Journal of Law, Information and Science
|
|
Home
| Databases
| WorldLII
| Search
| Feedback
Journal of Law, Information and Science |
|
ANDREW CHRISTIE[2] AND KENNETH FONG[3]
The infringement of copyright in a computer program by unauthorised copying of its code is a relatively straightforward legal issue. More complex is the issue of how far copyright extends to protect against copying of non-code elements of the program, such as its purpose and structure, its functional modules, and its algorithms and data structures. This article explores the extent to which such elements are and should be the subject matter of copyright protection. The authors postulate a framework for defining and categorising non-code elements of software, drawing a key distinction between those elements which are conceptual and those which are functional. Further distinctions are drawn between user interface and non-user interface elements, and between user interface elements which are visual and those which are not. Using this framework, the key international cases on copyright protection for non-code elements of software are analysed, and the outcomes of these cases are compared with the situation pertaining in Australia in relation to each type of non-code element. The analyses and comparisons undertaken by the authors suggest that conceptual non-code elements of software are not the proper subject matter of copyright. They also suggest that functional non-code elements of software should only be protected to the extent that copyright subsists in them in their own right; there should be no indirect protection of these non-code elements by way of the principle of non-literal infringement of copyright in the code implementing them.
The following diagram shows the elements of computer software, in a hierarchy of embodiments ranging from the most abstract to the most concrete, and separated into code and non-code elements:
Figure 1 - Elements of Computer Software
The hierarchy of embodiments may be separated into three levels, labelled respectively the 'conceptual' level, the 'functional' level and the 'implementational' level. At the highest level, the conceptual level, the software may be embodied in either a general statement of its purpose or in a general statement of the structure of the computer program which will implement this purpose. At the middle level, the functional level, the software may be embodied in either a description of the functional modules of which it consists or in a description of the algorithms and data structures which comprise those functional modules. At the lowest level, the implementational level, software may be embodied in either high-level code (eg. source code) or executable code (eg. machine code or microcode). At this level the software is said to be in code form, and thus the software is said to be comprised of code elements. At the other two levels, the software is said to be in non-code form and thus comprised of non-code elements.
The question of infringement of code elements of software is relatively straight forward. Code elements of software undoubtedly constitute a 'computer program' as that term is defined in the Copyright Act 1968.[4] Because the Act expressly provides that a computer program is a literary work for copyright purposes,[5] it is clear that copyright subsists in the published and unpublished code elements of software.[6] It is also clear that the general principles of infringement of copyright in literary works apply to the code elements of software. These principles include the principle that infringement may occur by copying the whole or a substantial part of a protected work (literal infringement), or by producing another work which is substantially similar to the whole or a substantial part of a protected work (non-literal infringement). The only difficult question arises in relation to non-literal infringement, and concerns the nature of substantially similarity. That is a difficulty which applies to non-literal infringement of all literary works, not just computer programs.
The issue of infringement of copyright in non-code elements is much more contentious. It is far from clear the extent to which the non-code elements satisfy the definition of computer program and hence are protected as literary works under the Copyright Act. It is also unclear the extent to which some of the non-code elements of software satisfy the definition of any of the other categories of protected subject matter, such as 'cinematograph films'. Finally, even where it appears that copyright does subsist in certain non-code elements of software, it is unclear the extent to which the principle of non-literal infringement applies to those elements.
It is the purpose of this article to explore the extent to which the non-code elements of software are protected by copyright, and the extent of any such copyright protection. In particular, the article seeks to categorise different non-code elements of software and determine whether these elements are ones in which Australian and overseas courts have found copyright to subsist. In addition, the article identifies the extent to which the principle of non-literal infringement applies to those elements of non-code software in which copyright does subsist.
The following diagram shows a particular categorisation of the non-code elements of computer software, prepared by the authors as a means for analysing the Australian and international cases dealing with copyright protection for such subject matter. In this categorisation, the non-code elements have been separated into the 'conceptual' and 'functional' elements described above. Each type of element has then been further separated into elements that relate to the user interface of the software and elements that do not. User interface elements have been further separated into visual and non-visual components.
Figure 2 - Non-code Elements of Software
Some examples may be given for most of the categories of non-code elements of software shown in the above diagram. Visual user interface elements at the conceptual level include such generalised ideas as the use of visual metaphors for function tasks and the sequences of various visual displays. Examples of visual user interface elements at the functional level are particular screen displays, such as icons and forms. Non-visual user interface elements at the conceptual level include the ideas of the sequence of functions and the choice of operations to be provided. Examples of such elements at the functional level are particular menu hierarchies and key assignments.
The term 'non-user interface elements' is used to refer to those elements of the program which do not interface with the user - that is, which are transparent to the user. An example of such elements at the functional level are particular file formats for representing data, and particular mathematical formulae used in computation.
The outcomes of a number of the main international cases on protection of non-code elements of software are discussed and summarised, according to the postulated categorisation of the non-code elements of the software. This discussion does not purport to comprehensively review all of the overseas decisions on this topic. Rather, it seeks to illustrate the general trends which may be observed in the two leading common law jurisdictions, the U.S. and the U.K., by noting the outcomes in a selection of key cases.
A significant decision on the issue of copyright protection for visual user interface elements at the conceptual level is provided by the Court of Appeals in the U.S. case of Apple Computer, Inc v Microsoft Corp.[7] Apple sought copyright protection for its Macintosh graphical user interface (GUI) by claiming it constituted an audio-visual work, and that Microsoft’s Windows program was substantially similar to it. Apple’s claim to infringement was restricted by the fact that it had granted Microsoft a licence to use the Macintosh’s visual displays and their individual elements. For the Macintosh GUI, Apple had selected elements such as icons and windows which were arranged on the computer screen to resemble a desktop - the so called ‘desktop metaphor’.
The Court of Appeals for the Ninth Circuit held that there were five basic ideas, other than those elements which had been subject to the licence, embodied in the desktop metaphor: overlapping windows; iconic representation; object opening and closing; use of menus; and iconic manipulation. The court considered these to be basic ideas expressed in an obvious manner, and hence they could not attract copyright protection in their own right. [8] Essentially Apple’s claim was too abstract to attract copyright protection, and could have only succeeded if copyright subsisted in the individual visual elements in their own right.
No cases have been decided on the issue of copyright protection for the conceptual aspects of user interface elements that are non-visual in nature. It is submitted that no copyright would subsist in these elements. These elements would seem to be even more abstract than the visual elements of the user interface at the conceptual level. Following the reasoning of Apple v Microsoft, such elements are at the ideas end of the idea-expression continuum and are not in their own right protected by copyright.
Generally these elements are not protectable unless a separate copyright exists independent from the literary copyright in the program code which implements them. In Digital Communications Associates Inc v Softklone,[9] the U.S. District Court held that because identical screen displays could be generated by programs with different codes, the copying of a program’s screen display did not necessarily infringe the copyright in the program code as a literary work. It further held that the plaintiff’s screen display, a status screen which listed the variable information required by the program and which enabled the user to enter that information, could attract copyright protection in its own right, as a compilation of command terms.[10] The plaintiff’s compilation here consisted of the assembling and arranging of command terms on screen; this was held to be sufficient to constitute an original work of authorship.
Generally these elements are not protectable if their adoption by a competitor would be necessitated by external factors such as efficiency, de facto industry standards, the need for compatibility or some other explicable public policy consideration. The desirability of compatibility between programs was examined in Lotus Development Corp v Borland International Inc.[11] The U.S. Court of Appeals for the First Circuit denied copyright protection to Lotus’s menu command hierarchy. This menu command hierarchy was the program’s system of selecting and operating commands such as ‘Open’, ‘Save’ and ‘Delete’. Commands were linked in a hierarchy of menus and sub-menus. Each menu consisted of a list of commands. The user selected a particular command by highlighting it on the screen menu or by typing its first letter. To make its competing program easily used by those familiar with the Lotus program, Borland copied the Lotus menu command hierarchy’s words and structures, but not the underlying code. Thus Lotus users could switch to the Borland program without having to learn new commands or having to rewrite their Lotus macros.
The court held that the Lotus menu command hierarchy was a ‘method of operation’ and therefore did not attract copyright protection.[12] The court noted the policy considerations in favour of promoting compatibility between programs. Justice Stahl considered that if the menu command hierarchy were copyrightable, then users of several different programs would have to learn how to perform the same operation in a different way for each program used; Justice Stahl thought this would be absurd. Also, the Lotus command hierarchy was used in writing macros for the Lotus program. If the menu command hierarchy was subject to copyright, users switching from Lotus to Borland would have to rewrite their macros, despite the fact that those macros were their own work. These policy considerations led the court to conclude that the Lotus menu command hierarchy was a non-code element of software in which copyright did not subsist.
Copyright protection of the functional elements of the user interface was also denied by the English High Court in John Richardson Computers Ltd v Flanders.[13] The plaintiff’s program, used by pharmacists to label drugs and to keep stock control, was designed to run on BBC computers. The defendant produced a version of the plaintiff’s program to run on IBM computers. The plaintiff alleged that the defendant had taken the general scheme of the BBC program. The court found no relevant similarity between the main menus. Although the defendant’s menu made available functions of the same nature as the plaintiff’s menu, this subject matter was found to be a concept or idea, and thus not the subject of copyright protection. Also held to be an irrelevant similarity was the fact that both programs allowed the user to select between the various options by pressing a key identified by a letter against each option; this was a standard method of choice in programs, and so not protected by copyright. Also, the plaintiff’s unconventional use of the ‘Escape’ key was held to be not the proper subject matter of copyright.
Generally the main purpose of a program is an idea, and so is not protected by copyright. Also, the basic function of each program module will nearly always be an unprotectable idea or process. In Computer Associates International, Inc v Altai, Inc, [14] the U.S. Court of Appeals for the Second Circuit had to consider whether two sub-programs which performed the same translating function, enabling the main program to run on different operating systems, were substantially similar. The court devised a three step test to be applied when assessing the substantial similarity of non-literal elements of program structure. This test has become known as the abstraction-filtration-comparison test.
The first step of the test, abstraction, aims to separate 'idea' from 'expression'. The court said that levels of abstraction in a program range in a continuum from the highest level, the basic idea of the program, to the lowest level, the literal code of the program. The court must dissect the plaintiff’s program structure and isolate each level of abstraction contained within it, and then decide at what level an abstraction constitutes expression. Having decided what constitutes expression, as a second step the court must filter out the protectable expression from the unprotectable expression. Material which is dictated by efficiency, such as where the expression is necessarily incidental to the idea being expressed, is not protected. Also not protected is material which is dictated by factors external to the program itself such as compatibility requirements or widely accepted programming standards, or which is taken from the public domain. Once the court has filtered out the unprotectable expression from the plaintiff’s program, it must then undertake the third step of comparing the remaining core of the plaintiff’s protectable expression with the defendant’s program. The court’s comparison must focus on whether the defendant has copied any aspect of the plaintiff’s protected expression, as well as assess the copied portion’s relative importance in relation to the plaintiff’s program as a whole.
The first two stages of the test are clearly aimed at eliminating highly abstract levels of program conceptualisation from copyright protection. Abstraction sifts out elements of the plaintiff’s program which are mere ideas, whilst filtration leaves only the plaintiff’s more creative expression to be compared with the defendant’s program.
In the U.S., these elements are not protectable if dictated by industry standards, or by requirements of efficiency or compatibility. The courts have recognised that these extrinsic considerations may circumscribe a programmer’s freedom of design choice.
In Gates Rubber Co v Bando Chemical Industries [15] the U.S. Court of Appeals for the Tenth Circuit considered the protectability of elements dictated by the standards of the industry being served. Gates Rubber owned the copyright in Design Flex 4.0, a program which calculated the proper Gates belt to be used with a particular machine. The program used published formulae in conjunction with certain mathematical constants developed by Gates to determine the proper belt size. Bando marketed Chauffeur, a program similar to Design Flex. On appeal, Bando contended the District Court had erred by extending copyright protection to unprotectable elements, facts and ideas, in the Design Flex program.
In determining which elements of the program were protectable, the court applied and further defined the three step test from Computer Associates v Altai. At the abstractions level, the court suggested that a program can often be classified into at least six levels of abstraction, the first two of which equate with conceptual level, the second two of which equate with the functional level, and the last two of which equate with the implementational level in the hierarchy of software illustrated in Figure 1 of this article.
The court stated that, in general, the elements of a program at the conceptual level are unprotectable, being abstract ideas; the elements of a program at the implementational level are protectable, being code within the definition of ‘computer program’, and thus a ‘literary work’; but that elements of a program at the functional level are less prone to generalisation. In this particular case, the court considered that the control flow (the overall sequence of actions and events) and the data flow (the order in which data was manipulated) of the plaintiff’s program were probably in the nature of unprotectable ideas. Similarly, certain algorithms used in the plaintiff’s program were probably unprotectable, being either in the nature of a method of operation, or excluded by the doctrines of merger or scenes a faire.
The approach adopted here significantly narrows copyright protection for reasons of public policy: if the adoption of the same or similar elements at the functional level is dictated by de facto standards or requirements of efficiency or compatibility, then these elements are filtered out as unprotectable material. Programmers striving to maximise speed, efficiency or ease of use may ultimately recreate the structure of an existing program; therefore it is undesirable to protect elements dictated by those goals.
In the U.K., in contrast, protection of these elements is possible, by virtue of them constituting a compilation of subroutines and modules. This approach was adopted by the English High Court in Ibcos Computers Ltd v Barclays Finance Ltd.[16] The court held that in regarding the whole of the plaintiff’s package as a copyright work, it was correct to consider the package’s structure as part of that work. Accordingly, copyright not only subsisted in the individual programs within the plaintiff’s package, but also in the whole suite of programs as a compilation. Copyright in such a compilation could be, and in that case was, infringed by another producing a substantially similar compilation of functional subroutines and modules, even though the particular code for such were different. In addition, there was copyright infringement by the use in file transfer programs of the same date file formats, even though such use was necessary for the defendant's program to compete with the plaintiff's program.
The law on copyright protection for non-code elements of computer software is relatively undeveloped in Australia. In this section of the article the authors discuss those deliberations of the Copyright Law Review Committee, as set out in its 1995 Final Report on Computer Software Protection ('Report'),[17] which touch upon this issue.[18] The authors also consider the one Australian case to date on this topic, the first instance decision of the Federal Court of Australia in Data Access v Powerflex.[19]
In its Report, the CLRC gave its interpretation of what the law is, both in Australia and overseas, in relation to some matters concerning copyright protection for non-code elements of software. To the extent to which it considered the law on these matters in Australia to be unresolved or inappropriate, the CLRC also gave its recommendations as to what that law should be in the future.
The CLRC drew a fundamental distinction between the code of a program on the one hand, and the behaviour of a program implemented by that code on the other hand.[20] In drawing this distinction, the CLRC was implicitly adopting the distinction drawn in this article between the code and non-code elements of software - the phrase 'program behaviour' being the CLRC's way of describing the non-code elements of software. The CLRC considered that both program code and program behaviour may be said to have their own 'structure, sequence and organisation' of elements. The CLRC was of the view that the structure, sequence and organisation of elements of program code is analogous to the organisation and layout of a traditional book, and hence concluded that these elements of program code should and do receive the same degree of copyright protection as do the organisation and layout of a traditional book. There was no need for changes to the current law on this issue. [21]
The CLRC interpreted the structure, sequence and organisation of the various elements of a program's behaviour to equate with the so-called 'look and feel' of the program.[22] These elements include individual screen icons, screen displays and the sequence of screen displays.[23] The CLRC noted two possible approaches to the protection of 'look and feel', as exemplified by the U.S. decisions of Digital Communications Associates v Softklone Distributing[24] and Lotus Development v Paperback Software International.[25] It found preferable the approach adopted in Digital Communications, wherein a clear distinction is made between the program code which gives rise to the look and feel of the program on the one hand, and any visual user interfaces such as screen displays which result from the program code. The program code is protected by copyright as a literary work, whereas resulting screen displays attract separate copyright protection as a compilation.
It is possible to draw from the general approach of the CLRC, as described above, some conclusions about how the CLRC believes the various non-code elements of software are or should be treated under Australian copyright law.
Although the CLRC does not address this issue expressly, it is implicit in its discussion that it considers both user and non-user interface elements at the conceptual level, whether visual or non-visual, to be not protected by copyright law in Australia, on the ground that at this level these elements are merely ideas, and hence subject matter in which copyright does not subsist. Such a view is consistent with the CLRC's express approval of the three-step test enunciated by the U.S. Court of Appeals in Computer Associates International v Altai.[26]
In relation to the visual aspects of the user interface elements at the functional level, the CLRC's expressed preference for the Digital Communications approach would lead to the result that visual aspects are not protected pursuant to the principle of non-literal infringement of any copyright in the program code which implements them; these aspects will only be protected if copyright subsists in the visual interface itself. That is to say, in the absence of copyright subsisting in the visual interface per se, reproducing another program's visual interface is not an infringement of copyright if the particular code which implements that interface has not been copied.
The CLRC explored the possibility of copyright subsisting in a visual interface such as a screen display per se, by way of such being either an artistic work or a cinematograph film, as those two categories of protected subject matter are defined in the Copyright Act 1968. On the possibility of categorising a screen display as an artistic work, the CLRC considered this to be unfeasible on the ground that the screen display does not satisfy the (unwritten) requirement of the Act that an artistic work be in a material form. [27] It stated that the need for a material form 'requires a form of storage, and the ephemeral nature of computer-generated screen displays would appear to fail this requirement'.[28] With respect to categorising a screen display as a cinematograph film, the CLRC considered that the definition of the subject matter as being an 'aggregate of visual images embodied in an article or thing' would not be satisfied, because it could not be said that the visual images of the screen display were actually embodied in anything.[29]
The correctness of the CLRC's conclusions on this last point must be doubted. Prior to the release of the Report, sufficient embodiment of dynamic (ie. non-static) visual images of a screen display had been found in the program code which implements that display, by a South African court interpreting a similar definition of cinematograph film in copyright legislation.[30] Since the release of the Report, the South African approach has been adopted by an Australian court.[31] The CLRC would regard this development in the law in relation to dynamic screen displays as undesirable, since its recommendation was that no additional form of protection be given to any screen displays or any other aspect of the visual user interface.[32]
In relation to the non-visual aspects of the user interface elements at the functional level, the CLRC considered that these elements of software constitute program behaviour and thus are not protected under the principle of non-literal infringement of any copyright in the program code which implements them. As with the visual aspects of a program's user interface, the non-visual aspects will only be protected if copyright subsists in the non-visual interface itself. That is to say, in the absence of copyright subsisting in the non-visual interface per se, reproducing another program's non-visual interface is not an infringement of copyright if the particular code which implements that interface has not been copied. The policy reason for this is the fact that the same program behaviour can be implemented by infinitely different program code. To hold that there could be copyright infringement of program code by the implementation in different code of a non-visual interface would be to give copyright protection to program function. That, in the CLRC's opinion, is not and should not be an object of copyright protection.[33] Hence the CLRC recommended no changes to the law on this issue.[34]
Whilst the CLRC does not make any express reference to the non-user interface elements of software at the functional level, as those elements have been defined in this article, it would seem that those elements are included in its concept of the 'structure, sequence and organisation of program behaviour'. The CLRC was of the opinion that the structure, sequence and organisation of program behaviour are protected in the same manner as is the organisation and layout of a traditional literary work, such as a book.[35] Unfortunately, the CLRC did not make clear its view as to how the organisation and layout of a book is protected by copyright. It may be that it considered these elements are protected as a compilation, along the lines of the reasoning adopted in the U.K. case of Ibcos Computers v Barclays Mercantile Highland Finance.[36] In that case the U.K. High Court held that copyright subsisted in a computer program as a compilation, being a compilation of individual subroutines and modules. This liberal approach to protecting non-code elements of the non-user interface differs from the approach which has been adopted in the U.S., whereby very limited protection is given against the reproduction of such elements that are dictated by requirements of efficiency or compatibility.[37]
At the time of the release of the CLRC's Report, there was no case in Australia which addressed this issue. Since the Report's release, however, the Federal Court of Australia has given its view on this and a number of related issues, in the case discussed below.
Data Access v Powerflex[38] is a first instance decision of the Federal Court of Australia. The case concerned the alleged infringement of copyright in a group of computer programs created by the applicant, Data Access, comprising a database application development system. The respondent, Powerflex, created a similar product which adopted a number of features of the applicant's programs, to enable it to compete with and become a substitute for the applicant's product. Whilst there was no literal copying of the applicant’s program code by the respondent, the applicant nevertheless alleged infringement of copyright in the following six particular subject matters contained in its group of programs: (i) the words of the development language; (ii) the command macros; (iii) the error text table; (iv) the data compression table; (v) the function key words; and (vi) the file structure.
The decision of the court on the claim of copyright infringement in each of these aspects of the applicant's programs is summarised below. In addition, the category of non-code elements of software into which the particular subject matter falls is identified, and the correctness of the court's decision in light of this categorisation is analysed.
The applicant's development language was comprised of 225 words - such as 'filelist', 'graphic' and 'plot' - each of which caused a computer running a program developed in the language to perform a particular function. The respondent's development language used 192 of these words, each of which performed the same function as the corresponding word in the applicant’s language. The court found that each word in the applicant’s program language was a 'computer program', because it was an expression of a set of computer instructions. The court held that copyright in each word was infringed by the respondent's adoption of the same word in its computer language. Use of the same word was a reproduction of the applicant's word, even though the instructions represented by each word were implemented by the respondent in different program code. The court also held that if it were wrong in finding each word to be a computer program, nevertheless there was infringement, because each word was a substantial part of the computer program comprising the applicant's application development system, and the respondent's adoption of the same words was a version, and hence an adaptation, of those parts.
It is submitted that it is both incorrect in law and undesirable as a matter of policy to give copyright protection to each word. A word is not a computer program - rather it is a short character label naming a sub-routine.[39] A word is similar in nature to a key assignment, and hence may be categorised as a non-visual aspect of the user interface at the functional level. Under the approach adopted in the U.S. and preferred by the CLRC for Australia, such elements of software are not subject matter in which copyright subsists per se, and are not protected by any copyright in the code implementing them in the absence of copying of that code.
The applicant alleged infringement of copyright in three 'command macros', labelled 'report', 'entergroup' and 'enter', each one of which was a grouping of a series of commands more complex than any individual word in its development language. The respondent's development language had three command macros performing the same function as the applicant's macros, implemented in different computer code but identified by similar although not identical labels. The court found that each macro was a computer program because it was an expression of a set of computer instructions. It held that copyright in each macro was infringed by the applicant's adoption of similar macros, because the 'strong objective similarity' of the respondent's program code to the applicant's code led to the conclusion that the respondent's code was an adaptation of the applicant's code.
The court's reference to the 'strong objective similarity' of the respondent's and applicant's code implementing the macros suggests that the court found that the respondent had copied the applicant's code, albeit by writing it in another language. If that is the case, then this aspect of the judgment deals only with protection for code elements of software, and says nothing about protection for non-code elements. If, however, the court merely found that there was copying of the function of the code, rather than the code itself, then the court’s conclusion of copyright infringement of the command macros is as flawed, and flawed for the same reasons, as the court’s previous conclusion of copyright infringement of the words of the development language.
The applicant alleged infringement of copyright in its error text table, being a table storing the text of 100 error messages corresponding to particular errors in operation of the application development system. The respondent's system had an error text table with a text entry for each of the 100 errors tabulated in the applicant's system. The text for many, but not all, of the tabulated errors was identical to the corresponding text for that error in the applicant's table. The court held that the error text table of the applicant's computer program was data, rather than a computer program in its own right. Whilst the respondent was found to have reproduced the applicant's table, there was no infringement of copyright in the applicant's computer program. This was because the data was not a substantial part of the applicant's computer program, since it could run without the table. The court considered whether copyright subsisted in the table as a compilation, and held that it did not. This was because each of the reproduced error text messages was one of only a quite limited number of ways of expressing the idea of the error, and so constituted subject matter for which the idea was inseparable from its function.
The text of each error message in the applicant's and respondent’s tables was correctly identified by the court as data, rather than a computer program. That text is thus a non-code element of the applicant's software, being an example of a non-visual aspect of the user interface, at the functional level. The court was right to hold that even slavish copying of such subject matter is not an infringement of copyright to the extent to which the idea of such subject matter is inseparable from its expression. To the extent to which the idea of such subject matter is not inseparable from its expression, however, the court seemed willing to hold that copyright could subsist in such subject matter, and that such copyright would be infringed by slavish copying.
The applicant's system contained a Huffman compression table, by which frequently occurring characters, normally represented in code of eight bits per character, were assigned a shorter string of numbers, with the most frequently occurring characters assigned the shortest strings. The respondent's system contained an identical Huffman compression table, in that the respondent's table assigned to each character in the applicant's table the same compressed string. The court accepted the applicant's argument that copyright subsisted in its Huffman compression table, as a compilation. In contrast to the text of the error table, the contents of the applicant's Huffman compression table did not constitute subject matter for which its idea merged with its expression. The court found the copyright in the table was infringed by the defendant's slavish copying of it. The court did not accept that the respondent's need for complete identity of tables - a need arising from the respondent’s desire to make its product compatible with the applicant's product - excused this copying.
The data compression table, like the error text table, is data. It is a non-code element of the software, being a non-user interface element at the functional level. The approach of the court is contrary to that adopted in the U.S., where copying of such elements is excused from infringement of copyright if such copying is necessary to achieve efficiency or compatibility.
The applicant's application development software language assigned to each of 16 computer keyboard function keys a particular function and a word suggestive of that function. The respondent's language assigned to the same 16 function keys that same function and word. The court held that each of the 16 words was a computer program because each was an expression of a set of instructions which, by pressing the correlative function key, caused a computer to perform a particular function. Copyright in each of these 'computer programs' was infringed by the respondent's allocation of the same word to the same function key.
The function key word allocations are non-code elements of the applicant's software, being non-visual aspects of the user interface at the functional level. The approach adopted in the U.S., and accepted as appropriate for Australia by the CLRC, is that there should be no copyright protection against the reproduction of such non-code elements in the absence of subsistence of copyright in those elements in their own right. The court in this case in fact held that copyright subsisted in these elements of software in their own right. It is submitted, however, that this finding, based on the conclusion that a function key word assignment is a computer program in its own right is as flawed as its earlier conclusions that each word of the development language and each command macro word is a computer program.
The applicant alleged that the respondent had infringed its copyright by the use in the respondent's program of file structures identical to those in the applicant's programs. The court held that copyright subsisted in the 'set of instructions which determine file structure'. Even though there was no objective similarity between the source code or object code implementing the file structures in the applicant's and respondent's programs, the court found that the respondent's code was an infringing adaptation of the applicant's code, due to the identity of function in defining the file structures achieved by the respondent's code.
The file structures are non-code elements of the applicant’s software, being non-user interface elements at the functional level. Finding copyright infringement of these elements when their copying is necessitated by the desire for compatibility is contrary to the approach of the U.S. courts. Furthermore, it is not consistent with the approach adopted by the U.K. courts and arguably preferred by the CLRC. That approach rests on an initial finding that copyright subsists in a particular choice of such non-code elements as a whole, by way of constituting a compilation. This reasoning is absent from the judgment in the Data Access case.
Previous discussions of copyright protection for non-code elements of software have often been tainted by a confusion as to what, precisely, is the subject matter in issue.[40] This article has sought to avoid that confusion, and instead shed light on the debate, by proposing a conceptual framework for defining and categorising the various elements of software that are correctly considered to be non-code in nature. That framework recognises that non-code elements of software exist at two levels, the conceptual level and the functional level, and may relate to either the user interface of the program or rather may relate to aspects of the program transparent to the user. That framework further recognises that non-code user interface elements of software may be either visual or non-visual. The recognition of these distinctions contributes to the debate, since copyright law itself distinguishes between subject matter which is visual in nature from that which is not, between subject matter which is perceptible by humans from that which is not, and subject matter which is abstract in nature from that which is more functional.
Having postulated a particular framework for defining and categorising non-code elements of software, the article has reviewed a number of the key cases in the U.S. and the U.K., with a view to discerning if there are any general trends regarding the protection granted by copyright law to each of these non-code elements of software. That review has found that the approach of both the U.S. and the U.K. courts to date has been generally consistent. That approach may be summarised, as follows.
Non-code elements of software at the conceptual level, whether user interface elements or non-user interface elements, are not subject matter per se in which copyright subsists, being properly considered to be too abstract in nature. Since copyright does not subsist in those elements in their own right, it would be inappropriate as a matter of policy to hold that copying those elements without copying the code which implements them is nevertheless an infringement of copyright by virtue of being a non-literal reproduction of the code. Accordingly, the overseas law is to the effect that copying these elements, in the absence of the copying of the implementing code, is permitted.
With respect to non-code elements at the functional level, the approach is less straightforward, although to a similar effect. As a general rule, these elements are protected by copyright only if copyright subsists in them in their own right; copyright protection of these elements is not obtained pursuant to the principle of non-literal infringement of the copyright which subsists in the program code implementing them. The key issue is whether copyright subsists in these non-code elements. Copyright has been found to subsist in some visual interface elements, but not in non-visual interface elements due to public policy considerations. In relation to non-user interface elements, there has been a divergence of approach between the U.S. and the U.K. courts. The U.S. approach has been to find that copyright does not subsist in these elements in their own right, at least where the adoption of these elements by a competitor would be necessitated by the objective of achieving compatibility between the two programs. The U.K. approach has been to find that copyright can subsist in these elements in their own right, where the collection of these elements in the software package as a whole constitutes an original compilation.
The article has also considered the law in Australia. In particular, it has reviewed the deliberations of the relevant law reform committee, and analysed the decision in the one case to date. The views on the law expressed by the CLRC in its Final Report on Computer Software Protection are largely consistent with the approach adopted by the courts in the U.S. and the U.K. The views on the law as expressed in the first instance decision of the Federal Court in Data Access v Powerflex, however, are almost completely at odds with the overseas jurisprudence. Given that the authors are largely in agreement with the reasoning of the overseas courts on these matters, it can only be hoped that the eagerly awaited decision of the Full Court of the Federal Court on appeal in the Data Access case adopts and applies the approaches of these jurisdictions. To do so would thereby give Australia a law of copyright protection for non-code elements of software that is consistent with the other leading common law jurisdictions, and that permits its local software industry to produce software products that are compatible with or substitutable for those produced by others, so long as there is no literal copying of the code elements of those products. Both of those outcomes are to be desired.
[1] This article is based on a paper presented by Andrew Christie to the Business Law Education Centre one-day seminar on Developments in Information Technology Law, in Melbourne on 11 June 1996 and in Sydney on 17 June 1996.
[2] Senior Lecturer, Law School, University of Melbourne.
[3] Research Fellow, Law School, University of Melbourne.
[4] See the s. 10(1) definition of 'computer program'.
[5] Copyright Act, s. 10(1) definition of 'literary work'.
[6] Copyright Act, ss 32(1) and (2).
[7] 35 F 3d 1435 (9th Cir 1994).
[8] Ibid. at 1444. For example, the court found that representing a document with an iconic image shaped like a page was an obvious choice and not subject to copyright protection in its own right.
[9] 659 F Supp 49 (ND Ga 1987); (1987) 10 IPR 1.
[10] As a compilation, it was in essence a literary work rather than an audio-visual work.
[11] [1995] USCA11 37; 49 F 3d 807 (1st Cir 1995); [1995] USCA11 37; (1995) 33 IPR 233, affirmed by U.S. Supreme Court in [1996] USSC 5; (1996) 116 S. Ct. 804, petition for rehearing denied in (1996) 116 S. Ct. 1062.
[12] Section 102(b) of the U.S. Copyright Act 1976 expressly excludes a method of operation from copyright protection. The menu command hierarchy was likened by the court to the control buttons on a video cassette recorder (VCR), without which it would be impossible to operate the VCR.
[14] [1992] USCA2 1158; 982 F 2d 693 (2nd Cir 1992); (1992) 23 IPR 385.
[15] 9 F 3d 823 (10th Cir 1993).
[17] Copyright Law Review Committee, Computer Software Protection, 1995, Office of Legal Information and Publishing, Attorney-General's Department, Canberra (ISBN 0-642-20830-1).
[18] A consideration of many of the other issues addressed by the CLRC in its Report is undertaken in Christie, 'The Future of Australian Copyright Protection for Computer Programs and Related Products', (1996) 7 Journal of Law and Information Science 87.
[19] (1996) 33 IPR 194. This decision was appealed to the Full Court of the Federal Court. The appeal was heard in 1996, but the judgment of the Full Court was still pending as at mid-January 1997.
[20] Report, para. 9.23.
[21] Ibid.
[22] Report, para. 9.20.
[23] Report, para. 9.19.
[24] 659 F Supp 499 (ND Ga 1987).
[25] 740 F Supp 37 (D Mass 1990).
[26] Report, paras 9.25-9.27.
[27] There is no provision in the Copyright Act which expressly states that an artistic work, or indeed any other copyright work, must be in a material form, but it is generally accepted that such a requirement is implicit in the various references in the Act, including ss 32(1) and (2), to artistic and other works being 'made' or 'published'.
[28] Report, para. 9.45.
[29] Ibid.
[30] Nintendo v Golden China TV-Game Centre (1993) 28 IPR 313.
[31] Sega Enterprises v Galaxy Electronics (1996) 35 IPR 161.
[32] Report, para. 9.48.
[33] Report, para. 9.38.
[34] Report, para. 9.42.
[35] Report, para. 9.23.
[37] See Gates Rubber Co v Bando Chemical Industries 9 F.3d 823 (10th Cir 1993).
[39] Whether or not the sub-routine is a computer program is a separate question, not in issue in this article.
[40] See, for example, the criticisms levelled in Hunter, ‘Limitations on the Proposals for Reform of Non-Literal Copyright and Reverse Engineering of Computer Software’, (1993) 4 Journal of Law and Information Science 247.
AustLII:
Copyright Policy
|
Disclaimers
|
Privacy Policy
|
Feedback
URL: http://www.austlii.edu.au/au/journals/JlLawInfoSci/1996/13.html