The vanishing author in computer-generated works: a critical analysis of recent Australian case law.

• Jurisdiction: Australia
• Author: McCutcheon, Jani
• Date: 01 December 2012

B Consideration of the Argument in IceTV and Phone Directories

In IceTV, only the Gummow judgment alluded to the argument, when it suggested that the lack of evidence of 'who was responsible for designing the Nine Database' so as to achieve the function of generating the schedule in its final form may indicate that the author 'was unknown'. (136) This suggests that had that person been identified, their conduct in designing the database may be authorial.

The argument was not directly considered in the Phone Directories cases. (137) This was apparently because '[t]he evidence did not establish who had created or customised, in whole or in part, the computer systems'. (138) It is clear that this heavy evidentiary burden will very often be difficult, if not impossible, to discharge and significantly contributes to the problems examined in this article.

The argument was most overtly raised in Telstra's application for special leave to appeal to the High Court. Telstra argued that there was 'human intellectual effort in choosing, customising, setting out specifications and formulating rules as to the operation of a computer system which effected the selection and arrangement in material form', (139) and that

the actual selection and arrangement, albeit done by a computer program, is a computer program that has been chosen and customised and then operated according to rules devised by human intellectual effort of a specific purpose of creating these published literary works. (140) This conduct of 'choosing', 'customising' and 'operating', not to mention 'setting out specifications and formulating rules' may be broader than the more targeted task of selecting software to achieve a preconceived design. However, some of that conduct is clearly relevant. In any event, as mentioned above, (141) in the special leave hearing, the High Court unfortunately did not address Telstra's arguments in respect of any of this conduct, simply finding that there had been no error of law by the Court below.

In Phone Directories (Appeal), Keane CJ and Perram J concentrated on the individuals running the software that extracted the relevant data from the database to create the 'book extract" being the first material formation of the directories in their final compiled form. (142) As discussed above, the Court dismissed this conduct as non-authorial because the extraction was done by software controlled, but not created, by these individuals and 'the form of the compilation [does not] originate with the individual who engages the mechanical processes to produce the compilation.' (143)

This begs the question: what humans were responsible for determining what the directories would look like in their final form and for choosing the software that would achieve that form? In principle, conceiving a design or choosing software to achieve that particular design should ascribe authorship to the designers/selectors.

In this regard, Perram J's metaphor in Phone Directories (Appeal) of an autopiloted plane 'flying itself' is perhaps unconvincing. Whether piloted by human or software, the plane is still flying to the designated destination, as directed. The question is: who set the coordinates? If we equate the plane's destination coordinates with the material form of the production, the pilot has 'fashioned the material form' of the journey but has outsourced the work of flying to the software. The autopilot is thus a mere labour-saving device. The end result--the destination--remains the same whether autopilot or manual control are applied. The difficulty faced by Telstra in Phone Directories was that the computer operators did not set the coordinates, and there was no evidence from those who did. (144)

In Phone Directories, the true designers of the material form were likely to be the 'ultimate' designers responsible for coordinating and directing all of the activities that, in combination, gave effect to the final form of the production. While it is compelling to claim that a work of such sophistication must have emanated ultimately from a human coordinator and ask a court, in the absence of direct evidence, to infer the existence of such an individual, for reasons explored below this may still not suffice to establish that person as an author. (145)

C Problems Ascribing Authorship to Software Selectors

1 Tenuous Causal Chain

Even if evidence from the 'ultimate selectors' had been presented to the Phone Directories Court, that conduct may not be authorial. Justice Yates in Phone Directories (Appeal) most overtly responded to (but rejected) the argument that humans selecting the computer software systems could be authors:

it is not to the point that the second appellants employees were also involved ... in selecting, customising, maintaining and operating the computer systems that were deployed in the production of the directories ... Those activities are akin to educating, training or instructing individuals, and maintaining a sufficient number of them, to carry out the discrete activities of selecting, ordering and arranging material to create the individual compilations. However, the two bodies of activity should not be confused for one another. (146) This implies that the selection and customisation of the software systems used to compile the directories is too removed from the actual compilation through the application of those systems to be authorial. The software selectors are merely responsible for 'setting up' or 'teaching' the conditions for the production of the compilation. They are decision-makers, organisers and trainers, but they are not authors.

However, it is not clear why this should be the case. If the persons running the software are effectively scribes of the software selectors (being 'educated, trained or instructed' by the selectors to perform a task), then, contrary to Yates J's assertion, the selectors' decisions relating to the software and what it does are very much 'to the point', since those decisions ultimately form the compilation. That others (the operators) were trained to effect the design of the selectors should be immaterial.

Nevertheless, the evidence may fail to demonstrate a true line of direction from an ultimate selector to the material fixers. The massive, complex and fragmented nature of many large productions may simply make it impossible for the authorial voice of an ultimate selector to reach the final fixers sufficiently undiluted. If the individuals running the software also contribute independent skill and effort that affects the particular form of the compilation, or if third parties step in and modify the original software design, are they then joint authors with the ultimate selector? It is difficult to fit this kind of relationship within the statutory definition of a work of joint authorship (147) (as interpreted by the courts). (148) The joint authorship claims in Phone Directories were rejected, (149) though not comprehensively considered. (150)

Further, the ultimate selectors will most likely perform a discrete design task, at some temporal (151) and practical distance from the downstream extractors. (152) An author must expend some kind of skill or effort directed to the originality of the expression of the work, even if that is through a scribe, (153) however, the ultimate selectors generally do not 'direct' the operation of the software to produce the production. Does there need to be a more proximate relationship between the ultimate selector and the downstream fixer? Or is it enough that the selector knows and intends that someone, at some time, will run the software and produce the designed form? In short, merely 'setting up' the conditions for some unknown person to later generate a contemplated work may fall short of the requisite 'direction'.

In this respect, the ultimate designers are perhaps more analogous to the authors of the selected software, who are also distant from the output created by their code. In Acohs, it was argued that the authors of the generated code were the programmers who wrote the software that caused its generation, since they understood what the source code would look like. However, Jessup J rejected this argument, for reasons that may equally apply to the ultimate selectors of software:

it would be artificial to regard the programmers as involved in the task of writing the source code for thousands of [safety data sheets] yet to take a material form merely because they wrote, and amended, the program which, when prompted, would put together a selection of the fragments of source code which they did write with other fragments later contributed by the authors. (154) 2 Insufficient Control over Particular Expression

As author, the ultimate selector must also be responsible for the particular expression of the output. (155) This highlights the indistinct region between computer-assisted and computer-generated material, discussed above. If the designers select software parameters that only shape the broad 'idea' of the work, then those fleshing out the detail would seem to be the true authors. (156) If the software takes a completed work and embellishes or modifies it, then, as discussed above, the degree to which the software dictates the particular expression may vary. If software autonomously fleshes out most or all of the detail, then can there be an author responsible for that 'particular' expression?

Phone Directories concerned a compilation, authorship of which related to the arrangement or selection of the compiled material, rather than any unoriginal elements populating the compilation. Therefore, it should have been immaterial that the ultimate designers of the phone directory could not foresee the actual details of the millions of subscribers entered by the data entrants. Provided that the selection or arrangement is not determined by the software, but predetermined by the software selector, then the selector sufficiently...