Introduction
In January 2005 the MPEG Licensing Administration (MPEG-LA) announced the terms of a joint patent portfolio license to be offered to implementers of the Open Mobile Alliance (OMA) DRM 1.0 specification. A royalty payment of $1 is due for every device that is issued using the OMA DRM specification and a further 1% of any transaction in which an end user pays for delivery of a digital asset (cf. MPEG-LA, 2005). From this patent portfolio, we are reviewing the patents that are related to rights expression languages (RELs), e.g. the European Patent EP 0 715 244 B8 respectively the US Patent 5,715,403 "System for controlling the distribution and use of digital works … utilizing a usage rights grammar" granted to the Xerox Corporation. Today the patent is controlled by the US Company ContentGuard which is owned by Microsoft, TimeWarner, and Thomson. These patents are obviously important in the MPEG-LA patent claim case, as the CEO of ContentGuard recently stated:

"The OMA didn’t choose to use our technology for implementing its Digital Rights Language for OMA 1.0, and instead chose to use a system developed by IPR systems in Australia. We told them that this wouldn’t mean that they could escape our patent portfolio and we’ve been telling them that all along" (cf. Faultine, 2005b).

The "system developed by IPR systems in Australia" identifies the Open Digital Rights Language (ODRL). For many years, the authors of this article developed with many collaborators the Open Digital Rights Language (ODRL). Version 1.1 of ODRL (part of the OMA 1.0 and OMA 2.0 specifications) has been implemented; license tools for ODRL have been created, and devices embedded with ODRL. With the experiences of this and other work, the ODRL Initiative members have improved and extended the ODRL data model and are creating profiles to integrate ODRL and adjacent metadata standards. The ODRL Initiative is about to publish these new research findings in ODRL Version 2.0. The new version meets all gathered requirements from the last years of experiences, and it will make the usage, implementation and processing of rights expression languages more efficient and less ambiguous and takes a step towards more interoperability between the different existing RELs. This ODRL research has mostly been supported by European and Australian research funding and was intended to be made freely available (under open licenses) for other researchers worldwide.

The MPEG LA patent claims state that every service provider or device seller that implements the open source rights expression language ODRL as per the OMA DRM specifications shall be obliged to pay a levy to the MPEG-LA consortium members. As ContentGuard claims to hold patents on any REL, not just only their own solution XrML (see also Böhle 2005 and Berlecon Research 2005), it is unclear on the extent of the patent claims to any version of ODRL (now or in the future) and other machine-based languages, such as the open and free Creative Commons licenses. At present, the MPEG LA patent claims seem targeted at OMA DRM implementations only and do not seem to apply to other (non-standard) DRM implementations on mobile devices and services.

The business model of open source is to freely distribute software and technical specifications and earn money with consulting and other services. If the implementation of ODRL or the simple usage of ODRL tools leads to potential royalty payments, the attractiveness of ODRL will shrink and the further work of the ODRL Initiative is seriously jeopardised. RELs like ODRL are gaining importance in University and European research projects for the creation of platforms to distribute learning material (script, slides, and examples) (cf. EducaNext 2005).

Rights expression languages are only a small building block of the Digital Rights Management Technology but all other components of DRM systems are most likely affected by software patents in the same way. This article will focus on the specific case of above named patents on rights grammars to illustrate problems and confusions arising from software patents.

What is a rights grammar?
As ContentGuard claims to hold a patent on any rights grammar, at this point one has to pose the question: What exactly is a rights grammar? ContentGuard always uses the term "rights grammar" synonymously with "rights expression language". Linguists would probably not agree here in the first place, however, a rights expression language is a language to express usage or access rights for parties over assets. A simple rights expression in ODRL granting the user sguth the right display for the asset proceedings2005 looks as follows:

<rights>
 <agreement>
  <party>
     <context>
       <uid>sguth</uid>
     </context>
  </party>
  <permission>
      <display/>
  </permission>
  <asset>
    <context>
       <uid>proceedings2005</uid>
    </context>
  </asset>
 </agreement>
</rights>


But also consider these two examples:

(1) If the system administrator of a Linux Server sets the "group" rights of file Z to "read, write, execute", he is using some kind of rights grammar.

(2) If, in a running computer system, four software objects of an access control mechanism, e.g. the party object drmUser, the asset object musicFile, the permission object play, and the constraint object untilTomorrow are related to each other and therewith constitute a rights expression, the system uses some kind of object-oriented rights expression language, too.

Do the REL patents now apply to access control software in general? Numerous copyrighted works that describe access control mechanisms and their implementation, such as discretionary access control (access control lists, capabilities), role based access control, etc. were published long before the filing of the ContentGuard patents. Countless research papers also include formal models with respect to computer system security (cf. Landwehr 1981and Burrows et al. 1991).

Most rights expression languages existing today are technically defined in XML schemas. Not using an XML-based rights expression language simply means that you grant or deny access rights with former programming means, i.e. access control information that is captured in software objects or variables, as shown in the two examples above. The following section will try to clarify the coverage of so-called software patents, particularly with respect to the above mentioned patents on rights grammars.

Investigation of ContentGuard patents with regard to (European) patentable inventions

"European patents are granted for any inventions which are susceptible of industrial application, which are new and which involve an inventive step" (cf. European Patent Convention, 1973).

Please note that the proposed software patent directive (cf. European Commission, 2002) in the EU has not yet passed the EU parliament. The European patent of ContentGuard has been filed under the Convention on the Grant of European Patents of 1973.

In the current European jurisdiction "programs for computers" are not patentable unless they are new and provide a technical contribution or further technical effect to the prior art. Additionally, to be patentable, an invention must have technical character and must be non-obvious.

  • The invention must have overall technical character. This means that the invention must use technical features and solve a technical problem. For example, software running on a computer has technical character.
  • The invention must be a new technical contribution or further technical effect, i.e. a solution of a technical problem, e.g. an improvement of computing efficiency.
  • The invention must be non-obvious, i.e. the invention must be beyond state-of-the-art and non-trivial for an expert in the field.
  • The invention must be new, i.e. the applicant of the patent must be the originator of the invention (i.e. not prior art).
In the following paragraphs, we would like to address the named requirements with respect to the above mentioned patents on rights expression languages.

Technical character
What would a software patent in the EU have to look like that applies to any rights language if the invention must have overall technical character? Such an invention would have to include the underlying data model, the technical specification and implementation of all today’s and tomorrow’s rights expression languages. The current rights expression languages already serve different domains (music industry, publishing industry, education), have different focuses (licenses, tickets, contracts), thus have varying data models resulting in diverse technical specifications, XML schemas, and implementations. The above named patents need to be investigated with regard to this requirement.

Technical contribution and non-obvious invention
Some facts: In 1969 the first mark-up language (GML) was developed by Goldfarb, Mosher, and Lorie followed by the Standard Generalized Markup Language (SGML), becoming ISO standard in 1986 (cf. ISO 1986). The successor of SGML is XML (eXtensible Markup Language) respectively XML schema. Today, mark-up languages are widely used and state of the art. The important inventions and copyrighted work in the field of access control were published in the 1960s and 1970s.

Rights expressions have existed since humankind can talk. The example: "I lend this book to you until tomorrow" is a human readable example of a rights expression language. Is a machine readable example of this rights expression language a technical contribution to the state-of-the-art if neither the means of writing (XML) nor the content (access control expression) is new? Since mark-up languages are state-of-the-art, experts use them to describe all kinds of things: books, individuals, and also rights. As mentioned above, common rights expression languages, including MPEG REL, XrML and ODRL, use XML Schema for their serialization. XML Schema can be regarded as state-of-the-art technology. General, freely available XML parsers can be used to interpret and process MPEG REL, XrML and ODRL rights expressions.

New (prior art)
For a patent to be accepted, it must pass a number of key requirements as outlined above. These include that no "prior art" in this invention currently exists. The process to determine these requirements are "self determined". That is, the patent applicant argues in the proposal that the invention meets these requirements. The patent administrator must make a judgement call based only on this information as they are not experts in the area of the invention.

In the case of the ContentGuard patent (‘403) – filed on November 1994 - a number of prior art inventions where overlooked. In particular, the well-known work of Ted Nelson’s Xanadu project from the 1980s (cf. Samuelson & Glushko, 1991) clearly a decade before the ‘403 patent submission. Nelson’s work is "novel in proposing to use a contract-based scheme for commercial distribution of written texts" and was also novel in "charging for each and every use of their documents" rather then each copy. The Xanadu project used an "intuitive rights-to-do framework", that is, a system that enabled the user rights to be described for content that limited its use. Strangely, the ‘403 patent references only a 1994 work of Ted Nelson but does not discuss it.

The European ESPRIT Project "Copyright in Transmitted Electronic Documents" (cf. The CITED Project) from 1990-1993 developed a model that provided control, policing and remuneration, in respect of the use of copyrighted material stored and transmitted in digital form. The project demonstrated and implemented software with mechanisms such as "The Use Right Collector (URC) that collects and manages the use right data base and links the data with their associated rights".

Henry H. Perritt (1993) wrote about the concept of "permissions headers" in which rights information would be attached to every digital work distributed across networks. He indicated that "this representation problem may benefit from the use of some deontic logic, possibly in the form of a grammar developed for intellectual property permissions."

Summary
From this understanding, we conclude that a patent on any rights expression language is not possible under current European jurisdiction and would be highly doubtful worldwide. Such a patent would only be relevant to the specific implementation that it describes. The technical implementations of DRM systems today differ widely from the ideas over a decade ago. The original ‘403 patent authors at Xerox had a "print industry" view of the world and probably would not have even contemplated that a REL would one day be in a small mobile device managing music delivery. Therefore, open source rights expression languages must be untroubled by the software patent levy.

Examining the ContentGuard patent claims with regard to OMA DRM
MPEG-LA claims that their patent portfolio applies to the OMA DRM standards and a license must be obtained. MPEG-LA has not released the full list of the patents in question and how/where they apply to an OMA DRM implementation. One needs to take it "on faith" that all the patents do apply.

If we look at the details of the ‘403 REL patent as an example, some interesting facts are revealed:

Patent ‘403 claims that "Digital works and their attached usage rights are stored in repositories" and "The enforcement elements of the present invention are embodied in repositories" and defines repository functions to include "… store digital works, control access to digital works, bill for access to digital works, loan digital works or automatically handle the commercial reuse of digital works, and maintain the security and integrity of the system". The model is clearly one of advanced repositories undertaking the major functions of the DRM transactions. In the mobile world, it is unlikely that a DRM client on a mobile phone would fall into the category of a "repository" as defined by patent ‘403. Additionally, the "enforcement elements" of OMA DRM are provided by encrypting the content and storing the keys in separate protected licenses. The OMA DRM model does not use a "repository" to request and allow access to content as this is handled by a client application on the handset.

Patent ‘403 claims that "A key feature of the present invention is that usage rights are permanently attached to the digital work" and that "It is fundamental to the present invention that the usage rights are treated as part of the digital work." This is quite the opposite in OMA DRM implementations. The content and license (usage rights) are separate data files and are never "permanently attached" to the content. Additionally, with superdistribution in OMA DRM, the content is sometimes not associated with any "usage rights" until after purchase.

Patent ‘403 claims that "The usage rights language is based on the grammar described below. A grammar is a convenient means for defining valid sequence of symbols for a language." The grammar of the OMA DRM licenses is based on XML, and more formally on XML Schema, which has its basis on the DTD (Document Type Definition) from SGML developed in the 1980s, and does not resemble the grammar in the ‘403 patent.

These are just a few examples of aspects of the ‘403 patent that need careful analysis for their applicability to current implementations of DRM systems.

Discussion
Is the MPEG LA patent portfolio removing uncertainty?
The MPEG LA is pooling (DRM) patent owners and offers implementers of patent affected (DRM-)technology a patent portfolio for a certain price (such as $USD1 per device and 1% of the digital asset’s cost). The Vice President of MPEG LA states that "a patent portfolio assists in removing the uncertainty surrounding the ‘patent overhang’ " (cf. Horn, 2005), i.e. it is a convenient and efficient way to access the (DRM) technology. On the other hand the patent pooling makes the patent claim non-transparent. In the OMA DRM case it is not clear which patents apply to which parts of the OMA DRM specifications. What consequences would it have if the ‘403 patent claims with respect to rights grammars turn out to be unjustified? To what extent would that reduce the patent levy? What about all the other patents in the portfolio? Who provides an independent analysis of them? From this point of view the pooling of patents probably causes rather than removes uncertainty.

In the latest news, the Mobile Entertainment Forum (MEF) announced that is has issued a statement regarding the licensing program proposed by MPEG LA for Mobile Digital Rights Management (Mobile DRM) saying that: "The terms being considered by MPEG LA … could have a devastating effect on any business involved in mobile and wireless entertainment" and "that MPEG LA's proposed royalty rates are onerous, impractical and unclear" (cf. w/o, 2005).

What are the possible consequences for the customer?
It was the aim of the Open Mobile Alliance to develop an open standard, to minimise any impact of patents, and to eliminate royalty payments (cf. Buhse 2004). Hence, all players in the mobile industry had a high motivation to support and implement this standard. Now, device sellers and service providers have been faced with a new potential royalty payment to MPEG-LA - significant additional costs that have not been calculated in their business models. This may also lead some vendors to continue to implement their proprietary DRM systems as there is little benefit in moving to the "open" standard. In this case the customers would have to bear the consequences of non-interoperable mobile phones.

If the claims of MPEG LA can be imposed (and software patents continue to be granted), the additional costs of the above mentioned patent levy will be transferred to the customer. The latest news about the adoption of software patents in Europe (cf. The Copenhagen Post 2005) shows that software licensing will be an important future business of large software companies.

What are the consequences for the ODRL Initiative and Open Source Developers in general?
If the claims of MPEG LA are validated, the work of the ODRL Initiative and other RELs such as the Creative Commons Licenses will be critically endangered. No open source developer would have a motivation to work on new concepts and implementations for RELs if the royalties for their application are paid to a different organisation. This would mean that open source developers, researchers and universities would have to now constantly monitor and review software patents in future. They will have to apply for patents themselves in order to make research freely available for the public and other researchers. This is costly (maybe not realizable with the restricted budgets at universities) and time consuming (i.e. disables progress). Additionally, the many software patents are an unpleasant surprise for researchers (and others) that started their work in a software patent free environment and are then faced with the fact, that maybe a large part of their work is not sufficiently protected by copyright. This has happened in Standards groups previously and has earned the name "submarine patents’ – that could surface at any time in the future.

Bottom line
From the concrete case above we can summarize that more transparency is needed with respect to the legal basis of software patents and its application to DRM technologies. The MPEG-LA patents have the clear potential to disturb research and development in the field of RELs and other DRM technologies. Furthermore, the MPEG-LA patent claims have a negative effect on the growth of European mobile (entertainment) industry. This article is also an appeal to politicians and patent offices to rethink the proposed software patent directive in the EU.

Sources
Disclaimer
The content of this article was compiled with the greatest possible care but should not be taken as legal advice in patent matters. The authors cannot take any responsibility for its correctness or completeness. Names and terms used in this article may be trademarks.

About the authors: Susanne Guth received her master’s degree in information systems from the University of Essen (Germany) in 2000. She specialized in software techniques and production/operations management. Since August 2000, she has been an assistant professor in the department of information systems at the University of Economics and Business Administration in Vienna, Austria. For the last four and a half years, her research focus has been on contract and rights management for digital goods. She wrote her doctoral thesis in this field. Susanne Guth is a private, active member of the ODRL Initiative, leading the development of ODRL Version 2.0.

Renato Iannella is a researcher at National ICT Australia (NICTA). he has extensive experience in the development and standardization of Internet, Web, and mobile technologies and was a former member of the World Wide Web Consortium (W3C) Advisory Board. He is also visiting associate professor at the University of Hong Kong and was previously the chief scientist at IPR Systems and principal research scientist at the Distributed Systems Technology Centre (DSTC).

Status: first posted 22/03/05; licensed under Creative Commons
URL: http://www.indicare.org/tiki-read_article.php?articleId=90