The tortuous development of the law concerning patents for software-related inventions in Europe has had many people scratching their heads in puzzlement. The two sides to the debate over whether patents should be allowed in the field of software at all – the software industry and the open source lobby – have been at daggers drawn in many amphitheatres over more than a decade. What to do about it all has been equally controversial. Kitchin J’s recent decision in Astron Clinica & Ors v Comptroller General of Patents ( EWHC 85) certainly does not synthesise those fundamentally opposed positions, but it marks a clear intention to reconcile the UK’s position with that of the European Patent Office (“EPO”), a far more immediate concern in practice for anyone working with software in the UK. Unfortunately, it seems that his brother judges in the English Patents Court may not share that intent.
The problems arise from two sources, of unrelated natures. The first is the drafting of the European Patent Convention (“EPC”) in 1975, which included a list of kinds of matter excluded from patentability. Computer programs feature on the list, but are only excluded “as such”. The ambiguity inherent in those two words has left substantial room for arguments.
The second is the evolution of the treatment of software inventions in the USA where, following the landmark case of Diamond v Chakrabarty (447 US 303) more or less anything manmade is patentable. In consequence, software patents have been granted in their tens of thousands in the US, including, since 1994, claims to software on a storage medium (i.e. a floppy disk) (Beauregard 53 F.3d 1583 (Fed. Cir. 1995)). This form of claim is essential for software houses, which otherwise face the impossible task of suing end users for the infringement of claims to a computer loaded with or running the software, rather than the commercial enterprises making and distributing infringing software.
Where an invention found to be patentable in the USA is rejected in the European Patent Office, significant transatlantic tensions inevitably arise.
The law in Europe
The EPC sets out the requirements for patentability of inventions in Articles 52 - 57. The primary definition in Article 52(1) provides that:
“European patents shall be granted for inventions in all fields of technology which are susceptible of industrial application, which are new and which involve an inventive step.”
Article 52 also attempts to define what things are not patentable. The provisions which concern software and business method inventions are:
“(2) The following in particular shall not be regarded as inventions...
(c) schemes, rules and methods for performing mental acts, playing games or doing business and programs for computers;…
(3) The provisions of paragraph 2 shall exclude patentability of the subjectmatter or activities referred to in that provision only to the extent to which a European patent application or European patent relates to such subject-matter or activities as such.”
The question then arising is, when is a computer program a computer program as such and when can it be treated as something more, and therefore patentable? The EPO has tried to minimise the divergence from the US position by circumscribing the exclusion as closely as possible in reliance upon these final two words, ‘as such’.
The ‘technical effect’ doctrine
To define when a computer program is more than merely a computer program “as such”, the EPO developed a doctrine of ‘technical effect’ as the test for patentable subject matter in a software-related invention.
The doctrine was first developed in the case of Re Vicom (T208/84) concerning a new method of handling the images for a CAD (computer aided design) system. In this case, the Board of Appeal decided that:
‘… a mathematical method as such is an abstract concept prescribing how to operate on numbers, without producing any direct technical result. By contrast, where a mathematical method is used in a technical process, the process is carried out on a physical entity (for example, an image stored as an electrical signal) by some technical means which produces a change in that entity.’
This has been the basis for all subsequent decisions of the EPO in this area. Unfortunately, the EPO Board of Appeal has not been consistent in its statements as to when there is or is not the right kind of ‘technical effect’, or at which stage the examiner should consider the issue: before consideration of the substantive requirements of novelty and inventive step, or as part of the evaluation of inventive step?
The subsequent trend was for increasing liberalisation, with decisions that an invention comprising functional features implemented by software is not excluded from patentability under Article 52(2)(c) and (3), if technical considerations concerning particulars of the solution of the problem the invention solves are required in order to carry out that same invention (Sohei/Yamamoto’s Application) since this implies a technical problem to be solved by implicit technical features.
Finally, in the 1997 IBM Computer Program Product cases (T 0935/97 and T 1173/97), IBM had included claims for a computer program product which had the effect of making a computer carry out certain actions, as well as the more conventional claims to a programmed computer capable of carrying out these actions and to the method. The Board of Appeal pronounced that if a computer program either by itself or on a carrier (e.g. a computer disk) has the potential to produce a technical effect when loaded and run on a computer, such a program claimed by itself should not be excluded from patentability. The criterion to be applied was whether, when run on a computer, the program produces any further technical effect going beyond the normal physical interaction between a program and a computer. In effect Beauregard type claims were now patentable in the EPO – for programs capable of a suitable technical effect.
Unfortunately, even after the Computer Program Products cases, EPO case law continued to evolve. In Pension Benefits (T931/95), the invention related to running a pension scheme and calculating the relevant benefits on a computer. The Board ruled that a non-technical advance could not contribute to inventive step, and so the application was rejected. Unlike the Sohei decision, the fact that a technical problem was overcome was irrelevant because this was not solved by technical means. This was the first time that the EPO had regarded the form of the claim (as opposed to its substance) as being of significance. The situation was muddied further by subsequent decisions in Hitachi (T258/03), where a different Board of Appeal expressly disagreed with Pension Benefits and found that the presence in the claim of computers and a network was enough to bring the claim within patentability. The Board then treated method and apparatus claims identically but ruled that the issue of patentability of subject matter should only be addressed at the point where the invention was being assessed for inventive step, by considering for that purpose only features which contribute to a technical character. In yet a further decision, Microsoft/ Data Transfer in 2006 (T424/03), the approach of treating any physical apparatus as sufficient to bring a claim within patentability was reiterated.
But not all examiners or Boards of Appeal follow this line of recent cases.
Legislative attempts at reform
In March 1999 SACEPO (the Standing Advisory Committee before the European Patent Office) published a report recommending that subsections 2 and 3 of Article 52 be deleted.
This was followed in June 1999 by an intergovernmental conference, which launched a reform procedure, setting up working parties on reducing the cost of European patents and harmonising the cost of patent litigation. A draft revision of the EPC was put forward in October 2000 which also proposed deleting from Article 52(c) the reference to computer programs. However, the Munich Intergovernmental Conference which followed could not agree on this revision, and it was dropped, although other changes went through. The Conference passed a resolution which stated inter alia:
“Other subjects the Conference felt had not yet been sufficiently thought through for this to be the right time to make a binding decision on them. In particular, recognising the importance of the issue, the Conference agreed to maintain for the present the European Patent Convention’s current provisions on software. This will provide the opportunity for full consultations already underway on this issue to be completed, but the Conference is fully aware that this is not a subject that can be left open indefinitely.”
Nothing then happened. In an attempt to move things forward in the narrower assembly of the EU Member States, in February 2002 the European Commission published a Proposal for a Directive on the Patentability of Computer-Implemented Inventions. With one important exception, the draft Directive was not intended to change the law on patenting software but only to codify and clarify it. However, the draft met with concerted opposition from the open source movement, which lobbied the European Parliament very effectively to restrict what might be patentable still further. Ultimately, after several hardfought rounds of negotiation the project was dropped.
The English courts’ approach
It is not surprising that the national courts, faced with the problem of attempting to decide which of the numerous softwarerelated patents that come into dispute should be upheld or revoked, have had some difficulty in finding a consistent line through these conflicting approaches. In shopalotto.com’s Application ( EWHC 2416), the late Pumfrey J (as he then was) said
“Suppose a program written for a computer that enables an existing computer to process data in a new way and so to produce a beneficial effect, such as increased speed, or more rapid display of information. It is difficult to say these are not technical effects…… The real question is whether this is a relevant technical effect, or, more crudely, whether there is enough technical effect: is there a technical effect over and above that to be expected from the mere loading of a program into a computer? From this sort of consideration there has developed an approach that I consider to be well established on the authorities, which is to take the claimed computer, and ask what it contributes to the art over and above the fact that it covers a programmed computer. If there is a contribution outside the list of excluded matter, then the invention is patentable, but if the only contribution to the art lies in excluded matter, it is not patentable.”
In response to this amongst other decisions, the UKIPO changed its examination policy and adopted the hardware-based approach, considering technical effect only at the stage of review of inventive step.
In 2006 the joined cases of Aerotel/ Macrossan reached the Court of Appeal and gave it an opportunity to attempt to clarify the situation for the first time in many years in a higher court. But neither of the inventions which the Court had to consider in that case amounted to a computer program ‘as such’. Aerotel’s invention was a new configuration of telephone exchanges and Macrossan’s an automated, web-based application for the formation of companies. The Court nevertheless conducted a detailed review of both English Court of Appeal and EPO Boards of Appeal case law. The Court’s conclusions were that the approach the EPO appeared to have taken post- Computer Program Products amounted to a policy that the subject matter of an invention would be patentable so long as any hardware was involved in its use. Such an approach was incompatible with the existing English Court of Appeal decisions. These decisions bind the Court, although they do not read with as much conviction as might be hoped, being speckled with otherwise-learned judges’ comments as to not themselves understanding what sort of a technical effect might be being produced by the computer-related inventions under discussion.
The Court went on to consider whether the exclusion of computer programs in Article 52 was intended to include computer programs on a storage medium. In a passage susceptible of more than one interpretation, Jacob LJ concluded that it was:
“A narrow view is that [computer program] means just the set of instructions as an abstract thing albeit they could be written down on a piece of paper. A wider view is that the term covers also the instructions on some form of media (floppy disk, CD or harddrive for instance) which causes a computer to execute the program – a program which works. This court and the earlier Boards of Appeal decisions clearly take the latter view…”.
The Court of Appeal’s comments were, of course, made in passing and not formally part of its findings on the case before it. Nevertheless, the Court set out a new four-step test for patentability of software inventions under English law: –
(1) properly construe the claim;
(2) identify the actual contribution;
(3) ask whether it falls solely within the excluded subject matter; and
(4) check whether the actual or alleged contribution is actually technical in nature.
This fourth step may not be necessary – the answer to it should have fallen out from assessing the third step. But it was necessary to include it for consistency with existing Court of Appeal case law which is binding on this court.
The judges also made a recommendation to the EPO to get the Enlarged Board of Appeal to look at the vexatious question in order to set clear principles as to patentability of software once and for all. Disappointingly, the EPO firmly declined, being satisfied with the present arrangement. Nevertheless, in response to Aerotel/Macrossan the UKIPO changed its practice once again, reverting to refusing all applications claiming simply a program on a medium.
The first instance judges are now attempting to interpret and apply Aerotel/Macrossan principles in the ongoing trickle of cases contesting UKIPO decisions in this area. In Astron Clinica, a January 2008 decision, Kitchin J has concluded that the UKIPO’s interpretation of Aerotel/Macrossan is wrong. The judge looks carefully at the wording of the Court of Appeal’s discussion, and interprets it to be saying that a computer program on a medium is within the exclusion of Article 52(2), but that this finding simply does not address the correct application of the proviso “as such”. Kitchin J therefore finds the way left open to conclude that while a program on a storage medium is within the exclusion, the effect of the proviso is to bring such a program back within patentability – in line with the EPO’s practice in Computer Program Products, and with the general trend of EPO decisions since then. This interpretation has the benefit of permitting the UKIPO to fall back into line with the EPO, and it has immediately taken the opportunity to do just that.
Doubts must remain, however, as to whether this was what the Court of Appeal actually intended by its comments. Jacob LJ commented that the narrow view is that adopted by the EPO in its decisions in Pension Benefits, Hitachi and Microsoft Data Transfer which are in his view, wrong:
“so to limit the meaning of “computer program” would be to render the exclusion without real content. We think the framers of the EPC really meant to exclude computer programs in a practical and operable form. They meant to exclude real computer programs, not just an abstract series of instructions.”
At this point, Jacob LJ seems to be contemplating that the exclusion should be effective against ‘real computer programs’, and the minimum degree of reality a computer program can have is a recording on some form of storage medium – whether paper, magnetic tape or disc or optical. If so, then to interpret the proviso “as such” to bring precisely those minimally-real programs back into patentability would go directly contrary to the Court of Appeal’s intention.
Certainly the first High Court decision following Aerotel/Macrossan appeared to follow that intention. In Oneida Indian Nation’s Application ( EWHC 954 (Pat)) Christopher Floyd QC2 ruled on a claim to a program for an offsite gaming system which was alleged to improve previous systems because the number of processing steps per bet is reduced so the number of data transmission steps is reduced. The system also dispenses with the need to make an account check for every bet, and as a result the system is more secure and robust. He held it was not patentable, because the inventive contribution, if any, lay in the excluded matter of the business method.
Nevertheless, in case found on appeal to have been wrong about this he also considered whether the program, if patentable as a method, would have been patentable as a series of instructions stored on a medium. He concluded, in reliance on Aerotel, that it is not. No technical problem is solved by doing so and no technical effect is produced. In Astron Clinica, Kitchin J distinguishes this comment by reference to Mr Floyd’s analysis earlier in his judgment that the invention’s advantages were solely those resulting from putting the method on a computer. This is true, but it is not the reason Mr Floyd gives at the relevant point in his judgment.
The most recent cases have given two of the remaining Assigned Judges the chance to consider the issues as well. In the latest, Patten J heard a dispute between Symbian Ltd and the UKIPO ( EWHC 518 (Pat)) where the patent examiner refused to grant a patent for Symbian’s improved interface for dynamic link libraries (“DLLs”), saying that the invention was excluded subject matter since it was nothing more than software controlling the operation of other software. The invention in issue was a method of improving reliability of a computer operating executable programs in the face of changes in functionality implemented by elements of code which do not form part of the program itself but are stored for access by any program running on the computer in a DLL. There are two established means for accessing such program elements: by name, or by numerical identifier (“ordinal”). The latter is more efficient in terms of computer resource usage since the calling program goes direct to the location in the DLL where the required ordinal is found, but the former uses an interface and lookup table which is easily updated, and so avoids errors which may occur if the elements are updated or the library expanded by incorporation of additional functionality. If this happens, the location of any given element in the library may be modified. The invention is to divide the DLL into two parts, one of which comprises functions with ordinal access whose locations cannot be changed, the other of functions which may be subject to third party modification or relocation. These are then accessed by a mapping interface. The result is increased efficiency without the increased risk of errors.
In evidence, the UKIPO explained that their approach to the Aerotel/Macrossan four step test was to focus on the answering of the second question, since in their view the answers to the remaining questions would fall out naturally once this issue had been resolved. However, the judge was not satisfied that this is a safe assumption to make. In particular, a narrow focus on identifying a particular contribution risks failing to ask the question as to whether the invention makes a relevant, technical contribution to the art. In Patten J’s view, it was relatively unimportant whether this question is considered at stage 2, 3 or 4 of the four step test – but it is critical that it be expressly addressed at some point in the process. This, the UKIPO appeared not to have done. Rather, their conclusion had been reached essentially upon the basis that the invention was not one which changed the role of the DLL in the operation of the computer but only in the manner in which it is accessed. The question of technical contribution was nowhere explicitly considered.
In the judge’s view, a contribution such as that of the invention to make the overall operation of the computer more robust was sufficient to amount to a technical contribution so that the means of achieving it, through software, was not excluded from patentability. In reaching this conclusion, the judge considered briefly Lewison J’s analysis in a case decided only immediately before Symbian, Autonomy Corp’s Application ( EWHC 146 (Pat)). Lewison J declined to find patentable an improved computeruser interface on the basis it did not require
“any new hardware or arrangement of hardware, did not fix any perceived technical shortcoming in the computer itself, and was purely concerned with the processing of data… done and done only by a computer program.”
This, Patten J felt, encapsulated the issue: the existing mechanisms for updating the DLL did suffer from perceived technical shortcomings, and Symbian’s invention alleviated these. Accordingly, although it consisted of a software fix for a software problem, Symbian’s invention was patentable where Autonomy’s was not.
Where are we now?
For practitioners, then, there is some hope of consistency between Patten J and Lewison J’s approaches, and even that of Floyd J, on the one hand, and the Court of Appeal on the other. It is not the end of the story however. The UKIPO did not appeal Astron Clinica but is appealing Symbian, preferring its own (and Kitchin J’s) former interpretation of Aerotel/Macrossan. This will give Jacob LJ a further chance to clarify what he meant. But ultimately, the controversy can only be decided by the House of Lords who, in the 30 or so years since software patents came into dispute, have managed never yet to have the pleasure of hearing a single case argued on the subject.