Read between the lines
The European Court of Justice decision that computer processes which read and make temporary copies of published work should be regarded as breaching copyright is illogical, says Paul Stanley NO
In Case C-5/08 Infopaq (4th Chamber, 16 July 2009) the ECJ had to grapple with a fascinating question concerning the interaction of copyright and computer processing of data.
Infopaq offered a media monitoring service in Denmark. It took printed newspaper reports, 'read them' (by converting them first to computer image files, and then to text files using optical character recognition) and searched the text file for predefined terms. If the terms were found, it copied short extracts (with five words either side of the search term) and produced printed summaries which contained these 11-word extracts identifying the publication. All intermediate files were deleted, so that at the end of the process Infopaq had only the (original) printed publications, which it had purchased and processed, and the files containing the brief extracts, which it had produced.
There is no doubt that a human being could do something similar without breaching copyright. If a person reads publications, identifies articles of interest and then produces a short summary or précis in original language, no breach of copyright occurs. On this the parties agreed. But Infopaq's process was different in two respects.
First, it involved two points at which a complete copy of the published text was made (first as an image file and then as a text file generated from that image), albeit the file in question was made only in order to be processed and was not kept longer than necessary for that purpose. This process was required simply because of the different ways in which a computer and a human 'reads' printed text. Was this a breach of copyright?
Secondly, the process ultimately produced not a précis or summary produced by the human mind, but a short extract or series of extracts putting the key word in context. Was that a breach of copyright?
These questions were raised in proceedings brought in Denmark, between Infopaq and an organisation representing various newspapers, and referred to the ECJ. The answer turned on the interpretation of various provisions of Directive 2001/29 on copyright harmonisation.
Originality and expression of intellectual creation
The ECJ started by considering the 11-word extracts. Did they infringe copyright? This turned on whether they could be described as 'reproduction... in part' of the original articles. The reproduction of a single word would not constitute reproduction even of 'part' of a work: it is the selection and arrangement of words which constitutes the originality and expression of intellectual creation that copyright protects. Could an 11-word extract, arbitrarily selected around a search term, contain enough of what was distinctive and original about the work itself to amount to some form of reproduction?
The answer the ECJ gave was: possibly. It would depend on the particular words. An extract such as: 'Chamber, 16 July 2009) the ECJ had to grapple with a' is meaningless. But, given a happy coincidence of search-term and context, 11 words may mean a lot: 'To be, or not to be: that is the question' is just ten. And, as the ECJ pointed out, if sufficient 'hits' were produced, a succession of 11-word extracts might reproduce much of the article. The essential question is whether, looking at the extracts, their effect is to contain sufficient elements to 'express the intellectual creation of the author'.
'Temporary acts of reproduction'
The ECJ then turned its attention to the other question. The extracts might involve a breach of copyright, but the process of creating image and text files certainly did, unless some exemption could be found. Infopaq relied on article 5(5) of the directive, which exempts 'temporary acts of reproduction ... which are transient or incidental [and] an integral and essential part of a technological process ... whose sole purpose is to enable ... a lawful use ... of a work ... and which have no independent economic significance'.
The ECJ interpreted these provisions strictly. In order to protect the right-holder it insisted that the 'transience' of the copying must be guaranteed, and not left to human intervention. So, for instance, if I were to make image and text files on my computer manually and then search and delete them, that would in the ECJ's view fall outside the exemption '“ because although I might in fact delete the files, there would be no inherent guarantee of this happening. The process must be fully automated.
Infopaq might have been able to survive this hurdle. But, at the end of the process, 11-word extracts were printed out. This, in the ECJ's view, was fatal, because it meant that the final stage of the process preserved some of the information that had been copied earlier by printing the extracts in paper form.
With respect to the ECJ, this conclusion is absurd. The acceptability of the intermediate processes only matters if the 11-word extracts do not themselves infringe. What one then has is a process whose every step is designed to enable the automated production of groups of words which are not to be regarded as a reproduction even of 'part' of the work. If that is so, there would seem to be no difficulty in saying that the entire process up to that point has been, with respect to the copyright work, transient. Nothing of the work remains at the end. The only reason any copy is even made is to meet the technical needs of a process whose 'output' is not a copy of the work or any part of it. These, it is suggested, are the very cases to which article 5(5) should apply.