LEGAL IMPLICATIONS OF ELECTRONIC INFORMATION
Rights Management:
Issues and Solutions
Rights Management
Bill Tuck
Independent Consultant
Tuck Segal Associates London UK
Recent months have seen a number of information sources, such as the Encyclopedia Britannica and the New York Times, become available on a commercial basis. Many more will follow and we will begin to see an explosion in the quantity of copyright material accessible on a fee paying basis. This will be handled in a number of ways, including subscription based services and pay--per--view.
This paper will report on recent developments in this area and discuss some of the economic, technical and ethical issues associated with the sort of information used and collected by electronic rights management systems.
As we move into an all pervasive “information economy”, intellectual property becomes of increasing significance. Electronic copyright management systems (ECMS) for rights management are emerging as one of the most important components in the control and administration of intellectual property rights.
One feature that is immediately apparent, however, is the diversity of views associated with the problem of electronic copyright management. These range all the way from statements to the effect that the computer (or, more precisely, digital technology) will allow total control over the production and distribution of cultural artefacts — whether print, images, audio or film — to the opposite extreme of saying that digitisation renders the very idea of copyright meaningless. There may be some truth in both points of view. What is very clear is that the issue is a complex one.
Anything may be digitised. For this reason, managing copyright control should be possible uniformly over the whole range of media types — text, images, audio, film, etc. Each object may be represented as a sequence of bits, so that any method capable of encrypting, marking or identifying the corresponding binary file should be applicable to all types. It is on these grounds that some believe the problem to be universally soluble — “the answer is in the machine”.1
The contrary view is less optimistic. The universal availability of digital technology — scanning, digitising, processing, network publishing, etc. — means that it is increasingly difficult to exert control over the processes of copying, modifying, and redistributing data2. As an example, once information is converted into an analogue form (such as by printing to paper) much of this digital control may be lost, as it can then be copied, distributed (e.g. by fax) or re--digitised unhindered.
There is no simple technical solution. For this reason, an important element of copyright management will continue to be moral persuasion. Even here, of course, there is considerable scope for the use of technology. For example, if documents are to be scanned for digital transmission as part of a document delivery service, then it is relatively simple to incorporate a copyright notice into the digital page image of the document. This might also include information on the supplier and intended recipient, along with other details. Although such a notice is unlikely to prove any serious technical constraint to misuse (re--scanning and a minimum of digital manipulation might easily eliminate it) it does provide an element of moral persuasion. In addition, this is likely to be endorsed by law, making the deliberate removal of such rights management information a legal offence.
In many situations, however, the effectiveness of copyright control may depend more on the ease of compliance than on the security of constraint. This applies, in particular, to the procedures necessary to gain copyright clearance, either from publishers (or other rights holders) or copyright clearance centres. If clearance is onerous, time--consuming or expensive, then users are much more likely to avoid formal procedures and simply take the chance that they will not get caught. Illegal copying for private or classroom use will generally fall into this category. The question is whether technology can be applied to make compliance easier.
In the longer term, initiatives such as national agreements on site licensing may simplify the problem. The recent contracts negotiated between the UK academic community and four major publishers serves as an important model. In exchange for a guarantee of subscription levels being maintained, a considerable degree of liberty can be afforded to the user as to what they may do with the material, including unlimited copying for classroom use.
For those outside such an agreement, however, the options may be more limiting. While pay--per--use may seem the logical strategy, it is not yet clear whether this approach is viable for low use academic material (compared to high--value “entertainment” such as films, records or games). A period of experimentation may be necessary before an answer emerges.
What is ECMS?
Electronic Copyright Management is the general term used here to cover all aspects of applying digital technology as a means of controlling copyright. It covers identification of copyrighted items and their attribution (author, publisher, etc.) as well as techniques to control access (i.e. prevent copyright infringement) and to track usage (along with billing procedures, where appropriate). In all three areas digital technology will have an important role to play.
Identification and attribution
Legislation on copyright is detailed, complex and extensive. As the name implies, it concerns the right to copy items. Its intention is to protect the skill that an author, artist or other creator expends in creating an original piece of work. The work so protected may be literary, musical, a painting, a photograph, a TV programme or whatever.
There is a difference between ownership of copyright and of “moral rights”. The latter ensure the right of the author of a work to be acknowledged as such and not to have the work subjected to derogatory treatment. They also give to authors the right not to be falsely attributed with authorship of work not created by them. While copyright may be assigned to another person, or organisation, such as a publisher, moral rights always remain with the original author, they cannot be re--assigned. Again unlike copyright, which is automatic, moral rights must be asserted.
For any form of copyright control to be exercised it must first be possible to uniquely identify the work and assign its attributions — author, publisher, copyright owner, etc. In the case of recordings, paintings, and photographs, this may not be a trivial problem. Even for “standard” textual works, or “documents”, attribution and identification may not always be easy.
One obvious problem is that of granularity. While it may be relatively easy to identify an entire work such as a book or article, say, on the basis of its cover information or “front matter”, an isolated chapter (let alone paragraph or sentence) could be very difficult to place. Digital marking might help, but even that must break down at some point. At the single word level it becomes absurd.
The problem is, nevertheless, quite real. Sampled sounds, fragments of prose or pictures are often reassembled into new works. How can the attribution of these components be properly assigned and the inherent copyrights protected? Pictures used to illustrate a text are a good example of how easily the problem can become insuperable. To copy such a document (for classroom use, say) might entail many hours of effort and weeks of delay in obtaining all the necessary permissions.
A number of different methods have been proposed for marking or identifying documents or images. Many rely on clever techniques for incorporating hidden codes within the document image — so--called watermarking or “steganography” — to serve as a sort of registration number.3 The intention of these codes is to enable the document, its origins and rightful recipient to be traced.
One example of such a technique is the “hidden line” encoding developed by AT&T as a possible control technology for the Right Pages document supply service. In this method, by varying the spacing between the lines of text, information may be concealed from casual view, but will persist through several generations of photocopy4. Unfortunately, it would not survive the more radical transform of OCR.
Another suggested scheme involves the registration of a characteristic digital signature generated by “hash encoding” the individual sentences of the text5. The set of numbers corresponding to the document (its “signature”) would be held in some central archive along with details of rights ownership. The problem here is one of administrative cost.
Perhaps the most highly developed schemes are based on the assumption that a strong measure of control over the user's display device is feasible. This is intended to prevent access to the document in unencrypted form, so that it cannot be copied or re--distributed in an uncontrolled way. The difficulty with this approach is that to be acceptable it will require adoption of a universal encoding standard and compatibility over a wide range of equipment manufacturers. While this might be feasible for CD video players, it will not be easy to achieve in the PC world.
One serious problem with all these technologies is that they are still experimental and doubt remains as to the real level of security provided by any of the suggested encoding schemes. The main problem, however, is the cost of tagging items in this way, relative to their intrinsic value. While images with a high tradable value (such as those used in the advertising industry) may warrant such measures, in the case of academic journal articles this additional cost may be difficult to justify. Here the simpler method of overprinting the image with its identifying label and provenance could be both cheaper and more effective.
Access control
The second component in the logical management of copyright is control over access. If a database is licenced to a particular organisation or university site, then the terms of that licence will generally exclude persons who are not members of the organisation or located on the named site from accessing the database. Techniques for authentication of network users are well--developed for this purpose.
In the case of print material, on the other hand, many libraries will generally operate a policy of open access, even to members of the general public. In addition, there are seldom any restrictions on who may make copies of such material, provided the normal rules of copyright, or “fair dealing”, are observed.
In other words, access control in the case of printed material is not seen as an issue, whereas for electronically held material it is generally regarded as fundamental. This inconsistency is perhaps understandable, but it serves to point out the kind of problem that is likely to arise as information is increasingly held in electronic form.
A public library might subscribe to a printed journal and allow anyone into the library to read it or make copies. It could not in the same way buy an electronic subscription to a journal and allow unrestricted network access on a global basis (via the Internet, say). Whereas access to print is limited by the physical need to travel to the library, no such constraint operates in the case of network access to an electronic document.
The logical conclusion is that “public” network resources can only provide access to “free” material. No publisher would sell to a library on an open access basis since, at least in principle, just one copy would suffice for the entire world. This is an extreme example of the problem of arbitrage — how do you ensure that when pricing a product for one market it is not simply transferred to another? Libraries offering products at no charge will inevitably undercut others offering the same product at a higher price. The normal solution is product differentiation, but this may be difficult in a networked environment in which one Web server is much the same as any other.6
The present Internet is, of course, a a kind of “public library” in which virtually all information is free and freely available (apart from network service charges). The downside is that, as with free newspapers, much of the information is simply publicity or advertising material, or else may be of doubtful provenance. This presents the public library “open access” principle with something of a dilemma — how to ensure equitable access to networked information without undermining commercial markets or disenfranchising less wealthy users. Access control is essential if the service is to provide network access to paid--for material, such as journals or databases for which a subscription is charged. This does not mean that the end--user must also pay, but simply that access must in some way be restricted or controlled. The simplest form of access control is to limit the network to the local area or even to a single workstation. Wider access can be retricted to known network addresses (e.g. Internet IP addresses) or registered users.
Tracking and billing
Finally, tracking the usage of material and, where appropriate, organising billing procedures are both fundamental operations in the management of copyright. Clearly, however, the approach to tracking and billing will depend on the particular service. It will be different for a network publishing operation than for a document delivery service, or a copyright licensing service for on--demand publishing of course packs. Even in the case where the material is freely accessible from a Web server, the need to monitor usage levels could be important, if only to satisfy advertisers.
It is at present unclear what level of monitoring may be necessary, and this will in any case depend on the charging policies in operation. There may also be important issues of privacy in this area, such as the legitimacy of the library knowing what individual users are reading, or printing. True, it may help them decide what electronic journals to stock, though this need not be done by recording individual usage.
A basic question is whether the service is to be supplied on a pay--as--you--go (pay--per--view) or on a subscription basis. This issue is complicated and will again depend on the particular service and how it is to be implemented. Subscription--based services are in general easier to organise and have the benefit of predictable costs (and revenues). Pay--as--you--go appears to benefit the casual or infrequent user and allows a more equitable allocation of costs based on resources actually used.
In situations where the utilisation rate is low, such as academic journal publishing, there is a strong case for subscription--only services. Average annual readership of journal articles from research publications may be less than 10 world--wide (and for some could be very much less). To support the cost of publication (the “first-- copy” costs) would require a very high charge for each access. Only by basing publication on subscription can the economic risks be offset.
Where the marginal costs of a service are high, on the other hand, (such as with document delivery from printed journals, using labour--intensive scan--on--demand) a pay--per--use policy is clearly preferable. Where the marginal costs are low, however, (such as document delivery from electronically stored journals) a subscription--based service could be justified.
The issue is further complicated by the chains of supply. A library might buy in electronic services from a third party or agency (such as a database host service), which in turn buys from a publisher. If the library--user is offered a subscription--based or unlimited access service (to an electronic journal, say), then it will be risky for the library to buy in these services on a pay--per--use basis from its own suppliers. On the other hand, supplying pay--per--use services to customers from material bought in on subscription may make good economic sense.
From the point of view of electronic copyright management, subscription--based services are much easier to deal with as the system does not have to monitor usage, keep accounts or bill customers. The overall cost should therefore be lower. At the same time, there are strong moves towards development of so-- called “micro--billing” systems, where numerous small transactions — each perhaps at a fraction of a cent — can be accumulated for eventual billing. Such systems will be essential if commercial information services on the Internet are to be feasible.
These problems run in parallel to the basic question of Internet charging itself. Current network access services are generally supplied on a flat rate basis, leading to inevitable congestion on the network. It is argued that usage sensitive or transaction based charging will be necessary if the quality of service is to be improved.7
Digital information services
Documents may be encoded digitally either by capturing them in digital form at source (e.g. as a PostScript or PDF file, or as an SGML/HTML document) or by scanning from paper (giving a linked set of TIFF images, for example). The former will usually be distributed via some sort of electronic publishing system (such as a Web server) while the latter may form the basis of a document delivery service.
Similar considerations will apply to other forms of information. Colour images, audio sequences or video clips may either be originated in digital form or converted by digitisation of the analogue medium — film transparency, tape recording or video playout. The technologies may differ but the basic principles are the same.
The present discussion is largely confined to the case of conventional documents — either print on paper or electronic — since these form the principal content--type of our libraries and information services at the moment. At the same time, the arguments are easily extended to non--traditional material, such as sound or video.
There are three easily identifiable services based on documents of this kind: electronic document delivery, electronic journal publication and on--demand publishing (eg. of reserve holdings or coursepacks). Each is subject to somewhat different constraints and regulations as regards copyright. The following sections discuss each in turn.
Electronic document delivery
Current document delivery services are generally based on either photocopy and post or fax. Electronic document delivery is the use of a digital network to transmit the article. The best known example is probably the Ariel system developed by RLG. With Ariel, documents are in the form of TIFF--encoded page images, together with a header file encoded according to the GEDI (Group on Electronic Document Interchange) standard.8 Information in the GEDI header is used to create a cover page for the printed document. This would normally include copyright information (such as any statutory copyright declaration notice) along with the names of the supplier and the requester, together with bibliographic details.
Of course, a copyright notice on the cover page does little in practical terms to prevent copyright abuse. The front cover is easily removed, and the rest may be photocopied with impunity. Nevertheless it does provide an explicit warning and thereby serves a useful educative function. In this sense, as a minimal requirement, it goes some way towards alleviating the mistrust of publishers.
To seriously tackle the copyright control issue will need something more than a detachable notice on the cover page. One suggestion mentioned earlier is watermarking or fingerprinting. In the form of hidden codes — either by line spacing, character modification, or image transform encoding — these could be introduced into the page images of the document itself. It is questionable, however, whether they would serve any useful purpose.
A much better option — and simpler to implement — would be to print the copyright control information legibly on the front page of the article itself. There are a number of ways in which this could be done. The simplest is to print the notice as a text header or footer on the page. To do this, the corresponding text would need to be converted into a bitmap (or TIFF image) with the same configuration parameters as the original document file. This page overlay would then be merged with the TIFF image of the original page (or pages). The next step might be to merge it as a “shadow print” into the background of the document.
Other possible elements for copyright control — such as document content registration, or universal article identifiers — seem to be of little immediate value in the context of academic electronic document delivery. In any case, they are really part of the pre--print process and therefore the responsibility of the publishers themselves. If the original article came already printed with a barcode containing the identifier information, as has been suggested, this might be read by the scanner and the information incorporated into the GEDI header. It might then be useful within the tracking and billing procedures.
Electronic journals
Many publishers are now beginning to issue their journals as dual publications. A typical example is that of the IEE, which from early 1996 has made several publications available for distribution via OCLC. Material must be SGML encoded, and the receiving station needs the OCLC Guidon software (later versions will use “standard” Web browsers, such as Netscape, and Java to extend HTML to full SGML). There are over a dozen titles now available through this channel. Other publishers are looking at PDF as the distribution format.
Copyright controls on the current generation of electronic journals (including the dual publication, standard titles), appear to be minimal. Some Internet journals are subscription based, for which each subscriber gets an individual password. The titles published through OCLC (including those of IEE) tend to be site licensed, for which the entire site may have just the one password together with the constraint of only one concurrent access (or multiple simultaneous accesses for a higher fee). ACM and AMS (Association for Computing Machinery and American Mathematical Society) are believed to offer individual subscriptions as well as multi--access site licences, with the latter based on IP address groups.
The copyright issue in this case generally revolves around the question of site licensing and access authorisation. What is a site? How do off--site users get access? How are passwords to be managed? etc. The present arrangements seem simplistic and unreasonable and will surely change under pressure from users, as well as from publishers seeking better forms of copyright control.
Commercial interests are moving rapidly into this area and a number of different charging and distribution models are being tested. Some 20 or more publishers, including the New York Times, have already signed up to contribute to an initiative from Netscape to deliver personalised news directly to end users via email. The service will be subscription based, although for a trial period it is free.
Pay--per--use access to electronic publications on the Internet is less common. Some experiments in this area have been carried out by the IEE as part of the EU--funded COPINET project, but as with most such services, successful deployment will depend on the development of a secure micro--payments facility.9 This is in marked contrast to the traditional online industry where time--based or volume--based charging (including pay--per--view) is the norm.
In the UK a site licence agreement has been obtained between the universities and four major serials publishers (Academic Press, Blackwell Publishers, Blackwell Scientific and the Institute of Physics Publishing). This represents an important development in the provision of journals to the academic sector. The agreement covers printed journals as well as their electronic counterparts, as the latter become available. AP and IOPP, for example are providing access to all their journals in electronic versions from January 1996.
The licensing arrangements allow articles to be included in study packs, and impose no restrictions on “non--profit” photocopying or electrocopying. Conventional interlibrary loans based on the original paper versions are also permitted, although it is difficult to see why any library would be interested in supplying documents at high marginal cost in a scan--on--demand operation rather than at zero marginal cost from an electronic copy. Clearly what is needed is a mechanism to allow “secondary licensing” whereby members of one library may get access to material held by another on a temporary basis. This would also cover the case of “unregistered” users and should, in principle, solve the document delivery problem.
On--demand publishing
On--demand publishing has a different set of copyright requirements from those of document delivery systems or electronic journals. Where the material to be published comes from existing publications, copyright clearance will be needed. If the material is originated in--house, then clearance is not required. In either case, however, some copyright control over the final product will be necessary — if only in the form of a copyright notice inside the front cover to inhibit re--copying.
Clearance procedures are onerous. Royalty payment costs also appear high. The combination of the two (since the introduction of revised CLA regulations in 1993) has severely inhibited the growth of course--pack publishing at higher education institutions in the UK. There is an obvious need to use technology to improve clearance procedures, eventually towards implementing a full--scale EDI based system between the universities (or other customer sites) and the copyright licensing agencies.
The other factor, high royalty costs, can only really be resolved by negotiations with the publishers. The problem with this is that it can be time--consuming and complicated. If each publisher has to be contacted and negotiated with separately, then the process is likely to be long drawn out and generally unsatisfactory. Collective agreements are clearly required.
Lastly, copyright control on the course--pack itself is probably best managed by conventional means. Experience has shown that they are highly susceptible to informal photocopying. This is, of course, illegal — though students may not be aware of that, since it does not seem the same thing as copying a book. The only constraint available is a copyright notice and adequate warning of the penalties of infringement. Whether this is likely to have any effect is another matter.
If students themselves are charged for the printing, and the printing is done to individual demand rather than in batch, then controls may be easier to implement. Network printing and mechanisms for page charging are becoming more widespread. It should not be difficult to add on an additional fee for copyright material held in a digitised reserve.
In the longer term, however, the preferred solution may lie once again with site licensing schemes such as that being adopted in the UK. Course--pack publishing and electronic reserves, along with document delivery, then fall under the subscription arrangements rather than as pay--per--use services.
Conclusions
The general conclusion of the copyright study on which this paper is based was that the technology of this area is still relatively immature. Formally published technical work on ECMS and related technology, as distinct from speculative commentary, is limited. There is also little evidence of any serious attempt at comparative evaluation. For this reason it is difficult at this time to place very much confidence in the few proposed solutions.
At the same time, there is ample evidence of considerable activity in the field. Given the importance that is at present attached to finding ways of enabling secure commercial transactions over the Internet, it is likely that workable solutions and products will emerge in the fairly near term. Most of these products, however, will be targetted at securing access to high value items such as mainstream films, games, financial publications, etc. Methods appropriate for dealing with material of low per--item value (such as academic research publications) are less likely to emerge in the short term, at least from this source.
In the case of high value items, the method of control tends to be based on the assumption of secure access and a considerable degree of control over the terminal equipment (TV, multimedia system or workstation) on which the product (whether film, audio or interactive game) is to be played.
In the case of academic material, such a degree of control may be both inappropriate and unattainable. This is due in part to the nature of academic work and in part to the context in which it normally occurs. As a (admittedly simplistic) generalisation, “study” involves the dismembering of an object — whether machine, organism or document — for the purpose of analysing its workings, which are then re--presented as an integral part of the explanation. Clearly, dealing with published documents (whether texts, audio or film) in this manner may be difficult to reconcile with strong control over copyright.
The context of academic enquiry must encourage free access to the material of enquiry. To some extent this principle has been enshrined in the concepts of “fair dealing” and “library privileges” which have guaranteed the unrestricted right of access “for purposes of private study or research”. It is on this basis that making copies of library--held documents is generally permitted. Strong access control, of the kind envisaged by some of the copyright management systems being proposed, may conflict with this right. Most such systems, following the CITED model, see the primary role of the ECMS as being to ensure the integrity of the document and the secure passage of royalty payments back to the rights holder.10 The notion of “making copies for private use” has little place within such a scheme.
For academic material, however, a much looser form of control may be desirable, not only from the point of view of readers, but also for publishers. Much of this control can be achieved by the relatively simple means of linking the document to explicit copyright declarations, making clear to the reader which actions are permitted and which forbidden. In general, permitted actions would include printing and copying for private use, but not the systematic downloading of entire journals or any commercial reselling of material (such as in course packs).
The proper aim of ECMS, as with copyright in general, is to find some formula wherein the author/publisher interest in receiving compensation can be balanced with the user/library interest in having access to information on fair and reasonable terms.11 Whether the information is to be provided on a pay--per--view or subscription basis should be part of this negotiation. It is not at all clear, for example, that marketing academic material on a per--article pay--per--view basis is even financially viable. The small number of existing empirical studies indicate that overall readership levels for this kind of material may be too low to allow for any reasonable level of cost recovery — at acceptable prices. If document delivery services were to wholly replace subscriptions, charges per article would have to be unrealistically high to replace lost revenue. In addition, publishers would have to carry the cost of implementing a tracking and billing system.
Subscription--based site licensing is therefore emerging as the preferred way for future publishing in this area. The national contract recently agreed by the UK academic community with a number of major publishers serves as an important model. In exchange for a guarantee of subscription levels being retained, a considerable degree of liberty can be granted to the user as to what they may do with the material.
The one area where such site licensing may have considerable impact on current services is document delivery. There is a general indication that publishers are unhappy with the systematic photocopying of documents for supply on demand, as this clearly threatens their subscription base. On the other hand, until the technology of network distribution improves, there is no realistic alternative. With full--scale electronic publication, however, the need for document supply services may dwindle.
Current offerings are generally subscription based. Nevertheless, a large number of experimental systems based on the pay--per--view approach are being developed. Although the general conclusion of this paper is that this approach is unlikely to prove viable, at least within the higher education sector, it will be important to continue to monitor this work. This situation may change as publications move from being primarily in print form to primarily electronic, with a corresponding shift in the underlying economics.
Acknowledgements
This paper is based on a study on ECMS carried out as part of the UK Electronic Libraries (eLib) programme and funded by the universities' Joint Information Systems Committee. Their sponsorship of this work is gratefully acknowleged. The full text of this report is available from the Library Information Technology Centre, South Bank University, Borough Rd, London SE1 (or online at URL: www.sbu.ac.uk/litc/copyright/ecms.htm).
References
1
Clark, C., The Copyright Environment for the Publisher in the Digital World,
ICSU Press–UNESCO, 1996.
Also at http:// www.lmcp.jussieu.fr/ICSU/Information/Proc
0296/clark.html
2
Rosenberg, V., “Will new information technology destroy copyright?”, The Electronic Library, vol. 12, no 5, October 1994, pp. 285–287.
3
Boland, F.M., O'Ruanaidh, J.J.K., and Dautzenberg, C., “Watermarking digital images for copyright protection”, Image Processing and its Applications, 4–6 July, 1995, (IEE Conference Publication No. 410), pp. 326--330.
4
Choudhury, A.K., Maxemchuk, N.F., Paul, S. and Schulzrinne, H.G., “Copyright Protection for Electronic Publishing over Computer Networks”, IEEE Network, vol. 9, no. 3, May/June 1995, pp. 12–20
5
Brin, S., Davis, J., Garcia--Molina, H., “Copy detection mechanisms for digital documents”, ACM SIGMOD 95, pp. 398–409.
6
Varian, H., “The economics of the Internet and academia”, in Butler,M. and Kingma, B. (eds), The Economics of Information in the Networked Environment, Proceedings of the Conference: Challenging Marketplace Solutions to Problems in the Economics of Information, Washington DC, September 18ä19, 1995, Association of Research Libraries, 1996.
7
McKnight, L. and Bailey, J. (eds.), “An introduction to Internet economics”,
in Special Issue on Internet Economics, Journal of Electronic Publishing,
Fall 1995.
Also at http:// www.press.umich.edu/jep/works/McKniIntro.html
8
GEDI, “Electronic Document Delivery: Towards Further Standardisation of
International Interchange — Agreements of the Group on Electronic Document
Interchange (GEDI), Version 2.1", Pica, Leiden, July 1995.
Also at http://
www.gu.edu.au/lib/iii/docdel/gediv21.htm
9
COPINET reports and information available from
http:// www.mari.co.uk/~copi/copinet
10
CITED, “Final Report of the CITED Project”, 1995.
Available from the British
Library Document Supply Centre, Boston Spa, West Yorkshire, LS23 7BQ, UK.
11
Samuelson, P., “Copyright and digital libraries”, Communications of the ACM, vol. 38, no. 3, April 1995, pp. 15–21, 110.