Graham Greenleaf
Faculty of Law, University of New South Wales
for publication in the 'Law and Information Technology' column
Australian Law Journal
7 April 1993
In Autodesk Inc. v Dyason [No 2] (1993) 111 ALR 385 the High Court refused to reopen its judgment in Autodesk Inc v Dyason (1991-92) 173 CLR 330; 104 ALR 563 22 IPR 163 ('Autodesk No 1') for the purpose of rehearing three issues which Autodesk submitted it had not had sufficient opportunity to address in the original hearing. The decision was by a majority of Dawson, Brennan and Gaudron JJ , with Mason CJ and Deane J dissenting. Although much of the interest in the decision is in the Court's discussion of the circumstances under which the interests of justice require that a judgment which has not yet been entered should be set aside, there was also considerable discussion of the substantive arguments concerning copyright protection of software and data. Autodesk No 2 weakens the authority of the decision in Autodesk No 1 and indicates unresolved divisions within the Court over some fundamental aspects of copyright. Even to the extent that the Court clarified certain matters in Autodesk No 2, the nature of the application meant that some matters were still not argued fully.
Since the Copyright Amendment Act 1984 (Cth) 'computer program' is defined to mean
'an expression, in any language, code or notation, of a set of instructions (whether with or without related information) intended, either directly or after either or both of the following: (a) conversion to another language, code or notation; (b) reproduction in a different material form; to cause a device having digital information processing capabilities to perform a particular function'.
The differences between courts, and within the High Court, are to a large degree differences over the role that considerations of the function(s) of a computer program should play in determining at least three different aspects of copyright, including what significance should be placed on that term's appearance as the final word of the definition.
The facts of the Autodesk cases (see this column, 63 ALJ 764) were, put very simply, that Autodesk's AutoCAD software required a hardware device, the AutoCAD lock, to be attached to the serial port of the computer on which it was used. The software regularly sent an electronic signal to the AutoCAD lock, which (if present) automatically sent the right electronic answer back to the program, without which it would cease to operate. The part of the AutoCAD program which controlled the sending and receipt of the signal, and thus the continuing functioning of the program, was the prosaically named 'Widget C'. The third respondent, Kelly, measured the input and output signals between the AutoCAD lock and the computer, and developed his own 'Auto-Key' 'lock' which would give the same responses as the AutoCAD lock. The two locks were otherwise quite different both in their physical nature (a hard-wired circuit compared with an EPROM chip) and logically. The AutoCAD lock contained an 8 `cell' shift register and an exclusive-OR gate which interacted with it to produce 127 different binary states in cell 6 (the output to Widget C) before it repeated. The Auto-Key lock was just an EPROM containing data in 127 cells, which matched the 127 outputs from the AutoCAD key. 127 bits is sufficient to represent a string of 16 English letters, or a moderately large number.
In Autodesk No 1 the High Court held (per Dawson J, with whom the rest of the Court agreed) that the computer program which had been copied was Widget C, considered alone. The only part of Widget C which could be considered to be reproduced in the Auto-Key lock was the 127 bit look-up table. Having conceded that the look up table is not a program in itself, because it is not a set of instructions, Dawson J concluded that it was `a substantial, indeed essential, part' of the computer program comprised in Widget C. In their additional reasons, Mason CJ and Brennan and Deane JJ assume similarly.
However, in Autodesk No 2, Mason CJ concludes that an `arguable' alternative view was that `the 127 bit look-up table is simply data or information which is accessed and used by the instructions that constitute the program Widget C'. He argues that although the instructions `are functionally useless without the data', in that Widget C would not fulfil its role without the look-up table, `this may not mean that the look-up table forms part of the instructions'. The inclusion in the definition of `computer program' of the words `instructions (whether with or without related information)' raised the question whether `related information', not itself being instructions, such as the look-up table here, `is brought within the scope of the protection or is impliedly excluded from [its] scope'. He does not express a decided view on the matter, but in raising it he clearly doubts whether it is the function of a program which determines what related information is to be considered to be part of the program.
In contrast, Gaudron J considered that the expression `set of instructions' `directs attention to an entire instruction, or, more accurately, an entire set of instructions, and not merely those parts that consist of bare commands'. That direction is `reinforced' by the parenthetical words in the definition, and is `in no way cut down'. Although Gaudron J does not explicitly state how to determine which data is to be considered part of a program, she stresses the reference to `function' in the definition, and that `in many cases it will be necessary for instructions to be accompanied by related information if those devices are to perform quite ordinary functions'. Brennan J is more direct: `If the look-up table were not part of Widget C, the electronic signals generated by Widget C would not cause the computer to perform in that way'. He rejects the distinction between instructions and data which he feels an electronic engineer might make, concluding that `[o]nce a discreet function of the computer is identified - here, the running of the AutoCAD application - it is necessary to identify the electronic signals which, as an entirety, cause the computer to perform that function', and that they constitute a `computer program' in their entirety. Dawson J considered that none of respondents' arguments have `sufficient prospect of success' to warrant reopening the judgment, even if he was so disposed, but he does not discuss the substance of the arguments.
Of course, it won't matter if data is considered to be part of a program if the data is in itself protected as a copyright work. In Autodesk No 1, Dawson J observes that `it would seem that... the look up table in Widget C is a table or compilation within the meaning of that definition' and that there can be `no doubt about the originality of authorship' of the look-up-table. His reference to `originality' suggests that he may have regarded the 127 bit number as not being too insubstantial to constitute a literary work. However he does not decide the question because `no reliance was placed upon it as a literary work in itself'. In Autodesk No 2, Mason CJ (who did not decide whether the look-up table was part of the program or not) regarded it as conceivable that the reproduction of the table was `the reproduction of something that was itself largely unoriginal'.
The originality of the 127 bit look-up table is, with respect, open to doubt. It is barely enough storage to represent the word `Exxon' three times, and could be argued to be too insubstantial to be original (see Exxon Corporation v Exxon Insurance Consultants International Ltd [1982] RPC 69). However, the originality claim is far weaker than with such an invented word. The 127 bit string represented in Widget C was a random number, generated by Autodesk's use of a pseudo-random number generator, the circuit in the AutoCAD lock. As author, Autodesk had no role in its `selection' beyond deciding how two wires in the AutoCAD lock would be connected, and no interest in its content beyond that it was random. If `Exxon' was not copyright, why is `vnklxcubmnpoqfnb', a random generation of 16 letters, copyright? `Exxon''s function in designating goods and services was not considered relevant, but it appears that this number's function is relevant.
In Autodesk No 1 Northrop J (at first instance) held that the two `locks' were in themselves computer programs (a view since rejected by the High Court), and that the fact that they performed identical function `constitutes the resemblance between the two programs.' In relation to computer programs he considered `the form of the reproduction ... is irrelevant', partly because of the use of `function' in the definition of `computer program'. This approach was criticised in this column as `form follows function', an abandonment of the traditional distinction between idea and expression (63 ALJ 764).
In Autodesk No 1 Dawson J saw `some confusion' in Northrop J's reasoning, stating that `the fact that they both perform the same function ... does not mean that there is any similarity between the two sets of instructions' (if they were instructions). While asserting that the `traditional dichotomy' between idea and expression was `fundamental', he said there is a `particular difficulty' in making the distinction in the case of `a utilitarian works, such as a computer program'. However, he refers with apparent approval to the approach taken in Whelan Associates Inc v Jaslow Dental Laboratory (1986) 797 F2d 1222, described as `that the idea of a utilitarian work is its purpose or function and that the method of arriving at that purpose or function is the expression of the idea', and clearly the `opposite approach' to that taken by Northrop J. While clearly rejecting similarity of `function' as the determinant of `resemblance' for copyright purposes, the High Court does not explain how the `purpose or function' of a program is to be determined for the purpose of this test. This aspect of Autodesk No 1 was not questioned in Autodesk No 2.
The Court in Whelan v Jaslow advanced a test (criticised in this column in (1988) 62 ALJ at 634) which was rather more sweeping than Dawson J's reference to the case would suggest, namely that `the purpose or function of a utilitarian work would be the work's idea, and everything that is not necessary to the purpose or function would be part of the expression of the idea' (emphasis added). In the United States the Whelan v Jaslow `test' has been subjected to extensive criticism, particularly on the grounds that it mistakenly assumes that only one `idea' underlies a program, and all else is expression (see Computer Associates Inc v Altai Inc (1991-92) 23 IPR 385 (vacated, on other points, by Amended Opinion of December 17, 1992, Reported at 1992 U.S. App. LEXIS 33369) and Apple Computer Inc v Microsoft Corporation (1992) 24 IPR 225). Neither the decision in Autodesk, not Dawson J's choice of words, support the conclusion that the High Court has yet adopted that part of the Whelan v Jaslow `test' italicised above.
Dawson J also frames the High Court's endorsement of the concept of `merger' of idea and expression (in relation to computer programs) in terms of `function': `when the expression of an idea is inseparable from its function, it forms part of the idea and is not entitled to the protection of copyright' (citing Lotus Development Corporation v Paperback Software International (1990) 18 IPR 1 at 25).
(This column is to be continued next month, with the questions `Does function determine substantiality?' and `Can function determine causation?')
Autodesk No 2 (continued) .... more functions for `function'
Graham Greenleaf
Faculty of Law, University of New South Wales
for publication in the 'Law and Information Technology' column
Australian Law Journal
Draft 3 April 1993
The previous column in this series (68 ALJ XXX) considered various aspects of the role played by the `function' of a computer program, in light of the High Court's decision in Autodesk Inc. v Dyason [No 2] (1993) 111 ALR 385.
In Autodesk No 1 Dawson J says `Widget C is a computer program and a substantial, indeed essential, part of that program is the look-up table by reference to which Widget C processes the information it receives from the AutoCAD lock'. It was held not necessary that a substantial part of a program should itself be a program. As the Court gave no further explanation as to why a 127 bit table was a substantial part of the Widget C program, the most likely interpretation is that a part which is `essential' for the function of a utilitarian item will therefore be regarded as `substantial'.
However, in Autodesk No 2, Mason CJ (dissenting) rejects Autodesk's argument that the look-up table must be regarded as an `essential' part because `the look-up table is essential to the operation of the AutoCAD locking mechanism'. Such an argument, Mason CJ says,
`misconceives the true nature of the enquiry and seeks to re-introduce by another avenue an emphasis on the copyright work's function. True it is that the look-up-table is essential to the functioning of the AutoCAD lock. However, in the context of copyright law, where emphasis is to be placed upon the 'originality' of the work's expression, the essential or material features of a work should be ascertained by considering the originality of the part allegedly taken. This is particularly important in the case of functional works ...'
He concludes that `it is perhaps possible that, in qualitative and not functional terms, the look-up table is not a material or key part of the program; the act of reproducing it may conceivably be akin to ... the reproduction of something that is itself largely unoriginal.' As discussed in the previous column in this series, the originality of a randomly generated equivalent of 16 letters must be open to question. Where an alleged `substantial part' is unoriginal, Mason CJ cites Lord Pearce with approval: `The reproduction of a part which by itself has no originality will not normally be a substantial part of a program' (Ladbroke (Football) Ltd v William Hill (Football) Ltd [1964] 1 WLR at 293) . The `substantial part' issue was the only matter which Deane J mentioned as warranting a rehearing.
Of the majority judges, Brennan J considers that the purpose or function of the program can determine what is a substantial part: `the series of digits in the look-up table is both original and critical to the set of instructions designed to cause the computer to run the AutoCAD application. When the running of the AutoCAD application is the purpose of the set of instructions expressed in that program, it would be difficult, if not impossible, to contend that the look-up table is not a substantial part of that program.' Gaudron J's words are more ambiguous, in that she says that the look-up table was 'the critical part of the instructions in that the other parts depended on and were made by reference to it' and there is therefore `no basis for an argument' that it was not a substantial part of Widget C. Deane J, while agreeing with Mason CJ that the parties did not have sufficient opportunity to argue the issue of substantiality, does not endorse the approach taken by any other member of the Court, and Dawson J does not clarify his previous judgment. While only Mason CJ clearly rejects stress on the function of the set of instructions to determine what is a substantial part, the question cannot yet be regarded as settled.
The `functional' approach to what is a substantial part of a program has been rejected in an English case, Total Information Processing Systems v Daman (1991-92) 22 IPR 71. The defendants copied a dozen field names from the plaintiff's costing program, so that the defendant's payroll program would interact with the plaintiff's program. Baker J (Chancery Division) held that no substantial part of the plaintiff's program was copied. He rejected the plaintiff's submission that 'the specification in the data division is fundamental to the costing program as a whole for without it the program cannot fetch instructions or put them out. Hence ... there cannot be anything more substantial than that which makes the whole system work.' (ie it was essential for the function). Instead, he found that `[w]hat was copied was a very small part of the commands in the program as a whole ... This is not a case where [the defendant] is seeking to copy the [costing] program at all; it is seeking to use it. ... The specification in high level language of fields and records in the data division tells one little or nothing about the costing program and so ... cannot be regarded as a substantial part...'. If a test such as `how much does it tell you about this program, considered in its entirety as a literary expression' had been applied to the look-up table in Autodesk, the answer would surely be `not much'. While this test does stress the expression, not the function, of a program, it does not seem to be clearly related to Mason CJ's stress on `originality'.
In Autodesk No 1, Dawson J held that `[t]he sequence of numbers put out by the AutoCAD lock is clearly a copy of the sequence in the look-up table in Widget C.' But it wasn't quite so clear. The respondents' final argument was that, even if the Auto Key lock was held to be the same as a substantial part of Widget C, there was no copying from Widget C involved. The basis of their argument, as explained by Gaudron J (the only member of the Court to address this issue), was that `the AutoCAD lock must have been devised first and the sequence of its signals later encrypted in the look-up table'. In other words, both the Auto Key lock and the look-up table were copied from the AutoCAD lock, rather than the Auto Key lock being indirectly copied from the look-up table via the AutoCAD lock. Gaudron J dismissed this approach with the argument that
`the AutoCAD lock and Widget C were devised in conjunction with each other and with the intention that they should complement each other by the lock's emission of a sequence of digits stored in Widget C. Perhaps the lock was wired first so as to generate the sequence eventually employed. But if so, that does not alter the fact that, just as with a conventional lock and key, the devising of one is necessarily the devising of the other.'
This lack of interest in the details and direction of causation is startling. Under what other circumstances are such matters to be ignored? Gaudron J's `lock and key' analogy suggests some type of `simultaneous creation', presumably meaning that causation is deemed to operate in both directions under such circumstances. While it is plausible to develop either a lock first and key second, or vice versa, the analogy fails here. To attempt to derive a circuit with an 8 cell shift register by starting with a random 127 bit string it is hoped to generate requires such a large number of possibilities to be investigated that it is `vanishingly unlikely' (in the words of the third respondent's submissions) to be the direction in which causation occurred.
While Widget C as a whole may have been `devised in conjunction' with the AutoCAD lock, it is more correct to say that the 127 bit look-up table was determined by the circuit configuration of the AutoCAD lock. The AutoCAD lock was not a copy of the look-up table. Of course, if the subsequent operation of the AutoCAD lock to had enabled Kelly to copy the look-up table (as Dawson J seems to have assumed at one point during argument), there could still be indirect reproduction. But this is simply not so, because no aspect of the AutoCAD lock's use involved copying the look-up table: the input to the AutoCAD lock from Widget C bore no resemblance to, and had no causal relationship with, the look-up table. In short, there was no causal link between the look-up table and the Auto-Key lock.
How are potential defendants to know when electronic components which are not copyright in themselves (the AutoCAD lock was not a program and it is difficult to see it as data either), but which determine elements in some copyright subject matter, cannot be copied because they are deemed to be the effect as well as the cause?
Gaudron J's above-quoted remarks were prefaced by her comment that
`Given the purpose and function of the AutoCAD lock in relation to the AutoCAD programs, it must have been obvious to the third respondent [Kelly] that the lock emitted a sequence which corresponded with something in the AutoCAD programs.
...
... unless the [Auto-Key] lock reproduced something within the program, "it would not do anything". And the evidence clearly shows that what it reproduced was the look-up table in Widget C. These matters all but compel a finding as to indirect copying of that look-up table.'
The shift from `corresponded' to `reproduced' in the two paragraphs is instructive. Widget C might have contained a formula which was capable of calculating whether the inputs it received satisfied it (the logical inverse of the AutoCAD lock). To a logician or mathematician the Auto Key EPROM might in some way `correspond' to this formula, but it is another step entirely for a copyright lawyer to say that it `reproduces' it. It does not seem that it should have been obvious to Kelly that he was `reproducing' anything, except of course the `purpose and function' of the AutoCAD lock, to which Gaudron J refers. Powerful stuff, function: it can all but compel time to stand still, causation to be suspended, and reproductions to be deemed!
While the Autodesk cases show that the High Court has not yet finally determined the role that the `function' of a computer program plays in software copyright (or in relation to other utilitarian items), and that there are differences within the Court on this question, the repeated stress placed on functional considerations by at least three members of the Court (Dawson, Gaudron and Brennan JJ) indicates that its role is likely to be a vital one.
The ways in which `function' has been used by various members of the Court to determine what is part of a program, what constitutes a `substantial' part, and what constitutes causation, all make it more likely that attempts to create inter-operable programs (as in Total Information Processing Systems v Daman ) or inter-operable devices (as in Autodesk) will be treated as infringements. On the other hand, the as yet unexplored role of `function' in the concept of `merger' of idea and expression (recognised explicitly by the Court in Autodesk No 1) gives some scope for restricting the role of copyright in the protection of utilitarian works.