The Court of Appeal’s decision in Tamiz v Google Inc ( EWCA Civ 68) comes at a moment when defamation on the internet is in the spotlight. It has been ‘the elephant in the room’ during the Leveson inquiry, it is being considered in relation to this year’s Defamation Bill, and has recently been the subject of decisions by Commonwealth courts which have diverged from the established position in England and Wales. The judgment is the first time the Court of Appeal has considered the topic.
The facts are set out in the Inforrm case comment on the decision at first instance. Briefly, comments appearing on the London Muslim Blog about the Claimant Appellant were sued on by him in 2011. Unable to identify the anonymous commenters, he was granted permission to serve the claim form on the Respondent in California on a without notice application on September 22. That decision was set aside by Eady J at first instance ( EWHC 449 (QB)).
Richards LJ, giving the judgment of the court, first addressed the question of whether Google Inc was a publisher of the comments, referring to Eady J’s analogy between Blogger and a graffiti wall “festooned overnight with defamatory graffiti”. He went on to assess Bunt v Tilley ( 1 WLR 1243) a decision which concerned ISPs who had not themselves hosted any website, where Eady J had held that a “passive instrumental role” in the process of publication was not enough, a “knowing involvement in the process of publication of the relevant words” being what was needed to create liability.
Eady J had applied a similar analysis in Metropolitan International Schools Ltd v Designtechnica Corpn ( 1 WLR 1743) in relation to defamatory comments which appeared in Google search results. Again the required mental element referred to in Bunt was not present, the search results being automated; after notification there was no liability for search results rather than mere web-hosting.
The Appellant made an elaborate attack on Bunt as applied in Metropolitan International Schools and here, submitting that the Respondent was liable both as a primary publisher, regardless of any notification, and as a secondary publisher or distributor. The Judge was not persuaded that Bunt was wrongly decided, but he did hold that it and the present case were not so similar as to call for the same conclusion.
Google provides a platform for the blogs along with advertising, and has the capacity for control over them. This does not, however, make it a primary publisher in the sense that it is vicariously liable for the users. Nor is it a secondary publisher facilitating publication in the manner of a distributor, the Judge noting the line of authority holding that a person involved only in dissemination is not to be treated as a publisher unless he knew or ought, by the exercise of reasonable care, to have known that the publication was likely to be defamatory.
However, after notification the considerations involved were different. The Judge was influenced here primarily by the Court of Appeal’s decision in Byrne v Deane  1 KB 81, in which defamatory material was left on the notice board of a golf club. Leaving the defamatory statement on the wall of the golf club amounted to participation in publication. Byrne had been considered in Godfrey v Demon Internet Ltd ( QB 201) and, more relevantly, in Davison v Habeeb and Others ( EWHC 3031 (QB)).
In Davison HHJ Parkes QC stated that Google’s role in relation to Blogger was that of “provider of a gigantic noticeboard on which others published defamatory material” making it arguable that, at some point post-notification it became liable on the basis of consent or acquiescence. The Judge found this noticeboard analogy far more useful than the graffiti-wall comparison, given Google’s relationship with Blogger and its users. An argument based on the difference in scale between the Blogger set-up and the club room in Byrne was not decisive in the Respondent’s favour, as the Appellant’s case only had to be an arguable one.
Although the period between notification and removal was short, just over five weeks on the findings of the trial judge (a point which the Appellant had attempted to dispute by adducing fresh evidence on the appeal), it was open to argument that this left room for an inference adverse to Google on the Byrne principles. The claim could not be dismissed on this basis.
The Judge then considered the defence offered by the cumulative conditions found at Section 1 of the Defamation Act 1996 (“the 1996 Act”). Richards LJ here agreed with Eady J’s assessment that Google Inc was not a publisher under section (1)(a) since it did not come within the definition of a commercial publisher in subsection (2). Prior to notification the conditions in subsections (b) and (c) were also satisfied. Post-notification, the Judge agreed with Eady J that, although dilatory, the response was reasonable having regard to the factors set out in subsection (5).
As to subsection (1)(c), on the question of whether the Respondent had knowledge or reasonable belief that it had contributed to a defamatory statement, the Judge disagreed with the brief reasoning given at first instance, and followed his analysis of common law liability under the Byrne principles. Agreeing again with the judge in Davison, he held that, at some point after notification, Google Inc knew or had reason to believe that its continued hosting of the material in question caused or contributed to the publication of a defamatory statement. Accordingly, section 1 of the 1996 Act would not provide Google with an unassailable defence if it were liable at common law.
The Judge then considered the argument for striking out made by the Respondent at first instance. This was to the effect that because of the limited period of publication between notification and removal, there had been no real and substantial tort, and the claim should be struck out as a Jameel abuse of process. The blog was published on 27 April, with the allegedly defamatory comments following on the 28 and 30 April.
Richards LJ held that the nature of the blog meant that the comments would have been followed by others, and “whilst still accessible, will have receded into history”. It was improbable that a significant number of readers would have accessed the blog in the period after notification, and damage to the Appellant’s reputation would have been trivial; the judge at first instance was right to conclude that the game would not be worth the candle.
The appeal failed and, on this basis, it was unnecessary to consider the Respondent’s defence under regulation 19 of the Electronic Commerce (EC Directive) Regulations 2002.
This is a decision which dips its toes in the icy waters of the liability of online publishers, but does not take the plunge. It is significant that the Court of Appeal held that Google could be liable as a publisher at common law, and that it does not have an unassailable defence under the 1996 Act. This is a more principled and clearer approach to the question of ISP liability which acknowledges the differences between the various entities which make the internet work. It is also one which is in line with a number of recent Commonwealth cases, including Trkulja v Google Inc (No.5)  VSC 533 from late last year, which had not adopted the muddled analysis arising out of Bunt, whereby legal thinking that developed in relation to the ‘back end’ of the internet had been used in relation to the ‘front end’.
There are, however, two apparent problems with the decision in the context of this developing and keenly observed area of the law. Firstly, the suggestion that comments left under the blog receded into history as they were followed by others does not seem to recognise the symbiotic relationship between articles and their comments. As keen readers of popular national newspaper websites could attest, comments are often the best (or worst) bit of the page containing the original publication. Here, the comments complained of were certainly the most sensational part of the page, and their impact should not be discounted simply because others were added. Unlike a forum or bulletin board, there is no other page to which they are displaced over time, and quantifying the extent of reputational damage online is a more nuanced question than the assessment here suggests.
The idea that the Appellant’s case failed because the comments did not amount to a real and substantial tort for the time for which they were published also fails to acknowledge the nature of online reputation. Given that the Appellant did not have legal representation for his initial claim against the Respondent, it is arguable that he did not have the opportunity properly to make out his case on the extent of publication.
Second, the decision fails to address the key ‘hosting’ defence on which website operators, and Google in particular, rely. It follows the HHJ Parkes QC line of reasoning on Blogger from Davison on the common law and the 1996 Act, but does not grapple with the question of what notice needs to be given to defeat this defence under Regulation 19 of the Ecommerce Regulations 2000. At first instance, and in the decision in Davison, it was held that the requirement under the Regulations for notice to a website operator to show why an allegation is unlawful means, in defamation terms, that they must demonstrate some awareness of the availability or strength of any defences. This follows the reasoning of the ECJ in L’Oréal SA and ors v eBay International AG and ors, 12 July 2011 which held, in a slightly different context, that “notifications of allegedly illegal activities or information may turn out to be insufficiently precise or inadequately substantiated”.
The question of unlawfulness is, as pointed out on this blog, one which is important to clause 5 of the Defamation Bill, which aims to provide a new defence for website operators. At present, the drafting requires only that notice is given of why the statement is defamatory, a lower threshold than the notice of unlawfulness required by the Regulations. The difference between these two standards is a source of uncertainty, and one which requires resolution, as do the mechanics of any ‘notice and takedown’ procedure set out in the proposed regulations under clause 5. Fundamentally, these will determine the balance between the rights of claimants and defendants where a disputed publication has taken place online. For all concerned, further clarification can’t come soon enough.
Gervase de Wilde is a pupil barrister at 5RB and former journalist at The Daily Telegraph