Peak 2000 Conference Reveals E-publishing's 'War for the Eyeballs'
Brian Rosenblum
Introduction
Librarians, publishers and economists gathered at the University of Michigan on 23-24 March 2000 for a two-day conference on economics and digital libraries. PEAK-2000: Economics and Usage of Digital Libraries grew out of the recognition that digital technology is transforming the way in which libraries and scholarly publishers conduct business and provide service. The conference was organized and sponsored by the University of Michigan's Program for Research on the Information Economy (PRIE),and the University Library and its Digital Library Initiatives program. The conference was hosted by Professor Jeff MacKie-Mason, founding director of PRIE,and Wendy Lougee, Associate Director of the University of Michigan Library, with responsibility for Digital Library Initiatives.
The two-day conference consisted of 16 presentations of about 20 minutes each, with a discussion following each presentation. Conference presenters included representatives from universities from the USA and Europe, publishers, corporate libraries, and non-profit organizations. Nearly 100 registered participants, plus many walk-in participants, attended the sessions during the course of the two days.
Many of the librarians opened their presentations with apologies for not being economists, and joked about not attending economics classes in school. While this became a predictable joke by the end of the conference, it illustrated how the digital environment is posing new and complex questions that the library profession is only now beginning to consider, and it pointed out the continued need to bring economic expertise into the library profession.
A number of topics emerged repeatedly during the presentations and in the often lively discussions following each presentation. The main topics were pricing models, usage studies and cost studies. The conference Web page has a complete listing of the presenters and their backgrounds, as well as the full text or slides of most of the papers presented[1].
Overview
Andrew Odlyzko, head of the mathematics and cryptography research department at AT&T Labs, set the stage in his engaging opening presentation by outlining some recent trends in the world of scholarly communication, and some possible implications of these trends. Many of the points he touched on emerged as themes that were repeated throughout the conference. Odlyzko began by pointing out that, while scholarly communication is changing rapidly in the way people access and exchange information, scholarly publishing is evolving much more slowly. The danger is that traditional journals and libraries "might be rapidly losing their value,and could become irrelevant".
He noted that the most interesting developments are taking place outside the traditional academic environment, and outside the traditional publication process, which is "badly flawed" because it does not provide the services scholars require. The incredible growth rates of online use and the declining use of print at a time when the demand for information is so high are indications of this.
Odlyzko noted the importance of looking at growth rates rather than absolute figures, and cited a number of studies and preliminary figures showing the extraordinary growth in the use of the Internet as a source for information, particularly in freely available resources, whether they be journals in electronic format, or non-traditional types of resources, such as online databases and research tools. Odlyzko cited growth rates of 50-100 per cent as a common rate for a number of resources.
Odlyzko cited data showing a direct relationship between how far someone lives from a library and how often he or she visits the library; there are similar factors in the electronic environment. Users are sensitive to these obstacles, and a poorly designed interface, a request for payment, or even a request to register or sign in can have a great effect on usage. Almost everything is replaceable, Odlyzko argues,and users are choosing to view articles they have easy access to, rather than spending more time and effort to retrieve possibly higher-quality resources. Thus, as Odlyzko puts it, "registration is out", and scholars and publishers,whether they like it or not, are engaged in a "war for the eyeballs". Anything that is not readily and easily available, such as articles that are not available online, will not be seen and thus could become irrelevant.
During the discussion after his presentation, Odlyzko was asked about differences in behavior among scholars in different academic fields, with some scholars finding more value in browsing in library stacks, while others are more readily attracted to electronic sources. Odlyzko predicted that digital technology will eventually obliterate these differences. He also said that in ten years, the main activity of library acquisitions departments would be negotiating license agreements.
General Trends
Two other presentations touched on a number of broad issues and could also have served as overviews of the issues covered in the conference in general.
Don King, a statistician who has spent 40 years evaluating information and communication systems, gave an overview of current trends in scholarly journal publishing and digital database pricing. He noted that the average number of times an article is read has increased, as has the number of articles a scientist reads per year, and concluded that the value of such information remains high. Why, he asked, has the price of journals gone up by nine times while the cost has gone down?Inflation has some effect, but does not explain the large disparity. Other factors include journals with a narrow focus and low subscription rate, as well as a high profit/revenue system created by large publishers.
King outlined a number of ways in which future studies can become more sophisticated and useful in their approach to looking at economics and usage of digital resources. It is important to look at publishing in the context of the entire system of knowledge generation, distribution and use, in which there are many variables affecting demand that have not been adequately examined. For instance, scientists' time is an important factor, but one which has not been measured. He also said it would be useful to separate two components of price/cost analysis: the article processing component and the distribution component.
Finally, an important point made by King during his presentation, and that he made again several times in his responses to other presentations throughout the conference, was that publishers and libraries should not think in either/or terms when it comes to choosing between digital and print formats, but should look at models that incorporate both. There are certain advantages to both formats, and it is important to take advantage of both. In any case, he said, there will be a diverse market of print, microform and online journals for a least several more years.
Paul Kantor, now on the faculty at Rutgers, reported on the Columbia Online Books project, a broad study that looked at a number of cost and usage issues in providing online content. In his 20 minutes Kantor could merely touch on the kinds of issues discussed in detail in the full 168-page report, but the study provides a good example of the complexity and number of factors involved when doing a cost and usage analysis of any large-scale digital project[2].
Kantor suggested that many libraries and publishers are engaged in digital projects now without knowing the full costs or benefits of what they are doing. He likened this to philosopher Robert Nozick's concept of symbolic utility, something that one does because it seems good, even if one cannot identify the effects of the activity. In specific cost terms, the Columbia project estimated that the cost for providing access to a print book for 30 years was $105, while providing access to an online book would be $77. In addition to these access costs, they found that production costs varied widely during the course of the project, partly due to a steep learning curve.
The Columbia project also measured a number of usage patterns, and found that their material was primarily used in linear or semi-linear fashion, but with some use following other patterns, such as random browsing, hyperlinking, bouncing back and forth from index to source, and so on. They found that when looking at small passages,users would read online, but when looking at an entire text, readers would print out a copy. The Columbia team was unable to get data on how individual users or groups used the source, and hope to look at such information in the future.
Significantly, the Columbia team found that users did not have a strong adoption rate of the electronic books. Use of the electronic sources did not continue to grow over time. Instead their use tended to drop off suddenly at some point. Even so, the overall use of the electronic books was three times greater than the use of the print books. Kantor also suggested, interestingly, that in the digital world,technology determined behavior, which then determined attitude. In a more traditional environment, according to Kantor, attitude determines behavior rather than the other way around.
Kantor ended his presentation as many of the presenters did, emphasizing that there remains a need for more in-depth studies of user behavior and more attention to the relation of behavior and value with regard to digital resources.
User Studies
Several presentations focused closely on the use of electronic resources. These studies showed the high growth rates that Odlyzko had pointed out earlier, while at the same time they pointed out the need for further and more focused user studies.
David Alsmeyer, head of the British Telecom (BT) library in Ipswich, England, discussed that library's recent move from a physical to mostly electronic environment, providing over 1,000 electronic journal titles to BT employees. The BT library has been keeping track of use data, and Alsmeyer presented some of the statistics that they have been looking at, such as repeat usage of an article (articles in the BT library are read an average of 1.4 times) and how many titles receive how much use. Alsmeyer questioned whether the 80/20 rule was still valid in the electronic environment (a number of conference participants cited data that indicated the rule did not hold in the electronic environment).
Alsmeyer also made some general observations about user behavior. There is an increased need for bundles of articles rather than specific journal titles because BT users want "some article" on a topic rather than, say, Nature magazine. He also noted that, in the BT system, if an article is not currently available online but will be available soon, the user is notified of this and given the option of ordering a hard copy (at no personal cost) or waiting for the online version. Many choose to wait for the online version.
The discussion after Alsmeyer's presentation focused on trends in corporate libraries. Many corporations (with the exception of pharmaceutical companies) are continuing to downsize their libraries, while at the same time having an increasing need for information. The trend may be for corporations to increasingly outsource library service.
Kevin Guthrie, president of JSTOR, a nonprofit organization that is providing an electronic archive of the back issues of scholarly journals, discussed JSTOR's background, and some of the usage data now available after its first five years. Formed in 1995, JSTOR now makes available back volumes of 117 titles from 15 academic disciplines (a total of five million pages) to 600 institutions in 25 countries. JSTOR has two missions that can often work against each other in the print world an archival mission to preserve retrospective documents while at the same time broadening dissemination and use. There are now several years of usage data available to indicate how well JSTOR is achieving its two missions.
A total of 52 percent of all JSTOR articles have been viewed at least once, a remarkable figure from an archival standpoint, although perhaps not as remarkable from a commercial viewpoint. Seventy percent of searches done in JSTOR are not only cross-title,but cross-discipline, reflecting the increasing interdisciplinary nature of scholarly research. The average age of viewed articles varied from discipline to discipline. In economics, the average age of viewed articles was 13 years, while in mathematics it was 32 years. These statistics can lead to some interesting insights into the nature of research in particular disciplines, but Guthrie cautioned that in some cases there is not yet enough data to make for good statistics. What is most remarkable about JSTOR to date is the extraordinary growth of its usage, even when factors such as the increasing number of titles and subscribing institutions are factored out.
These statistics give some idea of the kind of data which can now be tracked in some detail, and which will prove invaluable in the future as larger and longer data sets become available. Guthrie pointed out that some of the areas that should be studied in the future include the relation between usage and value (for example, the most viewed articles do not necessarily correspond to the most cited), user demographics,types of use, and the implications of this use for scholarly work.
Hannelore Rader also discussed the incredible growth in usage of electronic resources that her institution, the University of Louisville, has seen over the last few years. Twelve percent of the University of Louisville's acquisitions budget is now spent on electronic content, with that number expected to rise to 50 percent in the next five years. They currently offer over 8,000 electronic journals online,and have seen use increase by over 400 percent between 1999 and 2000. Many of these resources have been made available through a cooperative arrangement with the State of Kentucky, which has recently established digital initiatives to provide databases and online courses to all citizens of the state through a consortium at consortial pricing[3]. Rader presented this as an example of the way that cooperative and consortia-based initiatives can help libraries gain a more central place within the academic enterprise, while containing costs and expanding access.
Like her colleagues at the conference, Rader also noted that more evaluation is necessary, particularly in assessing the use of electronic information in terms of student learning and performance and faculty research outcomes. She also said there remained a need for more work with vendors to address user needs, such as creating a common interface among databases, which would be one less barrier for users to overcome.
Pricing Models
Another set of presentations focused on different pricing models for electronic resources. Michael Spinella of the American Association for the Advancement of Science discussed the online access models and pricing issues for the Association's primary publication, Science. Online in full text since 1996, Sciencehas 150,000 print subscribers and an online audience of 1.5 million, though most of those are not paying. Like many online journals, they use several access models simultaneously, which allows them to experiment. Their current options include selected free content, personal online subscriptions, institutional access (with three further types of pricing structures), a pay-per-article option, and a $10-per-day option which allows users full access to all content for one day. An online full-text subscription is only available if a print subscription, at $112 per year, is also maintained.
Spinella made a number of observations about online publishing from the commercial publisher's commercial perspective. In the digital world, issues involving competition are different from the print world. Signals from competitors are often unclear, with many new,different, and untested strategies emerging. There are also low barriers to new entrants, because of the low costs of distribution in an online environment. Online publishing can be especially beneficial for focused publications with low circulation that depend largely on library subscriptions. Spinella cautioned that pricing schemes are still tentative because the cost structure for online production is still evolving and has not stabilized. Finally, Spinella noted, in response to a question from the audience, when negotiating with third-party content suppliers it is important to limit in some way what is provided. Thus,if full-text is provided, then the context, look, and feel of Sciencewill not be provided.
While Spinella focused on the issues facing a large, successful print journal moving to an online environment, others focused on more experimental efforts. Leah Halliday of Loughborough University in the English midlands described a project which uses the Ithink modeling software to explore activity-based pricing models for online journals. Like Spinella, Halliday warned that such models may need to be used with caution, because activities related to producing online journals have not stabilized.
Halliday's project involved three models. The first follows a traditional publishing model but moves the editorial process and delivery mechanisms online. Second is a free access model, in which the cost of publication is recovered from the authors published in the journal, and content is provided to users at no charge. This idea met with some skepticism from the American audience, who were convinced that the idea of authors paying to have their articles published would not be successful in this country. In Halliday's UK-based project, the New Journal of Physics was formed around this model, charging $500 to the authors of papers that it publishes. In the 18 months since the journal began, it has published 19 papers[4].
The third model in Halliday's research project is a market model. This model is based on the assumption that the current supply-demand chain for scholarly journals is distorted, and it tries to inject market aspects into the production process,including the editing and refereeing of journals, which is often done for free. In this model, access to articles is restricted, forcing end-users to prioritize their needs. Authors are paid on a royalty basis, which is supposed to provide them an incentive to publish higher-quality work.
During the discussion after her presentation, both Halliday and the audience pointed out that popularity that is, high use was not necessarily an indication of quality, and that authors could be tempted merely to write high-quality abstracts in an attempt to get users to choose their articles.
These models are all in their initial stage, and in the future Halliday and her team will review and revise the models and try to define more clearly production, administrative,maintenance and staff costs. Like all the other studies presented at the conference, it is useful for developing understanding, eliciting feedback, and providing a framework for future research.
PEAK
PEAK (pricing electronic access to knowledge) was a three-year project which gave university and research facilities a choice of several pricing options for access to 1,200 Elsevier Science journals, and whose completion marks the occasion of this conference. PEAK, sponsored by the University of Michigan Library and PRIE, in cooperation with Elsevier Science, explored a number of issues, among them how to form a library-publisher partnership to provide access to digital content, how to develop and manage a host site for very large digital collections and, finally,how to apply bundling and other nonlinear pricing opportunities to electronic content. Three conference presentations discussed some of the findings of the PEAK project.
Wendy Lougee outlined some of the difficulties in organizing such a large-scale project; she noted that it was a challenge for the library to serve as a mediating host for content and services in this environment, and that the library did not always feel comfortable marketing this product to other institutions. Nevertheless,enthusiasm for the more novel models of PEAK has grown during and after the project, as project data have been shared with the larger community[5]. John Price-Wilkin, head of the library's Digital Library Production Service, talked about some of the system-building issues. The total cost of the project was about $400,000, of which 40 percent was for technical infrastructure and 60 percent was for staff support. During the length of the experiment, obstacles arose involving transferring data among the numerous participating parties,providing authentication mechanisms, and ensuring that content was always available, as well as other technical concerns. It was clear that developing and maintaining a project of such scope requires a large institution with sufficient resources.
Bob Gazzale, a PhD student in Economics at the University of Michigan, then followed up with some of his analysis of how user cost affects behavior. Gazzale's findings confirmed what Andrew Odlyzko suggested at the beginning of the conference, that user costs heavily affect access patterns; any obstacle to use dramatically reduces actual use.
Gazzale began by describing the three pricing schemes that were offered to participating institutions:
Per article Individual users can purchase unlimited access to a specific article for a fixed price.
Traditional subscription Institutions can purchase unlimited access to a set of articles that correspond to Elsevier print-on-paper journal titles.
Generalized subscription Institutional users can purchase unlimited access to a set of 120 articles selected after the fact by the users.
In studying response to these models, one interesting finding was that users' expectations have a substantial effect on their willingness to use a resource. Users accustomed to free access will be less willing to purchase articles, even on an infrequent basis, when that free access expires. Users expecting to pay from the beginning are more willing to pay. Gazzale also noted that, regardless of how many titles an institution subscribes to, whether 200 or 800, the percentage of requests for articles from non-subscribed titles remained steady at about 50 percent. Gazzale repeated the now-familiar observation that, while we are starting to see much data come in, more analysis is needed about who is using the material, how they are using it, and the relationship between use and value.
Karen Hunter of Elsevier concluded this section of the conference with Elsevier's perspective on the PEAK experiment. In the PEAK experiment itself, Elsevier provided the content, but the University of Michigan handled all the marketing and providing of service. Simultaneously, Elsevier began offering host services for their 1,200 Science Direct journals more broadly, with their own pricing schemes. The options ranged from print price plus 15 percent for complete access to titles, to electronic access only at 90 percent of the print price. Some of the lessons learned were that, as in PEAK, a significant amount of use came from non-subscribed titles,so that libraries highly valued having access to the entire database. They also concluded that any hurdle cuts down dramatically on use, and any pricing model based on individual transactions would not be feasible. A discussion ensued about Elsevier's future plans, in particular whether it plans to give up print entirely. Although there are no plans to do so at the moment, Hunter predicted that in several years Elsevier could start moving in this direction. She speculated that if Elsevier got completely out of the paper business, prices could drop by as much as 30 percent, something the librarians in attendance seemed happy to hear, but reluctant to believe.
Market Power
Before the conference,co-host Wendy Lougee speculated on the future of publishing, noting, "Some predict the disappearance of for-profit publishers, but others see a trend toward growth and increased market power of publishers". The PEAK conference gave evidence of both developments.
Mark McCabe, an economist at Georgia Tech, formerly with the US Department of Justice Antitrust Division,gave a presentation heavy on economic analysis, focusing on a study of how the market has been affected by recent mergers of large publishers, including Elsevier. In any anti-trust case, he noted, lawyers and economists begin by defining the market. Libraries follow a portfolio model, that is, they buy portfolios of information rather than individual journal titles. To maximize the quality of their holdings, libraries rank (or should rank) journals on a cost-per-use basis. Looking at data from 1,000 core biomedical journals from between 1988 and 1998, McCabe found that the cost per use doubled between these years. His study showed that small firms tend to price their products lower than big firms, and that, according to his portfolio model, publishers find it profitable to increase the size of their portfolios via mergers and takeovers. Increases in market power due to such mergers were substantial.
Part of the problem lies in low demand elasticity of libraries, unlike individuals who are sensitive to price and will stop using a source if it is too inconvenient or expensive. Even when the price of journals tripled, library demand remained steady.
In perhaps the most discordant moment of the conference, Karen Hunter from Elsevier responded to McCabe's presentation by saying that his portfolio-based price model was wrong and artificially imposed from outside the market. McCabe said it was as inside as one could get, having been derived from discussions with hundreds of librarians. Hunter then further explained that sometimes small publishers want to sell because they do not make money, and that large publishers could provide the choice in pricing and access options that libraries want. McCabe gave a counter-example that showed smaller publishers could be successful, and said there were options other than mergers that would in the end be better for libraries.
As a response to the"monopoly-like system of scholarly communication" dominated by large publishers,who saw profits increase by up to 140 percent between 1973 and 1987, SPARC, the Scholarly Publishing and Academic Resources Coalition, was formed "to introduce more competition and cost-based pricing into the marketplace". Mary Case of ARL(Association of Research Libraries) presented the case for SPARC. This nonprofit organization tries to encourage new entrants into the marketplace by lowering some of the barriers to entry. Electronic publishing is helping to lower some of these barriers because of low distribution costs. SPARC is also helping new journals get established by soliciting pledges of support for new journals,working with faculty to make them aware of new journals, and securing distinguished names to sit on editorial boards of journals.
During the discussion,however, the audience raised some critical questions about SPARC's ability to make a difference in the market in the long run. There are questions about scalability. SPARC currently publishes four titles, a seemingly insignificant amount compared with Elsevier's 1,200. How would they be able to support their campaign as their title list grew large?
But Mark McCabe replied that bringing journal prices down needs to start with initiatives such as these. With limited resources, a choice for a lower-priced journal means a choice against a higher-priced journal. He also noted that SPARC was not alone. There were other nonprofit, grassroots initiatives to bring serial prices down, such as a campaign to persuade economists not to referee expensive journal titles.
An example of a different approach to distributing information to scholars was provided by Thomas Krichel,a lecturer in Economics at the University of Surrey, who argued in favor of the open library concept and talked about RePEc, his own effort to help promote the free exchange of information between economics scholars. He noted that economics is a highly peer-reviewed field with up to four years' delay before publication of an article, and that there is a great need for informal dissemination channels to make working papers available. RePEc attempts to fulfill this need by providing a resource that is well organized but with low requirements on the part of contributors. Rather than storing documents, RePEc is a collection of searchable metadata which serve as pointers to documents elsewhere on the Web. There is no monetary cost to contribute to or use the resource. There are currently pointers to 22,000 papers in 120 different economics archives[6].
So far, most of the use of RePEc has been coming from graduate students and underdeveloped countries, where the content is used for educational rather than research purposes. But Krichel said that users are beginning to bypass published versions and to use (and cite)free versions of papers when they are available.
Krichel is a strong advocate of the open library concept, and he compared it to the open source movement in software development. He described the open library as the application of open source theories to bibliographic metadata. Library and information professionals will have a chance to make use of such a resource first-hand when ReLIS, a clone of RePEc designed for library science information, becomes available in the next month.
Cost Studies
A final major focus of the conference was examining the costs to libraries of providing digital content. Bruce Kingma, a professor of Economics at the University at Albany, looked at the costs involved in a large-scale project to digitize and provide access to rare manuscript materials, while librarians at Drexel University provided a wide-ranging overview of how their move to an electronic journal environment affected activities and costs across the library.
Kingma tried to answer the question of "When is digitization cost effective?" by studying the Early Canadiana Online project, which involved the conversion from microfilm to digital format of 3,000 titles of the Canadian Institute of Historical Microreproductions collection. Kingma tried to pin down the exact production,storage, and access costs of print, microfilm, and digital titles and came up with precise figures for a number of different situations. For instance, he calculated that production costs for microfiche were $143.10 per title, compared with $111.19 for digital. By contrast, annual access costs were much higher for digital resources, at $16.84 per title compared to just 7 cents for microfiche,and $2.53 for printed material.
Kingma spent much of his presentation fielding questions from the audience about exactly how he calculated the costs, and although he claimed that he took into consideration all costs associated with the production, cataloging, and sales of texts in microfiche or digital format, it was not always clear exactly how these costs were factored in[7].
An important conclusion of Kingma's study was that for digital resources the cost of access decreases by increasing the output, so that digitization can provide significant savings if shared by a sufficient number of subscribing institutions. However, other issues that still need to be addressed are pricing models (whether to collect revenues through subscriptions, donations, sponsorships, or grants), and the implications of the distinction between access and ownership. What does it mean for an institution to purchase access to a digital resource, rather than buy a print or microform copy for their own permanent collection?
Carol Montgomery and JoAnne Sparks discussed the library-wide cost effects of Drexel University's move to a digital environment. Drexel provides a good model because of the scale of their effort. Owing to a number of complex factors and a willingness to take risks, Drexel had the "institutional readiness" to move to an online environment. Striving to eliminate their print journals and provide as much online content as possible, the Drexel library has undergone a rapid and substantial change.
Montgomery and Sparks found that developing such a collection was more complex than developing a print collection, and tracking all the data needed to study the process was also complex. The monetary costs involved in purchasing journal titles were $335,000 for 4,950 electronic journals (at $66 per title), compared with $219,775 for 1,475 print journals (at $149 per title). Overall cost for purchasing journal titles went up, but price per individual journal title decreased significantly,and use has risen dramatically, with the result that there is a significant drop in the cost per use. However, when looking at the effect on staffing and activities library-wide, other costs become apparent. Negotiating contracts and managing change turned out to be time-intensive and costly, bringing to mind Andrew Odlyzko's prediction at the beginning of the conference that this kind of negotiation would be the primary activity of library acquisitions departments in the future. Drexel also saw an overall increase in ongoing annual salary costs due to the technical nature of systems maintenance requirements. There was a decline in some activities, such as circulation, binding, checkout, and reshelving, and an increase in other activities, such as systems maintenance,license negotiation, and printing and reference services. A complete overview of their study is available on the conference Web site.
Montgomery and Sparks ended with what had become the main theme of this conference, that their study was not definitive, that there was a great need for more in-depth usage and cost studies, and that their example provided a good start and base on which to expand further research.
Lessons
Jeff MacKie-Mason provided some closing comments at the end of the conference. Thanking the participants for their "delightful" discussion, he noted a few conclusions to be drawn from this series of presentations.
He noted that the 80/20 rule (80 percent of the use comes from 20 percent of the titles) can also be applied to content within a title. PEAK and other projects showed that at least half of the most-used articles at an institution come from outside subscribed titles. This applies equally to institutions with large or small subscription lists. This is further evidence that the journal title is becoming an increasingly arbitrary way of gathering articles together, and publishers and libraries should look at new models of bundling articles.
MacKie-Mason also noted that usage numbers are higher than even the most optimistic predictions. Users are sensitive to access barriers, so eliminating these barriers (this includes system and interface design as well as payment mechanisms) could result in even greater use. This observation should also have a great effect on future pricing models.
The conference was evidence that rich data sets are becoming increasingly available. One of the next frontiers will be figuring out what to do with all the data. Although usage is high, more study is still needed. Who is using these resources and how are they using them? What value is being attached to the resources, and how does the use differ from the use of printed material?
Finally, MacKie-Mason stressed that continued communication is needed to let the library and publishing community know that digital technology is not a cost-saving revolution, but a value-adding one. In many cases, the costs to libraries and content providers are significantly greater than they expected, and greater than working in a traditional print environment. But the high usage of such material suggests that, while the total cost has remained high or even increased, the cost per use has decreased. It remains to be determined what value to put on that use, and at what level to invest in digital resources.
Notes
2. The project site and full documentation can be viewed at http://www.columbia.edu/cu/libraries/digital/texts/about.html
3. See the Kentucky virtual library http://www.kcvl.org and virtual university at http://www.kcvu.org
5. The home page of PEAK is at http:// www.lib.umich.edu/libhome/peak/ Complete reports on the design and usage of PEAK can be found in D-lib Magazine. See Bonn, M.S.,MacKie-Mason, J.K., Lougee, W.P. and Riveros, J.F. (1999), "A report on the PEAK experiment: context and design", D-Lib Magazine, Vol. 5, June, http://www.dlib.org/dlib/june99/06bonn.html and MacKie-Mason, J.K., Bonn, M.S.,Riveros, J.F. and Lougee, W.P. (1999), "A report on the PEAK experiment: usage and economic behavior", D-Lib Magazine, Vol. 5, July/August, http://www.dlib.org/dlib/july99/mackie-mason/07mackie-mason.html
7. Kingma's full study can be viewed at http://www.albany.edu/~bk797The Canadiana Online project is at http://www.canadiana.org
Brian Rosenblumis Digital Projects Librarian, University of Michigan, Ann Arbor. brianlee@umich.edu
