Federation
Council on Library and Information Resources (CLIR)

The use of technologies, including the use of those that underlie emerging digital libraries, are subject to what historian Edward Tenner calls a rearranging effect. Malcolm Gladwell called attention to the phenomenon in a commentary earlier this year in the New Yorker. He observed that anyone standing on a city subway platform on a hot summer day experiences a rearranging effect.

Subway platforms seem as if they ought to be cool places, since they are underground and are shielded from the sun. Actually, they’re anything but. Come summer, they can be as much as ten degrees hotter than the street above, in part because the air-conditioners inside subway cars pump out so much hot air that they turn the rest of the subway system into an oven. In other words, we need air-conditioners on subway cars because air-conditioners on subway cars have made stations so hot that subway cars need to be air-conditioned. It’s a bit like the definition the Viennese writer Karl Kraus famously gave of psychoanalysis: “the disease of which it purports to be the cure.”

Not all technological advances result in this kind of problem, of course. But it happens often enough so that when someone comes along making spectacular claims in behalf of a new technology it’s worth asking whether that technology really solves the problem or simply rearranges the hot air from the car to the platform (Gladwell 1997: 7).

And so it is with digital library technologies. Popular rhetoric attributes considerable transforming effects to them. Yet, digital information and the technologies on which they depend are extremely fragile. Their fragility makes it highly uncertain that digital libraries can endure over time and it makes one wonder about the durability of their supposed benefits. Does the emergence of digital libraries, rather than helping to transform research, learning, and other forms of scholarly communication, promise merely to rearrange the hot air from car to platform, shifting the balance of information services from enduring to immediate access? If such a rearranging effect is indeed at work, then systems of scholarly communication will have to pay dearly to compensate for the loss of the information that they generate and on which the quality of future scholarship depends. At a time when higher education is already under fire for, among other things, its soaring costs, we must avoid such a perilous outcome. But how?

We have heard this morning and, just now, from Deanna a variety of steps that we need to take, can take, or are taking to preserve digital information and to avoid the rearranging effect in digital libraries. In the next few minutes, I want to focus your attention specifically on the argument of the Task Force on the Archiving of Digital Information that the problem of preserving digital information for the future is not only — or even primarily — a problem of fine-tuning a narrow set of technical variables. Rather, in the words of the Task Force report, it is a problem of organizing ourselves over time and as a society to maneuver effectively in a digital landscape. It is a problem of building — almost from scratch — the various systematic supports, or “deep infrastructure,” that will enable us to tame our anxieties and move our cultural records naturally and confidently into the future (Task Force 1996: 6).

The Challenge of Creating Deep Infrastructure

In thinking about the steps necessary to build a “deep infrastructure,” or system for digital archiving, I suggest we start with a series of assumptions. First, I am assuming that the pressures to transform systems of scholarly communications do not arise solely, or even primarily, from technological imperatives. Rather, new digital technologies give us tools with which to respond to profound political and social impulses in the emerging knowledge economy. Peter Drucker (1993) and others have identified and analyzed these impulses. Our success in responding to them depends on how astute we are in identifying and understanding the key organizing principles of the knowledge economy and then in designing appropriate means of applying the technologies as part of its overall development.

Second, I assume that one of the key organizing principles of the knowledge economy is that libraries play an essential preservation role in the pursuit of knowledge. Libraries can, of course, take a variety of organizational forms, including those that individuals manage for their own personal use, as well as more elaborate entities such as public, corporate, academic, and research libraries. The timeframe over which libraries preserve works of knowledge can also vary from the relatively short span of an individual’s personal library to the centuries covered by a large national or university research library. Regardless of their particular form or the timeframe of their preservation objective, what libraries preserve is the integrity of the works they contain so that these works are reliably and economically available to the individuals or communities whose pursuits of knowledge they support. Depending on their scale, libraries preserve the use and usability of works of knowledge through the more or less elaborate management of a set of operating features such as the selection of particular works for the collection, as well as their acquisition, cataloging, storage, and circulation.

As a new variant of the form, digital libraries manage collections of digital works. Like libraries of other kinds, digital libraries must organize themselves to preserve the integrity of the works they manage for use over time by the individuals or communities that they support in the overall knowledge economy. However, the fragile nature and other distinctive features of digital information give special shape to the essential preservation function of digital libraries and to the core features of their operations.

My third assumption is that preservation means different things to different people. A significant challenge of the digital environment is for communities of common interest to ensure that the information crucial to their pursuits of knowledge endures. One can make great sense of the apparent chaos of the World Wide Web by focusing on the various, and sometimes interlocked, communities of interest that have found a home there (see, for example, Hagel and Armstrong 1997, and Hof 1997). So, too, one can chart the building of the “deep infrastructure” needed to preserve the use and usability of digital information in the emerging knowledge economy by distinguishing and tracking the development of the various communities of interest in the pursuit of knowledge. These communities differ, of course, and are even changing, in the nature and subjects of their common interests, in their uses of digital information, and in the corresponding development of their digital libraries. However, for our purposes here, let us focus on the uses of digital libraries in the communities we know best, namely those embraced by research universities.

Let us also remember the distinctive characteristics of digital information. One can readily copy digital information without changing the original and one can transmit it widely over networks. These and other features have a variety of implications, but among other things they mean that the use and preservation of digital information are subject to economies of scale and to organization in digital libraries that differ from those in libraries of paper-based materials. Many of the functional attributes — the need for circulation, cataloging, and reference, for example — are the same. However, the division of labor to realize digital economies of scale can and almost certainly will result in digital libraries that effectively manage their collections by allocating functional responsibilities for their operation largely outside their organization, in ways that are quite different from how we are presently accustomed to seeing them. Indeed, if we look closely at the research university, we can see that the political, economic, and other conditions that shape the use of digital information in this community of our common interest are giving rise, before our eyes, to new and distinctive kinds of library organizations.

Developing the Deep Infrastructure for Digital Archiving

To illustrate how the “deep infrastructure” for the enduring use of digital libraries is developing in and affecting the research university community — and our research libraries — I focus on three specific issues. First, I examine the distributed organization of repositories of digital information, which must be the locus of any digital preservation activity. Second, I turn to the dependence on contract, rather than copyright law, as the basis for using and preserving the information in these repositories. Third, I conclude by reviewing the evolution of integrated systems of discovery and retrieval, which must support the information that we aim to preserve.

Distributed repositories.

Two years ago, when I last talked to you about these matters, I suggested that the problems in the scholarly communication process, which appear as a spiral of escalating prices and journal cancellations are, at least in large part, problems of ensuring the durability of the electronic record of knowledge (see Waters 1995). In increasing numbers, scholarly societies, such as those in high energy physics, astrophysics, engineering, and computing, are recognizing the dis-economies of the present system and have accepted responsibility for setting up and maintaining electronic channels of communication for their disciplines. Conventional publishers, such as Elsevier, Springer-Verlag, and Academic Press, are also opening electronic channels for disseminating their publications. Colleges and universities, usually through the agency of the library, or through consortiums of libraries acting as buying clubs, ensure that the information in these newly opened channels reach the researchers and students in their institutions.

The dynamics of this shift from conventional printed publication to the electronic medium raises a number of intriguing issues. Not only is the transition incomplete, but it is highly uneven across disciplines, meaning that the scholarly community is today experiencing significant transition costs (see Okerson 1995). Stakeholders maintain both print and electronic streams of publication as they decide when to shift permanently to the electronic medium. The unevenness of the transition, however, is also a positive signal, at least in this case, because there is tremendous competitive pressure to take advantage of the benefits of the electronic medium. The differences in skill and products among the scholarly societies and publishers provide an intense learning environment, which under substantial competitive pressure generates a highly productive “leap-frog” effect. Each new investment builds on known solutions, but also advances the field by offering competitive solutions to problems still remaining. The cumulative result is a rapid, sometimes disorienting, advance in the quality and reliability of the electronic systems.

Given the rapidity of the changes underway, the dust may not settle for some time on the transition from print-based to electronic scholarly communication. Still, the transition is far enough along in many disciplines to discern that there is demonstrably little economic benefit for libraries each to manage the costly logistics of disk storage, software compatibilities, and forward migration of scholarly works that publishers distribute to them in electronic form (see, for example, Task Force 1996: 29-35). Instead, economies of scale in the electronic medium favor a model for the management of the scholarly record over time that is network-based and centers the responsibility, at least initially, with creators, providers and publishers. Thus, scholarly societies and publishers are today creating and will presumably manage over time — either by themselves or via an outsourced repository management agent — their own electronic repositories of scholarly works. From these repositories, they distribute virtual copies to users over networks using standard browser software.

Organizing repositories in this way closer to the point of the creation of electronic works than to the point of their use vests long-term care of the works with those who, at least initially, care most about them. Such an organization also opens numerous possibilities for redesigning the scholarly publication process so that labor in tasks that are currently bundled in the process may be divided, the quality of work in these tasks improved and the overall costs of the process reduced. As many of you know, Paul Ginsparg has, from his considerable experience in managing the pre-print archives in high energy physics, imagined a network of discipline-based repositories for which scholarly societies are responsible and into which scholars deposit their works as they complete them. The repositories could help simplify at least two separate components of the complex publication process. Editors could manage the certification of quality through peer review by simply pointing reviewers to works rather than physically distributing them. Similarly, they could simplify the publication process itself by compiling pointers to the network of repositories for the works that they wish to disseminate to the targeted audiences they serve, rather than repackaging those works (Ginsparg 1997). Further development of these processes might even suggest that certification of the quality of scholarly works would be more effectively and efficiently managed separately from, rather than as part of, the publication stream that leads to the dissemination of works to targeted audiences.

Organizing repositories close to the point of the creation of scholarly works has significant implications not only for the publication process but also for the preservation of digital information in digital archives. Some libraries might position themselves to serve as mirror sites for the electronic repositories of publishers. Others might contract with scholarly societies, as Stanford University’s High Wire Press has done, to provide repository and other publication services. Still others might follow the lead of the Mellon Foundation’s JSTOR and position themselves to serve as a fail-safe repository for back issue works that publishers themselves no longer wish to manage. Generally, however, economies of scale in the storage and accessibility of digital works mean that the challenge of collection management in digital libraries will be to preserve the use and usability of works that are stored not locally and under the direct control of the digital library, but remotely and under the control of various and widely distributed agents. In the community of research universities, libraries are meeting that challenge on two major fronts: by developing licenses for digital works and by redesigning systems of discovery and retrieval.

Dependence on contract. The common means today for libraries to acquire the rights to use scholarly works in digital form is to execute a detailed contract, or license. For owners and providers, the general protections that copyright law presently affords in the sale of intellectual property seem completely inadequate in the digital realm. There, copying is technically so easy that it seems to put at great risk of loss the rights that owners have in their property. Thus, in the application of contract law, they seek directly from users the greater protections they feel they need.

Libraries, on the other hand, have regarded the resort to licensing suspiciously. They have viewed it as an attempt to subvert or sidestep the general protections, such as fair use, which copyright law affords to users of intellectual property in the interest of promoting “the progress of science and the useful arts.” Indeed, in the worst case, licensing may actually constrain the rights of users in substantial ways. For example, if libraries terminate a license and the provider withdraws previously licensed information, the prior investment becomes worthless and the user experiences a loss of the record of knowledge. Licenses may also extend the liability of users in the case of misuse, while limiting penalties for the provider if the information products fail to endure over time or otherwise to perform as advertised (see Okerson 1996a: 68-69).

Although licenses for the use of digital works appear to put libraries and their users at a disadvantage, growing experience of both providers and libraries in creating them suggests that contracts are not only appropriate to present circumstances in the digital arena, but can actually benefit both parties. Contracts provide the means for divergent interests to meet in times of uncertainty, high risk, and great promise. By engaging in the negotiation of content licenses, publishers and libraries are not just forging agreements, they are crafting the durable and trustworthy relationships that are so necessary to sustain the electronic information products they both need. In their contract-making, publishers and libraries are, as Ann Okerson has put it so well, “making [their] own peace, thoughtfully and responsibly, one step at a time” (Okerson 1996b).

When publishers and libraries are as unsure of one another as they are today in the digital environment, one way that licenses enable them to “make peace” is to define formally and legally who the parties to the agreement are. Licenses also enable the parties to specify their mutual responsibilities and provide for ways to settle disputes, should they arise. A growing trend in the creation of content licenses is for libraries joined by regional affiliation or other common features to define themselves as a consortium or buying club in relation to a publisher for the purpose of acquiring a license to use a digital work or set of works. Such an arrangement benefits publishers because it reduces the overhead of marketing products separately to each institution. For the libraries, the benefits include discounts on the purchase price, or inclusion in the purchase of more works for all the institutions than any institution acting separately could have afforded. Perhaps more importantly, licenses that define groups of libraries as buying clubs are beneficial because they serve to align library interests where previously there was disunity. Such an alignment is especially critical in an environment in which repositories of digital information are widely distributed and under external control. To assert any influence over the information products, libraries must be able to act in concert with themselves and the publishers. Buying clubs provide libraries an identity, defined by contract, for such concerted action.

Content licenses also provide a way for libraries and publishers to “make peace” with one another when the stakes are high. The retooling necessary to traffic in electronic, rather than print, publications is expensive for both parties. Publishers suddenly need to invest in systems engineering of processes that have been relatively stable for years and they have to provide annual budgets for research and development where previously they had little or none. Research libraries, who generally have had systems engineers on staff for years, regard with concern the relative naivet� of publishers. Libraries worry about the five to ten percent of their collection budgets that they currently have at risk in electronic information.

Publishers seek to limit their risk under contract by defining and limiting the community of users who can use the electronic information they provide to libraries. The categories of authorized users generally tend to exclude alumni and corporate partners in the research university community, the very categories to which these institutions are looking to extend their services. Libraries, however, focus in the development of content licenses on what authorized categories of users can do. Can they make copies? On what terms can they make copies for colleagues or readers who do not have a license to this particular resource? How can the resource be used in the rapidly evolving world of classroom access? Libraries are finding that, within the defined categories of users, the answers to these questions lead both parties back to the provisions of fair use and other rights generally afforded under copyright (ibid.).

Finally, content licenses afford a way, when the need and promise is as great as it is in the emerging knowledge economy, for libraries and publishers together to create and develop new markets and ways of conducting business. They provide a space to experiment and explore, affording a pragmatic way to achieve change and to build the necessary infrastructure for improving the quality, lowering the cost, and expanding the reach of education in research universities. One of the areas most in need of such attention is how new electronic information products integrate into the larger mix of resources that digital libraries provide their users. In the present jargon, the question is how do electronic systems of discovery and retrieval “inter-operate.”

Systems of discovery and retrieval. The information that digital libraries in research universities are licensing for use from various and distributed repositories is composed of works that have diverse document and data structures and depend on various search engines and vocabularies for access. The heterogeneity of the information in structure and form significantly challenges users in their ability to identify, retrieve, and discern the quality of needed information. Designing and constructing systems that lower the barriers for discovery and retrieval of these heterogeneous materials is essential for the enduring use of digital libraries and is, as more and more materials become available to members of the research university communities, an increasingly urgent task.

To reliably stimulate the attention of users over time, information from a distributed set of repositories must integrate into the information space of a digital library. Such a space typically consists of four areas: a catalogue of the works selected for the library or set of libraries; a series of index structures that describe works in greater detail than the catalog or in ways that the catalog does not permit; the works themselves; and tools with which to use or analyze the works.

The library catalogue describes and organizes at either an item or collection level much of the material selected for a library and judged to be most pertinent to the community of users it serves. MARC is the standard interchange format in which catalogue records are represented electronically. Concern about the complexity of the MARC record and its inability to represent complex hierarchical and certain other kinds of relationships among source works has led in recent years to the development of the so-called “core” record and the exploration of alternatives such as the Dublin Core and the use of SGML for tagging fields. MARC has proved remarkably durable, however. The development of the 856 field in the MARC record, which links to related objects in digital form, the Z39.50 protocol, and World Wide Web interfaces to the protocol, have made it possible for MARC-based library catalogues to integrate seamlessly into a networked environment.

Although crucial, library catalogues simply are not sufficient for digital libraries to provide intellectual access to the world of knowledge. Traditional abstract and index files have long been available electronically to provide detailed information about the contents of journals, and providers are moving quickly to place them in the networked environment. Significant work recently on the Encoded Archival Description (EAD) has produced a standard method for detailed, online descriptions of archival collections. Similarly, work is progressing under the auspices of the Inter-university Consortium for Political and Social Research (ICPSR) for a standard means of encoding the data dictionaries, or codebooks, for data files. Less advanced, but now moving rapidly, are efforts to organize the methods for describing and classifying visual resources, such as photographs and works of art. Finally, inverted indices for full-text documents and the means of searching them are also figuring prominently in the digital landscape.

As catalogues and index structures for various types of works — books, serials, archives, data files and visual resources — all appear on-line in a distributed networked environment, so, too, are the sources, or digitized surrogates of them, appearing there in greater quantities. In some cases, the library may have licensed the sources; in other cases, it may own and hold them locally. In addition, tools for on-line textual, data, and image analysis and manipulation also exist online and have become increasingly sophisticated in function.

Given the short span of time since the invention of the World Wide Web, it is almost miraculous that so many of these components — catalogues, index structures, sources, and tools — now live on the Web and that together they comprise a fully navigable information space for discovery, retrieval, and use. One can, for example, search a catalogue, find a record, and link from it through the Web directly to the source book. One can also search, find a catalogue record for an archival collection, link to the EAD for collection, traverse the finding aid and then link again to a surrogate of a photograph contained in the collection. Alternatively, one could skip the catalogue search and start directly by searching an EAD or collection of them. Perhaps more complex is the example of student searching the catalog, finding a data file of survey results, linking to the relevant online codebook, and using it to extract a subset of data for analysis using a favorite statistics program.

Experiencing for the first time an integrated online information space for discovery, retrieval, and use in these ways can leave one breathless with excitement. However, it is mostly an excitement of anticipating a promising future rather than of having it realized. As publishers deploy their repositories into this online space, they are learning about weaknesses in design and how to correct them. For example, some publishers provide journal titles that are directly addressable from a catalogue or other index structure. Many, however, do not, and this deficiency in effect cripples the navigation mechanism for the reader who cannot move from a catalogue record or journal index directly to the title. Publishers, such as Academic Press, in fact designed their system to force the reader upon entry to it to initiate another search to find the title, regardless of the information the reader brings to the repository from other searches elsewhere.

A design problem like this one clearly prevents a body of work from being well integrated into a prevailing information space, and the appropriate solutions are not always obvious. The development of supporting mechanisms, such as standard protocols for repository access and facilities for searching across a multitude of index structures, would facilitate such integration and is presently the subject of sustained research on agent architectures at the University of Michigan and at Stanford University under the auspices of the National Science Foundation’s Digital Library Initiative. Another hot area of research and development is how to support dynamic links among citations in electronic works through standard document identifiers, such as the SICI, and other means. In the end, however, the success of such research and of its eventual implementation cannot be measured solely by how well or poorly it supports integrated systems of discovery and retrieval. Rather, the measure is against a higher standard: Does it lower, rather than raise, the barriers to effective use of digital libraries by those in the communities they serve? And does it thereby fulfill the essential library goal of preserving for future members of those communities the use and usability of the record of knowledge?

Conclusion

Will Rogers is alleged once to have said that a revolution is like one cocktail — it just gets you organized for the next. The unfolding of the knowledge economy in what Peter Drucker calls post-capitalist society is not the first revolution in the organization of knowledge. How well did those earlier cocktails organize us for this one? Are we just drunk now with hype about the significance of current developments? To return to the question with which I opened: Are we simply moving hot air from car to platform?

The development of the digital library in all its aspects — technical, economic, political, and social — amounts to nothing if it does not generate literate citizens capable of engaging sensibly and productively in the discourse of the world in which they live. The literate citizen in the emerging knowledge economy is smart about how to use information and has it ready at hand or, in the current jargon, at one’s desktop. Allow me now to close by offering a description of such a desktop.

With this desk a man absolutely has no excuse for slovenly habits in the disposal of his numerous papers, and the man of method may here realize that pleasure and comfort which is only to be attained in the verification of the maxim: a place for everything and everything in its place. The operator having arranged and classified his books, papers, etc., seats himself for business at the writing table and realizes at once that he is master of the situation. Every portion of his desk is accessible without change of position and all immediately before the eye. Here he discovers that perfect system and order can be attained; confusion avoided; time saved and vexation spared; dispatch in the transaction of business facilitated and peace of mind promoted in the daily routine of business (quoted in Cooper 1983).

Don’t you want one of these? This advertisement appeared in 1880 and describes the Wooton Patent Desk. The desk is a lovely piece of furniture but, as Donald Norman points out in Things That Make Us Smart, it proved for a variety of reasons to be relatively useless (Norman 1993: 158-159). It did not spare vexations. It created more and worse ones.

As we aspire to the high purpose of preserving the use and usability of the record of knowledge in digital form, and as our technical, political, economic, and social strategies for doing so lead us to create and maintain digital libraries, let us resolve at least to learn from one past drunken mistake and avoid the fabrication of more Wooton Patent Desks.

References

Bronson, Po. 1996. “George Gilder.” Wired 4.03 (Mar.): 122-126, 186-195.

Cooper, Deborah. 1983. “Evolution of Wooton Patent Desks.” In Wooton Patent Desks: A Place for Everything and Everything in its Place, edited by J. Camille Showalter and Janice Driesbach. Indianapolis and Oakland: Indiana State Museum and the Oakland Museum.

Gilder, George. 1995. “Angst and Awe on the Internet.” Forbes ASAP (4 Dec.). Available at: http://www.seas.upenn.edu/~gaj1/ggindex.html.

Gladstone, Malcolm. 1997. “Chip Thrills.” New Yorker (20 Jan.): 7-8.

Drucker, Peter. 1993. Post-Capitalist Society. New York: HarperBusiness.

Ginsparg, Paul. 1997. “Winners and Losers in the Global Research Village.” Paper presented at the International Conference of the Academia Europaea on The Impact of Electronic Publishing on the Academic Community. Stockholm, Sweden, April 16-20.

Graham, Peter. 1995. “Requirements for the Digital Research Library.” College and Research Libraries 56(4): 331-339.

Grauer, Neil. 1995. Remember Laughter: A Life of James Thurber. Lincoln: University of Nebraska Press.

Hagel, John and Arthur Armstrong. 1997. Net Gain: Expanding Markets Through Virtual Communities. Cambridge: Harvard Business School Press.

Hof, Robert. 1997. “Internet Communities.” Business Week (5 May): 64-80.

Lanham, Richard A. 1993. The Electronic Word: Democracy, Technology and the Arts. Chicago: The University of Chicago Press.

Lehmann, Klaus-Dieter. 1996. “Making the Transitory Permanent: The Intellectual Heritage in a Digitized World of Knowledge.” Daedalus 125(4): 307-329.

Lynch, Clifford A. 1995. “The Technological Framework for Library Planning in the Next Decade.” In “Information Technology and the Remaking of theUniversity Library,” edited by Beverly P. Lynch, New Directions for Higher Education 90 (Summer), 93-106. San Francisco: Jossey-Bass Publishers.

National Research Council. 1995(a). Preserving Scientific Data on Our Physical Universe: A New Strategy for Archiving the Nation�s Scientific Information Resources. Washington, D.C.: National Academy Press.

National Research Council. 1995(b). Study on the Long-Term Retention of Selected Scientific and Technical Records of the Federal Government: Working Papers. Washington, D.C.: National Academy Press.

Norman, Donald. 1988. The Psychology of Everyday Things. New York: Basic Books, Inc.

Norman, Donald. 1993. Things That Make Us Smart: Defending Human Attributes in the Age of the Machine. Reading, MA: Addison-Wesley Publishing Co.

Okerson, Ann. 1995. “Can We Afford Digital Information? Libraries? An Early Assessment of Economic Prospects for Digital Publications.” In The Prospect of Digital Library, KIT International Roundtable for Library and Information Science, 9-10 Nov., 1995. The Japan Foundation: Library Center, Kanazawa Institute of Technology.

Okerson, Ann. 1996(a). “Licensing Perspectives: The Library View.” Paper presented at the ARL/CNI Licensing Symposium, San Francisco, 8 Dec. Available at http://www.library.yale.edu/~okerson/cni-license.html.

Okerson, Ann. 1996(b). “Buy or Lease? Two Models for Scholarly Information at the End (or the Beginning) of an Era.” Daedalus 125(4): 55-76. Also available at http://www.library.yale.edu/~okerson/daedalus.html.

Pelikan, Jarislov. 1992. The Idea of the University: A Reexamination. New Haven: Yale University Press. Task Force on Archiving of Digital Information. 1996. Preserving Digital Information. Report of the Task Force on Archiving of Digital Information. Washington, D.C.: Commission on Preservation and Access, and Mountain View, CA: The Research Libraries Group, 1 May. Also available at http://www.rlg.org/ArchTF/.

Waters, Donald J. 1996. Realizing Benefits from Inter-Institutional Agreements: The Implications of the Draft Report of the Task Force on Archiving of Digital Information. Washington, D.C.: The Commission on Preservation and Access. Also available at http://www.arl.org/arl/proceedings/127/waters.html.