JILT 1998 (2) - Dan L. Burk
Proprietary Rights in Hypertext Linkages
Plaintiffs in several recent court cases have claimed that unauthorised hypertext linkages infringe the intellectual property rights of the owner of the target material. Such claims could potentially chill the robust development of hypertext networks. However, such claims of legal liability based upon theories of copyright, trademark, or false designation of origin appear to be unfounded. When properly analysed against the operation of hypertext linkages, none of these theories of liability offers a sufficient ground for excluding such links.
Keywords: Copyright, Intellectual Property, Trademark, Contributory Infringement, Fair Use, Liability, Law
The author wishes to acknowledge helpful discussion by the members of the CYBERIA-L listserv, which helped clarify several of the points made in this paper.
This is a Refereed Article published on 30 June 1998.
Citation: Burk D L, 'Proprietary Rights in Hypertext Linkages', 1998 (2) The Journal of Information, Law and Technology (JILT). <http://elj.warwick.ac.uk/jilt/intprop/98_2burk/>. New citation as at 1/1/04: <http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1998_2/burk/>.
The explosive expansion of hypermedia usage, particularly via the Internet, has prompted the filing of at least three lawsuits involving the issue of unsolicited hyperlinking. Plaintiffs in these cases have alleged that hyperlinks referencing material at other sites on the World Wide Web infringe intellectual property rights in referenced material. In particular, infringement of copyright, trademark, and related business reputation rights have been claimed. Such legal claims have important ramifications for hypermedia: if the infringement claims are legally valid, development and use of hypermedia works may be curtailed by legal liability. Additionally, if such claims are valid, they may form the basis for proprietary control of information tagging and reference.
This article analyses the assertion that hyperlinks on the Web may infringe copyright, trademark, or related business reputational rights, concluding that such rights will not usually be implicated by such linkages. Because the majority of claims brought to date have been within the United States, the analysis in the article will focus on principles of U.S. law, while including appropriate reference to proprietary claims asserted in the United Kingdom. Based on presently established legal principles, the article concludes that such claims of infringement appear to be legally unsound, and should not be permitted to curtail the development of hypermedia works.
The first of the recent disputes over hypertext linking arose in the United Kingdom between rival newspapers in the Shetland islands, the Shetland Times and the Shetland News. Shetland Times v. Dr. Jonathan Wills and Zetnews Ltd.  (Hamilton, L., Edin Ct. Sess.). Each of the papers maintained a web site which carried the text of its printed publication. The Shetland News offered hypertext links to its competitor's news stories via 'hot linked' headlines taken verbatim from the rival Shetland Times. The Times successfully obtained an injunction against such linking when Scotlands Outer Court of the Court of Session held that the Times web page was a 'cable program' and that inclusion of the verbatim headlines constituted prima facie copyright infringement. However, no formal opinion was issued as the parties settled the case before final judgement.
Three subsequent disputes in the United States have raised similar issues. In the first, a World Wide Web site called 'TotalNews' offered a collection of hypertext links to other sites offering news content. The Washington Post Co. v. TotalNews, Inc. TotalNews itself offered no original news content, only the reference links to other sites. When users activated the proffered links, the content from the referenced site was displayed within a frame generated by TotalNews, which contained TotalNews advertising. Several of the referenced sites, including CNN, the Washington Post, and others, objected to this practice and filed suit against the operators of TotalNews, alleging breaches based on a variety of intellectual property principles. The dispute settled when TotalNews agreed not to 'frame' the content of sites that objected.
Subsequently, a linking dispute arose between Ticketmaster, a firm offering entertainment ticketing services, and Microsoft. Ticketmaster Corp. v. Microsoft Corp. As part of its online regional services, Microsoft sponsors 'Seattle Sidewalk,' a compendium of materials relating to the city of Seattle, Washington. 'Seattle Sidewalk' includes listings of arts and entertainment events in the city, and Microsoft provides a hypertext link to Ticketmaster's web site, for ease in ordering tickets to listed events. Ticketmaster has objected to the link, asserting in a legal pleading that the link creates a false impression of a business relationship between Microsoft and Ticketmaster.
Most recently, litigation over linkage to an online dental referral business produced two short opinions. Futuredontics, Inc. v Applied Anagramics, Inc. ; Futuredontics, Inc. v Applied Anagramics, Inc. . The plaintiff referral service, Futuredontics, licensed one of its service marks from defendant Applied Anagramics, which provided a hypertext link from it's own web site to that of Futuredontics. The Applied Anagramics link referenced Futuredontics web pages within a frame, prompting Futuredontic's multiple claims for copyright infringement and for unfair competition. The trial court in the ongoing case has thus far ruled that the plaintiff's claims are sufficient to withstand a motion to dismiss, but insufficient to warrant issuance of a preliminary injunction.
The practical implications of the claims raised in the lawsuits against the Shetland Times, TotalNews, Futuredontics, and Microsoft are not difficult to envision. If linkage to openly accessible digital files constitutes infringement of the file owner's intellectual property rights, then unauthorised creation of such linkages is fraught with potential liability. Hypertext authors who wish to reference the work of other authors would be required to either seek permission for each such link, or risk a lawsuit and possible damages award. Either option would significantly raise the cost of hypertext authoring, and as a matter of basic economics, when an activity becomes more expensive, people engage in less of that activity. As a consequence, adoption of the TotalNews or Ticketmaster theories of intellectual property liability holds the potential to curtail - and perhaps severely curtail - the production of robust hypertext.
However, the full implications of these theories may be far broader and far more serious than the obvious burden on hypertext authoring. Such implications become apparent when the nature of hypertext linkage is considered. Typical mark-up codes for hyperlinking include the HREF and the IMG codes, which prompt retrieval of, respectively, text or images. (Norderhaug and Oberding 1995). Retrieval may be automatic, or prompted by the user of the hypertext document. In the case of the IMG code, retrieved material will frequently be displayed 'inline,' that is, within the displayed document. But despite these superficial differences in usage, both codes perform essentially the same function. Both types of codes use the Universal Resource Locator (URL) as a reference to retrieve the linked material.
Thus, the hypertext link is in essence an automated version of a scholarly footnote or bibliographic reference: it tells the reader where to find the referenced material. In the case of hypertext, the user's browser or other application can then retrieve the material from its location, a process that is not only hidden from the user, but far more convenient than physically venturing into library stacks to retrieve hardcopy referenced in a plain footnote. Thus, as Clifford Lynch has observed, the TotalNews family of cases may be ultimately conceptualised as a set of disputes over whether an author can cite or refer to another author's work without permission. (Lynch 1997). These cases therefore have broad ramifications for scholarly citation, bibliographic indexing, and information cataloguing. And, in an environment where lookup and retrieval of the referenced work is automated, the cases have similarly broad ramifications for information access. In the emerging world of information technology, control of information references may be tantamount to control of information access.
The principal set of intellectual property rights implicated in hypertext media is copyright. Ownership of a copyright confers certain exclusive rights on the owner, including the right to reproduce, distribute, publicly perform, publicly display, or adapt the copyrighted work. United States Code Title XVII, s.106. Claims that hypertext linkage somehow infringes copyright must rest upon some rationale explaining first, how the subject of the link falls within the scope of copyright protection, and second, that hyperlinking implicates some act that infringes one of the copyright owner's exclusive rights. However, where digital media are concerned, it is exceedingly difficult to advance a coherent theory that meets these requirements.
Copyright extends to a wide variety of subject matter found in or associated with hypermedia, including text, graphics, photographs, motion pictures or other audio-visual works, musical compositions, sound recordings, and computer software. United States Code Title XVII, s.102(a). However, the scope of copyright in such digitised works is a matter of some dispute. To qualify for copyright, the material must be an original work of authorship, fixed for more than transitory duration in a tangible medium of expression. These qualifications for copyright protection, which were developed in the world of print media, do not always map well onto the world of digital media, particularly in a networked environment. (Litman 1996)
A pivotal question in the analysis of copyright in digital media is the issue of so-called 'RAM copies' (Hardy 1997; Lemley M 1997). As indicated above, copyright confers on the owner of a work the exclusive right to make and distribute copies. Although the parameters of this right are well established for print media, they are problematic for digital media. In automated information processing and retrieval systems, digitised information is reproduced and stored in computer memory whenever the information is accessed. For example, this storage may occur in RAM, or it may occur in temporary 'cache' files on magnetic media. In the case of networked computer communications, temporary copies may be made on several intermediate machines as packets are routed to their final destination.
Some courts have suggested that copies of a program loaded into RAM may be relevant for copyright purposes. MAI v. Peak Triad Systems Corp. v. Southeastern Express Co.. Because the RAM copies are accessible and endure for more than a transitory duration, these decisions hold that the copy in RAM may be an infringing copy. This in turn implies that intermediate copies, including browser caches, that are created during the operation of network applications, may be infringing copies. In turn, digital transmission of those copies over a network would presumably constitute a distribution of copies. To the extent that such reproduction and distribution is unauthorised, this implies that the normal operation of computers and computer networks results in repeated and widespread copyright infringement (Hardy 1997; Lemley M 1997).
These implications of the 'RAM copy' doctrine were articulated most clearly in a 'White Paper' study promulgated by a United States Department of Commerce Working Group on Intellectual Property (1995). That paper embraced these implications and subsequent bills introduced into Congress by the Clinton administration have attempted to amend the Copyright Act to reflect this position (H.R. 2281, 105th Cong., 1st Sess, (1997). S. 1121, 105th Cong., 1st Sess. (1997)). However, both the 'White Paper' and its attendant legislation have been severely criticised by a wide variety of commentators familiar with both copyright law and its application to digital media (Elkin-Koren 1995; Litman J 1994; Samuelson 1996). Aside from the possibly stifling effect that this doctrine could have on the growth of digital media, these critics point out that some courts have reached a contrary conclusion - that 'RAM copies' of programs are too transitory or ephemeral to be infringing - and that courts holding RAM copies to be infringing appear to directly contradict the language that Congress included in its explanation of the current copyright statute. Congress appears to have contemplated that the creation of transitory, ephemeral, or incidental copies in computers would not be an infringement. H.R. Rep. No. 1476, 94th Cong., 2d Sess. 53 (1976). Under this theory, the normal operation of computers and computers networks should not constitute copyright infringement.
Even if we accept the holding of some courts, that RAM copies are relevant for copyright purposes, at least two theories exist to ameliorate the potential effects of this holding. These theories fall under the respective rubrics of implied license and fair use. Implied licenses are contractual agreements that, although not written or formally negotiated, are inferred from the actions of two parties. Fair use encompasses a body of copyright law which permits certain socially necessary uses of copyrighted works, which would otherwise infringe the exclusive rights of the owner. For example, U.S. courts have held that a copy of computer software made incidentally in the course of reverse engineering uncopyrightable portions of the software is permissible as a fair use. Atari Games Corp. v. Nintendo of America, Inc. Sega Enterprises. Ltd v. Accolade, Inc.
Drawing on a theory of implied license, we might infer that the generation of RAM copies in a computer network is permitted under the terms of an unwritten contract for use of materials on a public network. Or, in the alternative, we might conclude that the creation of such copies is a fair use of the work. Either conclusion stems from the placement of the work in digital form onto a networked system. By placing the material on an open network, the owner must have intended that RAM copies be made and distributed, that digital transmissions occur. We conclude that there is some implied license for those actions to transpire as the material is accessed. Alternatively, under a fair use theory, we infer that the occurrence of such reproduction, distribution, performance, and display is a fair use of the work that occurs in the course of accessing it in the manner that the author apparently intended it to be accessed. If the owner of the work did not intend the work to be copied, distributed, performed, and displayed, he could well have restricted access to the material.
There is, of course, no legal requirement that a copyright holder take steps to secure or sequester material in order to claim protection. The onus of avoiding infringement is primarily on users, not on copyright holders - an infringement is an infringement whether the material was publicly available or password protected. But when the material is publicly available, the reasonable inference regarding the use of that material changes. If access to the material was restricted, then we might logically infer that reproduction, distribution, public performance, and public display of the work incidental to its use in a computer network is permissible only by those to whom access is granted, and under the conditions of access. When access is open to the world, then we logically infer that such incidental use is permissible by the world.
In the broadest sense, this problem of intermediate copying is not new, nor is the conclusion that such copying is permitted under either an implied license or under fair use. 'Copies' of a work are made in silicon memory during the operation of a computer, but we might also say that 'copies' of a work may be made in carbon memory whenever a human being reads or views the book. We have traditionally ignored such carbon-memory copies as permissible - rather severe First Amendment problems of free speech and thought control would arise if we did not believe such 'copying' to be permissible. It similarly follows that the RAM copies made in the course of browsing electronic materials are no different: the technology is such that one simply cannot avoid making such copies in the course of browsing online materials. Hence, we conclude again that the generation of these copies must be permissible under the copyright law - any other conclusion would effectively create a new exclusive right, which one commentator has dubbed 'the exclusive right to read' (Litman J 1994). The copyright statute lists no such broad and exclusive right for copyright holders in either the world of print or the world of bits.
Setting aside the problem of RAM copies, access to networked resources can also be considered under a separate copyright theory, which likewise highlights the difficulty of fitting digital media to older statutory models. One such example is found in the assertion of copyright liability from the British Shetland Times incident. The judge in that action classified the Shetland Times page as a 'cable programme' under British copyright law. This holding arose from a reading of the 1988 Copyright Designs and Patents Act, which defines a cable program service as one which 'consists wholly or mainly in sending visual images, sounds, or other information by means of a telecommunications system, other than by wireless telegraphy, for reception at two or more places . . . or for presentation of the subject.' Although the World Wide Web was likely not contemplated by the legislators who enacted this definition, the language is broad enough to encompass a web page.
This situation is by no means peculiar to the United Kingdom; a similar argument might be asserted under the United States copyright statute, that accessing digital material constitutes a public performance or a public display, which are among the exclusive rights of the copyright owner. Generally, these rights are linked to broadcast of copyrighted material, and it may not be immediately obvious how one would 'publicly perform' something like text by retrieving it over a computer network. But the statutory language is broad enough to encompass such use of digitised works, defining performance as rendering a work directly or by means of any device or process. Similarly, to display a work is to show a copy of it, either directly or by any device or process.
Under the U.S. statute, performances and displays are defined to be 'public' when they are transmitted or otherwise communicated to members of the public by any device or process whereby the public can receive the transmission or communication at the same time and place or at separate times in separate places. United States Code Title XVII, s.101. The statute defines a transmission as communication of a performance by any device or process whereby sounds or images are received at a place beyond the place from which they are sent. United States Code Title XVII, s.101. This language would appear to accurately describe the communication of digitised materials over computer networks, even if members of the public receive and view the materials separately on their individual home computers or workstations.
Thus, accessing materials via computer network appears to comprise acts that fall within the exclusive public performance and display rights of the copyright owner. However, even if such access occurs without the explicit permission of the copyright owner, such access does not necessarily constitute infringement. As with the creation of RAM copies, we may infer from the placement of the materials on an open network that the owner must have known, and certainly intended, that the materials would be publicly performed and displayed. We may therefore again infer that such performances and displays are part of an implied license, or are a fair use of the work.
The discussion of digital reproduction, distribution, performance, and display underscores an important consideration in delineating copyright for digital materials. Unlike previous media, digital media may simultaneously implicate all the exclusive rights of the copyright owner (Lemley 1997). Print materials, for example, may be reproduced and distributed by means of photocopiers -- but this activity is unlikely to raise questions about public performance and display. Unauthorised broadcast of television or radio programming may raise questions regarding public performance and display, but probably not regarding reproduction and distribution.
In a networked environment, however, copies of copyrighted material are routinely made and distributed as a necessary part of the network's operation. The materials are simultaneously publicly performed and publicly displayed. In some cases, they may also arguably be adapted. Consequently, when analysing possible theories of copyright infringement from hypertext linking, all the exclusive rights of the author must be considered.
Bearing in mind the particular problems of RAM copying and of overlapping rights, we can begin to assess the claim that hypertext links somehow violate the copyright interests of the owner of the linked materials. The basis for any such claim must be some unauthorised act that directly intrudes on the exclusive rights granted under copyright.
One possible theory of direct copyright infringement might be to assert that the reference used in the hypertext link is itself the subject of copyright, and so the creator of a link who uses the reference without permission is engaged in infringement. This theory is familiar to legal practitioners as it has been advanced by a major purveyor of legal content, West Publishing company. West publishes hardcopy and electronic compilations of judicial opinions, and many courts require lawyers to cite legal precedent as published in the readily available West format. While the opinions themselves are not original with West, and lie in the public domain, West has argued that cross-referencing of its materials by competing publishers constitutes copyright infringement (West Publishing Company v. Mead Data Central) This assertion is based on the premise that copyright protects original selection and arrangement of compiled materials, even if the materials themselves are unprotected and unprotectable. West argues that citations to their volumes embody the selection and arrangement of materials in those publications.
Although West initially won some courts over to this theory, the basis for their argument was largely undermined by the Supreme Court opinion in Feist v. Rural Telephone & Telegraph, holding telephone book white pages to be unprotectable as a matter of copyright law. Feist Publications, Inc. v. Rural Telephone Service Co.  499 U.S. 340. West has subsequently lost challenges to open use of citation to its materials, and the theory of copyright in citations appears to be largely defunct under current law (Matthew Bender & Co. v. West Publishing Co.; Matthew Bender & Co. v. West Publishing Co.).
A similar result would be expected with regard to hypertext linkages such as those in the TotalNews family of cases. The URL is essentially a location or identification number referencing a file somewhere on the Internet. Considered as a location, it seems unlikely that the URL can be subject to copyright protection by anyone, let alone the author or owner of the material referenced by the URL. As indicated above, ideas, facts, and short phrases cannot be copyrighted (United States Code Title XVII, s.102(b)). Because it is in essence an address, the reference is not the proper subject of copyright: it is simply an indicator of location, which is to say, a fact. If the reference is not copyrightable, then inclusion of it in a document cannot be infringement.
In many instances, the URL or other reference is not displayed, but the user sees rather some highlighted trigger. The trigger might be an alternative candidate for copyright; this was in essence the preliminary finding of the court in the Shetland Times case, where use of the Shetland Times' headlines as 'hot links' by the Shetland News was believed to be an infringement. However, this result is probably wrong under U.K. law, and is certainly wrong under U.S. law. The highlighted material seen by the user will generally be a poor candidate for copyright protection, as short words or phrases generally cannot be copyrighted: they either lack the originality necessary for copyright, or are 'merged' with the idea they express (Morrissey v. Proctor & Gamble Co.). If the visual trigger for the link is a button or other graphic design, it may in some instances be the proper subject for copyright, but probably not a copyright held by the owner of the referenced material -- rather, one held by the designer of the graphic.
If use of the URL or reference is not itself a source of infringement, then might its employment in retrieving the referenced document constitute infringement? This appears to be the theory relied upon in the TotalNews complaint: that the supplier of the hypertext link is supplying the referenced material without authorisation of the material's owner. However, at an operational level, it is extremely difficult to see how the hyperlink offeror might have directly infringed the exclusive rights of the owner of the material. Even if we accept the theory that RAM copies are relevant for purposes of copyright, no copy is ever made on the hyperlink site server. Rather, the user's computer takes the URL reference from the site, and uses that information to request copies from the site where the copyrighted material is resident. Copies might be made as the material is transferred from its resident site to the user's computer. But these copies neither originate from, nor pass through the site where the hyperlink is found. The seamless transition from the initial site to the site referenced is merely an illusion of the interface as seen by the user.
This analysis holds whether the link involved is manually activated - by pointing to and clicking on highlighted text or buttons in the interface - or whether the link is automatic, as is the usual case for the IMG code. Although in once instance the referenced material is called up at the user's discretion, and in the other it is called up automatically, in neither case does the creator of the link copy, distribute, perform, or display the referenced material. The same is true whether the referenced material is displayed 'inline' or not. The analysis similarly holds whether considered as a matter of reproduction and distribution of copies, or whether as a matter of public performance and display. Any copies made or transmissions performed originate from the server where the material is housed.
A alternative theory of copyright infringement might be to regard the act of accessing material via hyperlink as the creation of a derivative work. The right to adapt a work, that is, to create derivative works based on the copyrighted works, is one of the exclusive rights of a copyright holder. Even if we postulate an implied right to access material openly posted on a network, that implied right might not extend to the adaptation of the accessed material, or incorporation of the material into new works. This theory may seem particularly appropriate in instances, such as that involving TotalNews, where the accessed material is displayed within a 'frame' supplied by the site where the hyperlink is offered. The practice of 'inlining' might similarly appear to offer an example of adapting the referenced work into a new work without authorisation.
Such a theory is one way to conceptualise the complaint by CNN and other plaintiffs in the TotalNews dispute that the advertising on their sites was obscured by the TotalNews frame. We have discussed above why the complaint cannot credibly be grounded in the right of public display, but perhaps the display of the plaintiffs' material in an unforeseen manner - framed by TotalNews advertisements - could be regarded as an unauthorised adaptation of their web pages. This argument, if accepted, would have highly disturbing implications for the routine use of hypermedia, and indeed, for computer displays generally. Consider most off the shelf applications software: web browsers and other software applications routinely 'frame' displayed content in order to make menus, tool bars, and other locally generated aspects of interfaces readily available. Browsers and other 'frames' may even obscure or detract from advertising in displayed web pages. Under this theory, Microsoft, Netscape, and many other software publishers would be routinely liable in copyright for altering the presentation of displayed content. The practical absurdity of this result is underscored by the fact that the precise contours of a 'frame' are largely a function of monitor size and resolution, and can generally be altered by the recipient.
Outside the world of computer interfaces, this theory of infringement would similarly imply that someone who purchases a print, poster, or painting, would be liable to the artist for matting and framing the art (particularly if the frame were one of which the artist disapproved). This rather astonishing result has in fact been reached by at least one court, in a situation where a book of colour illustrations was lawfully purchased and the illustrations were removed from the book, matted, framed, and resold (Greenwich Workshop, Inc. v. Timber Creations, Inc.). More sensible courts have rejected this result, holding that framing of a lawfully acquired copy is not an adaptation of the work (Lee v. A.R.T. Co.). The latter result was reached in part so as not to create a new moral right of integrity that would give an author control over how a lawfully owned copy of a work was displayed. Although such moral rights are recognised in some parts of the world, they are not part of copyright in the United States.
These cases are instructive where computer displays are concerned; it is no more desirably to allow a copyright holder moral rights over a browser frame as over a wooden frame. In neither instance should the frame be considered an adaptation of the copyrighted work, because the frames in question are not permanent additions to the copyrighted work. Just as the frame and matte of a printed picture is fully separable from the framed material, so too is material that is 'inlined' or otherwise framed on HTML displays. The code of the displayed file is not altered and the digital object represented is fully separable from the other objects displayed, including 'frames'. Courts considering claims of adaptation for computer software have suggested that permanence and inseparability are necessary for an adaptation to be found. For example, in a case where 'add-on' software temporarily changed the output of a computer game cartridge, this was not considered to be an unauthorised adaptation because it did not incorporate the copyrighted material in a permanent form (Lewis Galoob Toys, Inc. v. Nintendo of America, Inc.). Similarly, the temporary display change occasioned by framing of linked material should not constitute an adaptation.
The preceding discussion indicates that the creator of a hypertext link should not be held liable for direct copyright infringement. However, not all copyright liability arises from direct infringement. In some instances, liability may attach to one who has not directly infringed any of the exclusive rights in a copyrighted work, but is nonetheless held responsible for the infringing acts of a third party. Even if the creator of a hypertext link does not directly infringe the copyright in the referenced work, an alternative theory of liability might be built upon the notion that the creator of a hypertext link is contributorily or vicariously liable for the infringement of those who employ the link.
The contributory theory of liability almost immediately faces a debilitating objection. The fundamental requirement for contributory or vicarious copyright liability is that there be some underlying direct infringement - by definition, one can only contribute to an infringement if there has been an infringement. Yet no direct infringement appears to occur when a hypertext link is employed. The discussion of direct infringement above indicates that there is no direct liability on the part of the link creator, and neither is there any infringement on the part of the link user.
The absence of a direct infringement by the link user stems directly from the analysis of RAM copies discussed above. We infer that the intervening RAM copies are either authorised copies, or a fair use of the material, and no direct infringement occurred. As suggested above, providing a hypertext reference is largely the equivalent of providing a citation in a footnote or bibliography. Providing a hyperlink should be no more objectionable than supplying someone with the call number of a book found on a physical library shelf. If the individual receiving the call number goes to the shelf, retrieves the book and reads its pages, no unauthorised copy has been made. There is no direct infringement, so there can be no contributory infringement.
Neither would the supplier of a reference or library call number be considered a contributory infringer if the recipient retrieves a book and makes unauthorised photocopies of the book's materials. This common sense conclusion is supported by the formal legal definition of contributory infringement. Contributory infringement can arise in three differing situations (Cavazos and Chao 1995). The first of these situations, called contributory liability, occurs when a party engages in conduct that actively supports or furthers an infringement, for example by financing, advertising, or managing infringing activity. The participating party must know or reasonably expect that his activity will result in infringement. But the supplier of a reference, even an automated reference, generally will not know and need not expect that supplying the reference will lead to infringing activity. Even if another's use of the reference eventually does lead to infringing activity, supplying the reference does not constitute active, knowing participation in the infringement.
In other cases, contributory infringement may occur when a party supplies the means to infringe, knowing that the materials supplied will be used to infringe copyright. This is potentially a very broad form of liability, as there are many types of materials - paper, paintbrushes, software, photocopy machines, tape recorders - that might be used to infringe a copyrighted work. However, the United States Supreme Court has limited this type of liability to situations where the materials supplied have no substantial non-infringing use (Sony Corp. of America v. Universal City Studios, Inc.). In other words, if the materials supplied could be used for any legitimate purpose other than copyright infringement, then the supplier will not be held liable.
A different theory of indirect infringement, called vicarious liability, may be applied when a party has authority or control over another who commits direct infringement. Such authority or control would typically occur in the context of employment or professional relationships. Vicarious liability arises out of an agency relationship; that is, a relationship where the infringer is an agent of a party who is responsible for the agent's actions. In such a situation, responsibility for the agent's actions may be attributed to the employer or supervisor.
None of these situations routinely occurs when a footnote or scholarly reference is supplied, even if the reference is used as a first step to infringement -- in general, the reference has legitimate non-infringing uses, the supplier of the reference is not a participant in the infringement, and has no agency relationship with the recipient of the reference. The same is true with respect to hypertext linkage. Certainly we have demonstrated above that no unauthorised reproduction, distribution, adaptation, public performance or public display occurs in the mere act of retrieval -- under either fair use or implied license, the link has substantial non-infringing uses. It may be that the user of the link will eventually engage in other infringing acts regarding the retrieved material, but the supplier of the link is not a participant in those acts, and will likely not stand in an agency relationship with the infringer. As a consequence, contributory infringement for hypertext linking is no more viable a theory of liability than direct infringement.
Although copyright appears to offer little ground for exclusion of unwanted hypertext linkages, other legal theories might also be relied upon, and require separate analysis. Plaintiffs in both the TotalNews and Ticketmaster cases relied upon multiple intellectual property theories, including claims of trademark infringement, which may be present entirely apart from copyright infringement. We therefore turn briefly to consideration of this alternative theory.
Trademarks may be constituted of any word, device, logo, design, or graphic that is associated with a particular business (McCarthy 1996). Unlike copyright protection, trademark law does not confers a true property right on the trademark owner. Rather, trademark law exists first to protect the goodwill and reputation of a commercial enterprise, and second, to protect the public from being confused as to the source of goods or services. Thus, to be a valid trademark, an insignia must identify the source of a good or service, not merely the type of good or service. In other words, the public must identify the mark with the reputation of a particular business. When this occurs, the mark is said to have 'secondary meaning.' (McCarthy 1996).
Trademark law might be implicated in hypertext linking if a word, logo, or other insignia which is associated by the public with a particular business is used as the graphic 'hot link' to that businesses online materials. For example, in the TotalNews situation, a button or highlighted link marked 'CNN' or 'Washington Post' used to access those companies sites, might constitute unauthorised use of those companies' marks. Additionally, it is by now well understood that domain names and other resource locators can be trademarks if they serve to identify the source of an item, and not simply its logical address. <http://www.urich.edu/~jolt/v1i1/burk.html> (1995). Under the proper circumstances, even URL strings or IP addresses can be trademarks. Hypertext links that incorporate such identifiers may therefore be making use of a trademark. Even hidden locators, such as 'meta-tags', can comprise trademarks, although they are not seen by hypertext users. In one recent case, Playboy enjoined the use of its name as a hidden search term, even though the name was not visible at the level of the user interface (Playboy Enterprises Inc. v. Calvin Designer Label). As a consequence, even though the visible link may not have secondary meaning, in some cases the underlying reference may, implicating trademark law.
The usual test for trademark infringement is the likelihood of consumer confusion: will a consumer be misled into believing that goods or services bearing the infringing mark originate with the trademark owner when in fact they do not? (McCarthy 1996). In the context of hypertext, display of a businesses mark in order to signal a link to its home page might also incorrectly signal a business relationship with the entity whose page displays the mark. Alternatively, the seamless transition between display of linked materials may be confusing to users who have difficulty keeping track of their logical 'location' within a web of hypertext linkages. Unsophisticated hypertext users may believe that linked materials all originate from the same source, or may be unaware that use of a hypertext link can call materials up from a different server. This problem can be exacerbated where 'framing' displays are used, as in the TotalNews case, because the URL displayed by a web browser may be that of the initial linking page, not the URL of the displayed material.
However, the requirements for trademark status are likely to limit the applicability of this theory as applied to hypertext. Unlike the TotalNews situation, the majority of 'hot links' do not necessarily include the name of a well-known business entity such as CNN or the Washington Post. A highlighted word, a graphic button or coloured URL string in a garden-variety hypertext document would appear not to embody business reputation or 'secondary meaning'. In those rare instances where a 'hot link' may carry secondary meaning, the linking page can be designed to make clear that there is no business relationship between the linked and linking pages. Similarly, to the extent that the automated retrieval of linked material may confuse the user as to where one document stops and another starts, the linking document can be designed to make clear that material from a separate entity is being retrieved.
Ironically, one of the best ways to make clear that linked material stems from a different source is to clearly label the link with the mark of the entity whose material will be retrieved. Such usage of another's mark should be permissible under established principles of trademark law. The law has long recognised a trademark usage exception for 'truthful labelling' of another's products or services. This common-sense rule prevents trademark infringement from bringing commercial discourse to a grinding halt. For example, an automotive shop specialising in the repair of Volkswagen vehicles can use the Volkswagen name to identify its specialty, so long as it does not mislead consumers into believing that it is affiliated with the Volkswagen manufacturer (Volskwagenwerk Aktiengesellschaft v. Church). Purveyors of brand-name merchandise can truthfully represent the name of the merchandise which they in fact sell. It would seem only logical to extend this rule to the labelling of hyperlinks; a link labelled CNN, which in fact refers a user to authentic CNN material, can hardly be said to be mislabelled or misleading. It is difficult to imagine what other label the link might bear.
Consequently, it would appear that in those relatively few instances where a trademark is used as a 'hot line', the use can be accommodated as a proper collateral use of the mark, especially if care is taken in the design of the hypertext document, so as to avoid commercial association with the owner of the mark. To the extent that a trademark claim may be proper in the context of a commercial situation such as the TotalNews or Ticketmaster cases, this need not extend broadly to routine creation of hypertext documents.
Closely related to trademark infringement is all the U.S. cases to date have included unfair competition claims, couched in term of the federal Lanham Act. Section 43 (a) of the Act prohibits false designations of origin or misleading representations that would tend to deceive consumers as to affiliations, connections, or associations between a the providers of goods or services and another (United States Code Title XV, s. 1125(a)). The gravamen of claims based on section 43(a) would be that the presence of the link between two sites would cause consumers to infer some authorised association or license between the linking site and the linked site. In the case of 'framing,' such consumer associations may be a particular problem, as the frame often displays the URL or IP address of the linking site, rather than that of the linked site.
The appearance of 'false designation' generally stems from the lack of clear designation, which in a hypertext environment appears to arise solely from the automated nature of the link. If linkage were considered in the context of traditional scholarly footnoting, a claim that reference to other materials confuses readers as to the origins of the referenced material seems absurd. Indeed, the traditional purpose of such references is to make clear the origins of the material drawn upon in the preparation of the referencing document. To be sure, in non-hypertext referencing, the time required for manual location and retrieval of the referenced material, as well as the physical space between the referencing material and the referenced material, makes clear that they are separate documents – in a hypertext environment, such temporal and physical spaces between the documents collapse. Additionally, in non-hypertext referencing, the information displayed in the reference itself differentiates referencing and referenced material, because complete reference and location information are necessary to enable the reader to locate the referenced materials. In a hypertext environment, however, the reference may be embedded behind a single word, a number or an icon, which may not convey the distinct origins of the referenced work.
Yet this lack of traditional origin indicators in hypertext references can be potentially confusing only if the reader or consumer expects physical and temporal space between the materials – to the extent that the reader is accustomed to a hypertext environment, he will instead expect that links may reference documents of different origins. Even in the case of framing, false designation of origin claims are far less plausible than they might initially appear. This becomes clear when considering the analogous off-line purchased a frame of a print at a framing shop. The frame may well bear the name of the framing shop or the designation 'made in China', but there is no reason to conclude that consumers will believe that the framed print was designed or printed at the framing shop, or that it originated in China. The parallel claim of confusion for online 'framing' references seems plausible only because consumers – and courts – are less familiar with web format, and may mentally 'bundle' the frame and the framed page as to origin. However, sophisticated web users undoubtedly already will know that online frames and framed material are not 'bundled,' and as the level of familiarity with the technology increases, the average web user may be assumed to recognise the potential for different origins of screen display elements just as she now recognises the potential for different origins of combined physical goods. In the interim, the potential for consumer confusion may be mitigated by notices or disclaimers indicating which links are internal, and which links have enabled transition between unrelated web sites.
This paper has analysed proprietary rights in the context of current hypermedia protocols that utilise URLs as references to locate linked materials. New protocols have been and will be developed, most likely within the system of Universal Resource Name (URN) 'tags' and 'handles' that will travel with a given chunk of information (Lynch 1997). Although the evolution of such protocols may change specific details of the legal analysis outlined here, the end result should be the same, in large part because copyright and trademark laws have been crafted with the policy of promoting public access to information. Control of information identifiers through proprietary legal rights would ultimately mean control of information in a digital networked environment. The Shetland Times, TotalNews and Ticketmaster legal disputes represent the first attempts at legal control over information tagging. However, as the analysis here indicates, current doctrines of copyright and trademark protection do not lend themselves to a proprietary interest in information tagging in the context of hypermedia.
The question of control over information referencing is in essence a version of the well known 'standards' problem in interoperability: interoperability is best achieved by adoption of a common standard, however, if that standard is proprietary, the owner of the standard may be able to control the development of an entire market sector (Farrell 1989; Friedman 1994). Ownership of patent, copyright, or similar exclusive rights in an industry standard pose particular problems, as they may facilitate undesirable acquisition of monopoly market power (Lee v. A.R.T. Co.). Thus, intellectual property law should optimally be interpreted so as to forestall future monopolisation of information tagging systems. Current attempts to develop sui generis forms of database protection should also be viewed with great caution; initiatives such as the Collections of Information Antipiracy Act, H.R. 2625, 105th Cong., 2d Sess. (1997), now being considered in the United States House of Representatives could unintentionally confer monopoly control over information referencing on networked systems.
United States Department of Commerce Information Infrastructure Task Force. (1995) 'Intellectual Property and The National Information Infrastructure: The Report of the Working Group on Intellectual Property Rights'