What the Defamation Bill means for the internet – Graham Smith

The Defamation Bill published last week after the Queen’s Speech contains four clauses of especial significance for the internet:

• Clause 5  a new defence for website operators in respect of third party posts.  In essence this significantly enhances website operators’ protection for posts by identifiable posters; and is also designed to encourage website operators voluntarily to disclose to defamation complainants the identity and contact details of the author of an anonymous defamatory post.
• Clause 8: a single publication rule for purpose of time-barring defamation actions.  This will protect internet republications and archives.  Some legislative action in this area was inevitable following the 2009 decision of the European Court of Human Rights in the Times v UK case.
• Clause 9: a forum bar on defamation actions against defendants domiciled outside the EU or EEA, unless of all places in which the statement has been published, England and Wales is clearly the most appropriate place in which to bring an action.  While this is framed as a general restriction on forum-shopping, it will have especial impact on actions founded on the mere accessibility of a foreign internet publication in England and Wales.
• Clause 10: a bar on defamation actions against secondary publishers unless it is not reasonably practicable to proceed against the author, editor (if any) or commercial publisher (if any).  Although presented mainly as a defence for booksellers, this clause will also benefit online intermediaries.

The most controversial and difficult of these provisions is Clause 5, which according to the government’s consultation response on the draft Bill was to deliver a greater degree of protection against liability for intermediaries.

To achieve this aim, the Clause 5 defence would have to be significantly more attractive to online intermediaries than the defences already provided by S1 of the Defamation Act 1996 and by the conduit, caching and hosting safe harbours under the Electronic Commerce Directive.

For identifiable posts Clause 5 does provide significantly greater protection than existing defences, since the defence is not defeated by any degree of knowledge so long as the operator can show that it did not post the statement.  However this may raise a question (which would also apply to non-identifiable posts) as to when moderation strays from knowledge into participation in posting.

For anonymous or pseudonymous posts, on the face of it Clause 5 mostly offers no more than the existing defences, and in some important respects is less attractive.  A proper assessment of the effect of Clause 5 on anonymous and pseudonymous posts is hindered by the unsatisfactory device of leaving significant elements to be prescribed by secondary legislation.  The clause reads as if the drafters gave up half way through and punted off the most difficult aspects to a statutory instrument.

Identifiable posts apart, the most significant increase in protection for online intermediaries is probably delivered by Clause 10, which should make it much more difficult to proceed against online intermediaries rather than the primary perpetrators of the libel.

Apart from its impact on website operators, Clause 5 is controversial in singling out anonymous and pseudonymous speech for special treatment.  The justification for this is said to be that people who defame others should not be able to hide behind a cloak of anonymity.

However, taken in isolation this is mere sloganeering.  Any serious policy-making has to take into account the benefits of anonymous and pseudonymous speech, including speech which someone aggrieved (and let’s not forget that libel claimants can be notoriously thin-skinned) may consider to be defamatory, even with the new threshold of ‘serious harm to reputation’.  Anyone who heard Rowan Davies of Mumsnet speak at the Westminster Legal Policy Forum on 15 March 2012 could hardly doubt the value of anonymous speech.

At a detailed level there are more reasons to be concerned about Clause 5.  Here are some that are immediately apparent.

Effect on common law publication

Clause 5 applies where an action for defamation is brought against a website operator in respect of a statement posted on the website.  It provides for a defence where the website operator can show that it did not post the statement on the website.

Since the provision is cast as a defence, it presupposes that a website operator would otherwise be at risk of liability for publishing a third party post.  However this is counter to the trend of the most recent defamation caselaw on liability for publication, namely Tamiz v Google.

That case (which is understood to be under appeal) held that Google was not a publisher of comments to a blog hosted on its Blogger platform.  Not only is someone who is not a publisher in no need of a defence, but a defence reverses the burden of proof.  It is for the claimant to prove that the defendant published the statement, whereas an affirmative defence places the burden on the defendant.

Clause 5 shares this vice with section 1 of the Defamation Act 1996, which is structured in similar terms and provides a defence for various secondary publishers, some of whom (such as conduits) would almost certainly not have been regarded as publishers at common law.  It is true that no-one to date has argued that the terms of the 1996 Act broadened or should influence the class of those taken to be publishers at common law, but then the recent law in this area has been built almost entirely on claims by litigants in person.

It would be most unfortunate if the effect of Clause 5 were to render website operators or other intermediaries potentially liable for publication of statements for which they may currently be able to argue that they are not publishers at common law.  At least it should be explicitly stated that Clause 5 does not affect any question of whether the defendant has published the statement at common law; and it would do no harm to take the opportunity to insert a similar provision in the 1996 Act.  It is doubtful whether Clause 14, stating that publish and publication have the meaning they have for the purposes of defamation law generally, is sufficient.

What is a website operator?

Clause 5 is silent as to what is meant by a website, or the operator of a website.  This is understandable to an extent, since attempts to define such terms are liable to be overly technology-specific and to create uncertainty through excessive precision (see Professor Chris Reed’s new book Making Laws for Cyberspace for an excellent discussion of these pitfalls).  However the terms ‘website’ and ‘operator’ are themselves hardly models of precision or certainty.

We can think of a core set of examples that would fall within the provision.  A business website with a discussion forum would be one.  A site such as Mumsnet would also be a good candidate.  But what about, say, the Blogger platform considered in Tamiz v Google and Davison v Habeeb?  Is the Blogger platform a website?  Is an individual blog hosted on the platform a website?  If so who is the operator: Google, the blogger, both?  Is Facebook a website?  Is Twitter a website?  Is World of Warcraft a website?  We could go on.  The clause gives no hint of what the thinking behind the drafting may have been.

Defeating the defence – inability to identify the identity of the poster

The Clause 5 defence is defeated if the claimant can show three things:

1.    That it was not possible for the claimant to identify the person who posted the statement

2.    The claimant gave the website operator a notice of complaint (containing various specified information); and

3.    That the website operator failed to respond to the notice of complaint in accordance with regulations to be made under secondary legislation

So if the claimant is able to identify the person who posted the statement, then the defence cannot be defeated regardless of whether the operator had knowledge of the defamatory or actionable nature of the posting. This is a significant enhancement over existing defences since it avoids website operators being put in the invidious position of acting as judge and jury over third party statements alleged to be defamatory.

Presumably the inability of the claimant to identify the person posting the statement is to be assessed by reference to the information available on the site, rather than taking into account e.g. the possibility of making a Norwich Pharmacal application against the website operator to ascertain this.  However nothing is said about what constitutes ‘identity’.  Does a pseudonym or an alias qualify?  Or does the clause only bite on anonymous speech?  What if the identity behind a pseudonym is apparent, to some degree of certainty, from material available elsewhere on the internet?

Defeating the defence – notice of complaint

A notice of complaint in respect of a post by a non-identifiable poster has to specify the complainant’s name, specify where on the website the statement was posted and contain such other information as may be specified in regulations (why leave this to secondary legislation?).

Most importantly, the notice must set out the statement concerned and explain why it is defamatory of the complainant.

Under Clause 5 the complainant has only to explain why the statement is defamatory, not why it is actionable.  Actionability brings into play possible defences such as (as reframed under the Bill) truth, honest opinion, responsible publication on matters of public interest and so on.

Since this is a notice to which a website operator has to consider how to respond, why should the notice not be required to address the whole context?  It is unclear why a website operator should potebtially be encouraged to reveal identity details or take down a post, even an anonymous post, on the basis of only half the story.

In this respect the provision repeats a feature of the 1996 Act secondary publisher defence, but which is absent in the Electronic Commerce Directive hosting exception which refers to knowledge of illegality.

On the other hand it appears that prior knowledge of the defamatory or illegal nature of a post by a non-identifiable poster would not defeat the Clause 5 defence, so long the website operator did not post the statement (see comment above regarding moderation) and upon receipt of a notice complying with the Clause 5 requirements follows the procedure laid down in Clause 5 and the regulations.

Defeating the defence – the website operator’s response

The government consultation response proposed that on receipt of a notice of complaint in respect of a non-identifiable poster, the intermediary would have to contact the author of the material or, if this did not prove possible, take the posting down.  If the matter remained in dispute after an exchange of correspondence, the intermediary would be required to provide details of the author to the complainant, who would then have to sue the author to secure removal of the material and could not pursue an action against the intermediary.

One attraction of this proposal to intermediaries would have been that, having initially acted as a liaison and then having provided identity and contact details to the complainant, the intermediary was out of the picture even if the disputed posting stayed on the website.  The existing defences under the 1996 Act and the Electronic Commerce Directive hosting exemption provide no route to safety other than taking down the allegedly offending material.

However, rather than clearly implement this escape route the Bill leaves it to secondary legislation, giving the government an unsettlingly wide range of future options.  The Bill merely provides that Regulations may make provision including “as to the action required to be taken by an operator of a website in response to a notice of complaint (which may in particular include action relating to the identity or contact details of the person who posted the statement and action relating to its removal)”.  There is no assurance here that taking down the posting would occur only if the intermediary could not contact the poster of the material.

Why is a new route to disclosure of identity needed at all?

A troubling aspect of the provisions regarding non-identifiable posts is why, when there is already a widely used judicial route (the Norwich Pharmacal order) to obtaining the identity of anonymous posters of defamatory material, any of this is necessary at all.  The court is uniquely able to balance competing considerations of reputational harm, privacy and freedom of speech.  So, for instance, in Sheffield Wednesday Football Club Ltd and Others v Hargreaves the court refused to order disclosure of identities for posts that it regarded as bordering on the trivial, saloon-bar moanings, or mild abuse, even if arguably defamatory.

The procedure under Clause 5 would, on the face of it, throw this balancing exercise back on website operators, or encourage them to reveal identity and contact details without regard to broader considerations, simply in order to preserve their own liability position.  It is difficult to see how this can serve a wider interest.  It may also not comply with EU law as interpreted in the Promusicae and Bonnier Audio ECJ decisions, even allowing for the generous latitude to Member States granted by those decisions.

The government consultation response stated that there would “be a need to ensure that appropriate safeguards were in place relating to the releasing of details of the author of the material to the complainant, taking into account concerns such as those expressed by the Committee in relation to whistleblowers.”  None of this is contained in Clause 5.  If it is intended to be addressed in secondary legislation, is that the right place to deal with issues of such importance?

What does ‘required’ mean?

As set out above, Clause 5 provides that Regulations may make provision as to action ‘required’ to be taken by a website operator in response to a notice of complaint in respect of a non-identifiable poster.   Since this is all in the context of defeating the defence provided by Clause 5, it is to be hoped that ‘required’ means no more than ‘required in order for the operator to rely on the defence’.  However this is not explicitly stated.  If the intention is indeed that the only consequence is that the operator be unable to rely on the Clause 5 defence, then ‘required’ is ambiguous and should be revisited.

If the intention is that a website operator should be subject to other sanctions if it does not do what is required, then that should be made explicit so that it can be fully debated and, it is to be hoped, rejected.

The notion that a website operator could be compelled to disclose third party identity and contact details, or face sanctions for not doing so, merely through receipt of a notice alleging defamation, with no opportunity for a court to review the full circumstances and exercise its discretion, would be highly controversial and quite likely contrary to Promusicae and Bonnier Audio.

It is questionable whether a website should even be encouraged voluntarily to disclose such details through the prospect of losing the benefit of this new defence.  At any rate the operator’s ability to rely on other existing defences, and to argue that it has not published at common law, would need to be fully preserved.

How does Clause 5 interact with Clause 10?

Clause 10 provides that a court does not have jurisdiction to hear and determine an action for defamation against a person who is not the author, editor or publisher (i.e. commercial publisher – all as defined in the 1996 Act) of the statement unless the court is satisfied that it is not reasonably practicable for an action to be brought against the author, editor or commercial publisher.

It is interesting to speculate on whether a website operator’s decision not to respond to a notice under Clause 5 would be held against it in the court’s evaluation of reasonable practicability, even though it is open to the claimant to pursue a Norwich Pharmacal application against the website operator to disclose the identity of the author.  Given all the problems with Clause 5, it is to be hoped not.

This post originally appeared on the Cyberleagle blog and is reproduced with permission and thanks.