Library copublication of electronic journals Frank Quinn (quinn@math.vt.edu) Gail McMillan (gailmac@vt.edu) Virginia Tech Blacksburg VA 24061 October 20, 1994 In this note we consider two aspects of library copublication of electronic or mixed-media journals. The first concerns the organization and cost of a library group undertaking electronic publication. The second is a proposal for library-publisher cooperation. The underlying theme is a broader view of libraries and serial librarians in the scholarly enterprize. Narrowly stated, the role of libraries is to procure journals and make them available to their user communities. But the journal system is showing stress, and is likely to be in considerable disarray before the end of the century. As Andrew Odlyzko has pointed out, capacity of the journal system is essentially saturated but scholarly output continues to rise. Either journals must become significantly more efficient in their presentation of information, or a significant amount of scholarly interchange must take place outside of traditional journals. Both changes will require a shift in library focus from procuring journals to support of more general information-providing structures. === Library copublicaton units === Here we suggest a structure for a small library department dedicated to acquisition and maintainence of electronic journal files and related materials. This is drawn from the experience of the Scholarly Communications Project at Virginia Tech (http://scholar.lib.vt.edu/). There are two misconceptions which should be addressed. The first is that electronic file maintainence is purely mechanical, and the only expense is the cost of a workstation stuck under a table somewhere. In fact it is still vital to have staff expertese in every part of the operation in order to have a robust and reliable system which can stay current. The second misconception is that there is a straightforward relation between the cost of the operation and the amount of material that can be handled. Our experience suggests instead that there are thresholds of staffing and experience. Passing a threshold permits great expansion of volume with little additional expense. === A model small electronic publication department === Our ideal small department would consist of five people: 1) Director: responsible for outreach, policy, communication with editors, and developing funding for special initiatives. Special initiatives undertaken at Virginia Tech include putting the local newspapers online, and digitizing collections such as the photographic collection of the Norfolk and Western railroad and the local TV news archives of text and video. 2) Systems administrator: responsible for maintaining and upgrading the hardware, Unix systems administration, security. 3) Software manager: responsible for selection and maintainence (upgrades) of database software, search and other software tools. Stays current with developments and anticipates new needs and opportunities. 4) Database manager: responsible for maintaining database, transferring and indexing new material, detecting and repairing file problems. 5) Office staff: responsible for routine functioning of office. Other duties as needed, for example scanning printed texts or images. These functions are quite distinct in the skills required. Our group is about half this size, with several of these covered by single people and others done part-time by people with other duties in the library. This has been workable as an experimental startup, but would not be satisfactory for full production. By the end of this year we will have 2 newspapers; 14 journals; abstracts of six more; and a large image database. This is not our maximum capacity, but in our present circumstances is not far from it. The group envisioned above, by contrast, should be able to handle files for hundreds of journals, image archives, and more. A principal bottleneck in our operations is the cleanness of the files delivered by editors for entry in the database. Currently it takes about 30 minutes to mount and test a journal issue. It should be possible to drastically reduce this as the editors gain experience and respond to feedback from the technical staff. But housebreaking of editors is a key issue. Problems in this area might require several people working on database management. At current levels in southwestern Virginia salaries for this ideal group would cost about $130K. Yearly outlays in equipment (not counting startup, and major upgrades) should be in the $40K range. Software purchase and licensing might cost $5K, and $15K should be budgeted for travel and continuing education. The total yearly cost would be in the $200K range. As noted above it is misleading to think of this in terms of marginal costs, but if files for 200 journals were supported this would be about $1,000 per journal per year. We cannot project beyond this level with any confidence, but do not see any technical reason why a slightly larger group would not be able to handle substantially more than 200 journals. For instance adding another database manager might increase capacity to 400 journals. We are thinking of these copublication units as supported by library funds, perhaps diverted from subscription budgets. This means they would scale in numbers rather than size. If 50 research libraries were to establish copublication groups in the $0.2--0.3M range they should be able to handle electronic files for 20,000 journals. Since the files would be freely accessible this means that the libraries would receive access to all of them for a net cost of under $15 per journal. The rest of the world would, in a sense, get a free ride, but we hope this can be seen as a bonus rather than a deficiency. === A model for copublication === The system envisioned above does not address the acquisition, reviewing and editing of journal files. These are functions traditionally done by editorial boards and publishers, and lie outside the domain of libraries. They are also not free. The proposed system would not cost publishers anything, but neither would it provide revenue to support these other functions. To date the only plausible mechanism for generating revenue seems to be some sort of subscription charge, and in most models this charge would be for access to the files. Charging for access is again not a function traditionally done by libraries, and particularly not if the proceeds are to be passed on to publishers. And even modest charges would qualitatively change the way the information can be used. We would prefer not to do this, and in this section we suggest a model in which the files remain free. This has been previously circulated in discussion groups as the "VT model," and is similar to a plan announced by Birkhaeuser for the Journal of Mathematical Systems, Estimation and Control. The proposal is that each journal should appear in two parts. One, already discussed, would be full text files archived in a library and freely available over the network. The other would be a paper (or eventually electronic) subscription journal consisting of brief summaries of the work. The two formats address different needs: the full files are needed by purposeful scholars interested in the work and aware of its existence and location. The paper version would be better suited for browsing and serendipidous discovery. The full text would be subject to standard peer review and editorial oversight for content. But since it will be offered free, processing beyond this must be minimized. The author would be expected to supply an electronic file, in a format format intelligible to machines and with content intelligible to humans. If assistance were needed in either file formatting or copy editing then the author would be subject to "page charges." The short summaries, in contrast, should be carefully edited for style and readability. In practice this would mean full text would not be copy edited. This suggestion seems to be radical enough to require further comment. First, it has to happen: resources available to support editing are simply being overwhelmed by the increasing volume of scholarship. Second, the great majority of readers will use the edited summary, and only purposeful and motivated scholars will have to contend with quirks in the full text. And they can contend: it is not true that scholars can only communicate with the assistance of copy editors. Finally, when deprived of this crutch perhaps scholars will learn to write well, just as most of them have learned to type. Economically this is supposed to work as follows: by passing the cost of archiving to libraries and concentrating expensive editing functions on short summaries, great savings should be realized, even allowing for reasonable publisher profits. It should cost 70-80% less to process a given volume of material than with the current system. Even with significantly increased volume there would be a reduction in total publication costs. These savings would be passed along to libraries. This should release enough of current subscription budgets to pay for the archives, and allow wide subscriptions to the paper summaries. This plan would also greatly reduce the flow of paper into libraries. === Steps to implementation === To get any significant fraction of the literature into this mode would require several important changes of viewpoint and professional strategy among librarians. For instance in this model rather than providing a small community with access to many journals, librarians would provide the global community with access to a smaller number. They would rely on similar activities in other libraries to provide local users with access to other journals. This means a greater reliance on collective action in the library community, and a more global view of the library mission. In another direction, librarians (and their funding sources) often think of themselves as serving a local community. This leads naturally to thinking of electronic journals in terms of support for local editors or outlets for local authors. But this is nearly a prescription for vanity presses. Serious quality control requires expert editors drawn from throughout the discipline. And archiving must be initiated by and under the control of editors, not authors. There are a few exceptions: Princeton may have enough faculty depth to be able to do a quality job with local editors, and MIT may generate enough quality material to support a quality local archive. But for the most part librarians must be happy to handle files for authors and editors unrelated to their institutions, and be ready to explain the appropriateness of this approach to their local users and funding sources. This is not to say they should abandon their local users. Our group, for instance, offers access to local newspapers and image collections. But in archiving, local users will be best served in the long run by careful support of quality-control mechanisms. Even greater changes would have to take place in the library-publisher relationship. The model we have presented calls for great economies in journals preparation costs and passing enough of the savings along to libraries to support archives. But this is not something publishers can take the initiative on: librarians must establish archives first so the mechanism is available when publishers develop the courage to use it. Also, publishers think of journals as competing for market share, and often price their products to maximize return. This "market" view pictures librarians as automata operating by very simple rules (e.g. "cancel the 'weakest' journals first, regardless of price"). Librarians would have to deliberately change the rules. For example suppose it was library policy to cancel the most expensive journals first, regardless of quality. Suppose some of this was done immediately to release resources for archive development and to acquire subscriptions to things like summary journals with dramatically better cost/volume value. This would encourage publishers to evolve in good directions. It would also make evolution possible: few publishers are willing to put their products at an immediate competetive disadvantage no matter what the long-term benefits might be. As long as the market rules remain unchanged publishers are locked in to the status quo, so rule changes enable behavior changes. We note a "cancel expensive first" strategy is likely to upset local users. Librarians would have to be prepared to explain why this is necessary to break out of a vicious circle, and encourage these users to direct their complaints to the publishers. Finally this model requires publishers to relinquish control of the information. Instead of denying access to non-subscribers their product would be improved access (the summaries) to information which is in principle freely available. This change flies in the face of centuries of custom which can be clearly seen embodied in the copyright laws. Librarians would have to demonstrate quickly and clearly that they will support this new product, and show publishers they can make money with it. === Conclusions === There is a dramatic contrast between visions of an electronic future and the nearly complete lack of movement in the publication industry. This can be understood as a lack of evolutionary pathways, or put another way that the traditional relationships between publishers, libraries and scholars have them locked in to the status quo. A common conclusion in discussions of electronic publication is that we will have to abandon the traditional system and simply start over in the electronic medium. In this paper we have suggested a model which might permit a more orderly transition. It seems to be economically feasible in that it does not require additional library resources and continues to generate revenue for publishers. But it would require very deliberate and coherent action on the part of librarians. === References === [Od] Andrew M. Odlyzko, "Tragic loss or good riddance? The impending demise of traditional scholarly journals", short version to appear in Notices of the AMS, January 1995. Long version by email: send message to netlib@@research.att.com, with message: send tragic.loss.txt from att/math/odlyzko [VT] The VT model for electronic copublication, in the vpiej-l discussion list http://borg.lib.vt.edu/ejournals/vpiej-l.html [H] Stevan Harnad, selected works available at: ftp://princeton.edu/pub/harnad/Harnad http://www.princeton.edu/~harnad/intpub.html gopher://gopher.princeton.edu:9000/1