How the Internet is Reshaping Defamation Laws
Defamation is an example of how the common law is continually updating a traditional doctrine to it relevant and applicable in the age of the Internet. Traditional notions of defamation law may not lend themselves readily to online speech and conduct, and Internet defamation claims now turn on a host of new developments to the elements required to prove a defamation claim.
This article explores how each of the three elements of defamation, namely publication, identification, and meaning, differs substantively and procedurally from their traditional roots when applied to defamation that happens on the Internet.
Some of us were born into the age of the Internet. For many others, the Internet, although very much a mainstay in our modern lives, remains a relatively new epitome of technological advancement. It is no longer a stretch to suggest that most of our activities (be they for work or leisure) are now carried out online. Along with this came a need for the law to regulate the transactions, conduct, and behaviour that takes place in the virtual realm. In order to keep up with the realities of the Internet, new laws had to be enacted (for example, those concerning privacy, data-protection, online payments etc). Existing legal doctrines may also have been amended or reformed.
Defamation is an example of how the common law is continually updating a traditional doctrine to make it more relevant and applicable in the age of cyberspace. Defamation is a venerable tort. For a long time, the law was fashioned to deal with traditional modes of communication such as print and broadcast. But much has changed in the last decade, during which the common law scrambled to cope with Internet-based defamation. While the elements required to make out a defamation claim remain the same as before, the ways in which these conditions must be proven have changed. New doctrines and defences have also arisen to address the influx of defamation cases emanating from postings on the Internet, including on social media platforms like Twitter and Facebook.
This article explores how defamation law in Singapore today, when applied to online content, differs substantively and procedurally from its traditional roots. It is not possible to detail all the differences; only the key ones that practitioners ought to be mindful of when conducting Internet defamation cases will be discussed.
1. Publication: There is No Presumption of Publication Just Because Content Exists Online
The first requirement for making out defamation is that of publication: the content in question must have been communicated to at least one third party. It is not sufficient for the content in question to simply exist; rather, a claimant must show that a third party has seen and comprehended the content.1Gatley on Libel and Slander (Alastair Mullis & Richard Parkes eds) (Sweet & Maxwell, 12th Ed, 2013) at para 6.1
Where Internet defamation is concerned, how publication is to be proven appears to vary according to the platform in question.
There have been instances where a court deemed the publication requirement as having been satisfied on the mere basis that content was posted on the Internet. Golden Season Pte Ltd and others v Kairos Singapore Holdings Pte Ltd and another  2 SLR 751 (Kairos) concerned a series of Facebook posts that were alleged to have contained defamatory content. The High Court did not analyse the requirement of publication in detail. Instead, it readily accepted that the publication requirement was satisfied by virtue of the Facebook posts having been made and posted. A reading of Kairos would thus suggest that posts made on social media sites such as Facebook and Instagram can be deemed published based on the just the existence of the post.
The position was radically different in an earlier High Court case – Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another  4 SLR 977 (Qingdao). Here, content of interest was found in articles put up on various websites. The High Court held that publication cannot be presumed just because material is uploaded or posted on the Internet. That is merely one component of the publication element. A second component must be satisfied, under which a plaintiff must prove that the material had been accessed and read by a third party in the jurisdiction where the plaintiff is suing.2Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977 at (34) to (41). Mere existence of content on website is not sufficient. This position has been widely accepted in a number of other Singapore and English decisions.3Al Amoudi v Brisard and another (2007) 1 WLR 113 (cited with approval in Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (27) and Zhu Yong Zhen v AIA Singapore Pte Ltd and another (2013) 2 SLR 478 at (44)).
One can perhaps reconcile Kairos and Qingdao by limiting the position in Kairos only to social media, where the presence of “likes” or comments made to a post can lead to an inference that the post has been accessed and interacted with. Absent such indications, the general proposition in Qingdao will apply.
How Does One Prove that Content Has been Accessed, Downloaded, and Read?
There are many ways to show that published web content has been accessed and read by third parties. This is an important inquiry: not only does such evidence prove “publication”, it also reveals the extent to which the content was circulated – a key factor used to determine damages payable after liability is established.
The most straightforward way is for a plaintiff to call witnesses who can testify to the fact that they have personally accessed and read the online content in question. If a court accepts such evidence as credible, publication can be made out.4Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977 at (122) to (135) However, this approach may not practically feasible when the plaintiff believes that the content was read by a large number of people, or when the exact identities of the readers are unknown to the plaintiff.
Where direct proof is not possible, an inference can be made. An assortment of cases suggests that publication and its extent can be inferred using certain indicia common to the Internet. For example, courts have relied on statistics that detail the number of times a web page was visited to determine how widely content on that page was accessed.5Lee Hsien Loong v Ngerng Yi Ling Roy (2016) 1 SLR 1321 An inference of content having been accessed and read may also be made where a plaintiff can show evidence of the content showing up prominently on search results when certain key terms are keyed into an Internet search engine6Steinberg v Pritchard Englefield (a firm) (2005) EWCA Civ 288 (affirmed in Zhu Yong Zhen v AIA Singapore Pte Ltd and another (2013) 2 SLR 478 at (46); See also Qingdao at (138) to (141). (although this inference ought not be drawn if the content only surfaces after a highly specific combination of search terms are entered7Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (31).). In the case of content sent by e-mail, courts have accepted that, where the email was sent to a large number of people, it can be inferred that a “not insubstantial number of persons would have downloaded and read the email”.8Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (45). And where social media is concerned, courts have regarded the presence of follow-up comments left by third parties as proof that content has been accessed, read, and therefore published.9National Union of Bank Employees v Norzeela bt Latimah and anor suit (2014) 7 MLJ 31 at (127). Content that was shared and forwarded by third parties (such as in the case of a Twitter post that had been re-tweeted10Dato Mohamad Salim Fateh bin Fateh Din v Nadeswaran a/l Rajah (No 1) (2012) 10 MLJ 203 at (53).) have also been regarded by courts as having been extensively published.
The main observation on proving publication in Internet defamation is this. Direct testimony of individuals having accessed and read the content in question is good evidence of publication. In absence of such evidence, publication can still be inferred from other facts, but what these facts are will depend on the platform or medium in which the content is found or circulated.
The Extent of Publication No Longer Affects Only Assessment of Damages
Traditional defamation law imputes liability on a defendant as long as the plaintiff can show that the offending content was published to just one other third party. How widely the content was published is a factor that goes towards assessing the amount of damages a plaintiff ought to be awarded, and does not affect any finding of liability.
This distinction between liability and damages was blurred by the English case of Jameel (Yousef) v Dow Jones and Co Inc  2 WLR 1614 (Jameel). In Jameel, the court struck out a plaintiff’s defamation on the basis that, while the content in question was deemed defamatory, its publication was extremely limited. Although the plaintiff might eventually succeed in establishing liability on the part of the defendant, the plaintiff would ultimately receive only very nominal damages because of the limited publication. This would in turn render the cost of the lawsuit out of proportion to the damages the plaintiff could receive. The English Court of Appeal regarded this as an abuse of process, and proceeded to strike out the plaintiff’s claim.
This principle, now commonly referred to as the Jameel Doctrine, has been adopted and applied as good law by both the Singapore High Court11Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977. and the Court of Appeal.12Yan Jun v Attorney-General (2014) 1 SLR 793. In fact, the High Court in Qingdao took this doctrine further by suggesting (albeit in obiter) that not only can a defamation suit be struck out (on the application of a defendant) based on the Jameel Doctrine, a court can rely on the Jameel Doctrine as a ground for dismissing a defamation claim.
In the wake of these developments, one can no longer say that the extent of publication affects only damages. The extent now strikes at the heart of liability. The Jameel Doctrine may be relevant where a plaintiff sues over content that was circulated to a closed and limited group of people, e.g. in a small WhatsApp chat group, on a private Facebook page, or over e-mail to a small number of recipients. A suit like this may be dismissed or struck out based on the Jameel Doctrine even though all the requirements traditionally needed to prove a defamation claim are satisfied.
It is common for social media users to restrict views to their accounts on a “friends-only” or “followers-only” setting. Users have the options to control who may view a particular post. These settings do not permit the world at large from viewing a user’s account or posts. What difference would it make, if a defamatory post was made on a “friends-only” page or post, as opposed to one that is publicly viewable by the world at large? Would the Jameel Doctrine intervene to render a closed or private post unactionable on the basis of limited publication?
The answer to that would likely depend on the exact number of “friends” or “followers” who accessed and read the post. Celebrities can command a friend or follower list of thousands or even millions on their social media accounts. It is also common for the average person to have at least a few hundred friends on social media.
In contrast, the Jameel Doctrine, where successfully applied, involved situations where five or less people had read and accessed the defamatory content. In Jameel itself, the publication was read by five people. In Qingdao, the publication was deemed by the High Court to have been read by one person. In Kesavan Engineering & Construction Pte Ltd v S P Powerassets Limited  SGDC 179, a District Court case that applied the Jameel Doctrine, four people read the defamatory content.
To date, there is no express judicial indication on how limited publication needs to be before the Jameel Doctrine is triggered. Based on the tenor of existing cases, it is safe to surmise that, where a defamatory statement gets posted on a private account that nevertheless has a following of hundreds, the Jameel Doctrine may not apply. In other words, having a private or closed account is by no means a safe harbour from potential defamation claims.
On the other hand, a defendant could potentially argue that not all of his 300 followers read the offending post (a fact that a plaintiff would find incredibly difficult to prove). For example, if a post was liked and/or commented on by only five out of a defendant’s 300 followers, it might still be open to the defendant to invoke the Jameel Doctrine based on a limited publication to five readers. Arguments of this nature have not been judicially accepted in our courts, but they do appear sensible given how it is not safe to presume that every post put up on social media account will invariably be read by every follower of that account.
2. Identification: No Reliance on External Content
The second requirement that must be satisfied in making out a defamation claim is that of identification. A statement that a plaintiff claims is defamatory must have referred to the plaintiff. Such reference can either be direct or indirect. In any case, general defamation law provides that this identification element is present where an “ordinary reasonable person who, at the material time, was aware of the relevant circumstances or special facts (if any) would reasonably understand the plaintiff to be referred to by the offending words”.13Review Publishing Co Ltd and anor v Lee Hsien Loong and anor Appeal (2010) 1 SLR 52 at (49). This was further clarified as whether “a reasonable person who was acquainted with the plaintiffs, on reading the article, would come to the conclusion that the offending words referred to the plaintiffs even though the plaintiffs were not expressly identified in the article.” The plaintiff thus “need not be expressly referred to by name in the offending words; it is also immaterial whether or not the defendant intended to refer to the plaintiff.”
Identification poses special concerns for Internet defamation in cases where the original content contains no references to the plaintiff, but external content associated with the original does. It is not difficult to see how this can often happen on the Internet, where content from different sources and across different platforms or sites can be easily linked. Take, for instance, where John writes an article on a webpage but does not identify who the article is referring to. Mary comes along independently of John, and leaves a comment on the article identifying Hank as the subject of article. Can John be made liable for defamation if Hank sues John?
Case law in this area does not provide a definitive answer, likely because of the numerous ways and manners in which cross-publication identification can arise. That having been said, Kairos offers some clarity into whether a plaintiff can rely on identification in one publication to show defamation in another publication.
Kairos concerned two Facebook posts. The first comprised the original post and a follow-up comment, both of which were made by the same defendant. These was no reference to the plaintiff in the first post. The second comprised the original post made by a defendant, and, among other things, a comment made by the plaintiff. The plaintiff identified himself by leaving the comment to the second post.
The facts of Kairos bring to fore a number of questions. First, the High Court held that a Facebook post and its comments should constitute a single publication because they will likely be read together.14Golden Season Pte Ltd and others v Kairos Singapore Holdings Pte Ltd and another (2015) 2 SLR 751 at (55). The logical corollary (although not expressly stated by the High Court) is that, if a defendant makes a Facebook post that does not identify a plaintiff but subsequently leaves a follow-up comment that does so, the identification element would be deemed satisfied because the original post and the comments constitute one single publication.
In Kairos, identification did happen in the comments to the second post. However, it was the plaintiff who made the comment, and therefore confirmed his own identity. As a result, the defendant cannot be held liable for the plaintiff’s self-identification.
Still, this leaves the following question unanswered. Recall the John, Mary and Hank example above. If a defendant makes a post with no identification, but someone else leaves a comment independently of the defendant that contains identification, will the defendant be liable? Kairos did not address this point, but two possible approaches exist.
The first possible approach would be that the defendant should not be liable because the identification did not come from him (unless, of course, the comment containing the identification was made at the behest or with the knowledge of the defendant). The defendant may, nonetheless, be made liable after he was put on notice of the defamatory nature of the post and the presence of an identification comment, but refused to remove the post and/or the comment.15This is similar to the well-established common-law doctrine that holds forum owners, website administrators, and account owners liable for defamatory content on their platforms only after they are notified of the content but fail to remove the content or delay the removal. See, for example, Godfrey v Demon Internet Ltd (2001) QB 201; Davison v Habeeb (2011) EWHC 3031; Tamiz v Google Inc (2013) 1 WLR 215; Duffy v Google Inc (2015) SASC 170; Wishart v Murray (2013) NZCA 461.
The second possible approach is to hold the person who made the comment containing the identification liable for defamation, because he had “completed” the defamation. This approach may be problematic because the commenter ultimately did not author or make the offending statements. However, cases like Vischer v Maritime Union of Australia (No 6)  NSWSC 350 have suggested that if, for example, Hannah makes a defamatory post and Kenny shares a hyperlink to the post together with comments endorsing or approving of the post’s content, Kenny can be held liable for defamation. If leaving a comment containing identification could be likened to endorsing or approving a post that was otherwise silent on identification, the commenter might possibly be held liable.
The other key issue the the High Court in Kairos had to consider was whether the plaintiff could rely on the identification present in the second post (notwithstanding it having been a case of self-identification by the plaintiff) to establish identification in the first post (where there was no reference whatsoever to the plaintiff). In order to succeed, the plaintiff would have to prove that a reasonable reader of the first post would also have read the second post. And if the second post was made after the first, a reader would have to read the first post without knowledge that any second post would later be made, read the second post, then go back to the first post to make sense of it again.
The High Court held that, on a balance of probabilities, this sequence of events was unlikely to apply to all readers. Accordingly, a plaintiff should not be able to rely on a subsequent post to establish reference to himself in a preceding post.16Golden Season Pte Ltd and others v Kairos Singapore Holdings Pte Ltd and another (2015) 2 SLR 751 at (61).
Case law does not appear to have dealt with a situation in reverse: whether a plaintiff can rely on an earlier post to establish identification in a subsequent post. A possible view would be to adopt the reasoning in Kairos and to surmise that it is unlikely a reasonable reader would read the first post, keep it in mind, and when he eventually reads the second post, would know from his recollection of the first post that the second post refers to the plaintiff. Also, whether a plaintiff would succeed in a case like this will depend on a host of other factors such as how temporally far apart the two posts were, whether they were made on the same platform, whether the author of the posts provided express links between the two posts etc.
3. Meaning: Statements on Social Media May Not be as Defamatory as They Seem to be
A plaintiff wishing to succeed in a defamation claim must, finally, show that the statements complained of were defamatory of the plaintiff. Content may be defamatory in two general ways: first, in their natural and ordinary meaning, or second, in their innuendo meaning.17Review Publishing Co Ltd and anor v Lee Hsien Loong and anor Appeal (2010) 1 SLR 52 at (26)
A determination of whether content is defamatory relies heavily on the perspective of a hypothetical “right-thinking member of society”.18Sim v Stretch (1936) 52 TLR 669. The broad question often formulated and asked is this: would such a member of society think less of the a plaintiff after reading the statement?
This “reasonable-person”-based inquiry remains to be a starting point when determining if an online statement is defamatory. 19Stocker v Stocker (2019) 2 WLR 1033 at (36) and (37): “The touchstone remains what would the ordinary reasonable reader consider the words to mean…All of this, of course, emphasises that the primary role of the court is to focus on how the ordinary reasonable reader would construe the words.” However, some foreign cases have suggested that, in the online world, words and their meanings can have a very different import. One such recent authority is the English case of Stocker v Stocker  2 WLR 1033 (Stocker). This case concerned an exchange on Facebook between a man’s (the eponymous Mr Stocker) ex-wife and his new partner. In that exchange, the ex-wife claimed that, when she was married to Mr Stocker, Mr Stocker “tried to strangle” her. Mr Stocker sued his ex-wife for defamation based on these words: he claimed that the meaning intended by his ex-wife was that he had tried to kill her.
The UK Supreme Court ultimately decided that the statement was not defamatory. It highlighted a need for courts to be conscious of the context in which online statements are made. In the court’s words, “[a] judge tasked with deciding how a Facebook post or tweet on Twitter would be interpreted by a social media user must keep in mind the way in which such postings and tweets are made and read”.20Stocker v Stocker (2019) 2 WLR 1033 at (41). The court also observed that parsing a Facebook post for its “theoretically or logically deducible meaning”21Stocker v Stocker (2019) 2 WLR 1033 at (43). would be unwise given that Facebook and Twitter are “casual”22Stocker v Stocker (2019) 2 WLR 1033 at (43). and “fast moving”23Stocker v Stocker (2019) 2 WLR 1033 at (44). media, where users scroll through quickly, do not pause to reflect, do not ponder too deeply the meaning of a post, and ultimately derive only impressionistic and fleeting reactions to content.
Stocker has not been applied in Singapore, but is well worth studying in its entirety for anyone interested in the impact of social media on defamation law. Its key holding can be stated as follows: one would not read a social media post as one would a newspaper article; in that same vein, courts should determine the meaning of an allegedly defamatory statement while having regard to the fact that content on social media is perused rapidly, cursorily, and superficially. This has the effect of raising the defamatory threshold for content made on social media.
This proposition, however, should be limited only to social media. There are many other sources of online content such as news sites, blogs, and e-mails that attract higher levels of thought, attention, and analysis from its readers. The observations in Stocker cannot realistically apply to these other sources.
Change is the Only Constant
The need to balance freedom of expression with an individual’s right to reputation has made defamation law more nuanced than it appears to be. The advent of Internet and the added ease of expression (on top of freedom) it brings makes striking that balance trickier.
The law often attracts criticism for failing to evolve in tandem with changes in the real world. Yet, this cannot be said of the law concerning defamation. Courts around the world are continually trying to rationalise the way in which we interact and deal with Internet content. There are, in fact, many other aspects of defamation law that this article has not covered for the sake of brevity. A key area would be how the Internet may have changed the defences to a defamation claim. For example, can qualified privilege – a defence commonly deployed by news outlets and journalists – now be extended to private individuals acting as citizen journalists? Should courts expand the definition of public interest to cover social issues “trending” on the Internet when determining if the defence of fair comment should apply? As the elements of defamation evolve, so should defences, although these discussions must be left for another article.
We are beginning to realise that traditional notions of defamation may not lend themselves readily to online speech and conduct, and that how Internet defamation claims ought to be approached must now depend on a whole new host of factors, including the medium of publication, readership patterns, and even human psychology.
This trend is best summed up by the words of the United States Supreme Court in Packingham v North Carolina 562 U.S (2017):
“While we now may be coming to the realization that the Cyber Age is a revolution of historic proportions, we cannot appreciate yet its full dimensions and vast potential to alter how we think, express ourselves, and define who we want to be. The forces and directions of the Internet are so new, so protean, and so far reaching that courts must be conscious that what they say today might be obsolete tomorrow.”
Endnotes [ + ]
|1.||↑||Gatley on Libel and Slander (Alastair Mullis & Richard Parkes eds) (Sweet & Maxwell, 12th Ed, 2013) at para 6.1|
|2.||↑||Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977 at (34) to (41).|
|3.||↑||Al Amoudi v Brisard and another (2007) 1 WLR 113 (cited with approval in Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (27) and Zhu Yong Zhen v AIA Singapore Pte Ltd and another (2013) 2 SLR 478 at (44)).|
|4.||↑||Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977 at (122) to (135)|
|5.||↑||Lee Hsien Loong v Ngerng Yi Ling Roy (2016) 1 SLR 1321|
|6.||↑||Steinberg v Pritchard Englefield (a firm) (2005) EWCA Civ 288 (affirmed in Zhu Yong Zhen v AIA Singapore Pte Ltd and another (2013) 2 SLR 478 at (46); See also Qingdao at (138) to (141).|
|7.||↑||Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (31).|
|8.||↑||Ng Koo Kay Benedict v Zim Integrated Shipping Services (2010) 2 SLR 860 at (45).|
|9.||↑||National Union of Bank Employees v Norzeela bt Latimah and anor suit (2014) 7 MLJ 31 at (127).|
|10.||↑||Dato Mohamad Salim Fateh bin Fateh Din v Nadeswaran a/l Rajah (No 1) (2012) 10 MLJ 203 at (53).|
|11.||↑||Qingdao Bohai Construction Group Co Ltd and others v Goh Teck Beng and another (2016) 4 SLR 977.|
|12.||↑||Yan Jun v Attorney-General (2014) 1 SLR 793.|
|13.||↑||Review Publishing Co Ltd and anor v Lee Hsien Loong and anor Appeal (2010) 1 SLR 52 at (49). This was further clarified as whether “a reasonable person who was acquainted with the plaintiffs, on reading the article, would come to the conclusion that the offending words referred to the plaintiffs even though the plaintiffs were not expressly identified in the article.” The plaintiff thus “need not be expressly referred to by name in the offending words; it is also immaterial whether or not the defendant intended to refer to the plaintiff.”|
|14.||↑||Golden Season Pte Ltd and others v Kairos Singapore Holdings Pte Ltd and another (2015) 2 SLR 751 at (55).|
|15.||↑||This is similar to the well-established common-law doctrine that holds forum owners, website administrators, and account owners liable for defamatory content on their platforms only after they are notified of the content but fail to remove the content or delay the removal. See, for example, Godfrey v Demon Internet Ltd (2001) QB 201; Davison v Habeeb (2011) EWHC 3031; Tamiz v Google Inc (2013) 1 WLR 215; Duffy v Google Inc (2015) SASC 170; Wishart v Murray (2013) NZCA 461.|
|16.||↑||Golden Season Pte Ltd and others v Kairos Singapore Holdings Pte Ltd and another (2015) 2 SLR 751 at (61).|
|17.||↑||Review Publishing Co Ltd and anor v Lee Hsien Loong and anor Appeal (2010) 1 SLR 52 at (26)|
|18.||↑||Sim v Stretch (1936) 52 TLR 669.|
|19.||↑||Stocker v Stocker (2019) 2 WLR 1033 at (36) and (37): “The touchstone remains what would the ordinary reasonable reader consider the words to mean…All of this, of course, emphasises that the primary role of the court is to focus on how the ordinary reasonable reader would construe the words.”|
|20.||↑||Stocker v Stocker (2019) 2 WLR 1033 at (41).|
|21.||↑||Stocker v Stocker (2019) 2 WLR 1033 at (43).|
|22.||↑||Stocker v Stocker (2019) 2 WLR 1033 at (43).|
|23.||↑||Stocker v Stocker (2019) 2 WLR 1033 at (44).|