Patent Jurisprudence on a Slippery Slope

A simple look at the results of patenting software in Europe and the economic studies of their effects on innovation and competition is enough to make most people realise that something is going utterly wrong here. Now some people will claim that all this is just a problem of improving prior art search capabilities of patent offices. Others point out that, even if the decade-old novelty examination problem could be solved tomorrow, this would not help much: meaningful non-obviousness standards can hardly be formulated and would, even if they were found, be difficult to uphold against the normal tendencies of patent offices to "deliver good value for the money of our customers", as a high european patent official recently put it.

Yet, even if reliable standards of novelty and non-obviousness could be defined and consistently upheld in practice, that would not be enough. We will not get around the question what should be patentable and what not. The patent system needs a limitation of statutory subject matter. Should new schemes of social engineering be patentable? Musical ideas? Ways of doing business? Argumentation chains? Anything useful? Anything practical? Anything economically exploitable? Anything for which patent offices decide to hire examiners? Where exactly are the limits? Traditionally this question has been answered by the concept of "technical invention", i.e. the requirement that an invention should not teach us a new idea but rather a new way of putting an idea to work by using controllable forces of nature.

There are many reasons for this traditional line-drawing between between logical ideas and physical implementations.

We have on one side questions asked to the real world, or, more narrowly, the world of controllable natural forces. Any new solution that is an answer to such a question will be fairly difficult to find, and it will entail the use of material objects both for experimentation and for utilisation, so that there will be an overhead cost to which patent fees can be added without much of a hassle.

On the other side we have questions asked to human reason, where the answer is a new rule of organisation or calculation, verifiable solely by logical deduction.

The separation between these two worlds was usually fairly straightforward 50 years ago and didn't create spectacular problems. Yet difficult borderline cases have always existed, and sophisticated reasoning for separating the mental from the material was already published and discussed by patent law experts in the 19th century.

However, gradually the two worlds, also called "world of matter" and "world of mind" by German patent law philosophy, have come closer together, and the most common mediating element between them is the universal computer.

So on one side there are people who say: if physical processes are patentable, why should it make any difference whether they run on special hardware or on general-purpose hardware?

On the other there are people who say: if rules or organisation and calculation are not patentable, why should it make any difference whether I run them in my head, with pencil and paper or with the normal tool of today's civilisation, which is the universal computer?

And in fact the second argument is right, because the economic rationale behind not granting patents on thoughts applies also here: abstract ideas are, regardless of their applicability to technical problems, produced without experimentation cost, applicable to an infinite range of problems, and replicated at zero cost with no overhead to which patent license fees could be added. The division of the extra cost of patents by the marginal cost (and long-term ideal price) of information goods is a division by zero. Moreover the deal between the inventor and the public, characterised as "monopoly on commercial implementation in return for disclosure of idea", is led ad absurdum: since between the idea and the application there is no invention, any adequate disclosure of the idea in turing-complete syntax risks to become an act of patent infringement.

These problems were intensely debated in the 60s and 70s among patent law scholars, especially in Germany. It has been explained in patent law text books and worked out by leading German law courts. Even possible future aberrations and their consequences were predicted by some of those texts. The only reason that the aberrations nonetheless occurred lies in the well-known weakness (ignorance and greed) of man. Patent lawyers are on the average quite ignorant about these questions, because being knowledgable about them does not bring any extra revenues into the law office. The same applies to patent offices such as the EPO, who exercise control over patent jurisdiction and at the same time finance themselves by revenues from the patents they grant.

According to the European Patent Convention, computer programs are not patentable inventions. This does not exclude that a patentable technical process may run under program control. However this process must be more than a rule (program) for using conventional computing equipment: it could reside in computer periphery which applies brake forces to tyres, and it should teach some new knowledge about causal relations in the use of forces of nature. The realm of pure reason, i.e. calculation, abstraction and programmation, must, according to the will of the legislators, remain free from patent claims. In its landmark decisions Dispositionsprogramm of 1976, the German Federal Court (BGH) discussed this subject extensively and confirmed the dichotomy between the "realm of reason" and the "realm of natural forces", saying that the patent system must stay confined to the latter and lawcourts do not have the right to extend it into the former, since this would be against the will of the legislator and could pose a serious danger to freedom and basic interests of society while not necessarily helping to promote the progress of mathematics or any other of the non-patentable realms of reasoning. The wisdom of the doctrine of "technical invention", as formulated in the Dispositionsprogramm decision and other decisions in Germany and other countries, has been corroborated by old and new studies in the economics of technological innovation. It still serves as an authoritative source for Patent Law commentaries such as the standard work of Dr. Georg Benkard, who explained this concept of technical character in his commentary of 1988 and even in later editions and in many treatises and court decisions about the limits of patentability until this day, be it as a model or as a sign of bad conscience.

At the same time, since the 1960s there has been an increasingly influential group of patent lawyers who were not willing to accept a limitation of the patent systems to the "world of matter", arguing that this would keep patents out of some of the most interesting spheres of future economic development. Under their influence, the erosion of patentability standards proceded at a quick pace. Control logic in the area of microelectronics was patented, and abstract claims to functionality gradually became acceptable, making it inconsistent to uphold the "discrimination against computer programs". The patent departments of Siemens and IBM made special contributions to the "evolution of caselaw", investing great efforts to bring about several landmark decisions by the EPO and the German Federal Court. Already by the early 90s, all items in the exclusion list of Art 52 EPC, including mathematical methods, computer programs, organisational methods and methods of presenting information, became de facto patentable, and the exclusion list was interpreted to apply only to meaningless "as such" entities. This means that on the one hand anything, even chemicals or pharmaceutics, could be put on the exclusion list without influencing the patentability of such items, and on the other hand there is no more any limit to the expansion of the patent system into the whole range of logical innovation, from software to music to social engineering or political schemes or business models. However these have to be claimed using the terminology of "information technology". Thus it can happen that a carelessly worded "pension benefit system" claim will fail before the EPO, because the applicant forgot to disclose enough "technical effects" at the application date. What is patentable becomes mainly a question of claim wording, as one leading expert attorney correctly remarked.

Until the mid 90s, the EPO was however still shying away from the utmost consequence of this adventure. Rules of organisation and calculation were patented only indirectly, more or less carefully dressed up as "technical" processes. Whoever distributed a program on the market could still not infringe on any patent claim. The public at large therefore did not notice what was happening. Meanwhile the patent community gradually became more bold. In 1998 the EPO's technical board of appeal (not competent in questions of law) defined that a "computer program with a technical effect" is not a "computer program as such" and started granting patent claims to informational entities ("computer program product", computer program, data structure). By an "editor's remark" under the document the EPO's president made it clear that this was the new policy of the Office.

Meanwhile, the caselaw of the EPO has indeed been pushing the boundaries of what is technical ever wider. According to its Examination Guidelines of 2002 "A further technical effect which lends technical character to a computer program may be found e.g. in the control of an industrial process or in processing data which represent physical entities or in the internal functioning of the computer itself or its interfaces under the influence of the program and could, for example, affect the efficiency or security of a process, the management of computer resources required or the rate of data transfer in a communication link."

According to rulings on EP 0689133, even the "economical use of the resource area on the screen" is "technical".

And if that doesn't cover the program, then "processing which is based on considerations of how a computer works is technical". Thus in the Sohei case a patent the EPO Board of Appeal upheld a patent for using the same input form to update two databases, namely an inventory database and a billing database, because it implied (but didn't actually limit itself by specifying how to do it) the handling of files containing different types of information, which is technical.

However the patents granted on this basis are of an uncertain value. Independent lawcourts (such as the Swedish Highest Court as well as the 17th senate of the German Federal Patent Court) regularly find contradictions in the reasoning of EPO and BGH and reject claims to computer programs or logical ideas. Other European lawcourts can be expected to point out these contradictions any time, as there is not very much authoritative case law e.g. in infringement procedings. Thus the patent community is very keen to change the existing laws so as to adapt them to the new practice of the EPO. In fact, at the time of the introduction of this practice, they had already assured themselves of the European Commission's support for a planned change of law. This failed however, at least preliminarily, in November 2000, due to vocal opposition from software creators and lack of support from European governments. Nevertheless, the EPO seems unimpressed. In 2001-10, it published new examination guidelines adapted to the 1998 decision, as if nothing had happened. Moreover, in the upcoming months a directive proposal from the European Commission is expected, and so far few in Brussels or in the national governments have shown any willingness to criticise the policy of the EPO. However even if the EPO and its friends eventually succeed in pushing their agenda through, this will not clarify the rules, but only bring the law in line with EPO jurisprudence, i.e. institutionalise legal uncertainty: US-style unlimited patentability veiled in a mist of politically motivated patent newspeak, with the effect that unimpressed American and Japanese companies will continue to obtain most of those patents which the EPO in theory does not grant.

At the end of the nineties, the terminology around the "technical invention" was still in use at the EPO. However it only rarely served to exclude patentable subject matter. Most of the time it served to play political games. It allowed the EPO to pretend that it was doing what the law and the politicians wanted it to do. From this purpose, a complex Doublespeak developped around terms such as "technical problem", "technical contribution", "computer-implemented invention" etc. The analysis of this Doublespeak is a task of sociolinguistics which is beyond the scope of this overview.