1 The present appeal involves two competing publishers of telephone directories. The plaintiff in the suit below, Global Yellow Pages Limited (“GYP”), appeals against the decision of the High Court Judge (“the Judge”) to dismiss its claim for breach of copyright by the defendant, Promedia Directories Pte Ltd (“Promedia”). The Judge’s judgment is reported as Global Yellow Pages Limited v Promedia Directories Pte Ltd [2016] 2 SLR 165 (the “Judgment”). At issue is the nature of the protection from copying that is afforded by copyright law.

2 GYP claimed that Promedia infringed its copyright in the Internet Yellow Pages (an online directory built around a search engine and maintained at http://www.yellowpages.com.sg) (the “Online Directory”), as well as in seven editions (being those from 2003/04 to 2009/10) of its three print directories, namely, the Business Listings (the “BL”), the Yellow Pages Business and the Yellow Pages Consumer (collectively the “YP”). The BL is a white pages directory in which listings of businesses are presented in alphabetical order, while the YP comprises two classified directories that contain listings of businesses arranged within various classifications.

3 At trial, GYP alleged that its claimed copyright in these works had been infringed by Promedia in three directories produced or maintained by the latter, namely, the Green Book (a print directory), the Green Book CD-ROM (a digital directory), and the Green Book Directory (an online directory maintained at http://www.thegreenbook.com) (collectively the “GB”). On appeal, GYP also argues that its copyright was infringed by Promedia’s use and deployment of material from GYP’s directories in the former’s temporary database. Specifically, Promedia is alleged to have photocopied or scanned pages from the BL (at least in 2004, 2006 and 2008), to have copied the listings in the YP when updating its own listings using a computer programme called the YP Advertiser Module, and to have saved or printed the web pages from the Online Directory. This temporary database, in fact, was the focus of GYP’s arguments before us at least insofar as infringement was concerned.

4 The Judge has set out the facts and the parties’ business processes in considerable detail at [7]–[69] and [306]–[361] of the Judgment and we do not propose to repeat them. It suffices that we summarise the most essential facts.

5 As noted above, the works in question were essentially telephone directories of one sort or another. The production of these directories begins with the obtaining of a large mass of data. Essentially, GYP entered into an exclusive agreement with Singapore Telecommunications Limited (“Singtel”), one of the principal telephone service providers in Singapore, pursuant to which it was entitled to receive, on a daily basis, information on new subscribers or changes pertaining to existing subscribers of landlines in mainland Singapore and mobile lines in Pulau Ubin. The production processes that were performed by GYP upon receiving such information were largely computerised. Specifically, such information, which arrives as raw data, is manipulated into a format that is suitable for publication through various processes that entail the verification, classification and embellishment of that data.

6 After the raw data has been verified, subscribers are classified into business types. In this regard, GYP operates a free listing policy, which affords each Singtel subscriber only one free listing in the BL and in either YP directory (the details vary yearly but are immaterial for present purposes). This classification is based on the primary nature of the subscriber’s business. To determine this, subscribers are first assigned main classification headings (“MCHs”), which are back-end categories that are generated by GYP for the purpose of organising the raw data and which are not visible to the end user of the directories. The MCHs are used as a tool to help determine the actual classifications which are eventually published in the directories, and these are visible to the end user. Initially, this task of assigning the MCHs and then determining the actual classification was manually performed by a team which corresponded with subscribers to understand the nature of their businesses. From about November 2006, it was outsourced to a third party, to whom GYP had furnished a map correlating the various commercial class codes maintained by the Accounting and Corporate Regulatory Authority (“ACRA”) to the MCHs used by GYP. Approximately 80% of the subscribers were classified by that third party. From that time (that is, November 2006), GYP’s direct input was confined to manually assigning MCHs to certain limited classes of subscribers being those (a) who conceivably fell within “controlled classifications” (these were classifications in which subscribers could list only if they met certain specified criteria); (b) who disagreed with the matching performed by the third party; or (c) whom the third party was unable to match successfully. The subscriber in turn could embellish its single listing by paying for more prominent listings or a “cross-listing” (which is a listing under multiple classifications). This usually followed the persuasive efforts of GYP’s sales staff.

7 Eventually, the basic subscriber data (that is, the subscriber’s name, telephone numbers and address), any embellishments the subscriber may have paid for, and the classifications under which the subscriber would fall would be extracted, sorted, typeset and printed or published online, as the case may be.

8 Promedia’s production processes, in comparison, were considerably less automated. Its business processes were reduced to an extensive set of standard operating procedures (“SOP”), each of which governed a specific task that employees were to perform. Promedia could not obtain subscriber information directly from Singtel. Instead it collected data from multiple third-party sources such as delivery orders, marketing collateral, field surveys, telephone calls and company websites. However, by far the most significant source of the information to be included in its directories were the telephone directories of competing publishers. One of Promedia’s SOPs required employees to take the data in the most recent edition of the BL and YP and merge it with their existing database by keying in the data through a computer program in what was essentially a form-filling exercise. Important for the purposes of this appeal is the fact that, in some years, this task involved the saving or printing of the Internet Yellow Pages entries and the photocopying or scanning of the BL into what parties refer to as Promedia’s “temporary database”. Subsequent to this exercise, the SOPs required employees to perform a few other tasks, of which three are notable. First, they used the data for “market intelligence” exercises to develop Promedia’s business. This entailed identifying the prominence of a subscriber’s listing in a competing directory (for instance, whether it had a bold entry or a boxed entry) in order to enable Promedia to assess whether it was likely to also pay Promedia a premium for a more prominent listing. Second, they had to verify and update the entries in the database by calling subscribers. Third, in that same call, the employees had to understand the subscriber’s businesses and product or service offerings in greater detail, with a view to listing them under multiple business types or offerings where applicable. This was modelled after the “Thomas register”, a directory in the USA well known for its detailed and specific classifications. After these steps, the data in the database would be manipulated to generate the various GB directories.

9 The result of Promedia’s efforts was a set of GB directories that was visibly different from GYP’s corresponding directories. The evidence showed that, in essence, what was copied by Promedia was the data in GYP’s directories. There is no evidence, however, that such data as was copied was then reproduced in the same form in which it had appeared in GYP’s directories. Indeed, as evident from the table at [348] of the Judgment, GYP’s directories had no more than 1.07 classifications per subscriber on average whereas GB’s corresponding directory had no less than 3.10 — this figure in fact rose to 7.45 in 2009. It may also be noted that there were far more classifications in the corresponding directories produced by Promedia as compared to those produced by GYP in which classifications were used.

10 However, not all the SOPs were executed to the letter. Promedia’s employees failed to verify some subscriber listings in the BL or YP. That is how some “seeds”, which are fictitious listings designed to detect copying, made their way into the CD-ROM and online versions of the GB. And that, in turn, was how GYP discovered that Promedia had been copying the data in its directories.

11 It is against this factual backdrop that GYP commenced suit against Promedia alleging copyright infringement. At the trial below, GYP claimed that copyright subsisted in three categories of works:

(a) the “enhanced data” in its directories (which, it suffices to say for now, refers to individual listings after they had been verified, embellished, arranged and classified);

(b) each of its directories “in whole or in part”, as compilations that constitute intellectual creations by the selection and arrangement of their content; and

(c) individual seeds (fictitious listings introduced to detect copying).

12 To assist us in our consideration of the issues, we appointed Professor David Llewelyn as amicus curiae. We found both his written and his oral submissions to be admirably succinct, clear and illuminating and of great assistance. For this, we are deeply grateful.

13 This case turns on a few key principles of copyright law which we think helpful to set out at the outset.

14 The first is simply that liability for the infringement of copyright turns on the related issues of subsistence and infringement. The orthodox method of analysis poses three questions in sequence: whether copyright subsists in a work; whether the copyright has prima facie been infringed; and whether any defences apply.

15 The second principle is that copyright protects not ideas, facts or data, but the expression thereof. As the Judge observed at [70] of the Judgment, this principle appears more difficult to apply in the context of factual compilations. However, ss 4, 7A and 27 of the Copyright Act (Cap 63, 2006 Rev Ed) (“CA”) make it clear that copyright subsists in an original published literary work, which includes a compilation that, “by reason of the selection or arrangement of its contents, constitutes an intellectual creation” (emphasis added). Facts alone fall outside the protection of copyright law. Feist Publications, Inc v Rural Telephone Service Company, Inc 499 US 340 (1991) (“Feist”) — where the US Supreme Court decided that Feist did not infringe any copyright by copying Rural’s white pages listings — is a good example of why copyright law will afford no remedy when only data is copied. O’Connor J, on behalf of the court, held that facts and data were not copyrightable because they were not original works: that is, works independently created (as opposed to having been copied) by the author and possessing a minimal degree of creativity. On the other hand, a compilation of facts, specifically a selection or arrangement of the facts, was original and could, as such, conceivably be eligible for copyright protection, though the protection conferred would be “thin”.

16 The third principle is that the method of analysis is heavily shaped by what it is in a work that attracts copyright protection, especially where infringement and fair dealing are concerned. In other words, there is a nexus between the originality, skill and effort that goes into a work, and the substantiality of copying required to establish infringement. Thus, although copyright may subsist in a work as a whole, there would be no infringement of such copyright unless one copies the work as a whole, or a substantial portion of the part of the work that attracts copyright protection in the first place. This point was made by French CJ, Crennan and Kiefel JJ in the decision of the High Court of Australia in IceTV Pty Limited and another v Nine Network Australia Pty Ltd (2009) 239 CLR 458 (“IceTV”) at [30]–[32], [35]–[44] and [49]–[54]; see also the Judgment at [121]. For example, the copyright that subsists in a compilation of all the news articles published by a certain agency in a certain month by virtue only of its thematic arrangement would not be infringed by another compilation of the same selection of articles arranged chronologically because, even though the same underlying material might have been copied, the arrangement, in which copyright was found to subsist, was not. Also, the thinner the copyright protection, the more substantial the copying must be before a finding of infringement will be made.

17 The Judge clearly bore in mind these basic principles of copyright law; they explain why the Judge found that the copyright protection in GYP’s directories was very thin indeed, and consequently that its copyright had not been infringed.

18 On the issue of subsistence, the Judge held that for a compilation to attract copyright protection, the compiler must have exercised sufficient creativity or intellectual effort in the selection or arrangement of the material (see the Judgment at [208]). The “creativity” standard was not only required by the CA (and its drafting history) in the light of the precedents, but also rested on policy justifications (see the Judgment at [189]–[203]). Any such copyright protection would extend only to the original expression in the form of the selection or arrangement of data or material, but not to the composite parts of the compilation; that is to say, it would not extend to the facts or data contained therein (see the Judgment at [208]). Thus, the Judge found that:

(a) Copyright did not subsist in the individual listings. The form of expression was unoriginal. In fact, the fact-expression dichotomy dissolved entirely as the fact was essentially coterminous with the expression. As for the selection or arrangement of the information within a subscriber listing, this lacked the minimum level of creativity or intellectual effort necessary to make it an original work. The effort, skill and judgment in collecting, verifying, enhancing and arranging the data was not directed at a particular form of expression. Instead, this was directed at ensuring the accuracy of the underlying facts. In these circumstances, granting copyright here would amount to granting an impermissible monopoly over the use of bare facts (see the Judgment at [219]–[223]).

(b) Copyright did not subsist in the selection or the arrangement of the listings within each classification. Any creativity or intellectual effort expended in this connection was directed at the discovery of facts, for instance, the nature of the subscriber’s business, and not at reducing the work to its final form of expression through selecting and arranging the listings within the classifications. Copyright also did not subsist in the classification headings. The effort expended fell far short of the required level of originality. Each heading was hardly a work in its own right; it was commonplace and there was insufficient effort in terms either of selection or labelling to make it an intellectual creation (see the Judgment at [226]–[233]).

(c) As to the directories:

(i) Copyright subsisted in each BL compilation as a whole (including the introductory material and the selection and alphabetical arrangement of listings and other material such as the section on emergency/important/useful/Government numbers), but this would not extend to or protect just the selection and arrangement of listings in the BL (which was done alphabetically only) as a compilation. There was no creativity in reducing the BL to its final form of expression as a compilation because the selection was exhaustive and the alphabetical arrangement that was used was entirely commonplace. Further, it lacked human authorship (see the Judgment at [262]–[270] and [369]).

(ii) Copyright subsisted in each YP compilation as a whole (which meant the selection and arrangement of the various classifications and listings within the directory together with the introductory material). The intellectual effort in identifying and selecting classifications and assigning listings to these was creative enough to cloak the compilation of classifications and listings as a whole in each directory with copyright. But again, this did not protect the individual listings, the listings within each classification, or the individual classifications (see the Judgment at [251]–[260]).

(iii) Copyright subsisted in the Online Directory as a whole (meaning the arrangement of the listings through the entire scheme setting in place the various categories (including drill-downs and sub-categories), filters and keywords that were employed in the Online Directory). This was because GYP’s employees had exercised sufficient creativity and intellectual effort in the arrangement of the listings in the Online Directory in maintaining and introducing new category headings, deciding the depth of the drill-downs and sub-categories, and determining the filters used in the directory (see the Judgment at [273]–[284]).

(d) Copyright did not subsist in each individual seed. They were not original literary works; the skill in devising them was insignificant and insufficient. Copyright also did not subsist in the totality of the seeds as an original compilation (see the Judgment at [302]–[304]).

19 On the issue of infringement, the Judge found that:

(a) Copyright in each BL compilation had not been infringed. The photocopying or scanning of the BL compilation into the temporary database might have entailed voluminous unauthorised reproduction but it was not “substantial in the copyright sense” because only the listings had been photocopied or scanned in their entirety (see the Judgment at [366]–[371] and [373]).

(b) Copyright in each YP compilation and the Online Directory had not been infringed. Promedia may have referred to the information in the listings and classifications and copied much of the basic data without further verification, and might have recorded other classification headings as part of its research, but this was not “substantial” taking. There were sufficient differences in Promedia’s method of selecting or arranging material in its directories such that it did not reproduce a substantial part of the original features of GYP’s YP compilations and Online Directory (see the Judgment at [372]–[373]).

20 The issue of defences became moot because the Judge found no copyright infringement, but the Judge briefly discussed his views. He reserved his views on the public interest defence (see the Judgment at [403]–[407]), and stated that he would have rejected both the innocent infringement defence (see the Judgment at [380]–[386]) and the argument that injunctive relief was unavailable by virtue of laches, delay or acquiescence (see the Judgment at [375]–[379]). However, the Judge also said that he would have accepted the fair dealing defence, on the assumption that any infringement would involve either Promedia’s competing directories (on the basis of substantial taking) or the copying of GYP’s directories during Promedia’s production process. The scanning or photocopying of GYP’s BL directories during the production process was a commercial dealing and made referencing and use of the data easier, but this did not tip the balance against a finding of fair dealing. To the extent Promedia copied classification headings as part of its research, it was fair dealing (see the Judgment at [396]–[402]).

21 Finally, the Judge held that Promedia succeeded on its counterclaim for groundless threat of copyright infringement, with damages to be assessed. The letters of demand amounted to threats, and liability followed as a matter of course once GYP’s claim failed. In any event, the threats were objectively unjustifiable because there was no clear binding authority and the threats went beyond even the “industrious collection” approach in Singapore Land Authority v Virtual Map (Singapore) Pte Ltd [2007] SGDC 216 (“Virtual Map (DC)”) and were the very type of threats that were meant to be deterred (see the Judgment at [410]–[418]).

22 The same basic principles of copyright law also inform our conclusion that GYP’s appeal is essentially without merit. Fundamentally, what Promedia took from GYP was data. Data is not copyrightable material even if, as was the case here, it was a valuable commodity for which GYP had paid a substantial amount. We will address the parties’ submissions in the course of our discussion on the various issues that arise. Leaving aside the substantive issues in the case, we note that Promedia also filed a summons seeking to strike out the Appellant’s Reply for non-compliance with O 57 r 9A(5A)–(5B) of the Rules of Court (Cap 322, R 5, 2014 Rev Ed) (“ROC”). We make no order on this because we were inclined to consider the arguments raised in the reply in any event.

23 The popular view is that there are two main approaches to determine whether a compilation is original: the “creativity” approach and the “sweat of the brow” approach. These have been described in more detail at [74]–[78] of the Judgment. Essentially, the two approaches have differing views on the quality and the object of the effort in producing a work. The latter considers the author’s labour and industry to be relevant, and admits consideration of the effort taken in the preparatory steps leading to the reduction of the work to its final form. As against this, the former places emphasis on the end product of the work and considers efforts applied towards the formulation of how that end product will be expressed.

24 We agree with the amicus curiae, Prof Llewelyn, that for copyright to subsist in any literary work, there must be an authorial creation that is causally connected with the engagement of the human intellect. By the human intellect, we mean the application of intellectual effort, creativity, or the exercise of mental labour, skill or judgment. Effort (even intellectual) that is applied not towards the authorial creation but towards other ends such as the verification of facts will not be relevant in this context even if such verified facts might be the eventual subject of the authorial creation. Therefore, in the context of compilations, we agree with the Judge’s view (see the Judgment at [208]) that the “creativity” approach is the correct one. It follows from this that the compiler must exercise sufficient creativity in selecting or arranging the material within the compilation; and if the compiler does so, the resulting copyright will only protect the original expression in the form of the selection or arrangement of the material, as the case may be.

25 The Judge, in arriving at his decision, undertook an extensive discussion of the CA provisions and their legislative history (including the treaties on which their language or substance was based), of the authorities (from the US, the UK, Australia, Canada and Singapore), of the policy considerations and of first principles (see the Judgment at [70]–[208]). We largely agree with the Judge’s analysis and will only add or emphasise four important points.

26 First, as the Judge observed, there has been a noticeable retreat from the “sweat of the brow” approach that once featured in the decisions of the Australian and English courts. In Australia, IceTV marked a shift away from the decision of the Full Bench of the Federal Court of Australia in Desktop Marketing Systems Pty Ltd v Telstra Corporation Ltd (2002) 192 ALR 433, which the Judge described as the “high watermark of the ‘sweat of the brow’” approach (see the Judgment at [110]). And in the UK, recent cases have placed greater emphasis on the effort and labour of authorship (see, for example, Ibcos Computers Ltd v Barclays Mercantile Highland Finance Ltd [1994] FSR 275 (“Ibcos Computers”) at 302 per Jacob J; and Autospin (Oil Seals) Ltd v Beehive Spinning (A Firm) [1995] RPC 683 at 698 per Laddie J). Although copyright law had at times been stretched to protect the labour and expense invested in the creation of databases, this tension was eased after the introduction of a sui generis database right in the Copyright and Rights in Databases Regulations 1997 (SI 1997 No 3022), which gave effect to the European Union Directive 96/9/EC of the European Parliament and Council of 11 March 1996 on the Legal Protection of Databases. These regulations allowed copyright law in UK to perform the function for which it was originally designated. In Singapore, Virtual Map (DC), which on one view adopted the “sweat of the brow” approach, has not been endorsed by any higher court specifically on the issue of the proper approach to determining originality. Instead, it has been overshadowed by the later decision of this court in Asia Pacific Publishing Pte Ltd v Pioneers & Leaders (Publishers) Pte Ltd [2011] 4 SLR 381 (“Pioneers & Leaders”), where V K Rajah JA expressed a clear preference for the creativity approach. In our judgment, Virtual Map (DC) is incorrect insofar as it adopted the “sweat of the brow” approach, and is defensible only on other bases such as that suggested by Prof Ng-Loy Wee Loon SC in Law of Intellectual Property of Singapore (Academy Publishing, 2nd Ed, 2014) at para 10.1.35, namely, that what was protected was not merely Singapore’s topographical information, but its expression in a vector format (which, on the evidence, would have been more accurate than using GPS technology).

27 Second, we also agree with Prof Llewelyn that there is no meaningful difference in the standards used by the various jurisdictions today to describe the requisite standard of creativity directed at the expression in question; the differences in language are essentially semantic. Although the language used by the High Court of Australia in IceTV decision was that of “independent intellectual effort” (at [33] per French CJ, Crennan and Kiefel JJ) or “sufficient effort of a literary nature” (at [99] per Gummow, Hayne and Heydon JJ), the decision approved that of the US Supreme Court in Feist, which espoused the “minimal degree of creativity” standard (at 345 per O’Connor J). Similarly, it is the “skill, labour and judgment” in creating a work that attracts copyright in the UK (see Ibcos Computers at 302). Even the position taken by the Supreme Court of Canada in CCH Canadian Ltd v Law Society of Upper Canada [2004] SCC 13 (“CCH Canadian”) requires an “exercise of skill and judgment” in the expression of an idea which cannot be so trivial as to amount to a purely mechanical exercise. We note in passing that CCH Canadian has been said by some to represent an “intermediate position” between the “creativity” and “sweat of the brow” schools of thought. Nothing turns on this, in our view, because, in substance, the standard it adopts equally places the emphasis on the creativity that is directed towards the eventual expression.

28 Third, the sufficiency of intellectual effort will often be, as Rajah JA observed in Pioneers & Leaders at [34], a question of “fact and degree”, and in our judgment it would be practically impossible to prescribe prospectively a minimum level of such effort. Indeed, our view that the differences across the various jurisdictions are, in substance, semantic stems largely out of the seemingly shared sense that it is impossible to prescribe a minimum level of creativity. What is common across the jurisdictions, however, is that each court seeks to characterise the human acts done and decisions made in reducing the universe of raw information into a work. There are, on the one hand, managerial decisions (whether to publish or update a phone directory) which clearly do not merit copyright protection. Then there are purely mechanical tasks (such as the algorithmic collection or arrangement of data) that, taken alone, would not cross the creativity threshold. Then there are choices in between that require human judgment and do impact the selection (such as the geographic area covered, the cut-off date, and the type of information that will be published) or the arrangement of the material (such as the order in which entries are sorted, and the order in which various fields of information are presented), and which might more properly be regarded as authorial in nature.

29 There will be some cases in which the decisions or actions to be taken contain elements of discovering or creating facts and these must be analysed to determine if the acts are truly authorial. There will be other cases in which the fact or idea and its expression are virtually coterminous, and decisions and actions to express that fact or idea are so obvious or inevitable that they cannot be said to have been creative.

30 In the final analysis, it will be a question of fact and degree whether such acts and choices demonstrate sufficient intellectual effort and bear a sufficient causal nexus with the final work such that it attracts copyright. As we have stated above, in every aspect of this analysis, what guides the court is the fundamental principle that copyright law protects not ideas, facts or data but the expression thereof.

31 The cases show how nuanced this exercise can be. For example, the Court of Appeals for the Eleventh Circuit sitting en banc in Bellsouth Advertising & Publishing Corporation v Donnelley Information Publishing, Inc 999 F.2d 1436 (1993) (“Bellsouth”) held by a 6-1 majority that copyright did not subsist in a phone directory. Hatchett J, dissenting, considered at least four aspects of the claimant’s publication to be original: the selection of the geographic area to be covered, the selection of businesses with only business telephone numbers, the selection of the 7,000 classified headings, and the arrangement of about 106,400 business listings within those headings. Conversely, the majority thought that the selection of the geographic area and the arrangement exercise were not acts of authorship but techniques for the discovery of facts. The former was practically inevitable for a business directory while the latter depended at least partly on the subscribers’ willingness to purchase listings in response to cross-selling efforts.

32 Another example is CCH Canadian, which concerned law reports and where copyright was claimed in, among other things, reported judicial decisions. McLachlin CJ held at [33]–[35] that although the reported judicial decisions (which contained both a headnote and the edited judicial reasons) were copyrighted works, the judicial reasons in and of themselves (that is, without the accompanying headnote) were not works in which the publishers could claim copyright despite the editing they had performed, because the effort applied was relatively trivial and more properly characterised as an essentially mechanical exercise. The changes included the addition of basic factual information (such as the date of the judgment, the court and the panel hearing the case, counsel for each party, lists of cases, statutes and parallel citations), and the correction of minor grammatical or spelling errors. One can imagine, however, that if a work has been more substantially edited, the edited copy of the work could enjoy its own copyright.

33 In this context, we note that in every decision of which we are aware, the alphabetical arrangement of data has not crossed the threshold of creativity. Instead, it has been described as “unoriginal” and “practically inevitable” (Feist at 363, per O’Connor J) or “obvious” (Nine Network Australia Pty Ltd v IceTV Pty Ltd and Another [2008] FCAFC 71 at [119]), and these all suggest that such an arrangement exhibits insufficient creativity or intellectual effort for copyright to subsist.

34 Finally, this appeal throws the spotlight on how the law is to deal with “free-riding”, or the appropriation of data or facts that represents the fruit of an investment. We agree with Prof Llewelyn that even if one might frown on the commercial immorality underlying such conduct, this is simply not within the purview of copyright law. Instead, this is more properly the province of a sui generis database right, which has been recognised in the European Union but not in Singapore. In fact, Parliament consciously decided not to introduce such a right at the time the 1998 amendments to the CA were debated. It was put in this way in Singapore Parliamentary Debates, Official Report (19 February 1998) vol 68 cols 321–322 (Prof S Jayakumar, Minister for Law):

35 It is not our function to supplant the legislature, as GYP attempted to persuade us to do, by adopting a lower standard of creativity in the case of factual compilations in comparison to other authors’ works in order to provide the sort of protection that Parliament considered but then declined to provide. In any event, it remains an open question whether database creators are severely disadvantaged by the lack of a sui generis database right. Prof Llewelyn observed that a measure of protection may be achieved through the law of contract.

36 With these principles in mind, we turn to the facts before us.

37 The list of works in which copyright is said to subsist has been narrowed on appeal to three classes: the listings arranged within each classification, the listings in the BL, and the seeds. Even so, we are not persuaded that copyright subsists in any of these, save for a very narrow instance where the BL is concerned.

38 We agree with Promedia and the Judge that copyright cannot subsist in the listings arranged within each classification, because the selection of the contents lacked creativity or was in reality a fact-discovery exercise. GYP claimed that it had selected or arranged these listings creatively in three ways, which we examine in turn.

39 First, it submits that it exercised creativity in assigning the MCHs, especially for subscribers with multiple or ambiguous lines of activity. In such cases, GYP would offer two choices to each subscriber; if the subscriber chose neither or was uncontactable, it would not be listed. Additionally, GYP would review the classifications every few years. In our judgment, the only authorial act is the creation of the map that correlates the ACRA class codes with GYP’s MCHs. Most subscribers were categorised mechanically using this map; the rest were classified based on what was thought to best describe their business, and the fact that GYP gave each of them a choice leads us to characterise this exercise primarily as one of joint fact discovery rather than one of authorship. We observe that the entire list of MCHs or classifications could also attract copyright, but this does not appear to have been GYP’s case and, in any event, it was not GYP’s case that Promedia took a substantial portion of GYP’s MCHs or classifications.

40 Next, GYP says that it exercised creativity in selling cross-listings (see [6] above), and that the subscribers’ choices were confined to those offered by GYP. In our judgment, this cannot be characterised as authorial. First, as Promedia points out, the ultimate decision to purchase cross-listings lay with the subscriber even though its choices might have been circumscribed by GYP. In substance, this was an exercise in marketing and, if anything, it was an instance of joint fact discovery, namely, discovering further details of the subscriber’s business and whether this could bring it within other listings. As observed in Bellsouth (at 1441), cross-selling may also be seen as a technique to discover whether a subscriber was willing to pay to be listed under certain classifications. GYP and Promedia each had different perspectives on what underlay this exercise. GYP contends that cross-selling was driven by its desire to enhance the user experience and so to benefit the public while Promedia alleges that it was to maximise profit. In our judgment, nothing turns on this.

41 Finally, GYP says that it exercised creativity in using “controlled classifications”, which are classifications under which a business could list only if it met certain criteria (for example, holding a National Environment Agency (“NEA”) licence for a “pest control” listing). In our judgment, this, again, is an exercise in fact discovery. For example, the fact discovered in relation to the “pest control” listing is whether a subscriber is principally involved in pest control and is licenced by the NEA as a pest controller. The only possible creative act would be determining which classifications ought to be “controlled” and the conditions that a subscriber would have to meet to be listed under such classifications. However, many of these, such as vector control, caterers and employment agencies, would have been statutorily prescribed in the first place. Although GYP, at its own initiative, created other such classifications like “social escorts” and determined the qualifying conditions, the only way in which GYP believed itself to have departed from the existing guidelines promulgated by the Criminal Investigation Department was to require a non-residential address. That, in our judgment, falls short of the minimum level of creativity required.

42 In these circumstances, we hold that the Judge was correct to have found that copyright did not subsist in the listings arranged within the classifications.

43 We largely agree with Promedia and the Judge that copyright cannot subsist in the BL by virtue of the selection of its contents, as that lacked creativity. However, we agree with GYP that the arrangement of the BL’s contents exhibited sufficient creativity, but only barely so.

44 First, GYP says it exercised creativity in selecting the listings because of the use of prescribed guidelines to determine the subscribers whose information it would publish. By this GYP means that it was interested only in mainland subscribers with land lines and subscribers in Pulau Ubin with mobile lines. However, in our judgment, there was no creativity in this. Singtel was obligated both by a contract with GYP and by requirements imposed by the Infocomm Development Authority of Singapore to provide GYP with information of subscribers with fixed lines, while the only mode of communication for subscribers in Pulau Ubin was through mobile numbers. As the court in Feist observed at 362–363, a selection of subscribers which had been prescribed by law would be unoriginal. In our judgment, it is not a material difference that the obligation was one not on the telephone directory publisher to publish certain information but on the telecommunications service provider to provide that information, because the central point on either analysis is that the selection of subscribers was unoriginal. GYP’s claim that these guidelines balanced comprehensiveness and readability was nothing more than an endeavour to rationalise its choices, which, in truth, were obvious.

45 GYP’s second claim to creativity in its selection is its free listing policy (described above at [5]). The policy varies yearly, but generally it determines the parcels of information that is published for free or publishable at an extra fee. In our judgment, this policy is best characterised as a technique for discovering facts, for example, the category or categories of businesses under which the subscriber may be and is willing to be listed.

46 The final claim to creativity in its selection, that GYP advances, is its streamlining of information. This entails suppressing additional phone numbers belonging to a single subscriber (usually to leave such numbers free for internal use) or suppressing “ditto lines”, which are repetitions of a single address to which multiple phone numbers pertain (to improve readability and save printing costs). The suppression of phone numbers is done when a subscriber pays for it and in our judgment should be seen as the discovery of facts (namely, the numbers through which a subscriber wishes to be contacted by the public), while the suppression of ditto lines is in our judgment a far cry from meeting the requisite level of creativity. Indeed, we are not able to see how any of these three claims can amount to creative expressions of an idea or fact.

47 The situation as regards arrangement is a little more difficult to assess. The Judge took the view that the sorting rules were meant to reinforce rather than depart from the alphabetical arrangement and that the arrangement therefore did not meet the requisite level of creativity (see the Judgment at [263]–[265]).

48 In principle, the sorting rules may sometimes be more properly seen as a choice to depart from the default rules of arrangement which a computer would apply; if the rules are sufficiently detailed and unusual, they could notionally cross the requisite threshold of creativity. For example, a very detailed system might address the treatment or relative priority of the letters of the alphabet, capital letters, accented letters of the alphabet (for example, ç, é, ñ and ÿ), compound characters (for example, æ), numbers, ordinals and fractions (for example, 1, 1st and ½), spaces, special characters (both those that resemble letters of the alphabet (for example, !, @, $, ¢, and £) and those that do not (for example, #, %, & and *)) and abbreviations (for example, Dr, Mr and Bros). GYP’s sorting rules can be expressed in five propositions:

(a) precedence is given to spaces, then numerals, then the letters of the alphabet;

(b) names beginning with numerals are listed immediately in front of entries bearing the first letter of the alphabet of the first digit (for example, some entries, when placed in order, might read “1 Country”, “11th Hour”, “1Com”, “1st Frontier” and “O C Company”);

(c) abbreviations are sorted “as is” (that is, according to the abbreviated form);

(d) names beginning with @ and & are sorted under “A”; and

(e) the characters ‘, “, (, [, and { at the beginning of a name are ignored.

49 There can only be (and were only) minor tweaks within the narrow confines of an alphabetical arrangement. Item (c) does not alter the default rules of sorting at all, whether this is computer-driven or manual. However, in our judgment, the sorting rules (taken in their entirety) are neither entirely obvious nor inevitable. The other rules taken together would, in our judgment, cross the creativity threshold but just barely. There is little, if any, creativity in items (a), (d) and (e). However, item (b), in our view, barely meets the requisite level of creativity required to cloak the compilation with copyright. But the resulting copyright protection is extremely thin. As we stated above, the infringement and fair dealing analyses are heavily shaped by what it is that attracts copyright. In other words, the substantiality of copying is affected by the nature and degree of originality of the work. Accordingly, we take the view that infringement would only be established by nothing less than near-wholesale taking of the BL listings, arranged exactly as they were in the BL.

50 It will be recalled that the seeds were fictitious entries designed principally to detect and prove copying. GYP’s directories contained about 55 seeds each. Each seed was tied to GYP’s address (including its PO Box) but contained fictitious names, telephone numbers and unit numbers. Each name contained an additional unique identifier (for example, “Byd” or “Buyo”) unique to each directory and edition, such that the directory and edition from which copying took place could be traced. For example, a seed in a given year might read “Best Care Movers Byd Pte Ltd”.

51 GYP argues that it authored the seeds with a view to mislead and entrap. In particular, it says it exercised creativity by devising a fictitious name that was realistic enough to avoid detection by competitors but unique enough to prevent conflicts with genuine listings, by changing the unique identifier annually, and by using a real address or PO Box to track who was using its directories. GYP also argues that the policy reason against recognising copyright in short works does not apply here — it would neither impede the development of free speech nor deprive the public from accessing information.

52 In our judgment, GYP’s claim fails for two reasons, both of which stem from the fact that the very purpose of the seeds was to be copied. The first is that copyright cannot subsist in a seed as it is not a literary work to begin with. The many examples raised at [297]–[298] of the Judgment show the courts’ general reluctance in cloaking short works with copyright. A seed certainly does not provide “information, pleasure or instruction”, which the Judge held is the purpose of a literary work (see the Judgment at [297]). If anything, it misinforms a user of a phone directory. We agree with the Judge that the creation of a fictitious name and telephone number falls well short of the requisite standard of creativity. So too do we think that the year-on-year changes to the unique identifier fails to meet that standard. Second, it would surprise us if GYP claims that the copying of the seeds, which provides the requisite evidence to prove the fact that its data had been copied, causes it any loss or Promedia any gain. It is not GYP’s case that Promedia, for example, created its own seeds based on the fictitious names or unique identifiers created by GYP.

53 We turn to address the issue of infringement. Of course, given our findings on subsistence of copyright, only the near-wholesale taking by Promedia of the BL listings, arranged exactly as they were found in the BL, would give rise to infringement. We will consider this below. Before us, GYP does not seriously attempt to upset the Judge’s finding, with which we agree, that none of the GB directories infringe any copyright in GYP’s directories. As we stated above at [9], all Promedia took was the data in GYP’s directories. Its central argument on appeal is that Promedia’s temporary database was an infringing work. However, it faces obstacles both in terms of the pleadings and on the merits.

54 Promedia submits that the Judge erred in holding that the temporary database was covered by GYP’s pleadings. It is now clear, following our recent decision in L Capital Jones Ltd and another v Maniach Pte Ltd [2017] 1 SLR 312 at [33]–[73] (see especially [65]–[67] and [72]–[73]), that Promedia is procedurally entitled to raise such an argument under O 57 r 9A(5) of the ROC, which allows it to contend, without having brought its own appeal, that the Judge’s ultimate decision should be affirmed on grounds other than those he relied upon.

55 We agree with Promedia’s argument. The statement of claim (“SOC”) states that the infringing works comprised Promedia’s directories, which are the Online Directory, the printed form directories and the Green Book CD-ROM. The relevant extract of the pleading on infringement is set out at [56] below. The use of “namely” (instead of “including”) suggests that the list of allegedly infringing works is exhaustive, while the lack of any reference or allusion to databases suggest that they were not regarded as infringing works. In this respect, we respectfully disagree with the Judge, who held at [367] of the Judgment that the SOC was framed in sufficiently broad and general terms such that the temporary database could be regarded as an infringing work. Indeed, the Judge himself dismissed an application by GYP after the trial to amend the SOC so as to include the temporary databases within the list of infringing works.

56 The relevant paragraphs of the SOC (with the proposed deletions struck through and additions in underline) read as follows:

57 We would observe that the amendment application was dismissed for good reason. GYP argued that that it emerged only late in the trial that Promedia had scanned or photocopied GYP’s phone directories for the purpose of compiling its temporary database. However, this fact emerged as a result of cross-examination on the SOPs, to which GYP had access since at least 2013. Moreover, although this fact emerged in late October 2014, GYP filed its summons to amend its Statement of Claim only on 15 December 2014 — this was not only more than six weeks later, but also four days before the parties’ closing submissions were due.

58 In these circumstances, we are satisfied that GYP cannot rely on Promedia’s temporary database as an infringing work. Nonetheless, we will explain why, in any case, it would not succeed on the merits.

59 On the basis that Promedia’s temporary database could be regarded as an infringing work in this case, we are satisfied that no copyright in GYP’s works was even prima facie infringed. While it is true that the listings in the BL were scanned or photocopied in their entirety, that leads nowhere, because of the view we take on fair dealing as explained below.

60 GYP says that the Judge was wrong to find that only the listings in the BL (but not the other material) had been scanned or photocopied. In this regard, it points to Promedia’s SOPs. SOP 1075 instructs employees to photocopy and enlarge “every page” of the BL while SOP 0535 features a table that requires employees to record the events (being the facts relating to cutting, photocopying, scanning and conversion, reviewing and sorting, and program processing) at each page starting from page 1. GYP also says that, in any event, its copyright would be infringed should copyright be found to subsist in the listings alone.

61 In our view, the Judge’s findings were largely correct. We are not persuaded by GYP’s argument that the entire BL directory was scanned or photocopied because Promedia’s witnesses were never cross-examined on what was meant by “every page” in SOP 1075 or page 1 in SOP 0535.

62 We reproduce the relevant parts of SOP 1075:

63 First, the reference to “all the companies” “in alpha order” and the listing of “required fields” suggest that Promedia was interested only in the listings. Second, the stated purpose of photocopying each page – “to make comparison easier” – somewhat suggests that what was being compared was the data in the listings. Third, the instruction to dedicate a separate book for “Government & Quasi-Government Listings” reinforces the idea that it was only the listings that Promedia was interested in. Hence, we see no reason to disturb the Judge’s finding that only the business subscriber listings had been scanned or photocopied.

64 As for SOP 0535, we take the view that the table — especially the fact that whatever was taken had to be treated the same way and, in particular, sorted — suggests that the information required is homogenous and thus of a sort that can only be found in the listings, but, as the Judge also observed, not in the “other sections of the [BL], such as the front matter or introductory material” (see the Judgment at [335]).

65 We accept that copyright in the listings could prima facie be infringed by the photocopying or scanning of the entire listings found in the BL (see [49] above). As to this, Promedia contends that the finding of infringement that could flow from this should be limited to the years 2004, 2006 and 2008, because scanning or photocopying of the BL listings had not been shown in any other year. GYP’s response is that any finding of infringement should extend to every edition of the BL in respect of which a claim was brought, because systematic copying from the BL listings was Promedia’s modus operandi and was a key source for its master database. We agree with GYP, because we find it difficult to believe that Promedia would keep changing its operating procedures from year to year. However, our discussion on fair dealing below at [72]–[91] will make it apparent that nothing ultimately turns on this finding.

66 GYP complains that copyright in the YP was infringed because Promedia copied the “essence” of the YP by reproducing the subscriber data in its database by means of the YP Advertiser Module, by copying the classification and then tagging the subscriber’s listing to that classification. To this end, it refers to SOP 0312 (issued in August 2002) and SOP 0894 (issued in November 2006), which were dubbed “market intelligence” reports and which instructed Promedia’s employees to tear the YP directory into sections and to key its contents (namely, the company name, classification and level of advertising) into the YP Advertiser Module based on the publication and edition.

67 We do not agree with GYP. The copying of the “essence” of a work is an uncertain concept that could potentially turn the law of copyright on its head; there is either infringement by virtue of substantial taking or no infringement. In this case, we consider that the Judge was correct in finding that Promedia took only the information in the listings and the classifications. The SOPs were meant to instruct employees how to collect data in a systematic manner so that market intelligence reports could be created for the sales executives. Such reports would show when and how much a company was spending on advertising with other phone directory publishers, and appears to have been based on data which had been keyed in using a program. Specifically, we were referred to an interface called “Company Advertisement Record”, which only allows users to key in specific fields of data. It is clear to us that data from the YP directory was amalgamated with data from multiple directories since 1992 and keyed in and stored in a format which was very different from the arrangement found in the YP compilation. In other words, Promedia took bare facts, in which copyright could not subsist. The taking was voluminous but it was not substantial in the copyright sense. Accordingly, there was no infringement.

68 Finally, GYP says that its Online Directory was reproduced on a wholesale or substantial basis when Promedia’s employees saved or printed the webpages of the directory. To this end, it relies on SOPs which instructed Promedia’s employees to save or print the webpage in GYP’s Online Directory.

69 The Judge held that copyright subsisted in the Online Directory as a whole by virtue of the arrangement of listings, which he found to be creative in essentially three ways: (i) maintaining and introducing new category headings, (ii) deciding the depth of drill-downs and sub-categories, and (iii) determining the filters and keywords. Copyright was accordingly limited to these aspects. Viewed in this light, the saving or printing of individual web pages was not substantial in the copyright sense. The reproduction of individual web pages (as opposed to the entire web site) would have broken the structure of the Online Directory and caused the expression of the categories and filters to have been lost. The only way in which GYP can establish infringement is by showing that Promedia saved or printed a substantial number of pages in the Online Directory and arranged them according to GYP’s categories. That was not borne out on the evidence. The situation is even clearer in relation to the filters. The taking of the filters would have been quite impossible in the sense that the algorithm was governed by a database that was accessible only through GYP’s Information Technology Department and not to the regular end users of the Online Directory. The position might have been different if GYP had claimed and established that each individual web page enjoyed its own copyright, but this was not the case here.

70 GYP also points to a time study, which Promedia conducted, comparing the efficiency of printing (as opposed to saving) web pages. That study identified the individual actions that saving or printing would entail (such as the act of printing or saving the web page, collecting the printed pages, and highlighting relevant information either on the printed pages or using word-processing software) and their estimated duration. Its conclusion was that the saving of web pages was preferable to printing even though it would take about eight minutes per web page (whereas printing would take only six). GYP submits that the very fact that this study was conducted suggests that the scale of copying was vast, presumably because the matters studied, which seem minor in one’s daily experience, would only make a significant difference to Promedia in the context of large-scale copying. In our judgment, this argument gives GYP hardly any mileage beyond what we have discussed above. We agree with Promedia that this, without more, is hardly probative of widespread saving or printing of web pages from YP’s Online Directory. It is difficult to tell how much saving or printing is required to warrant a time study of this nature being done. Even if there was copying, GYP cannot pinpoint what was copied. We note that Promedia had downloaded information not only from GYP’s Online Directory but also from other directories, company web sites and the ACRA web site as well.

71 In these circumstances, we do not think that Promedia infringed any copyright save to the extent that it photocopied or scanned the BL listings onto its temporary database, as to which there is the objection on the pleadings which we have noted at [54]–[58] above, and the defence of fair dealing, which we deal with in the next section.

72 Even if GYP can overcome the defect in its pleadings, in our judgment, Promedia’s taking of the BL listings by photocopying or scanning them into its temporary database amounts to fair dealing within s 35 of the CA.

73 For convenience, we reproduce the relevant portions of that section:

74 Copyright law is meant to promote creativity and innovation by granting exclusive rights to copyright holders, but there is equally “a public interest in not allowing copyright law to hinder creativity and innovation” (RecordTV Pte Ltd v MediaCorp TV Singapore Pte Ltd and others [2011] 1 SLR 830 at [69]). The fair dealing provisions in particular embody “the legislature’s concern to safeguard the interests of the public at large within a framework of strong effective protection for copyright subject matter” (George Wei, “A Look Back at Public Policy, the Legislature, the Courts and the Development of Copyright Law in Singapore: Twenty-Five Years On” (2012) 24 SAcLJ 867 at para 17).

75 As the Judge observed, the first iteration of the fair dealing provisions was introduced in ss 35–37 of the Copyright Act 1987 (Act 2 of 1987), which provided that a fair dealing for the purpose of “research or private study”, “reporting news or current events” and “criticism or review” would not amount to an infringement of copyright. Section 35, in particular, was modelled after s 40 of the Copyright Act 1968 (Australia) (“Australian CA”), save that s 40(2)(c) (which required the court to consider the possibility of obtaining the work within a reasonable time at an ordinary commercial price) was omitted because of concerns that it would hamper disadvantaged students’ access to textbooks (Report of the Select Committee on the Copyright Bill (Bill No 8 of 1986) (Parl 9 of 1986, 22 December 1986) at pp A87–A88, B56 and D11; Singapore Parliamentary Debates, Official Report (5 May 1986) vol 48 at cols 19–20 (J B Jeyaretnam). Conceptually, Singapore’s approach mirrored the Anglo-Australian approach of having closed categories within which the fair dealing defence could operate.

76 Concerns were raised, however, after our decision in Creative Technology Ltd v Aztech Systems Pte Ltd [1996] 3 SLR(R) 673, where the defendant had sought to develop a sound card in competition with the plaintiff’s by means of reverse engineering, which entailed operating the plaintiff’s computer program. The defendant, by virtue of having operated the plaintiff’s program, was found liable for making an infringing reproduction; it could not however avail itself of the fair dealing defence because s 35(5) then excluded commercial research from the scope of fair dealing. This led to the broadening of the fair dealing provisions in two ways. In 1998, the restriction on commercial research being part of fair dealing was lifted through the repeal of s 35(5). In Parliament’s view, this brought Singapore in line with the position in the United States, the United Kingdom, Australia, Hong Kong and continental Europe (Singapore Parliamentary Debates, Official Report (19 February 1998) vol 68 at col 314 (Prof S Jayakumar, Minister for Law)). In 2004, the scope of s 35 was broadened such that a fair dealing for “any purpose” (as opposed to merely for “research or private study”) might be held not to amount to an infringement of copyright. This was done by amending s 35(1), adding s 35(1A) and adding s 35(2)(e) — this latter addition restored s 40(2)(c) of the Australian CA which had been omitted from the original enactment. In Parliament’s view, this was intended to “create an environment conducive to the development of creative works, and also facilitate greater investment, research and development in the copyright industries in Singapore” (Singapore Parliamentary Debates, Official Report (16 November 2004) vol 78 at col 1052 (Prof S Jayakumar, Deputy Prime Minister and Minister for Law)). This also made Singapore’s fair dealing provisions more similar to its American counterpart, which is more open-textured. While the underlying factors remain of Australian origin, four of the five factors also mirror those found in s 107 of the Copyright Act 1976 (US). Accordingly, although there are very few reported local cases that consider in detail the scope of and relationship between the factors in s 35(2), both American and Anglo-Australian jurisprudence will be helpful in shaping our law on fair dealing. With this brief history of the provisions, we now turn to each of the five factors under s 35(2).

77 The first factor, the purpose and character of the dealing, has received more attention than the other factors. This factor requires the court to compare the purposes of the infringing work and the original work and assess whether the dealing was fair. In this regard, the inquiry is heavily shaped by what it was in a work that attracted copyright and what was done with that aspect of the work. This is a corollary of what we stated above at [16] that there would be no infringement of such copyright by copying a work save to the extent that one copied the work as a whole, or a substantial portion of the part of the work that attracted copyright protection in the first place. There are many perspectives from which the first factor may be analysed, corresponding to the varied factual matrices from which copyright actions arise.

78 In general, the fact that the original and infringing works share the same purpose will tend to weigh against a finding of fair dealing. In Longman Group Ltd v Carrington Technical Institute Board of Governors [1991] 2 NZLR 574 (“Longman”), a decision of the High Court of Auckland, Doogue J held that it was not a fair dealing for the defendants, an educational institution and one of its tutors, to extract 70% of a book published by the plaintiff publisher for use in classroom instruction and discussion. Even though the court need only have held that there was no fair dealing because the effective purpose was not “research or private study” but rather to save the tutor from having to prepare teaching material, it went further and found that the defendants had compiled a work relying on significant parts of the copyrighted work and used it for “the same general purpose for which [those copyrighted works] have in the main been designed, ie the teaching of technical drawing” (at p 588, lines 45–47) (emphasis added).

79 In the early English cases, this factor was premised on the notion that one might fairly adopt part of another’s work “for the promotion of science, and the benefit of the public” (Cary v Kearsley (1802) 4 Esp 168 at 170, per Lord Ellenborough). Later cases extended this to include cases where the defendant added to, recontextualised or transformed the parts taken (see, for example, Newspaper Licensing Agency v Marks & Spencer plc [1999] EMLR 369 at 380, per Lightman J; University of London Press Ltd v University Tutorial Press Ltd [1916] 2 Ch 601 at 613–614). The American cases on the “transformative use” doctrine are to similar effect. To determine the extent to which the new work is “transformative”, the court considers whether it “supersede[s] the objects” of the original creation, or adds something new, with a further purpose or different character (Campbell, aka Luke Skyywalker et al v Acuff-Rose Music, Inc 510 US 569 (1993) (“Campbell”) at 578–579, per Souter J (delivering the opinion of the court), referring to Folsom v Marsh 9 F Cas 342 (1841) (CCD Mass) at 348, per Story J).

80 As against this, in general terms, the fact that a dealing is driven by commercial exploitation will weigh against a finding of fair dealing. For example, fair dealing was not found by the High Court in RecordTV Pte Ltd v MediaCorp TV Singapore Pte Ltd and others [2010] 2 SLR 152, which essentially involved a commercial provider of time-shifting technology. The plaintiff owned an internet-based service which allowed users to request that free-to-air broadcasts be recorded on its servers for them to view these at a later time from their computers, usually at home. The defendants were the copyright owners of some of these works. After receiving cease and desist letters from the defendants, the plaintiff brought an action for groundless threats of copyright infringement. Among other things, the plaintiff relied on the fair dealing provision to make good its contention that it had not infringed the defendants’ copyright. On this latter point, Andrew Ang J held that s 109 of the CA (which is in pari materia with s 35 of the CA, save that it applies to audio-visual works) countenanced a balancing exercise between commercial gain and public benefit, but his conclusion that the fair dealing plea failed was premised on his view that the plaintiff’s service was for private profiteering. He was “greatly influenced” by the fact that the plaintiff’s dealing involved “a commercial project with the eventual goal of monetisation through advertisements and the licensing of advertising technology” (at [109]). While allowing an appeal on other grounds, we expressed no opinion on the fair dealing issue.

81 Having said this, we do not go as far as those cases which suggest that a commercial nature or purpose of the dealing will presumptively be regarded as unfair (see, for example, Re Brian Kelvin De Garis and another v Neville Jeffress Pidler Pty Ltd (1990) 18 IPR 292 (FCA) (“De Garis”) at [48]–[49] (“De Garis”); citing Pacific Southern Co Inc v Duncan 744 F.2d 1490 (11th Cir, 1984) at 1496, per Johnson J; Sony Corp of America v Universal City Studios, Inc 464 US 417 (1984) at 451 per Stevens J (with whom Burger CJ and Brennan, White and O’Connor JJ agreed)). In our judgment, the commerciality of the dealing is but one of the factors to be considered and it will not necessarily be fatal to a finding of fair dealing. As an example, the commerciality of the dealing was soundly trumped by transformative use in Campbell, where the US Supreme Court held that it was a fair dealing for a popular rap band to write a song, “Pretty Woman”, the lyrics of which satirised an original song, “Oh, Pretty Woman”. In the court’s view, parody had an obvious claim to transformative value (at 579), even if it did not raise a presumption of fair use (at 581). The court held (at 584):

82 The second factor, the nature of the work, draws on the “value of the materials used” (Campbell at 586, citing Folsom v Marsh 9 F Cas 342 (1841) (CCD Mass) at 348, per Story J). The point is that some works are “closer to the core of intended copyright protection than others, with the consequence that fair use is more difficult to establish when the former works are copied” (Campbell at 586). Thus, a finding of fair dealing is less likely in cases involving fiction-based (as opposed to fact-based) copyright works, because copyright protection is not as “thin”, and in cases involving unpublished works, because of the need to respect an author’s right to confidentiality and to release the work into the public domain (Ng-Loy Wee Loon, Law of Intellectual Property in Singapore (Sweet & Maxwell, 2nd Ed, 2014) at para 11.3.30).

83 The third factor, the amount and substantiality of the portion used in relation to the copyrighted work, asks whether the quantity and value (that is, its qualitative importance) of the materials used are reasonable in relation to the purpose of the copying (Campbell at 586–587).

84 The fourth factor, the effect of the use on the potential market for the value of the copyrighted work, requires the court to consider “not only the extent of market harm caused by” the alleged infringer’s action, but also whether the defendant’s conduct, if “unrestricted and widespread”, would “result in a substantially adverse impact on the potential market” for the original, taking into account not only harm to the original but also harm to the market for derivative works (Campbell at 590; Harper & Row Publishers, Inc. v Nation Enterprises 471 US 539 (1985) at 569).

85 The final factor, the possibility of obtaining the work within a reasonable time at an ordinary commercial price, contemplates that a finding of fair dealing is less likely if the defendant could have obtained the work on reasonable commercial terms. It has been suggested that this entails two inquiries: whether the defendant made reasonable investigations into the possibility of obtaining the work on reasonable terms and, if it was possible to obtain the work on such terms, whether he attempted to so obtain it (David Tan and Benjamin Foo, “The Unbearable Lightness of Fair Dealing: Towards an Autochthonous Approach in Singapore” (2016) 28 SAcLJ 124 at paras 44–47 (“Tan & Foo”)). This factor also relates to the fourth factor in the sense that both factors are premised on the notion that the market for the plaintiff’s work should not be usurped by the defendant without compensation.

86 In our judgment, the inquiry under s 35(2), in the final analysis, is necessarily fact-sensitive. As was said in Longman at 587–588, “the issue of whether there has been fair dealing is one of fact and impression having regard to all the circumstances”. In the light of the principles we have outlined above, we turn to consider whether it was fair dealing for Promedia to scan or photocopy the BL listings.

87 Crucially, it must be recalled that it was the arrangement of the listings, in particular the narrow ways in which the sorting rules departed from the default alphabetical (or even alphanumeric) arrangement, that attracted copyright in the first place (see above at [49]). However, we stated that such copyright was very thin and would be prima facie infringed only by a near-wholesale taking of the listings, arranged exactly as they were found in the BL. This was the case because the BL listings were reproduced by scanning or photocopying.

88 The purpose of such photocopying or scanning, as the Judge found at [399]–[400] of the Judgment, was simply to make it easier for Promedia’s employees to compare or identify listings not found in its database and to update its database. This was an internal exercise that is at best characterised as being incidental to commercial research; the scanned or photocopied version of the BL listings were never publicly distributed. Moreover, how the listings were arranged, which presumably was to help users locate the correct entry easily, was of no utility or consequence to Promedia. There is no evidence that GYP’s rules of arrangement were deployed in the master database into which Promedia saved the data it collected, or even in the GB. GYP submitted that this case was similar to De Garis where the Federal Court of Australia held that it was not fair dealing for the defendant to search for newspaper articles such as those written by the plaintiff and to supply them to customers for a fee, because such an exercise was more akin to an information audit than research (De Garis at [27]). We cannot agree with this analogy because, in that case, the copyright in question subsisted in the content of each article, and that very same content was supplied to customers for profit. That is quite unlike the present case where even though the scanning and photocopying took place in a commercial setting, the purpose of the dealing in this case was to access the data that was contained in the listings, and had nothing to do the use of the particular arrangement of that data, which is what attracted copyright in the first place. Accordingly, the purpose of the dealing in this case does not weigh against a finding of fair dealing.

89 It is important, in our judgment, that this work is largely fact-based and lies at the very edge of copyright protection, having acquired the barest of copyright only by virtue of the arrangement of listings. The nature of the work, therefore, favours a finding of fair dealing under s 35(2)(b). Similarly, given the narrow way in which copyright subsists, it would be inappropriate to place too much weight on the substantiality of the taking in relation to the whole work under s 35(2)(c).

90 We also do not consider that the potential market for GYP’s works would be usurped by such a dealing under s 35(2)(d). Had Promedia used multiple hard copies of the BL instead of photocopying or scanning it, Promedia would not have prima facie infringed any copyright. As the BL was distributed freely, we also do not consider that there was any point in Promedia seeking a licence from GYP to photocopy or scan a copy of its BL; in any case, what they were after was the factual content in the BL and this was simply not protected by copyright. Accordingly, s 35(2)(e) would not disfavour a finding of fair dealing as well. In the circumstances, we conclude that the scanning and photocopying of the BL listings was fair dealing.

91 As we agree with the Judge that the GB directories and the rest of the temporary database did not infringe any copyright in GYP’s directories, the issue of fair dealing does not arise in relation to those matters and it is not necessary for us to express any view on this.

92 On appeal, GYP also argues, based on ss 86 and 91 of the CA (popularly known as a publisher’s right or a copyright in the layout or typeset of a published work), that copyright subsisted in the published edition of its works and had been infringed. GYP stresses that this position had been pleaded and, if not, then it should be allowed to amend its statement of claim on the ground that Promedia would suffer no prejudice. Promedia’s reply essentially is that the position had not been pleaded and an amendment would cause irreparable prejudice; that in any event GYP had abandoned its claim for such a copyright; and further, that in any event the fair dealing defence would apply.

93 We are prepared to accept that the material facts concerning copyright in layout or typeset by virtue of s 91 of the CA have been pleaded, and even that the evidence on record suffices to prove that copyright in layout or typeset subsists by virtue of s 91 of the CA. However, GYP faces four more related and ultimately insurmountable obstacles.

94 First, this claim was not pursued below and should not be allowed to be pursued on appeal because it would prejudice Promedia. During cross-examination of GYP’s publishing director, counsel for GYP interjected to state in no uncertain terms that there was no claim for “layout”:

95 Second, there appears to be no reference in GYP’s pleadings or closing submissions to an infringement of copyright in the typeset or layout. The tenor of the pleadings and closing submissions suggests that the type of copyright that subsisted was by virtue of the arrangement and selection of contents in a compilation (under s 27(2) of the CA) rather than by virtue of typeset or layout (under s 91 of the CA). In its closing submissions, GYP set out the criteria for subsistence by virtue of s 27(2) CA and then sought to show that it had proved each element. The only reference to the Abacus System (which is the software used to typeset the directories) was to address Promedia’s potential counter-argument that the use of such software obviated the need for GYP’s employees to exercise skill, judgment or creativity. This, as noted above, went to copyright under s 27(2) of the CA.

96 Third, even assuming that GYP can establish infringement on the basis of photocopying and scanning of the BL listings, we would be minded to hold that the fair dealing provision under s 109(3) of the CA (which reproduces mutatis mutandis the language in s 35 of the CA) would apply. Crucially, the reproduction of the typeset was incidental and the scanned or photocopied version of the BL listings were never publicly distributed.

97 Finally and in any event, Promedia would not have been enriched by any infringement of the copyright in the typeset, nor would GYP have suffered any other loss.

98 For these reasons, the Judge was correct to hold that GYP’s claim ultimately failed in its entirety. We now turn to GYP’s appeal on Promedia’s counterclaim for groundless threat of copyright infringement.

99 GYP argues that it should not be liable to Promedia under s 200 of the CA for making groundless threats of copyright infringement. GYP argues that the letters of demand it sent were not threats and that, in any event, they were justifiable. Promedia, on the other hand, argues that the letters of demand were groundless threats and were intended to chill its legitimate activities and achieve an improper monopoly, the very mischief that the groundless threat provisions were meant to address. On this issue, we agree with GYP.

100 The law on groundless threats was recently restated in Singsung Pte Ltd v LG 26 Electronics Pte Ltd (trading as L S Electrical Trading) [2016] 4 SLR 86 (“Singsung”), which was decided between the time GYP’s and Promedia’s cases were filed but which GYP had the opportunity to address in its Appellant’s Reply. There, we stated that s 200 of the CA represents a balance between protecting copyright on one hand and preventing “bullying” tactics directed at competitors or their customers to chill their legitimate activities; it thus provides a statutory remedy for aggrieved parties whose business or reputation might be affected by threats emanating from another party without the need to prove the bad faith of the threatening party (at [129] and [133]). We then set out the approach and how it might be applied:

101 The crucial point in the foregoing passage is that relief under s 200(1) of the CA is discretionary. The court’s task is to consider whether, in all the circumstances, there is any reason for it to grant relief upon a claim of a groundless threat founded on a failed allegation of infringement. This will turn on the facts and circumstances of each case; some relevant considerations in particular are whether the action was warranted, whether any conceivable damage flowed from the demand having been made, what the costs consequences from the failed action were or would be, and whether relief was required at all.

102 The two letters of demand, in which the threats were allegedly found, were sent on 12 August 2009 and 9 September 2009 respectively. This was shortly before the action was commenced on 27 October 2009. The relevant claims in the first letter read:

103 The relevant portions of the second letter read:

104 In our judgment, Promedia does not succeed on its groundless threat claim. Essentially, we consider that the letters were neither unwarranted nor meant to chill Promedia’s legitimate activities. We are unimpressed by Promedia’s argument that GYP’s action was to bully other directory publishers to achieve a monopoly. As GYP points out, the three other actions it had brought against other publishers of telephone directories were settled on a “without prejudice” basis and since we are in no position to establish why this happened, this fact should be regarded as neutral. As for GYP’s action against Promedia, GYP had ample factual and legal basis to think that Promedia had copied a substantial amount of its works and could therefore be liable in copyright infringement.

105 Although we agree with Promedia and the Judge that the threats were strictly speaking overbroad, this does not change our conclusion for several reasons. First, there is little force, in our view, in Promedia’s argument that the letters in Singsung claimed copyright in specific works and contained highly particularised threats, whereas the copyright claims in the present letters covered a broad category of works. We do not hold this against GYP because a telephone directory (especially the Online Directory) is an iterative work with many components and sub-components in which different copyrights might independently exist.

106 Second, the reference to “artwork” was not taken seriously by either party after the letters had been sent. There was no claim made or evidence led for the subsistence or infringement of copyright in “artwork” and Promedia has not suffered any loss in this regard that is not compensable by costs if indeed even this can be proved.

107 Third, the claim for copyright in “information” arguably went beyond the industrious collection or “sweat of the brow” approach supposedly applied in Virtual Map (DC). However, we think it is plausible that “information”, when read in context, refers to the compilation of information, which could enjoy copyright on one interpretation of Virtual Map (DC). At any rate, we are not minded to read the letter as we would a statute.

108 Finally, in our respectful view, the Judge adopted an overly strict standard insofar as he held that a threat would be unjustifiable if it was not supported by “clear binding authority” (see the Judgment at [414]–[415]). The approach taken by Murphy J in Telstra Corporation Limited v Phone Directories Company Pty Ltd (2014) 316 ALR 590, on which the Judge relied, appears to have simply espoused a sliding-scale approach insofar as Murphy J held that the result depended on the “objective strength of the legal position underpinning the threat” (at [672]). What was shown in that case was that it was sufficient — not that it was necessary — that the threat was made on the back of “strong Full Court authority” for it to be justifiable (at [681]). Also, the “clear binding authority” standard would potentially render every threat in penumbral cases unjustifiable and might discourage even incremental changes to the law.

109 Promedia on the other hand submits that, on this view, the groundless threat provision would end up being overly narrow. However, that is precisely our view. The substance of this provision is not that it is a remedy that follows as a matter of course for a successful defendant. In this case, Virtual Map (DC) supports the “sweat of the brow” approach on one interpretation and even though it was only the District Court that had given full consideration to the issue, its analysis received no adverse comment from the High Court (which proceeded on the concession that copyright subsisted in the street directory or address point data of Singapore in vector format (Virtual Map (Singapore) Pte Ltd v Singapore Land Authority [2008] 3 SLR(R) 86) or the Court of Appeal (which denied leave to appeal on the basis that neither the District Court nor the High Court were prima facie in error (Virtual Map (Singapore) Pte Ltd v Singapore Land Authority and another application [2009] 2 SLR(R) 558)). We therefore accept GYP’s submission that its threats were not “unjustifiable” based on the state of the law at the time they were made. Essentially, GYP, as was the case for the claimant in Singsung, “sued and lost and there is nothing more that needs to be said or done in this regard”.

110 Accordingly, we conclude that the Judge was, with respect, wrong to have found GYP liable in groundless threats of copyright infringement under s 200 of the CA, and we reverse this finding accordingly. To be fair to the Judge, he had neither the benefit of reading the actual letters of demand, which were inadvertently omitted from the record before the High Court, nor the chance to consider our decision in Singsung, which was rendered after the Judgment was released.

111 For these reasons, GYP’s appeal is allowed only in relation to the issue of subsistence of copyright (though only insofar as it relates to the arrangement of the listings in the BL and even though this had no consequence to the outcome), and the issue of groundless threats of copyright infringement, and is otherwise dismissed.

112 Each party is to file, within two weeks of the date of this judgment, one set of written submissions (not exceeding 5 pages) in respect of the appropriate order as to the costs of the appeal and the application that was before us (including the appropriate quantum of costs).

113 We once again record our deepest gratitude to Prof Llewelyn, whose submissions were clear, concise and illuminating.