Thoughts on EPO G3/08

A current (June/2009) referral – Patentability of programs for computers, No G 3/08, by the European Patent Office to the Enlarged Board of Appeal – asks submitters of amicus briefs to answer the following questions:

  1. Can a computer program only be excluded as a computer program as such if it is explicity claimed as a computer program?
    1. Can a claim in the area of computer programs avoid exclusion under Art. 52(2)(c) and (3) merely by explicity mentioning the use of a computer or a computer-readable data storage medium?
    2. If question 2(a) is answered in the negative, is a further technical effect necessary to avoid exclusion, said effect going beyond those effects inherent in the use of a computer or data storage medium to respectively execute or store a computer program?
    1. Must a claimed feature cause a technical effect on a physical entity in the real world in order to contribute to the technical character of the claim?
    2. If question 3(a) is answered in the positive, is it sufficient that the physical entity be an unspecified computer?
    3. If question 3(a) is answered in the negative, can features contribute to the technical character of the claim if the only effects to which they contribute are independent of any particular hardware that may be used?
    1. Does the activity of programming a computer necessarily involve technical considerations?
    2. If question 4(a) is answered in the positive, do all features resulting from programming thus contribute to the technical character of a claim?
    3. If question 4(a) is answered in the negative, can features resulting from programming contribute to the technical character of a claim only when they contribute to a further technical effect when the program is executed?


These questions are an attempt to address issues with current interpretation of Article 52 of the European Patent Convention (EPC), which states:

  1. European patents shall be granted for any inventions, in all fields of technology, provided that they are new, involve an inventive step and are susceptible of industrial application.
  2. The following in particular shall not be regarded as inventions within the meaning of paragraph 1:
    1. discoveries, scientific theories and mathematical methods;
    2. aesthetic creations;
    3. schemes, rules and methods of performing mental acts, playing games or doing business, and programs for computers;
    4. presentations of information.
  3. Paragraph 2 shall exclude the patentability of the subject-matter or activities referred to therein only to the extent to which a European patent relates to such subject-matter or activities as such.

One could reasonably argue (as Red Hat does) that this is entirely unambiguous: you can't patent computer programs in Europe. Some confusion has been inserted into the situation by some past interpretations of the as such language but I would (with Red Hat) contend that the intent of paragraph 3 is merely to assert that a patent claim need not be excluded simply because it discusses or involves a computer program, even one essential to making the claimed invention work, so long as it does not claim the program itself as part of the patented subject matter. So long as what is claimed is not, in itself, excluded by paragraph 2 and does meet the requirements of paragraph 1, it is patent-eligible. As I elaborate below, this straightforward approach to Article 52 resolves all of the problems by the simple expedient of treating paragraph 2's exclusions in the same manner as the exclusions implied by paragraph 1's requirements that the invention be new, involve an inventive step and [be] susceptible of industrial application.

One major wrinkle in these questions is the problematic notion of technical effect, which appears to be a term of art among patent lawyers in Europe, related to paragraph 1's susceptible of industrial application requirement. Since, in practice, questions of whether something has technical effect shall be decided by lay juries (in at least some jurisdictions) it is quite certain that computer programs shall be found to be technical in at least some litigation, whatever the patent lawyers may think it means. It appears to be a component of patent eligibility and, as such, should be irrelevant to the discussion: computer programs are not patentable. However, patent lawyers endeavour to argue that involving a computer program in an invention, with other components, renders the whole technical (because the program is) even if the other components are not. Thus ordering flowers to be delivered to someone far away becomes technical if it is done using a computer program, where it was not if you merely did it over the telephone. As explained below, I consider this reasoning flawed.

I'm also inclined to criticise the questions for trying to discuss patent claims in isolation. This is a weakness: in practice, to make any sense of patent claims, one must ask what would be covered by the claim: if the patent is granted, what law-suits does it allow its holder to bring ? If I practice some useful art (for example, if I write a computer program and run it on my general-purpose computer), can the patent-holder sue me on the basis of my doing so ? These are the questions that actually matter when considering a patent claim: no examination of the meaning of patent claims can carry any value unless it addresses itself to these questions. Article 52 should, thus, be understood as implying that, if I have written a computer program and run it on a general-purpose computer (which, for present purposes, we may suppose to consist entirely of prior art and subject matter obvious to those normally proficient in relevant arts), then I do not infringe any claim of any patent.

Treat programs like prior art and obvious subject matter

The straigthforward treatment I advocate above is simply the same as the normal treatment of things that aren't patentable by virtue of having previously been described by others or by virtue of being obvious. A patent claim may freely intermix such things with the claimed invention so long as it does not purport to claim the obvious or previously known subject matter. I contend that paragraph 3 is merely saying we should apply the same to the subject matter excluded by paragraph 2.

A patent application is supposed to describe the claimed invention in sufficient detail that other folk working in the relevant industry can reasonably be expected to be able, from the text of the patent application, to implement the invention claimed. In order to meet this requirement, the application must, inter alia, recite a fair amount of subject matter that would be obvious to or previously known by those who work in relevant industries. The inclusion of such matter in the patent claim neither impedes the claim's eligibility for patentability nor, if the patent is granted, grants the patent-holder any control over how others may use such non-patentable subject matter as the patent incidentally involves.

The plain intent of EPC Article 52, paragraph 3, above, is to apply the same rule to computer programs. A patent application may, in order to adequately explain what it is claiming, need to give details of the operation of a computer program needed in order to make the invention operate as claimed, in the same way that patent claims have long been wont to include instructions for how a human operator is to make an invention operate as claimed. The as such language of paragraph 3 serves to allow the claim to give these details without prejudice to the patentability of the other subject matter discussed by the claim, in the same way that a claim's patentability is not impeded by discussion of prior art and things that are obvious. Like these last, any computer program described as part of a claim is understood as supporting material, not itself covered by the patent (if granted), but necessary in order to properly describe the claimed invention.

This approach folds computer programs (and the other excluded subject matter) into the existing framework already used to deal with the (more or less inevitable) presence of prior art (that is, what relevant folk already knew) and obvious subject matter in analysing patent claims. Roughly speaking, this means we should, in assessing patent-eligibility of a claim, set aside the effects that arise simply from the prior art, obvious subject matter and excluded subject matter. The ways these effects interact with other subject matter may, all the same, be included – although it may, in its turn, be obvious, previously known or excluded; likewise for the ways the claim combines its assorted subject matter; and for any indirect or secondary effects of the excluded subject matter, arising from the way they are so combined. This is not to say the examiner should ignore the excluded parts – a proper understanding of the invention claimed and how it works typically depends on them, else the claim would not have mentioned them – only that they do not contribute to the analysis of the claim's patent-eligibility. That analysis looks only at what remains: if there is nothing (not even a novel, non-obvious way of combining the other subject matter) then the claim fails.

Leaving out the excluded parts from eligibility analysis thus requires what remains to be patent-eligible subject matter that is non-obvious, novel and technical (whatever that may mean). It is well settled that combining prior art (which might not be obvious) with someting obvious (but novel) doesn't yield a patentable invention: the applicant can't point to the novelty of the obvious part and the non-obviousness of the prior art and claim that the whole is novel and non-obvious. It does not matter whether the prior art was obvious: it is excluded because it is prior art. It does not matter whether the obvious subject matter is novel: it is obvious, so excluded. Likewise, if we include a computer program in the mix, questions of whether the computer program is novel, obvious or technical are irrelevant: the program is excluded because it is software as such.

We may take this further: the program is a program, hence excluded. Running it on a computer is the most obvious thing to do with it and, in any case, by no means novel. Unless the patent is making claims about a non-obviously novel form of computer, we may take it as given that the computer on which our program runs is non-novel. Thus there is nothing patent-eligible in running any computer program on an ordinary computer. In particular, all the things that programs do when run on ordinary computers are inescapably excluded from patent eligibility: the effects of the program, as a program, are thus excluded as an inevitable extension of excluding the program itself.

Computers have long had hardware interfaces that can be used to control the voltages of different parts of connectors; and these interfaces may be driven by software, typically via reasonably standardised software interfaces. As such, having a computer program produce arbitrary patterns of variation in the voltages on such a connector must surely be excluded. However, an invention may involve novel hardware plugged in to those connectors, or to novel connectors: the computer then serves as an intermediary between users and (possibly several) such devices; programs serve to choreograph appropriate interactions and it is the results of these interactions that contain the potential for innovation outside the computer. Having the right program to do that choreography may be essential to making the invention work; and understanding what the program does may be essential to understanding how the invention works; but the patent-eligible subject matter is only the peripherals and the modes of interaction in which (thanks to the non-patentable control provided via the computer) they engage.

This is, indeed, a natural consequence of the separation of function that has lead to the existence of general-purpose computers in the first place: peripheral devices specialise in handling particular tasks while the general business of computing and mediating user interaction is concentrated in the computer. Standardised interface specifications between the computer and its peripherals make the resulting design modular, with all the benefits that flow from this. The manuracturer of peripherals need only make the peripheral work properly via the standard interfaces to make it usable by (and thus sellable to) all computer users, regardless of their operating system or, in most cases, their preferred hardware. An operating system can integrate its support for peripheral interfaces with the general scheme of how it does everything else, so as to make interaction with peripherals painless and intuitive to users by virtue of following common idioms. The results are generally good for users by stimulating competition and innovation in the markets for peripherals and for general-purpose computers.

Because the general-purpose computer is a standardised piece of equipment with a well-defined (albeit immensely diverse) range of potential behaviours, the greater part of its behaviour – in particular, the whole panoply of things that programs may direct general purpose computers to do – is a mere unfolding of the ramifications of the standards specifying what it can do. While programming involves great ingenuity, it is the ingenuity of a mathematician, not that of a mechanic.

Answers to the EPO's questions

  1. Can a computer program only be excluded as a computer program as such if it is explicity claimed as a computer program? No. If the claimed invention is, in fact, a computer program then it should not matter whether the applicant actually describes it as such. If a claimed invention is susceptible of instantiation as something which is patent-eligible subject matter, then it need not be excluded merely because it is also susceptible of instantiation as a computer program: however, in such a case, it must be understood that the patent, if granted, does not cover any computer program which practices the claimed invention.
    1. Can a claim in the area of computer programs avoid exclusion under Art. 52(2)(c) and (3) merely by explicity mentioning the use of a computer or a computer-readable data storage medium? No, in exactly the same sense and for exactly the same reasons as a claim to prior art or obvious subject matter cannot be made patent-eligible merely by mentioning some other subject matter, not even if that subject matter is novel and non-obvious (which is more than can be said of using a computer or its storage media). Such added subject matter may, indeed, be patent-eligible in its own right or even in the way that it exploits the effects of the computer program or prior art: all the same, a patent covering this added subject matter does not cover the prior art or computer program it involves (although it may cover the ways it exploits them).
    2. I have indeed answered that in the negative, so: … is a further technical effect necessary to avoid exclusion, said effect going beyond those effects inherent in the use of a computer or data storage medium to respectively execute or store a computer program? All discussion of avoiding exclusion is fruitless and distracting. The software itself is excluded and no application can avoid this exclusion; but a claim's involvement of the software need not cause the patent-eligible subject matter of the claim to be excluded – just as prior art (such as executing a program on a computer or storing it on some medium) cannot avoid exclusion – although a claim's discussion of it need not prejudice the eligibility of the claim, provided it contains patent-eligible subject matter which is novel, non-obvious and not excluded by paragraph 2. Such subject matter must, naturally, go beyond the effects inherent in such prior art and excluded subject matter as may be involved.
    1. Must a claimed feature cause a technical effect on a physical entity in the real world in order to contribute to the technical character of the claim? This question is wholly misguided. First, because the purported constraint it asks to impose is inadequately specified and fatuous in so far as it is specified – computers are physical entities in the real world and the effect on them of running a program may surely be called technical, although the question seems to be using a legal-technical meaning of the word technical which does not appear to have been adequately delineated. Second, because the features of a program are, by being natural consequences of the obvious use of excluded subject matter, not patent-eligible, nor can they be made patent-eligible by combining the program with other subject matter. A claim covers subject matter which is patent-eligible, non-obvious and novel: if its realisation happens to involve the deployment of obvious subject matter, any novelty and eligibility that the obvious subject matter might have enjoyed (had it not been obvious) does not contribute, in itself, to the novelty or eligibility of the claim; any non-obviousness or eligibility that any non-novel aspects of its realisation might have enjoyed (had they been novel) likewise do not contribute to the non-obviousness or eligibility of the claim; likewise, any ineligible subject matter, such as a computer program, cannot contribute to the novelty or non-obviousness of the claim. If the claim, when its obvious, non-novel and excluded parts are (in themselves) left out of consideration, lacks technical character, then the claim must fail; and, when a claim is granted, it does not cover the effects of any ineligible, obvious or non-novel subject matter it happens to incorporate, although it may cover the ways these effects interact with one another and with other subject matter of the claim, in so far as such interaction is itself eligible, novel and non-obvious.
    2. Had I answered that in the positive: … is it sufficient that the physical entity be an unspecified computer? An unspecified computer is, tacitly, a non-novel device; combining a computer program with one is an obvious mode of combination. There is thus nothing left of such a claim when its excluded program, non-novel computer and obvious mode of combination are all left out of account. The case would be otherwise with a specific computer: for example, one with some novel response to having particular data stored at some specific address. A program may save the right data to that address in memory in response to some given input: such is intrinsic to the excluded program and so (in itself) excluded; but, when the program is run on the specific hardware, then the program may play a useful rôle in causing an operator's behaviour (delivering the input) to produce a technical effect (the hardware's novel response). The composite may thus be patent-eligible, although any patent that may issue would not cover the computer program itself, only its use in conjunction with the other parts of the claimed invention.
    3. Had I answered it in the negative: can features contribute to the technical character of the claim if the only effects to which they contribute are independent of any particular hardware that may be used? Plainly not: since the effect does not depend on choice of hardware, it would be realised with a non-novel computer, so is covered by the fore-going. Furthermore, that it arises regardless of hardware implies that the effect is merely the natural unfolding of the excluded program's behaviour: it is thus, in all practical senses, part of the program and, thus, excluded.
    1. Does the activity of programming a computer necessarily involve technical considerations? Absent a clearer explanation of the legal-technical term technical, it is hard to say. One can generally avoid, however, such technical considerations as the physical characteristics of the hardware in use: programming languages provide an abstract machine in which to work, thereby transforming these technical considerations into abstract forms, while introducing assorted considerations relating to the abstract machine itself. These abstract considerations are indeed necessarily involved in the activity of programming a computer: in the vernacular they are considered technical. Whether such abstract considerations come under the term of art technical as used by patent law is another matter.
    2. In so far as the legal term of art embraces such abstract considerations: do all features resulting from programming thus contribute to the technical character of a claim? No: all features resulting from programming are results of the program, which is excluded from patentability. Just as a claim must found its claim to technical character outside any prior art or obvious subject matter that it exercises, so also must it find this outside any program it employs.
    3. In so far as it does not: can features resulting from programming contribute to the technical character of a claim only when they contribute to a further technical effect when the program is executed? Features which result solely from the programming cannot contribute to the claim. They may, however, ensure that the program works in such a way as to interact usefully with novel features of the rest of a composite system (including, potentially, the way that the components are combined). Thus a program may serve as a necessary component in a system eligible for patenting, although the whole must find its novelty, non-obviousness, eligibility and technical character in its other parts (such as novel ways of delivering input to the program or responding to output from it).

On technical effect

It is not uncommon, these days, for technical equipment to contain an embedded computer processor which serves as controller for the equipment. One of the increasingly common ways to provide for external interaction with this controller is in the form of a web server providing web forms: filling in a form and submitting it serves as the means of delivering instructions to the the controller. It should thus not be hard to see how forms in web pages may be said to have technical effect, or to be susceptible of industrial application.

As it happens, this web page ends with a form, comprising two buttons you can click on to ask the W3C whether this page conforms to their standards. I do not see that this can be construed as susceptible of industrial application, so I hope the legal term of art technical effect has a meaning under which such a form does not have technical effect, even though (to judge by how few pages on the internet manage to actually conform to the W3C's relevant standards) the question of conformance is (in the vernacular sense of the term) an entirely technical matter.

Thus a web form can serve as a fine example of how something can, by serving as an intermediary, have industrial applications (i.e. have a technical effect) even though the thing itself does not. A web form is, incidentally, also a fine example of how little distinction there is between data and program, on a computer. The bulk of this web page is incontestably data (whether it is also information, I leave each reader to decide) and the part that specifies a (very minimal) form is not so different in character, although it configures the way programs on your computer shall behave. Indeed, we may consider it configuration data for your web user agent (a.k.a. browser): and yet, it specifies (albeit at a very high level of abstraction) some specific ways in which your computer should behave, if you happen to click on the image that submits the form; as such, is it not also a program ?

Let me illustrate how this makes the boundary between data and program impossible to draw. If you opt to consider the form a program because it, by configuring how another program behaves, constitutes instructions (albeit at a high level of abstraction) for a computer to obey: then I shall point you to style sheets and images. They, also, configure how the web browser behaves (a style sheet instructs the computer to display certain parts of a document in certain ways, and it may instruct the computer to change how parts are displayed, in response to how the user interacts with the program; including an image in a web page instructs the computer to draw a certain pattern of colours on its screen, and an animated image instructs the computer to redraw that pattern in some specified manners after particular time intervals) and thus must be considered programs, too. Yet then the text of this paragraph is but an instruction to the computer as to what to draw on its screen, just like the image; how, then, is any data not a program ?

In the alternative, if the form is data, for all that it configures how a program behaves, I can point to the scripts in web pages (or any other program written in an interpreted language, especially when the interpreter runs inside a sandbox, to limit any harm it may do by malfunction). These are transmitted as part of the data constituting the web page (albeit possibly, like images, in response to secondary requests to web servers, rather than as part of the same response as contained the primary page); they are conventionally compiled to a byte-code, which is just data, albeit this data is specially constructed to configure the behaviour of an interpreter, a program which performs (largely) arbitrary computation in response to the instructions it receives as byte-code. Yet such scripts are written in a programming language which web browsers could (and increasingly often, now, do) compile to native object code to be executed by your computer's central processing unit; how, then, is the script not a program ? The only reasonable response to this dilemma is to say (following von Neumann) that there is no essential distinction between data and program: there are practical distinctiions which may be made at certain times by certain parties, but a given sequence of bytes may be understood as data in one context yet program in another; or, in a single context, be understood as data by one part of a system and program by another.

It is well established that information (and thus data) is not patent-eligible, regardless of any technical effect or susceptibility it might have to industrial application, albeit (a process or apparatus for) the industrial application of the information may be patent-eligible. Were computer programs to be opened up to patent eligibility, patent courts would then be subject to an entirely intractable problem of sophistry every time that they are obliged to determine whether a particular body of data is to be construed as data or a program.

Valid CSSValid HTML 4.01 Written by Eddy.