Date: Thu, 20 Feb 1992 14:00 EST From: BRIAN KAHIN Subject: Harvard/CNI project on scholarly communication; draft paper follows Dear Colleague: The file that follows is the draft background paper for the Harvard/Coalition for Networked Information project on Scholarly Communication in Network Environment. You responded to the announcement of the project or otherwise advised me or Paul Peters of your interest in the project. We hope to have your involvement in the computer conferences we are setting up. We are currently seeking volunteers to serve as co-chairs of these conferences in each of the six subject areas identified in the paper. We plan to convene a small workshop later this year, to which those who have made the most significant contributions to the project will be invited. Please get back to me or Paul with your feedback on the paper and the project. Brian Kahin Director, Information Infrastructure Project Science, Technology and Public Policy Program John F. Kennedy School of Government, Harvard University 79 John F. Kennedy St. Cambridge, MA 02138 617-495-8903 Fax: 617-495-5776 kahin@hulaw1.harvard.edu ===================================== Date: Thu, 20 Feb 1992 14:12 EST From: BRIAN KAHIN Subject: Scholarly Communication in the Network Environment (draft) DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT Scholarly Communication in the Network Environment Issues of Principle, Policy, and Practice Draft Background Paper February 18, 1992 (ASCII version with endnotes) by Brian Kahin Director, Information Infrastructure Project Science, Technology and Public Policy Program John F. Kennedy School of Government, Harvard University 79 John F. Kennedy St. Cambridge, MA 02138 617-495-8903 Fax: 617-495-5776 kahin@hulaw1.harvard.edu Copyright 1992 by Brian Kahin. The paper is publicly licensed so that it may be copied for further distribution, provided that it is copied and distributed in its entirety, including this title page. No copies may be made or distributed after September 1, 1992, without permission from the author. This paper has been prepared as a part of a joint project between the Information Infrastructure Project at Harvard and the Coalition for Networked Information. The project is partially supported by a grant from the National Science Foundation's Program on Ethics and Values Studies in Science and Technology. The globalization of innovation, industry, and commerce has spurred interest in intellectual property as a set of ground rules that may have profound distributive consequences for nations, industries, firms, or forms of economic activity.1 This attention has focused on the scope of underlying rights in scientific research2, but the digitization of information and the growth of research networking raise a distinct set of issues: How should the flow of research information be managed in the interests of scholarship and technological progress? Although electronic publishing has conventionally meant remote access to large databases on mainframe computers, the advent of distributed computing -- in the form of interconnected workstations and powerful microcomputers -- has led scholars and researchers to look both to electronic equivalents for the scholarly journal and to faster, less formal means of communicating new knowledge. The result is an increasingly articulated and multifaceted spectrum of scholarly communication, ranging from casual private messaging to complex software environments. At the same time, the greatly expanded scope and connectivity of the networks will enable publishing and resource- sharing on a sustainable scale. This paper looks at some of the critical points where these developments raise important legal and ethical issues. It moves roughly in sequence along the production and distribution cycle beginning with matters related to authorship and ending with distribution across national boundaries. Accordingly, the issues may be broken down into two principal areas: first, those arising from the greatly expanded opportunities and relationships within informal communication, pre-publication and formal publication; second, post-publication issues associated with the network as a distribution environment. The goal of the paper is to invite suggestions for policies and practices from the research community that produces and uses scholarly information, on the premise that the shared values within this community may be amenable to consensus. Indeed, the research community is unique in that revenues generated by publications do not flow to academic authors. To the contrary, academic authors are sometimes obliged to pay page charges from their research funds. Rather than monetary reward, the scope of dissemination within the community, including citations and word- of-mouth, motivates the author.3 Producers are also users, and scholarly communication is seen as a cooperative enterprise in which value is added to original work to the extent others cite it and build on it. Much primary and secondary information is published by organizations that are politically accountable to author and reader interests: academic and professional societies and, to a lesser extent, university presses. Although copyright defines a baseline of rights and responsibilities between owners and users of information,4 the basic framework it provides is overlaid by institutional policies, practices, and culture, as well as technological and market realities.5 The 1976 Copyright Act sought to be technology-neutral and in large measure succeeded through careful attention to principles. While there are a number of technical problems relating to use of digital information, there are no burning controversies such as those surrounding the scope of protection for computer programs. Where copyright is lacking, it should be possible to develop overlays of practice and expectation that could be adopted by organizations, publications, and individual projects within the academic research community. Communication, Pre-Publication, and Publication 1. Joint Authorship and Ownership Computer networks greatly facilitate collaboration between distant individuals or teams, including research projects in which multiple principal investigators may be thousands of miles apart.6 Differences over who has the right to claim authorship in different versions of research reports and articles are not unique to network-based remote collaboration; however, research teams have traditionally worked in close proximity under the direction of a senior figure who reviews all material intended for publication. Remote collaborations are likely to involve two or more individuals of relatively equal stature, each of whom is accustomed to acting independently, thereby increasing opportunities for misunderstanding and conflict. In addition, use of electronic mail tends to flatten hierarchies within organizations. This increases the risk that research results will be disseminated without explicit clearance from the top, especially since external networks facilitate distribution of material for review among an expanding set of formal and informal advisors. Interdisciplinary work which brings together researchers with different backgrounds and different sets of colleagues is particularly susceptible to leakage.7 Except for industry-funded research which contemplates patent filings8 or the submission of advance information to project sponsors, research projects often lack formal agreements or explicit policies on dissemination and publication.9 Individual responsibility for reporting research results may be distinct from the conduct of the research, or, indeed, the maintenance of project data or the discovery of patentable inventions. One aspect of a project may be owned by one person, while another aspect may be owned by a remote institution; a third may be owned jointly by an individual and a distant institution. Researchers on contract rather than salary are likely to end up with independent interests in their writings -- unless they are contractually obliged to assign copyright.10 Indeed, since universities allow faculty to retain copyright in their writings, there may be both a faculty interest and an institutional interest deriving from work performed by institutionally employed research assistants.11 A jointly authored work, or "joint work," results if it is "prepared by two or more authors with the intention that their contributions be merged into inseparable or interdependent parts of a unitary whole."12 Joint authorship means joint ownership, and a joint owner can legally license the work for publication -- or simply distribute it over the network without permission of the other joint owners. The joint owner who distributes the work is liable to the others only for their share of any money received. Thus, the more co-owners, the more parties with different interests are empowered to do what they wish with the work and the more it behaves as if it were in the public domain. However, as the number of co-owners increases, it becomes increasingly difficult to execute an assignment of copyright (which most publishers require) because all owners must agree to the assignment in writing. The publisher can be authorized to print the work by a non-exclusive license13 from any one of the joint owners, but only by receiving an assignment of copyright can the publisher be assured that the work will not be printed elsewhere.14 A related problem is that one co-owner can modify the jointly owned work on his or her own and thereby create a new derivative version in which he or she owns an exclusive copyright. The underlying work remains jointly owned but the modifications are the exclusive property of the modifier. For example, standards and associated documentation are typically the work of volunteers from different companies and institutions working together as joint authors. Some standards organizations hire reporters to synthesize the conclusions of working groups, so that the organization can claim copyright in the reporter's synthesis, which is published as the official version. It is easy to imagine situations where various jointly authored versions of a paper circulate among the authors, but one version has been modified by one of the authors for a special audience that the others are unaware of. Any reproduction or public performance of that version requires the permission of the person who modified it.15 2. Rights in Computer Conferencing In the conventional print environment, it was relatively easy to distinguish between publishing and person-to-person communication. Publication required a threshold investment in setting type and preparing plates, so it was not undertaken lightly. This threshold was lowered by a series of technologies -- mimeograph, offset printing, and photocopying. Even so, paper had to be distributed physically, with costs attributable to each copy and each delivery. Within the new network environment, these incremental costs virtually disappear or are borne by the end user in the form of storage and printing costs. There is an infinite spectrum between personal communication and publication; boundaries are defined by individuals and institutions rather than technology. Much of the territory between personal communication and conventional publication is covered by the computer conference -- in which all participants are potentially authors as well as audience. A distributed computer conference can be used to carry out a collaborative research project involving multiple institutions. Simple distributed computer conferences are known as "lists." They may be "moderated," in which case, contributions are reviewed by an editor before they are redistributed to the participants. In the case of "digests," contributions are edited and redistributed in batches, rather like print periodicals. An electronic journal may be peer reviewed like a conventional scholarly journal. At the other extreme are computer conferences based on public mail reflectors or "exploders" -- computer addresses set up so that incoming mail is automatically redistributed to a list of addresses. Thus, varying degrees of editorial control are possible. Computer conferences might seem almost by definition to be distributed creations, the product of many individuals contributing as equals. However, this is not necessarily the case. A course may be taught as a computer conference, or a research project may be conducted as a computer conference. In such cases, one or two individuals may be responsible for generating or synthesizing much of the content (but not for editing the contributions of others). A list, as opposed to a conference, may be a completely asymmetric one-to- many publication like a newsletter. A third issue is whether access to the conference is restricted, which is especially critical if there is an exploding list (and therefore no editorial controls). A list may be completely private in that the list is the personal creation of "list owner" who uses it to address a few highly regarded colleagues and no one else even knows that the list exists. Or a list may be publicly known and open for automatic subscription. There are many variants in between: The list-owner may have the opportunity to review subscribers or may require an application from prospective subscribers.16 ___________________________________________________________________________ selection edited <-----------> open creation centralized <-------> distributed access private <----------> public Characteristics of computer conferences ("lists"). ___________________________________________________________________________ In all of these respects, the production and publication of scholarly research has differed radically from the distributed computer conferencing. The production process is traditionally highly centralized, edited, and private. Even when others are brought in through the process of peer review, it is understood to be on a confidential basis. Only when a report or article has assumed a final, canonical form does it become public. Whereas research projects contemplate the production of singular reports and articles which synthesize the work of the team, computer conferences are aggregations of discrete individual contributions, ranging from single-sentence utterances to expositions the length of journal articles. While very short comments are not considered protectable by copyright, at some ill-defined point contributions become protectable expression. Such contributions remain the property of their authors, unless it is clearly understood that they are dedicated to the public domain, either by the statement of the authors or by an explicitly stated policy on contributions to the list. Since the United States acceded to the Berne Convention in 1989, the use of a copyright notice is no longer necessary to preserve copyright in a published work. Very few conferences have explicit policies on reuse of contributions. Individual contributors might reasonably be expected to label contributions as to whether and how they may be redistributed or cited. This is rarely done, although such notices could be effective public licenses since a redistributor needs the copyright owner's permission to copy the text and so must comply with the owner's conditions. There may also be a copyright in the computer conference as a compilation, if it is moderated and archived so that it expresses a degree of selectivity.17 A moderator holding a compilation copyright could conceivably claim control over the archive or even sequences of interaction among the conference participants. However, a case could also be made that the compilation copyright belongs to the contributors jointly or to the contributors and the moderator jointly. A mere "lurker," however, might have no rights to the compilation.18 But what is the determining context? Is it a particular strand of discussion? Or is it the conference as a whole? Ideally, conferences would have policies which would be dispositive of these issues and clearly advertised to anyone joining the conference. In addition, guidelines could be developed that reflect expectations based on the nature of the conference. In general, the more private the originating conference, the more one would expect redistribution to be restricted. Conversely, the more private the forum in which the material is redistributed, the more likely the original author would probably tolerate the redistribution. It is less clear how the degree of editorial control or the relative role of the owner of the originating conference should affect redistribution. However, the degree of editorial control in the redistributing conference may have serious repercussions: If the conference owner behaves more like a conventional publisher than a passive carrier and actively reviews and filters contributions, then he or she be liable for redistributing material owned by third parties. Anonymous FTP presents this end of the problem in a tightly focused way. Anonymous FTP is the posting of files for remote retrieval (using File Transfer Protocol) without a password in a publicly accessible directory. Anyone can log in as "anonymous" and retrieve the files in the public access directory. What assumptions, if any, should be made about files posted for anonymous FTP? Can it be assumed that the files may be freely copied and redistributed by anyone? Or can one only assume that files can be retrieved by anyone but redistributed only if the file itself explicitly permits redistribution? Anonymous FTP is a special case -- but an important one since it provides a frequently mechanism for distributing large quantities of information on the Internet. Assume for the moment that the copyright owner has not given the FTP site any rights in the file, and the site nonetheless posts the file for anonymous FTP. Unlike distribution of a file in a mailing, merely posting the file for remote retrieval by others does not plainly violate the rights of the copyright owner.19 However, each retrieval creates a copy, which presumably infringes the reproduction right. So the posting site may well be liable as a contributory infringer, especially if the posting is truly public.20 But is posting for anonymous FTP completely public the posting is tucked into an obscure directory or machine? Does this analysis change when a program like Archie comes along, which provides for automated scanning and cataloging of anonymous FTP sites? Aside from the liability of the posting site, what implications do these circumstances have for those who retrieve the files? Does the availability of Archie, in effect, create an expectation that the files are in the public domain? The more liberal the assumptions about reuse, the greater the liability risks for the posting site. 3. Derivative and Iterative Works A third set of problems arises from the tension between pre- publication and the formal publication, more frequent sequential and variant versions, and effects on the relationship between authors and publishers. It is a longstanding practice for journal publishers, including most noncommercial publishers, to require a full copyright assignment from authors. In theory, publishers are better able to manage exploitation of the work, including the sale or licensing of reprints, than the author. Certainly, publishers desire as much exclusivity as they get, and a copyright assignment appears to give them all they need. But since assignment of copyright includes the right to prepare derivative works, this precludes the author from revising or republishing his or her work. In effect, the work is limited to a single canonical journal publication, although the author is often granted the right to use the article and its contents in books or lectures. This traditional practice may be appropriate for reporting simple research projects (e.g., in which a hypothesis is tested by collection and analysis of data) within delineated specialities. But what if the research activity is complex, ongoing, and involves many participants at multiple locations? Under such circumstances, the reporting of research may be an iterative process rather than a final product. Results may be aired with the expectation that feedback from the field will contribute to the further design and conduct of the research. Peer review normally works as a conservative force that sharply differentiates between submission and acceptance -- and between unpublished and published. While publication in refereed journals drives the young scholar seeking tenure and promotion, it may be of less concern to established scholars, who though as eager to advance their reputations, are free to do so in less conventional ways. It is likewise of less concern to those working outside traditional disciplines and at the fringes of academia. In newly emerging fields, timeliness, turf preemption, and visibility may be important factors in expediting publication. Patent considerations may work to either hurry or slow publication,21 and a patent itself is also a publication which may have impact on promotion and tenure. Interdisciplinary research may merit multiple publications in multiple versions in order to reach its full audience.22 As the use of the network for scholarly communication increases, authors of specialized papers may become more expert on the potential audience for their work than publishers. The network and new software environments will invite constellations of preliminary, secondary, and tertiary publication that vary from field to field. Author-generated abstracts, synoptic publishing, and indexing will take on new forms and new significance. Authors, research programs, and scholarly organizations may find themselves negotiating with complex multi-publisher systems (the next generation of online vendors) to design dissemination programs that fit the institutional landscape, culture, and other information resources of their speciality. How much flexibility should authors (and research projects) have to recast their work in the light of new knowledge or for the benefit of new audiences? In the case of new knowledge, or a continuing flow of related knowledge, it would seem desirable to have it appear in the same periodical. What claims and what obligations should the original publisher have? To what extent will different kinds of research result less in discrete occasional products and more in the form of a service? Indeed, the textbook which is reissued in new editions every year or two may be a better model for some forms of network publication than the journal. On close inspection, the traditional assignment of copyright may not be as effective as it appears. Under the 1976 Copyright Act, copyright inheres in a work from the moment it is fixed in a tangible medium. There is no longer any distinction between the common law copyright in unpublished works that was determined by state law and the statutory copyright in published work determined by federal law. Instead, there is a copyright in every outline and every draft of a work as much as in the final published version. The later versions are derivative works based on underlying works, but the assignment of copyright in the final version does not automatically assign copyright in the underlying work. Thus an author may be able to go back to an earlier draft and construct another version based on the draft that does not infringe the publisher's copyright in the original published version. This is not to suggest that such assignments are necessarily ineffective. If, as part of the copyright assignment, the author warrants that the work is original that may be construed to mean that it does not derive from an underlying work. It is rather to suggest the complexity of determining ownership, especially when word processing computers and networks make it easy to create and disseminate variant versions. And it is to suggest that relationships between publishers and authors need to be recast in terms of realistic and shared expectations. The Distribution Environment The second set of issues revolves around the impact of the network on the distribution process, driven by the expected speed, functionality, and cost efficiency of electronic publication. No more than a dozen refereed electronic journals presently exist, principally university-based efforts initiated by individual editors.23 Until recently, their development has been constrained by policies on acceptable use of the academic networks, especially BITNET, that seemed to preclude charging for material sent over the network. However, BITNET's policies were revised in November, 1990 to permit fee-based services in support of "academic, research and educational purposes," provided there is no separate charge for use of the network.24 Policies on the Internet vary, but in most cases commercial use in support of education and research is permissible and increasingly accepted. These experiments have followed the traditional scholarly journal paradigm: i.e, individuals and libraries pay a fixed annual fee paid to a publisher who maintains an editorial apparatus, including a peer review process, linked to a journal name. This same system could deliver an irregular stream of individual articles -- since there is no compelling reason for delivering electronic articles in uniform bundles at even intervals. The continued value and identity of journals in the digital environments is unclear, since the individual article is the fundamental unit sought by the user, and the journal is merely an organizing unit and navigation aid suited to the economics and practicalities of print publication. Nonetheless, the journal title may remain an imprimatur of editorial integrity, quality, and intellectual continuity. But in the intelligent networked environment, the article is emancipated from its archival home in journal time and space. It is instead found and used in functional space, linked to related articles, reviews, commentary, abstracts, and references, and perhaps available in multiple forms and versions. The form and scope of the scholarly article may vary and the traditional concept of publication as instant in time may vanish. There may well be many ways for ordering the flow of information in the conduct of research prior to publication, including new forms and procedures for peer review.25 Electronic journals (or articles, as the case may be) also appear to offer significant cost savings in eliminating press runs, packaging, and physical delivery -- in short-circuiting a publication process which now almost always begins in electronic form. Accordingly, it is sometimes seen as a partial solution to the high prices for serials that are currently troubling research libraries. Indeed, networked digital information is sometimes envisioned as a leveling, democratizing force, capable of transcending any distance and rendering all information equally available to all points on the network. But the network is not ubiquitous, and if access is to be universal, print versions must also be available. An electronic document that is more than a simple ASCII text may require additional processing power, software, storage capacity, graphics capability, and other end-user technology. The overall efficiency and effectiveness of electronic publication will increase, certainly relative to costs of end-user equipment; but there will be disparities, especially in functionality of the publication and the user's ability to make effective use of it, that do not exist for print. However even those that have the technology to use a digital document on a primitive personal computer also have the power to reproduce and redistribute the document at virtually no cost. For this reason, publishers have had little interest in distributing articles in electronic form. Instead, publishers have focused on electronic publishing as an enhancement of bibliographic indexes. A pre-microcomputer market was developed in which computing power, intelligence, functionality, and storage capacity have been centrally provided to line-oriented, character-based terminals, seldom at more than 2400 baud. Even as PCs became ubiquitous, the old network (whether voice or X.25) remained a common low-bandwidth bottleneck, and it remained easy to assess usage charges based on connect time and to bundle in time-based telecommunications costs. Eventually elaborate algorithms were devised to charge for CPU cycles, searches, and retrieved records in order to even things out between those logging in at 1200 or 2400 baud and the laggards at 300 baud. But by marketing information retrieval as a low-bandwidth metered service, publishers and distributors have been able to charge prices keyed to use and perceived value while minimizing problems associated with downloading and reuse.26 Under the circumstances, wholesale downloading is expensive and noticeable, and anyone doing so cannot readily recreate the functionality of the vendor's software environment. Usage-based pricing is sometimes described as desirable (especially to small users) in that users are charged only for their use of the system and no more. However, flat-rate pricing is much easier to budget for, whether the organization is for- profit or non-profit. Usage-based charging also creates major disparities within the research community, within higher education as a whole, and between universities and the general public. The stratification within the research community is even more severe than it might appear: Instead of being included in overhead, usage-priced information is customarily treated as a direct cost of a research project subject to overhead, typically 50-70%. The effect is both to discourage and limit the use of metered information services -- and implicitly to cross-subsidize library services and other forms of unmetered information. Furthermore, offering information on a metered basis threatens to change the nature of the library. Instead of a public repository offering free access to resources it owns, it now becomes a mere gateway for information held by others. Copyright has traditionally controlled the printing of texts but not the use of texts once printed. Over time, various rights have been added to the reproduction right that is at the heart of copyright; these additions include the adaptation right, the distribution right, the public performance right, and the public display right. But these additional rights still do not limit ordinary use of the text. Under the "First Sale Doctrine," the distribution right does not apply to owned copies, so that a book can be publicly sold, rented, or loaned without the permission of the copyright owner.27 The First Sale Doctrine also limits the recently created public display right: There is no restriction on the ordinary display of owned texts. The public display right only comes into play when a remote or multiple display is created -- or if the embodiment of the text is not owned by the party creating the display.28 Copyright has encouraged the growth of libraries as an institutionalized system for allowing individuals the privilege of trading cost against convenience, of reading or borrowing rather than buying. In effect, this means that those who acquire and collect books and journals subsidize those who merely read them. Digital, networked information changes this, presenting a seductive vision of a library without walls but eroding the paradigms of print publishing at a very basic level. As network and storage resources grow in abundance, decline in cost, and are used for an increasing variety of purposes, the medium itself becomes increasingly non-specific, abstract, and trivial.29 The distinction between the medium (the book) and the content (information) disappears with the dematerialization of the medium and disembodiment of information.30 4. Control of Dissemination While publishers fear losing control of their property in an environment where reproduction and broadcast are trivial, they also look to this environment as a unitary market where they can sell access to information -- rather than selling copies into a market where the copies can be resold or loaned at will. Indeed, it offers an opportunity for direct marketing which eliminates all middlemen, distributors as well as libraries. Furthermore, a direct relationship with the end user makes it easier to rely on contracts to control access and use. Unlike copyright, which applies universally, regardless of any relationship between the parties, contracts must be individually established and contractual controls are very difficult to maintain through multiple intermediaries. But while a direct channel and relationship between publisher and user looks relatively secure and profitable, it may limit the publisher's market. A competitive marketplace with a full complement of distributors and retailers enables a publisher to reach the largest possible universe of users. In the network environment, intermediaries can add value to information in many ways that are not possible in the print environment, further increasing the scope and size of the market. Although from a private publisher's perspective this is a strategic issue, there are important policy implications, since single-channel distribution necessarily limits application and use of information. Spillover effects will be lost, as will opportunities for third-party enhancements through software and linkage to other information. Publishers may counter arguments for efficiency, diversity, and openness with that data integrity and quality can best be maintained by tightly controlling dissemination. In addition, close control allows publishers to subsidize certain users, such as educational institutions or users lacking advanced telecommunications facilities, and thereby justifying controls for reasons of equity. Debate has been taking place in two different contexts. First, in the matter whether government agencies, which are not able to assert copyright in the information they generate,31 should be able to control redissemination through the use of contract. In particular, much debate has centered on the pricing policies of the National Library of Medicine, where it is linked to complex questions of what sort of cost recovery should be encouraged or permitted.32 More recently, these issues have been raised in an antitrust context in Dialog's lawsuit against the American Chemical Society over (among other things) the restriction of portions of Chemical Abstracts to ACS's own system and network.33 This case raises policy questions about when privately created information services should be deemed "essential facilities" and therefore subject to special obligations. Also, the National Science Foundation provided initial funding for mounting Chemical Abstracts as a machine-readable database in the 1960s. While the issue in the lawsuit is linked to the specific terms of the original funding agreement, it raises questions about what the intellectual rights policies of NSF and other agencies should be when they fund the development of information resources. Finally, there is the question of whether any special obligation should attach to an organization's tax status -- i.e., whether an educational or charitable organization exempt under Section 501(c)(3) has obligations different from a private company.34 5. Site Licensing While publishers have considerable power to shape their rights with distributors in retail market depends on the status of the end user. Large institutions and corporations wield considerable buying power on behalf of their employees and students, they are vulnerable to infringement claims because they are visible targets with deep pockets and thousands of potentially disaffected employees. The Software Publishers Association has been very effective in enforcing copyright compliance for microcomputer software through a program that combines the use of free auditing software, an 800 reporting number, and litigation when necessary. The SPA now estimates compliance in major companies at greater than 90% (purchased software packages as a percentage of installed software). Publishers see site licensing as an opportunity to maintain such direct relationships at a reasonably high level with a responsible office. In contrast to large firms, individuals who work alone may abuse copyright with impunity. Nobody is going to turn them in and the cost of pursuing them in court is prohibitive. But the effect is to reinforce the hierarchy -- to encourage marketing to companies and not to individuals. It even suggests that publishers should price higher for individuals to compensate for losses due to illegal "sharing." Despite the changes in technology and the market, library operations could still be supported by taxing all potential users in order to subsidize actual users. This practice (i.e., treating library expenditures as common overhead) appeals to educational institutions because it encourages use of information resources and minimizes inequities among students. It appeals to all organizations in that it aggregates purchasing power, facilitates budgeting, and eliminates transaction costs. At the same time, the vanishing marginal cost of digital information encourages publishers to offer institution-level site licenses.35 Colleges and universities receive the best prices because of the expectation of winning future commercial users. However, licensee institutions are typically required to restrict use by outsiders -- which has not been the case for library books and journals. The unaffiliated may be left to fend for themselves. Restrictions on use by outsiders can put libraries in an uncomfortable positions. Such restrictions could ultimately jeopardize a library's ability to rely on the Copyright Act's Section 108 safe harbor provisions for archival reproductions and interlibrary loan. Section 108 is available only to libraries and archives whose collections are either open to the public or to unaffiliated persons doing research in a specialized field.36 Such restrictions can also be problematic for state universities, which are often obliged to extend library privileges to all citizens of the state. Even state libraries cannot reasonably be expected to site license electronic information for the state as a whole. Ironically, one way to resolve the problem is to resurrect the library walls and allow unrestricted use only within the library premises. In practice, site licensing of electronic resources is a kind of localized resource-sharing. It contrasts dramatically with the very technical delineation of acceptable resource- sharing in the print environment. Under Section 108, the photocopying of scholarly journals for interlibrary loan must not be "systematic" or "substitute" for subscriptions.37 Guidelines negotiated by librarians and publishers, promulgated by the Commission on New Technological Uses of Copyrighted Works in 197638, and endorsed by Congress, interpret this to mean that a library may request no more than five photocopies of articles from a given journal per year.39 Limitations on concurrent use can also provide an artificial constraint on "overly efficient" use of digital resources. Concurrent use limitations, supported by software lock-outs, are already becoming a standard method for licensing software programs on local area networks. Concurrent licensing is actually a step back toward the natural use limits of the book (as exemplified in reserve readings for college courses) -- but without giving the library the benefit of an owned copy. The lack of an owned copy means that if the library decides to terminate a subscription to a database service or a CD-ROM, it has nothing to offer its patrons.40 Although libraries have not normally retained outdated reference works, librarians perceive that publishers are not interested in or capable of adequately archiving what they publish. The controversy is not fully joined because the primary literature is not yet in electronic form. But although publishers have not asked libraries to give up their archival function for journals, the precedent being set for indexes and abstracts is troubling to libraries. 6. International Access The international nature of the Internet adds another dimension to the site licensing issue. If licensing can be managed with security and confidence, each country can be treated as a separate market, so prices and terms for information services can readily be set to reflect local economic conditions and the ability to pay. Where the marginal cost of production is much lower than the average cost, the ability to price discriminate is important to maximizing revenues, regardless of equity issues. For example, pharmaceutical drugs are typically sold are much lower prices in developing countries than in the U.S. Thus, a drug available at low cost in sub-Saharan Africa may be prohibitively expensive to an uninsured American. Lives may be less at stake in the publication of scholarly research, but the very openness of the network brings such equity issues out into the open. And while the network makes communication and publishing distance-independent, but different intellectual property laws and practices apply in different countries. In some cases, the result may be to shift certain publishing to favorable foreign havens. In addition, precedents set for free dissemination -- either of government information41 or as charity42 -- may color expectations about the use, and redistribution, of proprietary information. The consequences may be a source of tension with efforts to create a truly inclusive global Internet, where research is not disadvantaged by location. Next Steps: Statements of Policy and Principle Over the next six months, we will seek to involve individual researchers and scholars, libraries and library organizations, journal editors, and academic societies in debating the issues raised in this paper: Is joint authorship an appropriate paradigm for scholarly communication? Should derivative works by one of many joint authors be approved by the joint authors? Is it possible to craft policies on communication and publication of research results without appearing to infringe on freedom of speech? What would such policies look like? Should a contribution to a public list carry a public license that allows reposting unless the contribution provides a notice to the contrary? Should similar assumptions be made of files posted for anonymous FTP -- assuming that anyone can retrieve but only authorized individuals can post to the directory? How can lists and FTP site and directories be coded to convey warranties and licensing information? Should academic authors reserve the right to prepare and publish derivative works? What sort of publisher exclusivity is appropriate and how should it be expressed so as not to inhibit legitimate modification of the author's work? How should academic and society publishers license their information -- -- to commercial and noncommercial information services? -- to academic institutions, other organizations, and individual users? -- for use in less developed countries? The crafting of specific policy statements requires common reference points. What is a private list? What is a public list? What is a public FTP site? We are staking out the electronic frontier. The distances fool the eye, but it is not the flat expanse we first saw. Endnotes: (1) See, e.g., Francis W. Rushing and Carole Ganz Brown, Intellectual Property Rights in Science, Technology, and Economic Performance: International Comparisons (Boulder: Westview Press, 1990); Robert M. Sherwood, Intellectual Property and Economic Development (Boulder: Westview Press, 1990). A National Research Council conference, "Intellectual Property Rights in Science and Technology: The Global Dimension," was held January 8-9, 1992. (2) Weil and Snapper (1989); see also Dorothy Nelkin, Science as Intellectual Property: Who Controls Research? (New York: Macmillan, 1984). (3) See Merton (1988) and Byrd (1990) (4) Technically, copyright protects original expression of information, including selections or arrangements of facts, not facts or ideas. (5) Concerns are frequently voiced that copyright law has been antiquated by technology, but these concerns often reflect problems of enforcement or of the complex private arrangements that overlay copyright. In any case, no convincing alternative to copyright law for digital information has yet been proposed. One of the few attempts to do so was presented to the Library of Congress Network Advisory Committee. Robert J. Kost, "Useright," in Network Advisory Committee, Intellectual Property Issues in the Library Network Context, (Washington, DC: Library of Congress, 1989), pp. 71-76. However, The Committee concluded that the copyright system did not require radical revision. (6) Thus, the network can extend the boundaries and functions of the traditional laboratory to encompass a national or international "collaboratory". See Wulf(1988). (7) Clinical research and policy research might seem to present worst-case scenarios because they involve both practitioners and academics. However, professional codes and practices -- and learned discretion of policymakers -- presumably work to limit dissemination on the practitioner. In any case, the use of electronic mail is much less commonplace outside of higher education. (8) Industrial sponsors reserve the right to delay publication in order to perfect patent filings abroad. While U.S. patent allows inventors who publish information about an invention a one-year grace period in which to file a patent application, in most countries a patent application cannot be filed once the invention is publicly disclosed. (9) Another notable exception is research on human subjects where privacy is an issue. (10) From the institution's perspective, it is desirable that the contractor's contribution be considered "work made for hire." But this can only be achieved if the contractor agrees in writing and the work is specially ordered or commissioned for use as: -- a contribution to a collective work -- a part of a motion picture or other audiovisual work -- a translation -- a supplementary work -- a compilation -- an instructional text -- a test -- answer material for a test -- an atlas [17 USC 101: Definition of "work made for hire"] A standalone report does fall into any of these categories. Nor does a jointly authored paper which is not for a particular anthology or collection. Because the limited definition of "work made for hire" could possibly operate as a trap, it is considered sound legal practice to include an assignment of copyright in the contract in case the assertion that the work is "made for hire" is ineffective. An assignment is less desirable to the institution because it is subject to the author's termination rights after 35 years. (See 17 USC Section 203.) (11) Many universities permit graduate student research assistants to retain ownership of their writings. (12) 17 U.S.C. 101. (13) An exclusive license to all the rights inherent in the copyright (i.e., the rights of reproduction, adaptation, distribution, public performance, and public display) is treated as equivalent to an assignment of copyright. (14) However, an assignment of copyright does not necessarily mean that one of the joint owners has not already licensed the work for publication elsewhere, and a careful publisher will want some assurance that such is not the case. Naturally, such assurance must come from all the joint owners. (15) See Weissman v. Freeman, 868 F.2d 1313 (CA2 1989). (16) A related matter is whether the individual subscribers are aware of who the other subscribers are. In many cases there will be a publicly available subscriber record but subscribers may suppress their names from the public record. (17) Feist Publications, Inc vs. Rural Telephone Services, ___ US ___, 111 S.Ct. 1282, (1991) dispensed with the controversial "sweat of the brow" doctrine which permitted compilation copyright based on no more than industrious collection of facts. Some selectivity or arrangement is required to meet a threshold requirement of orginality. Archiving may be necessary to establish a fixation of the compilation, which is required to support a copyright. 17 USC 102(a). (18) The lurker might argue that the discussion sequence is not a product of selection but a historical fact and therefore not subject to copyright. (19) It is not on its face a violation of the distribution right since there is no delivery of a copy, nor of the display right since the content of the file is not visible. (20) Contributory infringement is not defined in the Copyright Act but was discussed at length in the "Betamax" case, Sony v. Universal City Studios, Inc., 464 U.S. 417, 104 S.Ct. 774 (1984). However, it could also be argued that anonymous FTP is equivalent to placing a journal in the library. It provides an opportunity for anybody to make photocopies, but that does not amount to contributory infringement. (21) Industry-funded research often provides that publication may be postponed to enable the preparation filing of patents before publication. While there is a 12-month "grace period" in the U.S., there generally is no grace period is other jurisdictions, so this practice preserves the opportunity to file for foreign patents. In other circumstances, it may be advisable to publish rapidly to ensure that nobody else is able to secure a patent that might jeopardize further funding or otherwise inhibit future research. (22) The cross-fertilization that can result from divergent publication is a strong argument for networking infrastructure. See Lewis M. Branscomb, "Information Infrastructure for the 1990's: A Public Policy Perspective," paper presented at the Kennedy School symposium, "Information Infrastructure for the 1990s," November 29, 1990. (23) Representatives of eight of these met at the Association of Research Libraries in October, 1990, and established themselves as the Association of Electronic Scholarly Journals. (24) There was considerable discussion on the POLICY-L list (BITNET) in early 1990 concerning subscription fees for a proposed electronic journal on research in hotel administration. Many of the participants in the discussion would have preferred a membership arrangement (wherein the members of the organization received a "free" journal) to a straightforward subscription fee. NSF, by contrast, has been very liberal in permitting use of the NSFNET backbone for any service that supports education and research. Charging is not an explicit factor, and even quali- fying for-profit uses have been permitted on an experimental basis. Furthermore, NSF-funded mid-level networks have been free to set their own policies. (25) See, Harnad, above. See also, Sharon J. Rogers and Charlene S. Hurt, "How Scholarly Communication Should Work in the 21st Century," Chronicle of Higher Education, October 18, 1989, p. A56. (26) Of course, publishers may be schizophrenic about downloading. See John A. Hearty and Barbara F. Polansky, "ACS Chemical Journals Online: Is It Being Downloaded, Do We Care?" in Intellectual Property Rights in an Electronic Age (Network Planning Paper No. 16), Library of Congress Network Advisory Committee, 1987, contrasting legal and marketing perspectives within the same organization. (27) The 1976 Copyright Act codified the First Sale Doctrine as 17 USC Section 109. In recent years, the First Doctrine has been modified with respect to phonorecords (1984) and computer programs (1989). Some European countries have limited the application of the First Sale Doctrine to libraries by creating a public lending right. (28) Thus, the copyright owner's permission is required to put a CD-ROM on a network because a remote display is created. In addition, many CD-ROMs are leased rather than sold, so that even a local display requires permission from the copyright owner. (29) Or, to look at it another way, the medium becomes the network -- an increasingly transparent, ubiquitous network -- and any storage device or display facility is merely an extension of the network. (30) In the copyright context, it is the distinction between the copy and the protectable expression that is disappearing. This means that prerogatives inherent in copies are disappearing. Special provisions for copies are found in Sections 110(1) ("lawfully made copies" under the "face-to-face teaching exemption") and 117 (owned copies of computer programs) as well as Section 109 (owned copies under First Sale Doctrine). (31) 17 USC Section 105. (32) See U.S. Congress, House Committee on Government Operations, Electronic Collection and Dissemination of Information by Federal Agencies: A Policy Overview, 99th Cong., 2d sess., 1986, House Report 99-560. (33) Dialog Information Services, Inc. v. American Chemical Society, Docket No. 90-1338, U.S. District Court for the District of Columbia. ACS's counterclaim for unpaid royalties illustrates the anxiety publishers feel toward vendors in the online environment. (34) While ACS is recognized a 501(c)(3) organization, exemptions for professional societies are currently granted under 501(c)(6), the category that applies to trade associations and business leagues. 501(c)(6) organizations are tax-exempt but not "tax-deductible" -- i.e., donations are not tax-deductible as charitable contributions and foundations cannot ordinarily make grants to 501(c)(6) organizations. Academic societies are more likely to qualify as 501(c)(3) organizations. (35) A pure site license permits unrestricted copying and use at a given site. However, site licenses often limit the number of copies or, in the case of network licenses, concurrent users. (36) 17 U.S.C. 108(a)(3). (37) 17 USC Section 108(g). (38) United States, National Commission on New Technological Uses of Copyrighted Works, Final Report (Washington, DC: Library of Congress, 1978), pp. 54-55; U.S. Congress, House of Repre- sentatives, Conference Report, Report No. 94-1733, September 20, 1976, pp. 72-74. (39) For the sixth and subsequent copy, many libraries pay the Copyright Clearance Center the fee indicated on the article. Some libraries feel that the fair use provisions in Section 107 of the Copyright Act cover at least some copying beyond that permitted under Section 108. In any case, the five-article limit, while cited approvingly by Congress, does not actually appear in Section 108. It is important to note that these guidelines only constrain requests by libraries. Individuals may request copies directly from the library holding the journal without any limitation on the number of times journal articles are copied. (40) CD-ROM licenses typically require the return of outdated CD-ROMs as new ones are received. (41) Although not subject to copyright within the U.S., works of the U.S. Government are protectable by copyright elsewhere. See U.S. House, Committee on the Judiciary, Copyright Law Revision: Report Together with Additional Views to Accompany S. 22, 94th Cong., 2d sess., 1976, Report 1476 pp. 58-60. Virtually all other countries protect government works by copyright, so under the reciprocity principles of the Berne and UCC conventions, those countries are obligated to provide protection to U.S. Government works on the same basis. In practice, international exchange agreements negotiated by NTIS often end up providing the information at no cost. (42) The United States Information Agency and private foundations, such as the Carnegie Corporation, have supported dissemination of print materials to countries in the developing world. There has yet been little concerted effort to make electronic information available on special terms, although Cornell has a project underway to provide core agricultural literature to the developing world on CD-ROM. Bibliography: American National Standards Institute, "Procedures for the Development and Coordination of American National Standards" (New York), 9 Sep 1987. Arnold, Stephen. "Electronic Information on CD -- A Product or a Service?" ONLINE (Nov 1987). Battin, Patricia. "The Real Electronic Library, " Information Technology Quarterly, IX (Spr 1990), 4. Baumol, William, et al., eds. Manual of Pricing and Cost Determination for Organizations Engaged in the Dissemination of Knowledge, National Science Foundation (Dec 1977). Berardo, Felex M. "Scientific Norms and Research Publication Issues and Professional Ethics," Sociological Inquiry, 59 (Sum 1989). Byrd, Gary D. "An Economics `Commons' Tragedy for Research Libraries: Scholarly Journal Publishing and Pricing Trends," College and Research Libraries (May 1990). Cecil, Joe Shelby, and Eugene Griffin. "The Role of Legal Policies in Data Sharing," in Stephen E. Fienberg, Margaret E. Martin, and Miron L. Straf, eds. Sharing Research Data (Washington D.C.: National Academy Press, 1985). Corporation for National Research Initiatives. "Knowbots in the Real World: Workshop on the Protection of Intellectual Property Rights in a Digital Library System" (Reston, VA: Corporation for National Research Initiatives, 1989). Copyright Clearance Center, Inc. "Text to Screen: Copyright Issues in the Electronic Age," 1989. Council on Library Resources. Communications in Support of Science and Engineering (Report to the National Science Foundation, Washington D.C., Aug 1990). DeLoughry, Thomas J. "Scholarly Journals in Electronic Form Seen as Means to Speed Pace of Publication and Promote Dialogue," The Chronicle of Higher Education (22 Mar 1989). Dodd, W. "Convergent Publication, or the Hybrid Journal: Paper Plus Telecommunications," Electronic Publishing, 3 (1990), 47-59. Dougherty, Richard M. "To Meet the Crisis in Journal Costs, Universities Must Reassert Their Role in Scholarly Publishing," The Chronicle of Higher Education (Apr 1989). Eisenberg, Daniel. "Problems of the Paperless Book," Scholarly Publishing (Oct 1989). Eres, Beth Krevett. "International Information Issues," Annual Review of Information Science and Technology (1989). Gabig, Jerome S. "Federal Research Grants: Who Owns the Intellectual Property?" Harvard Journal of Law and Public Policy (Sum 1986). Gardner, William. "The Electronic Archive: Scientific Publishing for the 1990s," Psychological Science, I (1991), 333-341. Gire, Judith. "Current Literature in Law/Science: Policy and Intellectual and Industrial Property," IDEA, 30 (1989), 51-94. Harnad, Steven. "Scholarly Skywriting and the Prepublication Continuum of Scientific Inquiry," Psychological Science, 1 (Nov 1990), 342-344. Hogan, Tom. "Dialog Sues ACS Over Access to Data," Information Today, 7 (Jul/Aug 1990), 1. Horowitz, Irving Louis. Communicating Ideas (New York: Oxford University Press, 1986). Jollymore, Nicholas J. "Ancillary Uses of Collective Works," Communications and the Law (Sep 1990), 21-47. Kahin, Brian. "Property and Propriety in the Digital Environment: Towards an Examination Copy License," EDUCOM Bulletin, 23 (Win 1988), 15-24. Kennedy, Donald. "On Academic Authorship" (Report to the Office of Scholarly Communication and Technology, Washington D.C., 18 Sep 1985), reprint no. 4. Koch, Susan E. "How Electronic Networks Are Changing Science: Implications for a National Research and Education Network" (Report to the Office of Technology Assessment, Washington D.C., 15 Dec 1989). Lederberg, Joshua. "Digital Communications and the Conduct of Science," Proceedings of the IEEE, 66 (1978), 1314-1319. Lederberg, Joshua, and Keith Uncapher, co-chairs. "Towards a National Collaboratory: Report of an Invitational Workshop at the Rockefeller University," 17-18 Mar 1989. Lyman, Peter. "The Library of the (Not-So-Distant) Future," Change, 23 (1991), 34-41. Merton, Robert K. "The Matthew Effect in Science, II: Cumulative Advantage and the Symbolism of Intellectual Property," ISIS, 79 (1988). McCain, Roger A. "Information as Property and as a Public Good: Perspectives from the Economic Theory of Property Rights," The Library Quarterly, 58 (Jul 1988), 265-282. McClure, Charles R., Ann Bishop, and Philip Doty. The National Research and Education Network: Research and Policy Perspectives (Norwood, NJ: Ablex, 1991). McDonald, Kim A. "Despite Benefits, Electronic Journals Will Not Replace Print, Experts Say," The Chronicle of Higher Education (27 Feb 1991). National Academy of Sciences, Committee on Science, Engineering, and Public Policy. Information Technology and the Conduct of Research: The User's View (Washington, D.C.: National Academy Press, 1989). National Federation of Abstracting and Information Services. "NFAIS Gateway Code of Practice," June 1987. Neavill, Gordon. "Electronic Publishing, Libraries, and the Survival of Information," Library Resources and Technical Services (Jan/Mar 1984). Okerson, Ann, and Kendon Stubbs. "The Library `Doomsday Machine,'" Publishers Weekly, 238 (8 Feb 1991). Ordover, Janusz A., and Robert D. Willig. "On the Optimal Provision of Journals Qua Sometimes Shared Goods," American Economic Review (June 1978). Osburn, Charles B. "The Place of the Journal in the Scholarly Communications System," Library Resources and Technical Services (Oct/Dec 1984). Osburn, Charles B. "The Structuring of the Scholarly Communication System," College and Research Libraries, 50 (1989), 277-286. Peek, Robin P., and Joan N. Burstyn. "In Pursuit of Improved Scholarly Communications," in Joan N. Burstyn, ed., Desktop Publishing in the University (Syracuse, 1991). Roberts, Leslie. "The Rush to Publish," Science, 251 (18 Jan 1991), 260-263. Rogers, Sharon J., and Charlene S. Hurt. "How Scholarly Communication Should Work in the 21st Century," The Chronicle of Higher Education (18 Oct 1989). Rosenberg, Victor, and Gretchen Whitney, eds. The Transfer of Scholarly, Scientific, and Technical Information Between North and South America (Meutchen, NJ: The Scarecrow Press, 1986). Sauer, Raymond. "Estimates of the Returns to Quality and Co-authorship in Economic Academia," Journal of Political Economy, 96 (Aug 1988). Shaver, Donna B., Nancy S. Hewison, and Leslie Wykoff. "Ethics for Online Intermediaries," Special Libraries, 76 (1985), 238-245. Solomon, Richard J. "Vanishing Intellectual Boundaries: Virtual Networking and the Loss of Sovereignty and Control," in Murray Abron, ed., "Telescience: Scientific Communication in the Information Age," Annals of the American Academy of Political and Social Sciences, 495 (1988), 40-48. Sterling, Theodor E. "Analysis and Reanalysis of Shared Scientific Data," in Murray Abron, ed., "Telescience: Scientific Communication in the Information Age," Annals of the American Academy of Political and Social Sciences, 495 (1988), 49-60. Turner, Judith Axler. "Enormous Changes in Scholarly Publishing Expected as a Result of Advances in Information Technology," The Chronicle of Higher Education (21 Nov 1990). Turner, Judith Axler. "Research Libraries Mount Campaign to Combat Escalation of Journal Prices," The Chronicle of Higher Education (24 May 1989). Turner, Judith Axler. "Technology, New Attitudes Called Threats to Publishing," The Chronicle of Higher Education (14 June 1989). Unruh, B. "Copyright Issues: The Primary/Secondary Journal Interface," Information Services and Use, 9 (1989), 363-371. Waldrop, M. Mitchell. "Creating the Future for Electronic Publishing," Science, 251 (1 Mar 1991). Weil, Ben H., Kurt D. Steele, Morton David Goldberg, M. Lynne Neufeld, and Lois W. Granick. "Positions of the Major Players in Abstracts," American Society for Information Science Bulletin, 9 (1983), 18-21. Weil, Vivian and John W. Snapper, eds. Owning Scientific and Technical Information: Value and Ethical Issues (New Brunswick, NJ: Rutgers University Press, 1989). Weingarten, Fred W., and D. Linda Garcia. "Public Policy Concerning the Exchange and Distribution of Scientific Information," in Murray Abron, ed., "Telescience: ScientificCommunication in the Information Age," Annals of the American Academy of Political and Social Sciences, 495 (1988), 61-72. Wilson, David. "Testing Time for Electronic Journals," The Chronicle of Higher Education (11 Sep 1991). Wulf, William. "The National Collaboratory: A White Paper," National Science Directorate for Computer and Information Science and Engineering, 20 Dec 1988. Yavarkovsky, Jerome. "A University-based Electronic Publishing Network," EDUCOM Review, 25 (1991), 14-20. Zahray, Paul, and Marvin Sirbu. "Provision of Scholarly Journals by Libraries Via Electronic Technologies: An Economic Analysis," Information Economics and Policy, forthcoming.