Certifying Uncertainty: Assessing the Proposed Directive on the Patentability of Computer Implemented Inventions

Luca Egitto
Lawyer
[email protected]

Abstract

The proposal for a Directive on the patentability of computer implemented inventions issued in 2002 by the European Commission was amended by the European Parliament last September 2003. On 18 May 2004 the Council discarded most of the Parliament’s amendments and forwarded a proposal that was more practical and less concerned with potential distortions of a broad patent protection scheme for software.  The specific features of software make it difficult to apply traditional IP rights such as copyright and patents and give rise to judicial controversies and challenging interpretations. This article will analyze the current legal framework for computer programs, the possible outcomes of the Directive, and the rift between the European Parliament and the Council.

Keywords : Copyright, patents, computer invented inventions.

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

Citation : Egitto, Certifying Uncertainty: Assessing the Proposed Directive on the Patentability of Computer Implemented Inventions ’, 2004 (3) The Journal of Information, Law and Technology (JILT).< http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/2004_3/egitto/>

1.  Introduction 

In February 2002 the European Commission introduced a proposal concerning the patentability of computer implemented inventions, otherwise known as software patents.  The main aim of the proposal was to harmonize the interpretation of the European Patent Convention by the national courts, and to promote innovation within the European market.[1]  In September 2003 the Proposal went to the European Parliament for its first reading and numerous amendments were made to ensure unlimited patentability was avoided.  At the end of 2003, the Council delayed examination of the amendments until January, whereupon the Irish Presidency immediately started working towards a straighter, faster, and broader patent protection scheme for computer programs.  Since then, the tension between the Council and Parliament has been mounting.  The precise guidelines drawn up under the Irish Presidency which discarded the vast majority of the amendments made by the Parliament were eventually agreed by the EU ministers of the Competitiveness Council.[2]  Of the original 120 amendments introduced by the Parliament, only 21 remain.[3]  At present the document is being sent to the European Parliament for its second reading, as part of the Co-decision Procedure.  The issue, which has increasingly become a matter of political debate, will be discussed by the ‘new’ Parliament resulting from the elections last June.  The roots of the tension underpinning this debate lie in the different conceptions of the inventiveness underlying software, and in the radically different views programmers, researchers and politicians have as to the appropriate tool to foster innovation.  Before dealing with these political and economic issues, it would be appropriate to analyze the existing scheme of legal protection and nature of computer programs

2. Software as Expression: Copyright or Copywrong?

Intellectual property always acted as guardian for innovative ideas and valuable technological improvements, especially since the amount of investment in research and development became a substantial part of corporate spending.  When computers were mainly mechanical devices whose instructions came from punch cards,[4] the whole area seemed clearly to belong to the realm of technological innovation for which patent protection could be granted.[5]  Soon this certainty disappeared, since the critical duties of the machine were based on the instructions that the program put forward to govern the computer, and such instructions were represented graphically and recorded on a medium to be read.[6]  When WIPO was asked at the beginning of the seventies to conceive a legal framework to protect computer programs, the final outcome was predominantly copyright-based.[7]  Throughout the eighties and early nineties legislatures turned to the copyright regime as the most appropriate intellectual property right for protecting software science – computer programs were regarded as literary works – an original literary expression. [8]  Of late however the boundaries between hardware and software have become less neat and ever changing, and doubts have begun to be raised as to the appropriateness of the copyright system within this area of technological innovation.

There are a number of conceptual and practical problems that affect the appropriateness of copyright as the main protection for computer programs.  The first conceptual problem concerns the nature of ‘literary work’ that all major regional and international regulations confer to computer programs.[9]  All works that constitute subject matter of copyright are normally intelligible and comprehensible to humans in the sense that they can be perceived and appraised by observation.  In the case of computer programs, this would apply to source code (that is human readable), but not to object code (that is the source code compiled into the machine in order to be read and directed to specific applications).  Within the current legal framework, this latter code is equated to the source code as the expression protected by copyright although it is not understandable or appraisable as such.  Also, it is not just the nature of these instructions that gives rise to doubts as to the appropriateness of copyright as a method of protection, it is their function, in that the function of the traditional works protected by copyright is to convey information to humans, irrespective of the reaction, while a computer program carries instructions aimed at causing a specific behaviour into a device.[10]

On a more practical level during the eighties and the nineties, problems concerning the originality of object codes, and the subsistence [11] of copyright in program codes [12] and spreadsheets [13] were discussed by judges on both sides of the Atlantic.[14]  The most problematic issues were the definition as to what constituted actual copying of a computer program, what constituted a substantial part,[15] and the meaning of literal and non-literal copying.[16]  Since in most cases direct copying was impossible to prove, the courts developed the doctrine of ‘look and feel’[17] to assess the overall functioning of the program,[18] and to understand whether there was actual copying of it.  Drawing upon the ‘look and feel’ doctrine, and the AFC test,[19] it was considered that even without line-by-line [20] copying there could still be misappropriation and infringement. At the same time in Europe, the ability to decompile software, provided in art.6 on the Directive on the legal protection of computer programs,[21] was an opportunity given to allow programs to use the same protocols without affecting the respective underlying rights. 

This liberty was granted because computer programs were assimilated to other copyrighted works that could be consulted without being copied.[22]  Software manufacturers began being doubtful about the effectiveness of copyright as soon as reverse-engineering exceptions and difficulties in proving copying made a stronger safeguard of their products desirable.  Although the ‘look and feel’ doctrine and the AFC test tried to address the issue of copying, the problem of protecting the expression alone and not the idea underlying the program led the right-holders to rethink their IP strategy.  The natural outcome of such evolution was the IP right that more closely protects an inventive idea: patents.

  3. Software Patents: The U.S. and the E.U.

Since one of the reasons purported to justify the introduction of the Directive [23] was that global harmonization of patent law was needed, it is worth discussing the most debated patent jurisdiction at the moment, the U.S.  U.S. legislation has a broad statutory definition of patentable inventions:

U.S. Code , Title 35, Part II, Chapter 10, Sec. 101. - Inventions patentable

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefore, subject to the conditions and requirements of this title .[24]

The negative limit was established by the Supreme Court in Diamond v Diehr :

We recognize, of course, that when a claim recites a mathematical formula (or scientific principle or phenomenon of nature), an inquiry must be made into whether the claim is seeking patent protection for that formula in the abstract.  A mathematical formula as such is not accorded the protection of our patent laws, Gottschalk v. Benson, 409 U.S. 63 (1972), and this principle cannot be circumvented by attempting to limit the use of the formula to a particular technological environment.[25]

However this same decision opened something of a Pandora’s box :

When a claim containing a mathematical formula implements or applies the formula in a structure or process which, when considered as a whole, is performing a function which the patent laws were designed to protect (e.g., transforming or reducing an article to a different state or thing), then the claim satisfies 101's requirements.

The broad statutory definition and the USPTO’s explicit willingness to allow patent protection for software shifted the question from ‘can software be patented’ to ‘to what extent can software be patented’.[26]  In fact, while the USPTO’s 1996 Guidelines gave certainty as to patentability of computer implemented inventions (referring to either the computer as a whole, or the memory that directs it, or the process put forward by the computer) case law lifted the bar even higher.[27]

While the Diamond v Diehr decision softened the line on the unpatentability of algorithms, the then Court of Customs and Patent Appeals developed a test (the FWA test) that practically allowed patentability by relaxing the physicality requirement:[28]

It is first determined whether a mathematical algorithm is recited directly or indirectly in the claim. If so, it is next determined whether the claimed invention as a whole is no more than the algorithm itself; that is, whether the claim is directed to a mathematical algorithm that is not applied to or limited by physical elements or process steps.[29]

After this physicality requirement was further softened by the In re Alappat decision,[30] the decision of the State Street Bank [31] case determined that the issue was whether the algorithm had been applied in such a way to produce tangible useful results.  Thus, the open ended statutory definition and the approach sensitive to improvements in efficiency and usefulness, gave the U.S. a reputation as a more liberal patent jurisdiction.  However, another view is that the threshold for patentability in the U.S. is generally too low, and the generous approach of the USPTO strengthened big corporations [32] by allowing broad claims and non inventive patents, like Amazon’s ‘One Click’.[33]  We will see below that the European legal framework differs substantially from the American one, but nonetheless this difference almost fades away thanks to the European Patent Office’s practice.

The Berne Convention and the CDPA define what a literary work is, while the WIPO treaty and the Council Directive 91/250 provide copyright protection for computer programs, ‘whatever the form or mode of their expression’.[34]  If we combine this with the provisions of EPC 52 (2) (‘the following shall not be regarded as inventions within the meaning of paragraph 1: … computer programs..’) we could say that at least formally in Europe, computer programs lie outside the range of patentable inventions.  In practice however this is not the case.  In fact, although this is still the current legislation, the EPO have granted some 30,000 patents for computer implemented inventions.[35]  It narrowed  the art.52 (2) exclusion of computer programs in the VICOM case:

[A] claim directed to a technical process which is carried out under the control of a program (whether by means of hardware or software) cannot be regarded as relating to a computer program as such .[36]

The outcome of the EPO’s practice was that patents began to be granted to computer programs, relying on the ‘technical effect’ doctrine.[37]  Computer programs that had a technical effect per se, or in the execution of their instructions, and had a physical effect as outcome of their operations were considered suitable for patent protection.[38]  After leading cases such as VICOM and IBM,[39] the way was actually open for patent protection for computer programs, eroding the significance of EPO’s original guidelines.[40]  Nevertheless the requirement for technical character allowed for different interpretations to emerge.

4. Technical Contribution: Magic Formula or Legal Compass?

According to the EPO’s practice,[41] if a program constitutes a ‘technical contribution’ to the state of the art, and it is new and capable of industrial application, it is patentable.  This may involve the solution of a problem by the invention, or an alternative solution for an already solved problem.  The VICOM decision, and the two IBM cases,[42] shaped this feature of ‘technical contribution’, defining as patentable those programs that have a ‘technical character’, sidestepping the exclusion provided by the Guidelines of the EPO itself.[43]  This exclusion implied that a computer program claimed by itself or on carrier could not be patented, but the interpretation put forward by EPO meant that this exclusion could be circumvented by the assessment of the further effects of the program.  In other words, the execution of the instructions of the program might produce ‘technical effects’ when the software solves a technical problem.  The rationale comes from the practice of German patent courts, and was used by EPO to justify the grant of patents to a subject matter formally excluded from such protection.  In Germany, while it is settled that patents cover not only the surface but also the theory underlying an invention, nonetheless the idea of patenting algorithms or methods to process data has always been opposed.[44]  Therefore, prima facie , to be afforded a patent for a computer program, it is necessary to link the software to an apparatus in order to materially acknowledge the existence of a product or process as subject of the claim.  However, this ‘technical contribution’ does not provide a physicality requirement as it may seem at first.  In fact the EPO decision in IBM/Data Processor Network [45] made it clear that this requirement may be otherwise satisfied.  It was held that even when the claim refers to a communication method between ‘program and data files’,[46] the ‘technical contribution’ requirement is satisfied if the invention solves a technical problem.[47]

The conclusion of this interpretative evolution came with the decision of the EPO Board in IBM/Computer program product [48] where the technical contribution was found in the mere fact that the program was claimed through its carrier medium.  With this decision the EPO concluded that computer programs able ‘to produce a technical effect’ are patentable subject matter within the meaning of 52(1) of the EPC.  This may be considered as the end of the software ‘as such’ exclusion.[49]  Despite the above mentioned tradition of German patent courts requiring the program to be attached to a physical process or device, the recent EPO’s practice basically allows for the so called ‘program claims’.  The interpretation of IBM/Computer program product suggests that the technical effect may be confined to known physical features, external to the program and simply kick-started by data processing methods.  In such cases, novelty would be relegated to merely abstract mathematical methods, giving rise to ‘program claims’ without an individual technical character.  It is now clear that the ‘technical contribution’ doctrine, rather than being a physicality requirement, refers to efficiency and usefulness criteria, thus quite closely resembling to the American system.

Not surprisingly, this ‘technical contribution’ doctrine is at the same time criticized and disputed as to its real meaning within the context of the Directive.  Council's draft recital 12 provides that :

[W]here it does not make a technical contribution to the state of the art, as would be the case, for example, where its specific contribution lacks a technical character, it will lack an inventive step and thus will not be patentable.

In the light of what was explained before, this definition may seem quite vague, and some critics of the Directive have called for the ‘technical character test’ to be applied before assessing the technical contribution to the state of the art, so to filter out non technical inventions (software as such).[50]  Activists such as Eurolinux and the Foundation for a Free Information Infrastructure argue that the ‘technical contribution’ has been found by EPO in a very broad range of cases, many of which were controversial as to their effective technicality.

While it is settled [51] that the processing of physical data such as images or parameters of industrial processes by computer programs meets the requirement, that is not the case when software merely changes the way a computer works with an unperceivable technical effect [52].  In the latter case, when a program saves memory space, or increases the speed of data processing, simple efficiency and security requirements are met,[53] but not necessarily a technicality one.[54]  In other words, acknowledging technical character with such a generous approach would confer the nature of technical contribution to a broad range of features, such as ‘the problem to be solved, the implementation of the solution, the function of the implementation and its effects’.[55]

The fact that the ‘technical contribution’ requirement is central to the draft Directive and that its meaning is so open-ended, explains much of the controversy around the proposal.  It does not eliminate the uncertainty that pushed the Commission to propose the Directive, but rather constitutes a codification of this uncertainty.  The use of the ‘technical contribution’ requirement made by the Council in the Draft Directive is more an incentive to broaden the range of patentable subject matter rather than a clarification of its limits, as asked by the Parliament.[56]  The rift between the Parliament and the Council of Europe refers specifically to the wording of the Directive, both in relation to ‘technical contribution’ and ‘algorithm’.

5. The Algorithm

Computer programs are essentially algorithms.  For this reason it is necessary to briefly describe them in order to understand why they are central to the debate, and what their role is during the functioning of the program.  An algorithm is a set of instructions understandable by humans, aimed at performing a specific task that, given a set of inputs, will result in some recognisable ‘end-state’.[57]  The algorithm tells the computer what steps to perform, in what order and how many times, to achieve a specific task: the order of this steps and each step is crucial to carry out a certain operation.  Therefore it can be said that the algorithm is the ‘intelligence’ of the program, whose features determine the nature of the operations performed under its governance.  The algorithm is responsible for the behaviour of software, and as it is accepted by EPO that the technical contribution may be found in its ‘further technical effects’[58] occurring into a physical non-novel part, the following conclusion may be drawn: the practice of EPO and its codification in the Council’s common position allows claims where the technicality is found in known technologies and whose inventive part consists solely of an algorithm.

If such broad definition of technical contribution remained, there could be claims for inventions consisting of novel algorithms operating in known technical environments, confining inventiveness to the way data structures are processed by software.  In such a case, the patent refers not to an object or a process, but to the description of how it is and how it works, and there being a novel algorithm operating within a common technical framework, such novelty turns out to be ‘literary’[59] rather than technical.  In this light, the requirement put forward by the Parliament, that the invention control physical forces ‘beyond digital representation of information’, becomes an attempt to prevent controversial outcomes such as the one described above.  Open source activists call these kinds of claims ‘information claims’ or ‘text claims’ to emphasise the monopoly over information conferred by such patents [60].

Although the draft Directive (13a)[61] defines algorithms as non patentable, it still considers patentable a ‘method involving the use of an algorithm’ and provides that the patent would not monopolise that algorithm or its use in other contexts.  However, it is still not clear what use of an algorithm would be permitted and what would constitute an actionable infringement.  Once again the same regulation sets forth a provision along with the means to circumvent it: algorithms are not patentable but ‘technical contribution’ has such a broad meaning that the first exclusion may be sidestepped.  It is the same as EPC art.52: the EPO used 52(3) to nullify 52(2) and may use the definition of ‘technical contribution’ [art. 2 (b) of the Council draft] to overcome the exclusion described above (13a of the Council Draft).  Given the premises of the Proposal put forward by the Commission in 2002, to eliminate uncertainty as to what is patentable,[62] this hardly appears to have been accomplished.

6. Parliament v Council

As we have just seen, it is specifically the wording of the draft Directive that creates tension between Parliament and the Council, the proposal of the Commission [63] creating a rift at the highest level of the European institutions.  The Parliament was lobbied by Eurolinux, while the Council, being composed of ministers of member states, was more sensitive to the needs of corporations.[64]  The Parliament amended the Commission’s proposal in order to secure that the current exclusion of art.52 remained, recommending a review of EPO’s practice, and stressing that the Directive should explicitly require novelty to the technical effect itself, so restricting patent protection in cases of data processing methods.[65]  The Council replied deeming as inappropriate and unacceptable the amendments made by the Parliament, and issued a proposal that showed a more pragmatic approach, actually refraining from classifying ex ante what is a patentable software invention, so to leave the doors open to future developments.

The law-making process involving the two institutions is called ‘co-decision procedure’ (art.251 EC), and was introduced by the EC Treaty and later amended by the Treaty of Amsterdam.[66]  According to the co-decision procedure, the Commission submits the proposal to the Council and the Parliament, the latter votes on its decision in plenary session, and then submits the outcome of its first reading to the Council.[67]  The Council may accept the outcome or may reject it and adopt its common position, and consequently submit the amended text to the Parliament for the second reading: if such text is not accepted by the Parliament, the Procedure is terminated, and the Act is not adopted.[68]  The two institutions clashed on specific issues and sections of the draft Directive.  What follows is an overview of their two positions, comparing the wording used by the Parliament [69] and the Council [70] in their respective proposals.

The European Commission proposed the Directive for competition purposes (that will be dealt with below) and to harmonize the way national laws interpret the European Patent Convention.  While its aim was formally to introduce software patents assuring that pure software would remain non-patentable, the outcome of the reading of Parliament and the Council proposal do not seem to agree on any of these grounds.  The Council drafted broad and open ended legislation aimed at encouraging more applications suggesting there is ground to enlarge the realm of patentable inventions.  The Parliament stands firm requiring clear limits to the patentability of computer programs and set out a list of detailed conditions for such purpose.  Now the Parliament is about to proceed to the second reading, that may wreck the Directive in case of rejection of the Council’s common position.  The contest is still open but the last political agreement reached by the Council puts forward a proposal that gives rise to several considerations.  One such consideration is that leaving the wording of the Directive as it is now may open the door to very broad claims over inventive areas that are now at a crucial stage of their evolution.  Applicants normally seek to exercise a wide monopoly, and this aim is achieved by obtaining a broad patent, and that is easier in absence of clear limits to patentability.  The result of changing EPO’s practice during the years is also due to talented drafters, who use clever techniques to describe the invention.  The drafter may identify the infringer or the infringing activity beforehand, have responses to possible objections embedded in the claim, set out more scopes per claim, assign multiple claims for each individual component and avoid descriptions that may recall mental activity.  Without clear limits and definitions of patentable subject matter, of technical contribution and interoperability, these clever drafting techniques may confer to the patentee an undeserved extent of monopoly.

7. Competitiveness and Innovation

The uncertainty within the EU as to what was patentable and what not, created inefficiency because most SMEs were unaware of what was available at EPO to protect software.  Most of the patents related to software granted by EPO were afforded to non European companies,[71] while the ambiguity regarding the interpretation of the legislation kept out of the market many small competitors.[72]  On a global level, the apparent flaw in the European legislation arguably denied companies of E.U. member states the same degree of IP protection enjoyed by their competitors in Japan and the U.S.  Mobile phone manufacturers are definitely among those suffering most from the lack of certainty about the degree of patent protection available in Europe.  While their competitors in U.S., Japan and South Korea operate in an environment where the patentability of software is not disputed, Nokia, Ericsson, Philips and Siemens [73] are still unable to carry out in full their business strategy for third generation mobiles in the E.U.[74]

At the same time, there is a vast number of programmers, small companies and individual users, who resist the idea of considering software as a subject matter of patent law.  The Open Source community, who heavily lobbied MEPs to counter the position of the Council, relies on the incredible success of its project to demonstrate that property rights are detrimental to the evolution of software.  The fact that a simple handling (by means of publication for example) of codes or algorithms may constitute an actionable infringement, is a dreadful prospect for those like Linux users, who systematically trade on shared knowledge. 

The critical point for competitiveness is not who is excluded from patent protection, but who is excluded by patent protection. While the intention of the Commission was to make patent protection accessible to SMEs, there is a danger that an ill drafted Directive may actually put small companies not just outside the legal protection but also outside the market.  Costs are the main issue, since application and consultancy costs are too high for small companies and litigation fees can be fatal: many small enterprises would immediately surrender to any threat.  If the limit of patent protection and the wording of the definitions of the Directive are not clarified, companies with a portfolio of strong patents and resources for litigation may simply suffocate small competitors with a ‘cease and desist’ letter.  Infringing activities, permitted acts and scope of claims should be defined in a way that prevents arbitrary use of patent law as weapon of competition warfare. 

A wider issue regarding competitiveness and innovation concerns the information architecture and its network related aspects.  A network whose applications, nodes and end user technologies are made to work on a single prototype of computer platform is inherently weak.  It is weak in its body that communicates with a limited number of languages and protocols, while its limbs are feeble because the end users have just a small number of choices of applications.  A network with more than one predominant language, platform and protocol is stronger because has wider capacity in terms of data, and more solid ends due to the possibility to shift from one system to another.

8. Conclusion

The tension between the European Parliament and the Council now involves the Commission. Internal Market Commissioner Bolkestein did not conceal his disappointment at the large number of amendments made by the Parliament, and while welcoming the Council’s common position, threatened to withdraw the proposal if an agreement is not reached.  He hinted at the possibility of a renegotiation of the EPC that would not concern the E.U. and could possibly end up in an even more pragmatic outcome.  In general it is not fair to deny a patent to an invention simply because it is based on a computer program, but at the same time software is a problematic issue because there doesn’t seem to exist an IP right fit for its combination of speech and behaviour.  We have seen that both copyright and patent law lack something in connection with basic features of software, and that these asymmetries still haven’t been settled by the existing legislation.  So far there has been an attempt to codify the requirement of ‘technical contribution’ in order to eliminate uncertainty as to the patentable subject matter of software inventions.  The problem is that the meaning of this requirement is far from being clear.  The assurance made by the Commission that what was non-patentable before would remain non-patentable, and what was patentable would remain patentable has little significance since the distinction was and still is impossible to draw as to software.  It is important to understand that the concept of what is capable of forming the subject matter of patents should be adapted to less physical parameters in the light of the decrease of volume of hardware and to assist the growth of new technologies.  At the same time it is necessary to spread the benefits of the protection of intellectual property over a broader range of companies rather than assisting large IT corporations in their attempts to set a definitive standard.  Patent protection should be accessible to SMEs, comprehensible to end users and compatible with the Linux platform.  That appears to be quite difficult at the moment, for two specific reasons.  The first regards the struggle between the two European institutions: the Parliament sensitive to an anti-IP lobby,[75] the Council of Ministers keen on favouring corporations with massive R&D expenditure.  The second regards the procedural path of the Directive that may still collapse, opening the way to a renegotiation of the European Patent Convention that would hardly help towards harmonization with the E.U.

Notes and References