The Management of Rights in the Digital Environment: Lessons from Legal Artificial Intelligence

 

 

Alan Cunningham
Herchel Smith Research Fellow, Queen Mary Intellectual Property Research Institute, University of London
a.cunningham@qmul.ac.uk

Abstract

An electronic licence constructed with a REL, however, might have difficulty in adequately representing and implementing legal rights and relationships central to copyright law. This difficulty is illustrated by examining how mandatory copyright exceptions under U.K. law for databases and computer programs might be incapable of being adequately represented and implemented, if such representation and implementation is undertaken by rights expression languages.

The problems a REL might have in describing exceptions for databases and computer programs might be representative of a larger theoretical difficulty with the digital management of legal rights and relationships. In particular, the rationale behind, and process of, rights expression encourages comparisons with certain aspects of the developement of legal expert systems within the legal artificial intelligence research domain. Examination of the difficulties inherent in formally representing legal knowledge for computational utilisation may be of use in appreciating the possible problems in using REL-constructed licences to manage on-line content distribution.

Keywords: Rights management systems, rights expression languages, copyright, legal artificial intelligence.

---

This is a refereed article published on: 15 December 2004.

Citation: Cunningham, 'The Management of Rights in the Digital Environment: Lessons from Legal Artificial Intelligence’, 2004 (3) The Journal of Information, Law and Technology (JILT).< http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/2004_3/cunningham/>

---

1. Introduction

1.1 The Digital Management of Rights: A Summary Definition

It is necessary to first have a clear and appropriate perspective of what the process of rights management is trying to achieve, and how it acheives it. Such a viewpoint allows, subsequently, for a proper and objective definition of the practice of rights management, which is important in the overall context of this article and a necessary place to begin.

Discriptive endeavours concerning the management of rights in the digital environment usually provide simple listings of the disparate and separate technologies that can be called upon to create a rights management system. In one respect, this methodology is somewhat appropriate, for a rights management system is not an independently conceived system, but rather it utilises a variety of hardware and software components in order to achieve an appropriate level of control and management for content distribution in a digital environment[1][1].

However, aside from this useful insight, it is my belief that the specific listing and description of the various technologies involved in rights management is irrelevant in trying to understand rights management and its implications. Far more beneficial, from a jurisprudential perspective as well as an organisational perspective, is the establishment of an objective notion of the substance of the practice. In this spirit, it is suggested that a generic rights management system will invariably be based upon two sets of objectives, the acheivement of such being a goal common to most, if not all such systems, irrespective of the disparate technologies involved.

The first set of objectives of a rights management system are the regulatory[2][2]objectives, and they are the fundamental basis of rights management. Everything else follows from the regulatory objectives, and without them the system would not be required. They are twofold: (a) the exclusion of free access to digital copyright works and (b) the control of the use of such works subsequent to access. The second set of objectives are the architectural[3][3]objectives, which are both required due to the regulatory objectives, but also seperately because of the distinct technological environment in which rights management is enacted. They are (a) the acheivement, technologically, of the exclusion of free access to digital copyright works and (b) the acheivement, technologically, of the control of the use of such works subsequent to access.

The two sets of objectives allow the proper formation of a definition of the process of rights management. Conceptually, the process has two aspects, a regulatory aspect, and an architectural aspect. The regulatory aspect (what is to be achieved) is concerned with managing rights, rights of access and rights of use. The architectural aspect (how it is to be achieved) is concerned with ensuring that such management has proper implementation in the distinct electronic, digital and on-line environment. As such, one can state that the practice in question is the creation of technical systems that are used to manage rights of access and use in the digital communications environment. This is a clear and appropriate perspective from which the rest of the paper can continue.

The process of clarifying the objectives, and as a result the definition, of rights management has been useful because this paper wishes to now focus on problems surrounding one aspect of the architectural objectives: the achievement, technologically, of the control of use of digital copyright works subsequent to access.

1.2 Rights Expression

One method of achieving, technologically, the control of the use of digital copyright works subsequent to access is by using rights expression, or rights representation, languages. The origins of the concept of rights expression can be traced to a project undertaken by researchers at Xerox Palo Alto Research Centre (PARC) regarding trusted computing systems. Emerging from concerns regarding computer security, trusted systems required a formal method for the specification and utilisation of conditions under which such systems could render, or display, content. Within the context of the requirements of trusted computing, Dr. Marc Stefik developed the idea of categories of rights relating to digital content[4][4]. Stefik originally derived four basic categories; render rights, transport rights, derivative work rights and utility rights. A render right would grant a user the right to represent or create content in a specific format, such as a printed hard copy or on a computer screen. Transport rights allow the sharing, moving or copying of digital content. Derivative work rights are concerned with allowing the use of digital content to create derivate works, such as the insertion of an original work into a new original work, perhaps as a quotation. Finally, the utility right refers to rights that must exist because of the distinctive features of the technology that is used for the storage, creation, distribution, sale and use of content; for example, the right to create a copy for the purpose of caching or in order to create a back-up. In this respect, and quite apart from the economic necessity[5][5] and benefit of applying rights expression in a digital copyright distribution scenario, the idea of rights relating to digital content can be viewed as more a result of the nature of digital technology and the use of information in computers and across distributed communication networks. For example, information can be viewed as an object within a closed computational system. Certain permissions are attached to the object which allow the user to do certain things with the object (or which allow certain users to do certain things, such as administrators of a network system). This permission/object model, central to the practice of sensibly using information in a computer system, applies with a certain ease when a right holder considers how to allow digital content to be distributed and sold across a computer network. In addition to the basic permission/object model and the rights categories developed by Stefik, Rosenblatt[6][6] has identified the concept of rights attributes, the idea that although particular general rights exist, such as the right to print, specific permissive or restrictive attributes can be associated with particular general rights. For example, the right to print might be restricted by the requirement for the compliance with a certain condition, such as payment. Stefiks general model, allied with the concept of rights attributes, obviously appeals to the pecuniary interests of those right holders who wish to develop the market for digital informational products. The central aspect of representing rights for use in a rights management system environment is the rights expression language (REL) or rights representation language. Any REL is simply a ‘language’ that can be implemented and ‘understood’ in a computational system. One particularly developed example of a REL is eXtensible Rights Mark-Up Language (XrML).

Specifying Rights - eXtensible Rights Mark-Up Language (XrML.)

There is an inherent problem with a rights management system that utilises the concept of rights expression as a method of achieving, technologically, the control of use of digital copyright works subsequent to access. It is that a specification language must formally represent both legal rights and relationships for electronic utilisation. Such rights and relationships, it can be argued, are fuzzy in both meaning and interpretation. Electronically represented rights, however, must be machine-readable, and therefore not fuzzy. Illustration of this problem can be provided by a brief exploration of a specific REL. Although a number of languages are in development the dominant rights specification standard is XrML[7][7]. XrML has linguistic links with another computer language, eXtensible Mark-up Language (XML). Some historical information regarding XML might allow the reader to better understand rights representation and any difficulties that emerge from its use.

However, in order to appreciate XML one must also look at some of its linguistic ancestors, specifically Hyper Text Mark-up Language (HTML) and Standard Generalised Mark-up Language (SGML). SGML was developed as a text based language for use mostly in the publishing industry. It was required to “mark-up” data. For example, a copy editor could mark a piece of copy to be published pointing out errors, changes, anything that needed reviewing. In other words SGML allowed the editor of a text document to edit the document with a self-descriptive language. SGML was extremely versatile and thus very complex, although it served its main purpose well. The next stage in the development of mark-up languages, however, and the beginning of the emergence of XML, was the increasing development of the Internet. Those involved in the development of the nascent Internet were attempting to add more character to the appearance of documents available on-line, which, in the early days of the Internet, were very drab. It was decided that a mark-up language was required for the development of web documents used on the Internet, so that increased functionality regarding the marking-up of appearance was available. SGML, however, was very useful, but too complex; a language was required that was much simpler to use so that anyone placing documents online could do so quickly and easily. The language developed was HTML, a linguistic derivative of SGML. HTML not only allowed the appearance of the text to be improved; it allowed the appearance mark-up data (or appearance metadata) for the web document to be attached to the document no matter which computer it was rendered on. Unfortunately, HTML was only concerned with the marking-up of appearance, and not with the semantic nature of the information itself. As Internet transactions became more complex, and the idea of the semantic web was being mooted, a new mark-up language was required that was able to describe and understand the data on the web, and not just describe how it should appear. This new language was XML, a new subset of the original complex mark-up language SGML. XrML, to close the linguistic ancestral loop, is based on XML and merely specifies the semantic mark-up purpose of XML. XrML wants to describe a certain type of information for use on the web: it wants to mark-up legal rights for use in connection with content[8][8]. XrML is defined generally as a ‘language to specify rights’[9][9]. More specifically, it has been defined as an ‘XML-based usage grammar for specifying rights and conditions to control the access to digital content and services’[10][10]. In less technical language, XrML simply purports to allow a programmer to model the form of something (in this instance, the form of legal rights), without modelling its implementation (without saying how it should be used). Rights owners can then later utilise formal XrML terms to construct licences determining how content is managed. A brief description of XrML language terms is useful in appreciating how formally it models terms for managing distribution of copyright works[11][11].

The core conceptual elements of the language are a) identification of parties allowed to use digital resources, b) identification of rights available to those parties and c), the terms and conditions under which such rights may be exercised. The XrML 2.0 data model consists of four central conceptual terms that add clarity to this very general framework and the relationship between those identities.

The central terms are Principal, Right, Resource and Condition.

• The principal term is the technical identifier for any party to whom rights are granted. Any person who is the principal would be identified using an associated authentication mechanism.

• Right is the technical identifier for a verb that the principal can be granted to exercise in reference to a particular resource. A right relates to an action or collection of actions that may be undertaken in connection with the resource. The XrML core contains both an abstract right element, to contain information about rights, and a set of commonly used specific rights relating to other rights, for example revoke.

• Resource is the technical identifier for the object in question. Using the previous terms we can state that a principal can be granted a right in order to use a resource. The principal element identifies the person to whom the grant is issued. The right element identifies the activity that that can be undertaken. The resource element identifies the object that can be used.

• Finally, Condition is the technical identifier for the terms, conditions, and obligations under which rights can be exercised. For example, a right may be issued to a principal in order to use a resource within a certain timeframe. The timeframe is the condition under which the right is used.

The four terms (Principal, Right, Resource, Condition), are the conceptual core of any data model. However, what conceptually defines and relates the four terms is the grant element. An XrML grant is issued to a principal; the grant is a specific right to do something, the grant specifies a particular resource and the grant is limited by conditions. In addition to the grant element, the conceptual licence construct builds on and incorporates all previous elements, and is, conceptually, the issuance of grants to their issuing parties. In this respect a licence will involve an issuer, a principal, a resource, rights, conditions, and grants.A basic utilisation of specific XrML terms (which differ from the conceptual terms illustrated above) in the management of on-line content distribution and use would be as follows, and it illustrates how the conceptual terms and framework exist in any licensing scenario.

Example 1– An XrML constructed licence to allow anyone to print a given piece of content available online. (My descriptive comments are underlined).

<license> (This is a licence....

<grant> (That provides a grant to.......

  <cx:print /> (print...

<cx:digitalWork> (A certain piece of work....

<cx:locator> (at a given location – see sample url.)

  <nonSecureIndirect URI='http://www.sample.com />

  </cx:locator>

  </cx:digitalWork>

  </grant>

  </license>[12][12]

Using the extensibility of XrML and the core foundational terms as described above, content owners can control the use of a work in a number of restrictive ways for efficient profit maximisation.

Example 2 – An XrML Licence that allows a user to print only on the basis that they have relevant encryption key.

<license> (This is a licence....

<grant> (that provides a grant....

<keyHolder> (to a certain key holder.....

<info> (who has the information regarding confirmation of keys....

<dsig:KeyValue>

<dsig:RSAKeyValue>

 <dsig:Modulus>Fa7wo6NYfmvGqy4ACSWcNmuQfbejSZx7aCibIgkYswUeTCrmS0h27GJrA15SS7TYZzSfaS0xR9lZdUEF0ThO4w==</dsig:Modulus>  <dsig:Exponent>AQABAA==</dsig:Exponent>

  </dsig:RSAKeyValue>

  </dsig:KeyValue>

  </info>

  </keyHolder>

  <cx:print /> (to print ...

<cx:digitalWork> (a piece of content at ....

<cx:locator> (the following location.....

  <nonSecureIndirect URI='http://www.sample.com />

  </cx:locator>

  </cx:digitalWork>

  </grant>

</license>[13][13]

Thus the essential rationale of XrML and REL, as illustrated above, is the distillation of legal rights and relationships. Practical difficulties in using XrML licensing technology, and REL generally, to distribute content on-line can be illustrated by looking at mandatory exceptions in U.K. copyright law.

2. Mandatory Database Exceptions

For the purpose of UK law, a database is defined as ‘a collection of independent works, data or other materials which (a) are arranged in a systematic or methodical way and (b) are individually accessible by electronic or other means.’[14][14]. Databases have the possibility of two modes of legal protection; copyright[15][15] and sui generis[16][16] protection.

2.1 Copyright Protection

Literary copyright exists in a database by virtue of the high test of originality. Once such legal rights have been conferred by copyright, protection relates to the original aspect of the work, namely, the selection or arrangement of the contents. In this respect, the rights in respect of a database prevent the reproduction of a substantial part of the selection or arrangement of the database[17][17]. Copyright in a database will only be infringed by reproduction of a substantial part of the selection or arrangement of the database, implying that reproduction of an insubstantial part will not constitute infringement. The question of substantiality is a matter of degree to be decided under normal legal principles.

2.1.1 Permitted Acts

The law allows for the reproduction of an insubstantial part of the selection or arrangement of a copyright protected database. It also permits fair dealing with a database for the variety of purposes contained in the Copyright, Designs & Patents Act, 1988[18][18]. However, if access and use of a database is constrained by licence or contract, it could easily be arranged that undertaking such permitted acts would constitute breach of contract. As exceptions only relate to statutory rights, the fact that an act does not infringe copyright protection does not mean that it cannot breach some other right or obligation[19][19].

2.1.2 Mandatory Exceptions

There is provision under U.K. law, however, for mandatory exceptions that cannot be overridden by contract. Section 50D (1) of the CDPA 1988 provides that it is not an infringement of copyright in a database to do anything that is necessary for access to and use of the contents of a database, if such acts are undertaken (a) by a lawful user of a database who has the right to do any of the acts restricted by copyright and (b) during the exercise of that right. Importantly, the right of such a person to do this applies regardless of contractual terms or conditions prohibiting such acts; indeed any such terms are void. Given this, an XrML licence that manages access and use of a copyright protected database online cannot prevent a person who has paid for any type of use to do anything necessary for access to and use of the contents during that use. It is unlikely that XrML could accommodate the multitude of potential necessities that are implicit in s. 50D (1). This issue, while of potentially marginal importance in the case of copyright protection of a database, appears of more importance when one considers sui generis protection of a database.

2.2 Sui Generis Protection

Sui generis protection is offered to a database if there has been substantial investment in obtaining, verifying or presenting the contents of a database. Sui generis protection ensures that it is prohibited for anyone but the owner to extract or re-utilise all or a substantial part of the contents of a database. Extraction, in relation to the contents of a database, is defined as the permanent or temporary transfer of such contents to another medium by any means or by any form[20][20]. This definition of extraction implies that the management of the digital use of a database protected by the database right might require a licence of permission to extract, because to view on screen would involve extraction. Re-utilisation is defined as any form of making available to the public any contents of a database[21][21].

2.2.1 Permitted Acts

It is implicitly permissible for a user of a database to extract or re-utilise an insubstantial part of a sui generis protected database (although repeated and systematic extraction/re-utilisation of insubstantial parts may be an infringement). In addition, section 20 of the 1997 Copyright and Rights in Databases Regulations provides that a lawful user (that is any person who, whether under licence to do any of the prohibited acts or otherwise, has a right to use the database) can fair deal with a substantial part of the contents of a database if it is extracted for the purpose of illustration for teaching or research and not for any commercial purpose and the source is indicated. Schedule 1 of the Regulations also list other acts that can be undertaken for the purposes of public administration such as use for parliamentary or judicial proceedings, Royal Commissions and statutory inquiries and public inspection.However, as with general copyright protection and exceptions, if access and use of a sui generis database is constrained by licence or contract, it could easily be arranged that undertaking such permitted acts would constitute breach of express contractual terms.

2.2.2 Mandatory Exceptions

The 1997 Database Regulations have, however, provided mandatory exceptions regarding sui generis protected databases. Section 19 provides that a lawful user (any person who, whether under licence to do any of the prohibited acts or otherwise, has a right to use the database) of a database that has been made available to the public in any manner can extract or reutilise insubstantial parts of the contents of the database for any purpose. This right cannot be prevented by any term or condition in the agreement regarding use. Any such term or agreement purporting to prevent the extraction or re-utilising of insubstantial parts of the database will be considered, by law, void. How would an XrML licence managing distribution and use of a sui generis protected database on-line reflect the exception outlined above? The law states that anyone who pays to utilise such a database to do any of the prohibited acts must be able to, and cannot be prevented from being allowed to, extract or reutilise insubstantial parts of the contents for any purpose. I am not confident of the ability of XrML to adequately represent the full complexity of the terms ‘insubstantial’ or ‘any purpose’. Such practical difficulties with XrML can be further demonstrated by examining mandatory exceptions in U.K. law concerning computer programs.

2.3 Computer Programs

A computer program is considered a literary work under the CDPA 1988[22][22]. Thus rights granted to an author/owner of a computer program allows the prevention of others from undertaking the same acts prohibited for other works by copyright. Similarly, the standard exceptions to the right for a computer program are the same as for any other protected literary work[23][23]. In addition, the proper exercise of an exception with regard to a computer program, while not an infringement of copyright protection, can be a breach of some other legal obligation such as a private law term restricting any acts permitted by statute[24][24].

2.3.1 Mandatory Exceptions

The CDPA 1988 provides for certain uses of a protected computer program that are allowed, irrespective of restrictive terms. Section 50A ensures that a lawful user of a computer program can make any back-up copy of the program which is necessary for him to have for the purposes of his lawful use. Back-up copies of computer programs are made usually for safety, and for a professional individual would constitute a necessary act. Since potentially, an owner of copyright in a computer program could restrict the exercise of this necessary act, the legislation provides that the right to back-up cannot be prohibited or restricted by any private law term or agreement. Admittedly, in the context of an XrML licence managing access and use for a computer program available on-line, it may be relatively easy for such a right to be incorporated as an XrML specification. Allowing one back-up copy to be made is generally an administrative act whether on-line or off. However, the possibility that an undefined number of back-up copies could be necessary for the purpose of a lawful purpose would conflict with such ease of formal representation.

The exceptions regarding the right of decompilation are rather more worrying. Section 50B of the CDPA 1988 provides that a lawful user of a computer program expressed in a low level language can (a) convert the computer program into a version expressed in a higher-level language, or (b) incidentally in the course of so converting the program, to copy, or decompile it[25][25]. The right can only be exercised if (a) such decompilation is necessary to obtain information necessary to create an independent program with can be operated with the program decompiled or with another program, and (b) the information is not used for any purpose other than the permitted objective[26][26]. Further reflected in statute is the fact that a lawful user may not exercise the right if (a) the information is readily available necessary to achieve the permitted objective and (b) the decompilation is not confined to such acts as are necessary to achieve the objective of interoperability[27][27].

If one briefly examines the exception in the light of a computer program available for use online and controlled by an XrML licence, major difficulties arise. First, if it were attempted to incorporate the condition that any allowed decompilation must be “necessary” to create an interoperable independent program, how would necessity be considered as a formal XrML term, or within the context of formal XrML terms? Another condition to be met for lawful decompilation is that information obtained in decompilation is not used for any purpose other than the necessary objective of creating an independent interoperable computer program. How is the management of this right to be undertaken with a rights specification language when a particular use may appear unnecessary (or be categorised as unnecessary by a REL) but is, in context, actually quite reasonably necessary? In exercise of the mandatory right such actions must be undertaken by the lawful user in light of consideration of actual context and not formalised as a term of specification in an XrML licence.

Finally, regarding computer programs, U.K. Regulations laid down in 2003[28][28] have transposed to the U.K. statute book a mandatory exception from the EU Computer Program Directive[29][29] regarding the observing, studying and testing of computer programs. The exception provides that a person having a right to use a copy of a computer program (a lawful user) is entitled, with out authorisation, to observe, study or test the functioning of the program in order to determine the ideas and principles which underlie any element of the program if he does so while performing any of the acts of loading, displaying, running, transmitting or storing the program which he is entitled to do. The difficulty that rights expression has in accomodating the complexities of legal language is also apparent here. For example, the lawful user might decide to test the functioning of his/her legally purchased program in order to determine the ideas and principles that underlie a particular element of the program, whilst performing an act that he or she is entitled to do. Not being a computer scientist, I am unsure what the term “test” encompasses, but I regard the intent of the term as being, as a term of investigative science, semantically wide. The full scope of the term “test” is always fully incorporated in a non-digital licence, specifically because it is a linguistic term, open to general interpretation, and contextual definition. In an XrML licence, however, a closed representation is required which by implication restricts the full operation of the law. For example, a scenario might occur where the lawful user has to undertake an action that he considers appropriate in order to comply with his interpretation of “test”, whilst, for example, loading the program. If the lawful user has a physical product protected by a conceptual non-electronic licence, the act can be undertaken, and as always, it is up to the right holder and the Courts to consider the issue of ex post infringement. If the lawful user is dealing with a digital product controlled by a permissive electronic XrML licence, this new consideration of what is required for “testing” of the software may not be accommodated (and thus not allowed because the legal terms are active codal controls). In addition, the result of other terms might be the restriction of the exercise of activities that allow this new form of “testing”.

In all of the scenarios illustrated above, both for databases and computer programs, the activities of the lawful user are severly restricted due to the formal nature of legal language once it has been distilled for use in a REL. The problematic nature of this is heightened by the fact that that the exceptions utilised in the illustration are mandatory. If a REL constructed licence cannot accommodate and adequately represent the potential offered by a mandatory exception, what chance has a general sense of equity in being reflected in the contractual management of digital copyright works? This problem is a reflection of a greater conceptual difficulty that underlies the idea of rights expression as a useful facet of a rights management system: the utilisation of the idea that the body of knowledge that is copyright law can be easily formalised. Some insight into the nature of this difficulty can be gained from a brief exploration of certain aspects of the body of research that is legal artificial intelligence.

3. Law and Artificial Intelligence

3.1 Expert Systems and Rights Expression: A (Somewhat) Common Rationale?

The rationale behind rights expression shows some controversial similarities with the rationale behind representing legal knowledge for expert system use. For example, XrML is a language that attempts formal semantic representation of legal rights and legal relationship, such as the legal right to print a copy of a work, and the legal relationship between a licensee and licensor. Knowledge representation, in an artificial intelligence context, attempts to formally represent legal knowledge processes. Common to the two is the philosophical and practical thread of formalism. Thus in appreciating the difficulties inherent in knowledge representation for expert systems one can appreciate the shortcomings of REL in attempting something quite similar.

Artificial intelligence, while an ambiguous term, can be said to mean involving the creation of computer systems that adequately undertake and complete tasks usually undertaken by human beings. A pure definition of A.I. would provide the addendum ‘tasks for which thought particular to humans is considered to be required’. An expert system is a particular manifestation of artificial intelligence research. It is a system that performs a specific (expert) human task. Thus expert systems have been developed in a number of separate expert fields, with some degree of success [30][30]. An expert system purports to contain both the knowledge (the bare facts) and the expertise (interpretation of facts) required for undertaking and completing tasks in the specific expert area. For example, an expert system dealing with diagnosis of a particular group of diseases would contain facts relating to the diseases and rules for interpretation of those facts for successful diagnosis. In the early 1980's major research was undertaken in the area of legal expert systems.

3.2 Knowledge Representation, Expert Systems & Law

The creation of an expert system can be categorised into three distinct areas, knowledge acquisition, knowledge representation and knowledge utilisation. I wish to focus in this paper on knowledge representation. Essentially, knowledge representation is an attempt to resolve the following problem: in order that an expert system function it must have information that can be utilised in a computer system. Therefore, the knowledge particular to a certain area must be represented in a form that can be understood by computer systems.

The fundamental problem in representing knowledge for utilisation in a computer system thus revolves around an interesting paradox. First, in order for an expert system to utilise knowledge, the information needs to be adequately represented. However, in order to adequately represent knowledge for use on a computer system, that knowledge must be represented in a certain fashion: a formal fashion. The paradox, the conflict with what one would expect, is that some areas of knowledge lose definition and meaning when formalised. Formal representation for use in expert systems does not necessarily lead to increased efficiency and clarity in understanding and interpreting areas of knowledge; in fact it sometimes results in the opposite. The formal representation is less of a problem in certain areas, more difficult in others. For example, the ‘empirically based causal, descriptive laws of the natural sciences’, such as medicine, geology and chemistry will undoubtedly have less fundamental problems with the transfer to formal representation, because formality is already an inherent part of their make-up[31][31].

What about Law? At first blush, the law may seem suitable for representation on an expert system. After all, the law can be described as ‘ a discipline concerned with the elaboration of the practical art of government through rules’ its concern being ‘prescriptive and technical’[32][32]. However, two factors act to repel this initial consideration. First, knowledge can be separated into factual knowledge, readily formal, and heuristic knowledge, which is informal and discretionary. Heuristic knowledge is developed through experience and assists the expert in the application of mere facts. It is a central aspect of a human legal experts “intelligence system”. If one applies the distinction to a practical area of law the distinction appears more apparent. For example, on picking up a copy of Laddies Modern Law of Copyright[33][33], one will discover the variety of statutory clauses that define exceptions to copyright protection. Also provided will be the rules, or facts, of precedent that have emanated from the Courts. However a true understanding of how the facts of statute and the rules of precedent could apply to a new and original case would only be available from someone who has practised in the field of copyright law and has expert judgment.

The second factor that would dispel any thoughts of law being suitable for formal representation is the social and political context that law operates in. Contextual considerations impart meaning to what the law is and how it develops. It is unlikely that formal computational representation can represent such important factors. The combination of these two factors means that the potential for misrepresentation, distortion and oversimplification of the way in which legal problems are dealt with is very high.

3.3 The Exception That Proves the Rule

One company’s commercial success with ‘expert systems’ can be seen as empirical evidence of the discussion above. SoftLaw Corporation Limited is an Australian company that supply what they describe as ‘software solutions for the administration of complex legislation, policy and procedure’. The ability of Softlaw to undertake these tasks is personified in their major product, STATUTE Expert. STATUTE Expert is an expert system tool that creates rule base models of legislation or policy guidelines, so allowing computational utilisation of the legislation or policy. Established in 1989, Softlaw have had considerable success in their field. Clients include the Australian Departments of Defence, Family and Community Services, and Veterans Affairs, Environment Australia and the South Australian Community Housing Authority.

Examination of some of the projects that SoftLaw has developed applications for provides some interesting illumination regarding their success. For example, the project undertaken for the Department of Family and Community Services is the EDGE Claims Processing Application System, a decision support system for the administration of family based payments under Australian social security law. The project completed for the Department of Veterans Affairs was the Compensation Claims Processing System, a system to improve the administration of compensation entitlements for military service veterans and their dependants. Without prejudice to either of these areas of legal affairs, one could suggest that accessing eligibility for benefits and calculating entitlements is hardly a matter that requires excellent legal knowledge or reasoning skills. Statutory provisions and administrative rules regarding such issues are in the main procedural and clear. In such areas of public administration the ‘distribution of justice’ has to be utilitarian in nature. Knowledge representation is unlikely to be problematic.

It could thus be argued that the nature of the legislative material tackled by Softlaw benefits any legal expert system they care to create. It does not seem disingenuous to suggest that SoftLaw systems are successful because they minimise the elements of risk that would arise from either the standard of heuristic knowledge required in a legal area or a contextual political difficulty. Softlaw have realised that some areas of law are less contentious and more technical than most. In such scenarios, the processing and calculation of a veterans benefits is probably better done by a so-called legal ‘expert system’ because it will free up experienced lawyers to handle cases that require knowledge experience and skill. Thus, STATUTE Expert is an exception that proves the rule and illustrates the paradox: formal representation of law will only work when law is formal itself.

3.4 Rights Expression & the Limitations of Legal Language

The issue that requires addressing is whether the contractual distribution, use and sale of digital copyright works should be (or can adequately be) managed by a technology such as REL, given their requirement for formal representation. A rights owner would respond affirmatively, because the private and formal management of his intellectual production makes as good an attempt as is possible at preventing un-authorized uses. From the perspective of a right owner managing rights is an administrative area of law, and should be so.

However, can the use of copyright works be separated so much from public policy considerations? The argument exists that states that contractual use of copyright works exists in the non-digital distribution environment, and it allows one to ask what is so very different with contractual control in the digital environment? In response, two points come to mind.First, in the analogue distribution environment, there is less of a requirement for complete control. In the analogue environment ‘All Rights Reserved’ is de jure required, but cannot be always be de facto enforced, because of the technologies of distribution. One can argue that this lack of complete control over the use of works has an important subsidisation effect and benefits both economic and social activity. In the distributed communications network, a work protected by, for example, XrML within a rights management system, allows for, and in fact requires, both de jure and de facto reservation of rights. No subsidisation is permitted.

Secondly, contractual arrangements in a non-digital distribution environment are subjected to both external scrutiny and flexible interpretation. Competition law and consumer law regulate private law terms regarding intellectual property. Even within the confines of contractual use of copyright material, policy considerations are imposed as illustrated by mandatory exception for databases and computer programs. All these examples provide support for the principle that while in the main, contractual use of copyright is a matter of private law, there is always the potential for contractual overridability. Technological implementation of contract minimises such potential.

4. Conclusion

The lesson, or at least one of the lessons that appears to have been learnt by those who wish to develop practical and commercially successful ‘expert systems’, is that in representing knowledge for utilisation in a closed computational system, the topic should ideally be administrative in nature and have an uncontroversial context. This requirement allows the technical formalisation of legal concepts and relationships to succeed because the topic demonstrates formal qualities naturally.

Those who want to use and those who develop REL constructed licences in rights management systems must expect and abide by the formalism inherent in such a representation process. However it appears that they do not realise, or are indifferent to, the important fact that the topic for representation will be changed by this aspect of the process.

Arguably, the law relating to how protected ‘expressions of information’ are distributed and used is too controversial and political for such representation to be equitable, effective or justifiable. Admittedly, private law regulation of such works in the analogue environment attempts to achieve a similar effect, and in theory would be expected to do so. However, as illustrated by the difficulties of XrML adequately representing mandatory exceptions for databases and computer programmes, even in private law regulation external considerations are borne upon copyright work use. The inadequacy of XrML in accommodating these theoretical considerations is a major flaw in XrML and REL generally. The flaw is inherent because a REL, as a representational language, is inherent formal. Law, even in its most rigid incarnation, such as contract, belies a contextual fluidity.

For the right owner, distributing content on the communications network is an entirely administrative act, or at least the right owner wants it to be. This is why right owners might favour REL constructed licences as a control mechanism within a rights management environment. However, the inherent formalism in rights expression must be addressed. No longer can the use of all copyright material be considered as a mere property right that can be traded and licensed to the ignorance of other considerations. Over the last 20 years competition and consumer law has made its presence felt in the previously isolated area of copyright law. More recently human rights and public interest considerations are bearing upon the use of copyright. REL constructed licences, and rights management systems, operate within a philosophical model that rejects the complexity of external political consideration. Given this, the management of rights in the digital environment must be thoroughly examined and remedies may be required where such management is, possibly, inequitable.

Notes and References