Originial: http://dialspace.dial.pipex.com/town/street/xge21/software.htm

Fry Heath & Spence

Patenting of Computer Software 

A view from the UK & EPO

 The aim of this article is to give a whistle-stop tour of the law relating to the patenting of software-related inventions as it applies in the UK and in the European Patent Office.

 There is a general impression amongst non-European attorneys (and some European attorneys) that software is unpatentable in Europe. We will start by considering the root cause of this opinion and move on to a selection of reported cases to see whether or not this feeling is justified. By the time you have finished this article, you should be in a position to make up your own mind.

 The Statutes

 The UK Patents Act 1977 and the European Patent Convention are the relevant Statutes governing patentability in the UK and the EPO. Insofar as they apply to software inventions, both use equivalent wording, as follows:

  "It is hereby declared that the following (among other things) are not inventions for the purposes of this Act, that is to say, anything which consists of - (a) a discovery, scientific theory or mathematical method;

(b) a literary, dramatic, musical or artistic work or any other aesthetic creation whatsoever;

(c) a scheme, rule or method for performing a mental act, playing a game or doing business, or a program for a computer;

(d) the presentation of information ...;"

[UK Patents Act 1977 Section 1(2)]

 The equivalent wording of the EPC leads many to think that the standards of patentability imposed in the UK and EPO are identical. This is by no means the case at present, although UK Courts are generally seeking to give decisions which at least do not conflict with EPO Board of Appeal decisions. As matters stand, the EPO is generally considered to be more lenient in its treatment of software inventions than the UK.

 The basic approach of Examiners is however visible from this wording. A Patent cannot be granted for a software invention, unless for some reason it can be exempted from the general prohibition. UK and European Patent Office case law is clear that it is up to an Examiner to show that the invention is unpatentable, but the mindset is still that software inventions are unpatentable unless the applicant can show patentability.

 There is however one important exception to this. Section 1(2) of the UK Act includes a proviso which reads as follows:

  "... but the foregoing provision shall prevent anything from being treated as an invention for the purposes of this Act only to the extent that a patent or application for a patent relates to that thing as such."

[UK Patents Act 1977 Section 1(2) cont/d]

(emphasis added)

 This exception also appears as paragraph (3) of Article 52 EPC. This exception is very important in practice and has been the subject of a gradually widening interpretation since it was brought into force. Thus, an Application that relates somehow to more than a computer program as such is allowable provided that it meets the usual requirements of novelty etc.

 The terminology "as such" is clearly vague and somewhat elastic. Much of the difficulty in interpretation in the law in this field stems from this. The last fifteen years of case law has been directed purely to establishing exactly what "as such" means and attempting to replace it with a test which is clear and straightforward to interpret.

 It is therefore clear that if we are to obtain grant of a Patent for a software-related invention we need somehow to escape from the apparent prohibition. There are essentially two ways in which we can do this. First, we can side-step the entire problem. If the invention resides in an item of software, but the use of the invention inevitably involves a physical thing in which we can claim novelty and inventive step, we can simply apply to patent that physical thing. The other alternative is to argue that the invention is not a computer program as such and is therefore not excluded from patentability.

 Claiming the Novel Article

 A good example of this approach comes from a UK case called Lux -v- Pike (1993 RPC 107).

The inventions were methods of co-ordinating traffic streams so as to obtain a smoother traffic flow. Two Patents were involved, for two different ideas. The first idea was that in certain circumstances, a traffic stream which had just been stopped was given another brief flash of a green light. In the other idea, the period between one green phase stopping and the other being started was varied according to certain conditions. One or other of these responses could be triggered in unusual circumstances, such as a vehicle stalling or stopping on the junction.

The idea was clearly clever, and gave measurably beneficial effects on the traffic flow. However, the basic concept as such is not patentable since it is clearly a computer program, or at the very least a mental act. However, the claim that was presented to the Patent Office was essentially a traffic light system comprising a central controller, a plurality of light signals, wherein the central controller causes the signals to light in a predetermined sequence, etc. etc. This Patent was granted, and litigated successfully.

Thus, presenting the claim:

"A traffic light timing method wherein ..."

would result in refusal of the Application or invalidity after grant, whilst the claim:

"A traffic light control system comprising a central processor controlling a plurality of lights wherein ..."

resulted in grant of a Patent and a successful infringement action.

This approach will often occur naturally, without any particular thought. Despite this, it is a powerful approach which is well worth considering in a problematic case.

The difficulty in taking this approach before the UK and EPO, at least for a US applicant, is that it is totally impossible to add matter to an Application unless that matter is present in the original text. The attitude of both Offices on this point is strict by US standards, particularly so in the EPO. The only solution to this is to allow the local patent attorney time to review the text as filed in the US and propose pre-filing amendments. In many cases, it is sufficient simply to add a new claim which can then be used as basis. There really is no alternative to this; if the agent receives the case immediately before the Paris Convention deadline then he will simply not have time to review it in any depth and will have to file it as it stands.

The side effect to this is that the PCT route is not ideal for this kind of invention. It is inherent in the PCT route that by the time the local attorney receives the text, the text as filed has already been fixed and cannot be extended. If for some reason a PCT filing is essential, it is advisable to send the text to the British or European attorney before making the PCT filing. He or she can then be invited to propose additional claims or description which may be useful later in the National or Regional phase.

 The "As Such" Exception

 The other alternative is to argue that the invention is not in fact excluded from patentability. The test that has been laid down in decided cases is referred to as the "Technical Effect" test. This is a non-statutory definition adopted initially by the EPO in order to separate unpatentable ideas in general from patentable ideas. It has since been accepted (with reservations) by the UK Courts. The argument of the EPO was that all of the various exclusions from patentability share a single common factor; they do not have a technical character. Therefore, so the argument goes, an invention which has technical character is patentable. The logical non sequitur in this reasoning has never fully been explored, but is now probably lost in the mists of time.

 According to the EPO, technical character is present if there is a technical effect. A technical effect is present if there is a technical problem which is somehow solved by the use of technical means. This is all of course somewhat circular, which is why the Technical Effect test is not greatly liked. However, the test is in force and routinely applied by the UK and European Patent Offices and a response to an objection of lack of patentability does need to be presented within the framework of the Technical Effect test. This runs up against the difficulty that the Technical Effect test does not boil down to a question of fact upon which we can consult some guiding authority or expert such as the inventor or a local professor. It is purely a matter of law and argument.

 It is therefore essential to select a local attorney who understands the test and can argue coherently. It has been suggested, somewhat less than seriously, that lack of clarity inherent in the Technical Effect test means that grant of a Patent is more likely to be decided by the inventiveness of the attorney rather than the inventiveness of the idea.

 The only real way to develop a feel for the Technical Effect test is to look at the case law, which is therefore the logical next step on our tour.

 EPO Case Law

 This is the obvious place to start, as it was the EPO that developed the test in the first place.

 The original EPO guidelines were very straightforward on the subject of software patents. They noted that the EPC did regard a computer program as such as a patentable invention, and therefore laid down that examination of such Applications simply required the Examiner to ignore any term in the claim which appeared to be software and decide the claim on the basis of the remaining features. By definition a software invention involves novel software running on known equipment. Therefore, the decision reached by Examiners in such cases was relatively easy to predict.

 This approach fundamentally changed after the Vicom Decision (T208/84 - OJ 1987, 14). Vicom established the Technical Effect test, and explicitly stated that the software could provide the inventive step, provided that a technical effect existed which lay in a non-excluded area. The case itself related to a digital image processing method. Data representing an image was captured and supplied to a computer program. The program then applied a novel algorithm to the data which filtered it to reveal the image in a form superior to that originally supplied. There was a prior art algorithm, but the claimed algorithm was much faster.

 When the Application was filed, the claims were drafted in the form of a method and apparatus for data handling. These claims were refused by the Examining Division, and the applicants appealed. The Board of Appeal suggested to the applicants that they may wish to file amended claims relating to a method and apparatus for digitally processing an image. The Board of Appeal then held that these claims did not relate to a computer program as such. The Application was then referred back to the Examining Division, but was never granted. Such is life.

 The crucial feature in the Board of Appeal Decision was that the amended claims no longer related to a process of handling numbers in the abstract. Instead, the numbers were a real electronic representation of a real world image. According to the invention, a process was being carried out which produced a change in that real representation. That process was faster than before, which was clearly a useful effect. Thus, an effect was imparted on a technical thing (a detectable electronic signal) in a technical way (by a computer). Therefore, there was a Technical Effect and the claim was allowable.

 The explanation that the Board gave was that even if the idea underlying an invention is a mathematical process (as in this case) a claim directed to a technical process using that method was still patentable. This was a clear over-ruling of the old practice.

 This reasoning can be illustrated as follows:

A wide claim which is limited only to the mathematical process or program steps will be refused. If the claim is narrowed so as to include some aspects of the real world then it is liable to be allowed. This is because the wider claim is directed to the mathematical method as such. However, as soon as it is narrowed then the claim relates to the application of that mathematical process to a particular real world context. Therefore, the effect of that mathematical process can be discerned on a claimed physical item, and provided that effect is of a technical nature then the claim should be allowable.

 The next case of interest to us is Bosch (T164/92 OJ 1995/305). This invention was a monitoring circuit for a computer. When the computer started, a watchdog circuit looked at selected values in the volatile memory store (RAM) and then compared them with standards that it maintained in a non-volatile memory. If there was a substantial correspondence, the circuit decided that the computer had just reset itself rather than having been turned on from scratch. This corresponds to the difference between switching on the power to computer and resetting it through software such as via the "CRTL/ALT/DEL" routine. This decision by the watchdog circuit was fed to the processor, but the claim did not elaborate on what the processor did with this information.

 The value of the invention was that the processor could skip reloading the various application programs when a software reset was carried out. This meant that the recovery of the computer to a software fault was significantly quicker, for obvious reasons. It was not however a term of the claim that the decision was actually used, merely a term of the claim that the decision was available to be used.

 The Board of Appeal held that this was patentable on the grounds that there was indeed a contribution to the art in a non-excluded field. The contribution was the better response time that was permitted by the invention. The physical transfer of files and data within a computer is clearly a technical problem, partly since obvious alternatives might be a faster hard disk or a decreased access time to memory. The Board of Appeal held that patentability was satisfied purely by the decision being communicated by the microprocessor. They explicitly stated that in the absence of that communication step, the claim would have been excluded.

 As an aside, the Board of Appeal in Bosch also decided that program listings were valid prior art citations. Before then, EPO practice had been to ignore such prior art on the grounds that they were clearly written in gibberish and nobody could understand them. The Board held that there were people who could understand the gibberish and that this was perfectly sufficient for the matter to be made available to those skilled in the art.

 The last EPO case of particular interest is the Sohei Application (T769/92 - OJ 1995, 525). In this case, the invention related to the implementation of a financial management method on a computer. The claim presented to the Board of Appeal was extremely long and involved, but essentially comprised definitions of internal system files and processing steps carried out on these files. The files on the processing steps were defined in terms of financial management. They enabled the computer to display a "transfer slip", which was of a conventional appearance and corresponded to a form of the same name apparently used to send and receive information in the context of a particular financial management method.

 The Examining Division argued that the invention was excluded because it was a method of doing business as such. Their reasoning was based mainly on the fact that the claims were drafted in terms which had a meaning only as part of a financial management system. The applicant argued that the claim would have been patentable if it had not been limited to financial management, and that a voluntary limitation of the scope by the applicant to one field in particular did not make an otherwise patentable claim unpatentable. This of course begs the question of why the claim was limited in this way. We cannot of course know for certain, but two options seem likely. It may have been that the local attorney received the Application in that form and did not have time to review or was not authorised to do so. The other possibility is simply that it is impossible to draft a clear and meaningful claim without some framework for the various file handling steps.

 The claim was held to be patentable by the Board of Appeal, despite the financial management context. The Board relied on existing case law which stated that a mix of technical and non-technical features was patentable as a whole if the technical features were otherwise sufficient to ensure patentability. Essentially, in doing so, they were agreeing with the applicant's argument.

 The Board decided that the technical features of the claim were sufficient because there were significant difficulties in computerising the financial management methods referred to in the Application. These difficulties are inherently technical in character and therefore the invention addressed a technical problem. The claimed invention solved those problems by virtue of its file structure and was therefore a technical solution. As a result, the Board held the invention to be patentable.

 These cases enable us to distil some examples of a technical effect. In the Vicom case, there was an observable effect on a technical entity which was caused by a claimed item, despite the item itself being excluded from patentability. In the Bosch and Sohei cases the invention overcame a technical problem through the use of a software solution.

 UK Case Law

 The EPO examines and grants Patents on the basis of the principles set out above. However, if the granted Patent is to be enforced against an infringer, it must be litigated in the National Courts. Their view of the validity of software Patents is therefore highly relevant.

 In the UK, there is a significant body of case law as to the meaning of the "as such" exclusion. Most relate to appeals from a Decision of the Patent Office refusing an Application, with the effect that the inventions concerned are all either unpatentable of dubious at best.

 It is important to bear in mind the level at which a precedent was decided when considering its effect. A Decision of the Patent Office gives a likely indication of the manner in which a National GB Application is likely to be considered. Such a precedent does not however have any force in the Courts, and will therefore have no effect on the consideration of a Patent after it has been granted by the EPO. A Court precedent, on the other hand, is binding on the Patent Office and is therefore relevant to a National GB Application. A Court of Appeal precedent will also be binding on the High Court when considering litigation, whilst a High Court precedent will be of relevance (but not binding) in later High Court litigation.

 The first and foremost case is Merrill Lynch (Court of Appeal: 1989 RPC 561). This was an Appeal by the New York securities firm Merrill Lynch Pierce Fenner & Smith, Inc. against a Patent Office refusal of their Application. The invention was an automated securities trading system. It was implemented on a suitably programmed computer which received buy and sell orders for the securities that they held details of. The computer then matched the orders where possible and selected for execution those orders which qualified on the basis of certain preconditions. The computer therefore established an operative market in the securities.

 This was held to be unpatentable by the Court of Appeal. In doing so, they specifically rejected the argument that computer programs and so on should be ignored for the purpose of examining novelty and inventive step, and therefore endorsed the decision in Vicom to allow the inventive step to reside in an excluded item. As part of their reasoning, they pointed out that a discovery as such is an excluded item, and that if the application of a discovered pharmaceutical needed to be independently patentable then this would be a major blow to this field of patenting.

 The correct test, according to the Court of Appeal, was that there must be a technical contribution to the art. Thus, one identifies the "contribution" made by the invention, and if this contribution is an excluded item such as a computer program then the Application should be refused. In the Merrill Lynch case, the contribution identified was a method of trading securities, i.e. an excluded business method.

 The Application was formally remitted back to the Patent Office, where the applicants limited the claims to define more specifically the manner in which electronic trading took place. This resulted in the grant of UK Patent No. GB 2 180 380 B. The reasoning for grant of this Patent was probably similar to that later shown by the EPO in the Sohei case.

 The Raytheon case (High Court; 1993 RPC 47) was a slightly unfortunate step in the development of UK case law on the subject. It concerned a system for automatically recognising outline patterns such as those of ships. The value of such a system is obvious, for example as an immediate friend or foe identification in war.

 According to the invention, the outline image of the object was captured in digital form. Staying with our example of a ship, the sample outline is;

 The computer then used this outline to generate a standard number of height vectors uniformly distributed along the outline, thus;

 The outline of the image was then removed leaving only the vectors;

 Finally, the vectors were scaled so that their spacing was set to a standard spacing and the height was normalised to a standard total; i.e.

 The result of this processing was that the only information left in the image was representative of the shape, rather than some combination of size, distance, aspect and so on. This image was then compared with a database of similarly processed images and the closest match found. The computer would then suggest the closest match as an identification.

 Somewhat surprisingly, this was held to be unpatentable using the "mental act" exclusion. The judge's reasoning was that when a person views an object they instinctively correct for size, distance etc. and compare this corrected version with images they have seen before. They may well carry out the correction in the same way. It is certainly true that they could carry out the correction in the same way if they wished. Therefore, the invention was a method which was equivalent to a mental act and which could be carried out by a human mind. Hence it is unpatentable.

 The precedent which has been set is somewhat unfortunate. The Patent Office has been forced to take the stance that an invention which consists of steps which could be carried out in the same way by a human mind are unpatentable. Note the qualification "could"; this means that the decision bites even if the human mind would be physically incapable of carrying out the steps within a reasonable time-frame. Taken to its logical limits, this precedent could be used to refuse any Application in which the invention lies in the software. By definition, the software has either been written by a human or written by a program which was written by a human. Therefore, one or more human minds have succeeded in generating these program steps. This means that the program steps could be carried out by a human mind equipped with a suitably large piece of paper and pencil for recording the system variables.

 The Decision clearly rules out of patentability many inventions which other case law has held to be patentable. It is however binding on the UK Patent Office and will be applied unless the agent can distinguish it in some way.

 The next case is Fujitsu (High Court: 1996 RPC 511). I understand that the Court of Appeal has now reviewed this case, and although the Decision has not yet been published I believe that the High Court Decision has been approved.

 It is common for materials scientists to propose novel crystal structures by modelling two known structures and superimposing them. This gives a first indication of whether the structure is likely to be viable and useful. The Fujitsu Application related to a method of automating this. A computer was programmed to accept a definition of a first and a second crystal structure, and the user then pinpointed an atom, lattice vector and lattice plane in each structure. The computer then projected an image of the two structures superimposed.

 The Application could of course have been refused on the grounds that the mere automation of a known process is intrinsically obvious. However, the case went to the High Court on the subject of inherent patentability. The Court upheld the Patent Office decision to refuse the Application. It explained that a computer was simply a tool for carrying out technical and other processes. Hence, to determine patentability one should not concentrate on the fact that a computer is involved, but direct attention to what the computer was doing. This of course follows neatly the question set in Merrill Lynch. This could be illustrated simply by considering the hypothetical example of a computer controlling a chemical plant. In that case, the Application would be considered as any other chemical process application. In the Fujitsu case, however, the computer is replicating a mental act. Hence, the contribution was simply the automation of a mental act and was therefore excluded from patentability.

 The last case regarding patentability which I wish to discuss is FMC Corporation's Application (Patent Office: 14 December 1994, O/87/95). This Application related to the computerised analysis of multi-bodied systems. The problem with such systems is that while a simple two-body system can be analysed easily, once more than two bodies are involved the mathematics becomes theoretically unsolvable. It is possible to generate numeric solutions of varying accuracy, but the arithmetic involved is extremely lengthy. According to the Application, the user inputted details of the system and its starting configuration. The computer devised equations of motion for the system and converted these into executable source code. It then allowed the user to run the code, and the result was displayed.

 The Application was refused by the Patent Office. The Principal Examiner involved felt that the Application dealt entirely with abstract numbers that were not directly linked to a real physical object, as in Vicom. Instead, they were simply abstract numbers that we chose to label as a force or position etc, but this label was merely our choice rather than the result of interaction with the real world. It therefore did not help patentability.

 There is one very interesting paragraph in the Decision, which provides my reason for citing the case. The Examiner stated that:

  "While I admire the efforts [the agent] has made during the prosecution of these Applications to introduce a real technical process into the claims, his difficulty it seems to me has been that the specifications stop short at the point where the information is produced and displayed to an operator, for example in the form of an animation"

The implication is clear; if the specification had provided basis for further claimed limitations then it might have been possible to grant a Patent. These limitations would of course have had to provide a technical effect in the style of Vicom, such as a use to which the information was put, or an article that has been made using the information. This limitation would have served to give a real technical effect in the form of a direct observable effect on a real world object. This would be exactly analogous to the claim allowed in Bosch where the final claimed step was to feed the information to a processor chip.

This case reinforces the need to supply the text to the local agents early and to authorise them to propose amendments. If the basis had been in that text then a Patent might have been granted, albeit with a slightly limited scope. Once the Application is filed in the UK or EPO, no matter can be added and no continuation in part can be filed.

 Patent Enforcement

It is clear from the above that the claim cannot relate to the software as such, since this would result in invalidity. The closest the claim is going to approach such a broad scope will be a reference to a system incorporating various hardware elements and software of a particular function and effect.

The effect of such a claim is illustrated neatly in the case of Quantel -v- Spaceward (1990 RPC 83). This case is theoretically of only passing interest to the question of patentability, as the "software as such" issue was not actually raised or commented on in the Decision. However, the relevance of the case is that the Patent was for a computer-aided design (CAD) system which stored details of images and displayed them. The patented system was an acknowledged market leader, and the litigation was directed against a rival CAD system. The particulars of infringement related to supply of the rival software as such. The Patent was enforced successfully, and an injunction was granted which restrained sale of the software itself. Whilst the claim did not define a computer program as such, the supply of the computer program was held to be the supply of a means relating to an essential element of the claim and therefore an indirect infringement.

The case therefore demonstrates that despite the words in the Statute, one can in fact protect software as such by virtue of a Patent. If this is not tantamount to patenting software as such, then I am not sure what is.

 Future Developments

An article such as this could not be complete without an indication of possible future trends. There is scope for significant adjustments to UK law on the subject, by virtue of Section 1(5) of the UK Patents Act 1977, which reads: The Secretary of State may by order vary the provisions of sub-section (2) above for the purpose of maintaining them in conformity with developments in science and technology ....

Such an Order does need to be approved by Parliament, but the law clearly envisages that such changes may be needed and has therefore provided for an easy legislative means to do so. No equivalent comment appears in the European Patent Convention, and a diplomatic conference would therefore need to be held in order to confirm the agreement of the member States of the EPO.

The mostly likely impetus to change is the GATT/TRIPS Agreement. This requires Patents to be granted in all fields of technology. There are voices which say that this requires the ban on patenting computer software to be lifted, but others say this depends on whether software is a technical field. This does of course rather beg the question.

The UK Patent Office was sufficiently concerned at this to hold a Discussion Conference in October 1994. Views were sought from industry, patent agents, and solicitors. The response was mixed; most of the legal profession were strongly in favour of allowing patents for software in general, as were a sizeable proportion of the industry. Some industry representatives were however strongly against software patents, on the grounds that large software firms might thereby acquire yet another powerful tactic for squashing small ones.

The "Technical Effect" test was roundly criticised by both the patent profession and certain staff within the UK Patent Office. Criticism was directed essentially at the difficulty in interpreting the test and the fact that it is not much more clear than the Statute which it seeks to clarify.

It was suggested that simple deletion of the reference to "a program for a computer" in the list of excluded matter would be a suitable way to proceed. This would still leave an exclusion of "literary works", which copyright legislation deems to include computer software. The result of this would be that the listing would be unpatentable but protectable through copyright, whilst any inventive method embodied in the program would be patentable. This suggestion is however yet to be taken up by any official body.

A conference was held in London during the recent UK Presidency of the European Union, specifically to invite views as to whether, and if so how, these provisions should be amended. So far no action has been taken or warned of by the UK Government or the EPO.

 Conclusions

 It seems clear that the scope for patenting software is very much wider than the bare statutory wording suggests. In fact, it is sometimes possible to obtain effective protection for the software as such. Care does however need to be taken in preparing the text prior to filing, and a pre-filing amendment by the local attorney may make the difference between grant and refusal. It is notable that many of the instances of successful software patent litigation have been by UK-based companies whose Applications were therefore drafted with the law and practice on this point in mind. Many of the refused cases emanated from the US or Japan where the Application was probably not drafted in this fashion.

 Michael Downing

Chartered Patent Attorney - European Patent Attorney

Fry Heath & Spence

 

Back to article index

 

Back to Home Page


Kopiervermerk:
We do not mind them [the articles] being distributed, provided that we are acknowledged as the source.

Fry, Heath & Spence