Arron Banks v Carole Cadwalladr

The claimant sued the defendant for libel in a talk (“the TED Talk”) and a tweet (“the Tweet”) each of which suggested that the claimant had secretly broken the law on electoral funding by taking money from a foreign power and lied about the matter. The TED Talk and the Tweet were published online to a substantial audience in this jurisdiction. By the time of trial official investigations had found no evidence that there had been any such breach of the law. A defence of truth had been abandoned. The defendant had apologised. But she relied on the statutory defence of publication on matters of public interest.

The trial judge dismissed both claims, holding that although the initial publication of the TED Talk had caused serious harm to the claimant’s reputation it was protected by the public interest defence; later publication of the TED Talk was not so protected but had not caused serious harm and was therefore not actionable; as for the Tweet, its publication would have been protected by the public interest defence to the same extent as the TED Talk but none of it was actionable anyway as it had not caused any serious harm to the claimant’s reputation. The claimant now appeals.

The appeal raises three issues about the interpretation and application of section 1(1) of the Defamation Act 2013 (“the 2013 Act”). This provides that “a statement is not defamatory unless its publication has caused or is likely to cause serious harm to the reputation of the claimant”. This is sometimes referred to as the serious harm requirement.

The first issue concerns the meaning of section 1(1) and its relationship with section 4(1) of the 2013 Act in a case of continuing publication. Section 4(1) provides that “[i]t is a defence to an action for defamation for the defendant to show that (a) the statement complained of was, or formed part of, a statement on a matter of public interest; and (b) the defendant reasonably believed that publishing the statement complained of was in the public interest.”. Where the defendant has a public interest defence which falls away, is the fact that the first, lawful phase of publication caused serious harm to the claimant’s reputation enough as a matter of law to justify a judgment for the claimant in respect of the second phase?

The trial judge said that it is not. She held that in such a situation it is necessary for the claimant to prove that that publication in the second phase has caused serious harm or is likely to do so. The claimant challenges that conclusion but in my view it was correct. The effect of section 1(1) of the 2013 Act is that a statement is only to be regarded as defamatory if and to the extent that its publication causes serious harm to reputation or is likely to do so; publication that does not cause serious harm and is not likely to do so is not actionable. The judge was therefore right to consider whether the claimant had shown that the second phase of publication of the TED Talk had caused serious harm to the claimant’s reputation or was likely to do so.

The second issue is whether the judge’s approach to the question of whether serious harm was established was wrong in law. The claimant makes three main points on this issue. He argues that the judge failed to focus on the actual scale of the publication with which she was concerned, looking instead at how it compared with the scale of other publications; that her conclusion that harm was diminished because most of those to whom the relevant publications were made were in the defendant’s “echo chamber” was legally wrong or untenable on the evidence; and that her finding that harm was reduced because many of the publishees were people whose opinion of the claimant was of “no consequence” to him was also wrong in law. The judge is said to have made all three mistakes when dealing with the TED Talk and with the Tweet.

I am not persuaded that the judge made the first of the alleged mistakes, but I have concluded that she did make the second and the third of them. If what the judge meant by the term “echo chamber” was that most of the publishees were people who disliked or had a generally low opinion of the claimant that was irrelevant to the question she had to decide. If, as I believe, what she meant was that in the minds of most publishees the claimant already had a bad reputation for the specific misconduct of taking foreign money in breach of electoral law and lying about it the evidence did not allow such a finding. The judge’s finding that harm to the claimant’s reputation in the eyes of these publishees was of “no consequence” to him was also unsustainable. If she meant that the claimant did not care what these publishees thought, that was legally irrelevant to the issue of whether serious reputational harm was established. There was no evidence to support a conclusion that others’ adverse opinions of the claimant were of “no consequence” to him in the sense that they could have no practical impact upon his life.

The third issue is whether these errors of principle invalidate the judge’s overall conclusions and her decision to dismiss both claims. The claimant argues that they do, and that on a proper application of the law to the facts of the case it was not open to the judge to dismiss either claim. He says the only legitimate conclusion is that the serious harm requirement was satisfied and so there should be judgment for the claimant. This argument is advanced in respect of the TED Talk and the Tweet.

I would accept these submissions so far as the TED Talk is concerned. The TED Talk conveyed a serious allegation, involving dishonesty and breach of electoral law, which was inherently likely to cause serious reputational harm. On any view there was extensive publication of the TED Talk in this jurisdiction in phase two. The judge rejected the defendant’s attempt to show that the claimant had a pre-existing bad reputation. Her own reasoning about an “echo chamber” and the lack of any “consequence” was unsound as I have said. There was nothing else to act as a counterweight to the natural inference that publication in phase two had caused serious harm to the reputation of the claimant. The precise measure of that harm remains to be assessed but it is not possible to conclude that it was not “serious”. To that extent, I would allow the appeal against the dismissal of the claim in respect of the TED Talk.

The position in respect of the Tweet is different. Although it conveyed the same serious imputation as the TED Talk the judge found that its publication peaked at or near the time it was first posted, after which it fell further and further down the defendant’s timeline, as one would expect. For 10 months any publication was protected by the public interest defence. In my judgment, whatever might be said about harm caused by the initial phase of publication, there was no basis for any inference that there was any publication of the Tweet in phase two that caused any serious reputational harm. To that extent the judge was clearly right. I would therefore uphold her decision in respect of the Tweet and dismiss that aspect of the appeal.

The background to the appeal and the reasons for my conclusions on all these issues are set out in more detail in the remainder of this judgment. I address the consequences of my conclusions at the very end of the judgment.

The claimant is a businessman who was a leader of the campaign for the UK to leave the EU. The defendant is a freelance journalist and writer. The TED Talk was given by the defendant on 15 April 2019 at the TED2019 Conference in Canada. It was recorded and thereafter published on the TED.com website. The words complained of were: “And I am not even going to get into the lies that Arron Banks has told about his covert relationship with the Russian Government.” The claimant’s solicitors wrote a pre-action letter complaining about the TED Talk. On 24 June 2019 the defendant posted the Tweet. The words complained of were: “Oh Arron. This is too tragic. Nigel Farage’s secret funder Arron Banks has sent me a pre-action letter this morning: he’s suing me over this TED talk. If you haven’t watched it please do. I say he lied about his contact with the Russian govt. Because he did.” The Tweet contained a hyperlink to the TED talk. The claimant brought this action.

The defendant has never disputed her responsibility for either publication. There was a trial of preliminary issues at which the natural and ordinary meaning of each was determined by Saini J. He held that the words complained of in the TED Talk meant, in their context, that: “On more than one occasion Mr Banks told untruths about a secret relationship he had with the Russian government in relation to acceptance of foreign funding of electoral campaigns in breach of the law on such funding”. He held that the Tweet bore the same meaning. This was because the hypothetical ordinary reasonable reader of the Tweet would have followed the link to the TED Talk. That meaning was held to be defamatory at common law.

Some months later, the defendant abandoned her pleaded defence of truth and apologised, saying that it was not her intention to make any such allegation which she accepted would be untrue.

She also dropped a limitation defence. But she did not concede liability. She relied, as she had from the start, on the public interest defence. Her case was that she had intended to convey less serious meanings than those found by Saini J, to the effect that the claimant had told lies about a secret relationship he had with the Russian government and there were questions to be asked about the legitimacy of the source of political donations he had made. She contended that this was a statement on a matter of public interest, that she believed that it was in the public interest to say it, and that her belief was reasonable.

At the trial in January 2022 before Steyn J, DBE the main issues on liability were, in relation to each publication, (1) whether the claimant had shown that the serious harm requirement was satisfied and if so (2) whether the defendant had made out her public interest defence.

There was a subsidiary aspect to the second issue. The TED Talk and the Tweet had both remained accessible online from the time of first publication until the time of the trial. It was common ground that in such a case a public interest defence once established can expire if the circumstances change in such a way that it is no longer reasonable for the defendant to believe that publication is in the public interest. Nicklin J so held in Lachaux v Independent Print Ltd [2021] EWHC 1797 (QB), [2022] EMLR 2 at [159], adopting the general approach of the common law to the Reynolds defence of responsible journalism (see Loutchansky v Times Newspapers Ltd [2001] EWCA Civ 1805, [2002] QB 783 at [79], Flood v Times Newspapers Ltd [2010] EWCA Civ 804, [2011] 1 WLR 153 at [77]-[78]). The judge agreed. She therefore proceeded on the basis that the defendant had to make out all the elements of the public interest defence with respect to the whole of the continuing publication, and that in making its assessment the court should “focus on whether there had been a significant change of circumstances.”

The claimant argued that there had been five significant changes of circumstance. The first was a statement released on 24 September 2019 by the National Crime Agency (“the NCA Statement”). An article published by the defendant in 2017 had prompted official investigations by the Information Commissioner’s Office (“ICO”) and the Electoral Commission. On 1 November 2018 the Electoral Commission announced that it had unveiled evidence of criminal offending falling outside its remit and that it had handed that evidence to the NCA. The NCA investigated. The NCA Statement announced the results of that investigation. These were that the NCA had found no evidence that any criminal offences under the Political Parties, Elections and Referendums Act 2000 or company law had been committed by (among others) the claimant and had not received any evidence to suggest that the claimant and his companies “received funding from any third party … or that he acted as an agent on behalf of a third party.” Another of the changes of circumstance relied on was a joint statement made by Electoral Commission and the claimant and others on 29 April 2020 (“the Joint Statement”) in which the Electoral Commission publicly confirmed its acceptance of the NCA’s findings.

The defendant had taken no steps to stop publication of the TED Talk or the Tweet or to attach any qualifying statement to either of them. At trial, she did not accept that any of the changes of circumstance relied on, including the NCA Statement and the Joint Statement, required her to do this or that they made it unreasonable for her to continue believing that publication was in the public interest. Alternatively, she argued that if there was a point in time at which what she said in the publications complained of was no longer covered by the defence of public interest under section 4(1), because it was unreasonable for her to continue believing that publication was in the public interest, then the claimant would need to prove that the serious harm requirement was satisfied in respect of any publications after that point in time.

After hearing evidence and argument over five days the judge handed down judgment on 13 June 2022. The judgment had three main sections, and fifteen sub-sections, running to 416 paragraphs. The narrow scope of the issues before us however means that I can concentrate on relatively small portions of the judgment.

Section I of the judgment identified the parties and gave an outline of the issues, the history of the case, and the evidence. At [9] the judge addressed the fact that the defendant had “repeatedly labelled this claim a SLAPP suit, that is a strategic lawsuit against public participation, designed to silence and intimidate her.” The judge said that label was “neither fair nor apt” as the claimant’s “attempt to seek vindication through these proceedings was … legitimate.”

Section II of the judgment dealt with the issue of serious harm. It began at paragraph [48] with a statement of the relevant legal principles. For the most part these were not in dispute. But there was “one contentious point” which arose for determination, that is the one I have identified above as the main issue on this appeal. The judge dealt with this issue at paragraphs [53]-[59].

As the judge recorded, the claimant argued that the question whether the test in section 1 of the 2013 Act is met falls for determination once and only once, from the date of the original publication and that serious harm should be assessed “by reference to the entire period of publication, even if the court has determined that the defendant has a defence to the libel claim in respect of the first part of that period.” For the defendant it was argued that if the public interest defence applied for a specific period of time but ceased to apply later due to a significant change of circumstances “the court would have to revisit the question whether serious harm was established from the date of the significant change of circumstances”. It would be illogical to take account of any harm caused by earlier lawful publication.

The judge concluded that the defendant must be right as a matter of logic, for three reasons:

The judge set out her decisions and reasons on the issue of serious harm between paragraphs [80] and [99]. She did so in three sections.

She dealt first with the TED Talk. She identified the meaning as one that accused the claimant of “serious, repeated dishonesty … about a secret relationship he had with a potentially hostile foreign power”. She accepted that it was inherently probable in the circumstances that there had been substantial publication of the talk within this jurisdiction. She observed that the evidence that it had any impact on the claimant’s reputation and prospects in the business sphere was “negligible”; but she was persuaded that “it can be inferred that a sizeable number of people who knew or would later come to know [the claimant] would have viewed the TED talk and believed what was said about him, lowering his reputation in their eyes.”

The judge rejected the defendant’s contention that in the eyes of most viewers the claimant would have had no or no meaningful general reputation to be harmed. She held that it was impermissible for the defendant to rely for this purpose on earlier publications by the defendant and others of the allegation that the claimant had lied about his contact with the Russian government. That was contrary to the principle affirmed by the Supreme Court in Lachaux SC at [24], that “damage to the claimant’s reputation done by earlier publications of the same matter is legally irrelevant” to the question arising under section 1, as it is when assessing harm to reputation at common law (Dingle v Associated Newspapers Ltd [1964] AC 371 (“Dingle”)). The judge also rejected a contention that investigations by the Electoral Commission and others, and media reporting, had given the claimant a general bad reputation. At [90] the judge held that the claimant had established that publication of the TED talk caused serious harm to his reputation, so that the onus fell on the defendant to show that she had a defence.

Turning to the Tweet, the judge observed at [91] that “most of the matters I have referred to above are the same” but there were “two crucial differences”. The first (at [92]) was that “the number of people within the jurisdiction to whom the Tweet was published was likely to be only a fraction of the number who viewed the TED Talk”. The normal inference would be that “access to a Tweet will peak at or shortly after the time of first publication” and there was no reason to think otherwise in this case. The Tweet was “likely to be a long way down the defendant’s timeline”. Secondly, the judge said at [93] “as [counsel for claimant] has acknowledged, it may reasonably be inferred that the vast majority of the defendant’s followers on Twitter ‘are likely to be persons within her own echo chamber’ and ‘it’s probably right that they wouldn’t have thought very much of the claimant by that time.’” The judge also said at [93] that those within the jurisdiction to whom the Tweet was published were “likely to consist of people whose opinion of the claimant was of no consequence to him”.

At [94], the judge said that she was not persuaded that she should infer from the gravity of the imputation and the extent of publication that the Tweet had caused or was likely to cause serious harm to the claimant’s reputation. She dismissed the claim in respect of the Tweet on that basis.

In paragraphs [95]-[99] the judge addressed “Continuing publication of the TED Talk from 29 April 2020”. She summarised her conclusions on the public interest defence, explaining that she had found that this was established in respect of publication of the TED Talk up to 29 April 2020 but not beyond; and that accordingly “the question whether the claimant has proved that the serious harm threshold is met in respect of the TED Talk falls to be reconsidered by reference to the period from 29 April 2020.” It will be convenient to refer to this period, that is, from 29 April 2020, as Phase Two, although the judge did not use that term.

The judge then considered whether the claimant had established that his reputation suffered serious harm as a result of publication of the TED Talk in Phase Two. She said that the gravity of the imputation remained the same and there was the same lack of evidence that the claimant had been shunned or lost business opportunities. His case was “dependent on drawing an inference of serious harm from the combination of the gravity of the imputation and the extent of publication.” Although she had been persuaded to draw that inference in relation to “the original publication of the TED Talk” she was not so persuaded in relation to Phase Two. She found therefore that the claimant had not established that this publication (i.e. the publication of the TED talk from 29 April 2020) “has caused (or is likely to cause) serious harm to his reputation” [the judge’s emphasis].

The judge gave three main reasons for this conclusion. First, the extent of worldwide publication of the TED talk in Phase Two was “close to a tenth of the figure from 15 April 2019” and publication in this jurisdiction was likely to be “a similarly small fraction of the whole” as in the earlier phase. “The normal inference … would be that views of the TED Talk will have peaked at or shortly after the time of first publication…” Secondly, the fact that the talk was about the EU referendum, which had taken place nearly six years before the trial, increased the likelihood that the extent of publication would continue to diminish. Thirdly, it could also reasonably be inferred that most of those in this jurisdiction who had viewed the TED Talk during Phase Two were (like the defendant’s followers on Twitter) “likely to be people supportive of her defence of this claim and ‘within her own echo chamber’ and likely to be “people whose opinion of the claimant was of no consequence to him.”

Section III of the judgment dealt with the defence of public interest. After a detailed examination of the evidence the judge held that the TED Talk and the Tweet were both publications on a matter of public interest, and that the defendant had believed throughout that it was in the public interest to say what she had. The judge accepted that this was a reasonable belief when the defendant gave her TED Talk in April 2019 and that it remained a reasonable belief when she posted the Tweet in June of that year, and for a further ten months or so thereafter. At [399] she concluded that “the defendant had established that her belief that publishing the TED Talk was in the public interest was reasonable” and that “for the same reasons, if I had found that the serious harm threshold was met in relation to the Tweet, I would also have found that the defendant’s belief that publishing the Tweet was in the public interest was reasonable.”

In paragraphs [400]-[412] the judge explained the conclusions she had reached on change of circumstances. These were that in the light of the Joint Statement on 29 April 2020 “it was no longer reasonable to believe that it was in the public interest to assert that there were grounds to investigate whether the source of [the claimant’s] donations was foreign funding, accepted in breach of the law” at least without any qualification referring to the NCA Statement and the Joint Statement.

However, as the judge explained at [413]-[414], her earlier conclusion that the claimant had failed to establish that the TED Talk had caused or was likely to cause serious harm to his reputation in Phase Two, or that serious harm had been caused by publication of the Tweet, meant that although the defendant had failed to establish the defence of public interest for the publications after the change of circumstances on 29 April 2020, no question of awarding damages arose.

There are three grounds of appeal. Ground 1 is that the judge “was wrong to hold that the issue of whether or not a statement has caused or is likely to cause serious harm to a claimant’s reputation needs to be determined again from the date on which the defendant’s belief that it is in the public interest to publish the statement ceases to be reasonable (and hence the s 4 defence falls away)”. Permission to appeal on this ground was granted by the judge herself. This gives rise to the first of the three issues I have already identified.

Ground 2 is that the judge “was wrong to hold that the appellant had not proved that the Tweet caused (or was likely to cause) serious harm to his reputation”. Ground 3 is that she “was wrong to hold that the publication of the TED Talk since 29 April 2020 had not caused and was not likely to cause serious harm to the claimant’s reputation; the Court ought to have held (assuming such a requirement is necessary) that it had been satisfied on the facts of this case.” Permission to appeal on Grounds 2 and 3 was refused by the judge, but the claimant’s renewed application was adjourned for hearing with the appeal on Ground 1 with the appeal to follow if permission was granted. I would grant permission to appeal on each of grounds 2 and 3 although, as will be clear from the Summary at the start of this judgment, I have found it more convenient to deal with the issues they raise in a different way and I have ultimately concluded that Ground 2 fails.

For the claimant, Mr Benjamin Williams KC submits that the judge’s conclusion and her reasons are all mistaken. His main points can be fairly summarised in this way.

It is also argued that the judge’s analysis leads to a number of undesirable consequences. Among them are the practical difficulties for a claimant in knowing in advance what he or she has to prove in relation to what period of time, in a case where the section 4 public interest defence is relied on, and the impracticality of assessing whether each and every individual publication caused serious reputational harm.

The points I have summarised have been skilfully developed in written and oral argument before us. For the most part they are an elaboration of the argument below which did not persuade the judge. I have not been persuaded either. In my view, the judge’s analysis of the law on the relationship between section 1 of the 2013 Act and section 4 of the that Act was right.

At common law a statement is defamatory if it conveys an imputation with an inherent tendency to cause substantial harm to a person’s reputation. The publication of such a statement is actionable without proof of actual damage. “Publication”, for this purpose, means the communication of the statement to someone other than the claimant. A single communication is actionable. Each such communication is a separate tort. Analytically, a mass publication such as a broadcast or online statement like the TED Talk or the Tweet gives rise to as many causes of action as there are viewers, listeners or readers. It does not give rise to a single cause of action, as is the position in some other jurisdictions where the principle that there is only a single cause of action is known as “the single publication rule”. Where a defence succeeds in part, damages will be awarded to compensate for all harm caused by all the actionable publications. Any harm caused by publications that have been held to be lawful is ignored for this purpose. These are all elementary propositions, but they form the legal context in which Parliament enacted the 2013 Act.

Section 1 of the 2013 Act was intended to modify the common law by requiring a claimant bringing a claim for defamation to prove as a fact that the publication complained of has caused the claimant actual reputational harm that is serious (or that the publication of the statement is likely to cause such harm). The means Parliament adopted to achieve this was to modify the pre-existing common law definition of the term “defamatory”. That term can no longer be applied to a statement just because the statement has the inherent qualities required by the common law. Parliament has provided that such a statement “is not defamatory unless” it also satisfies the additional statutory criterion that “its publication has caused … serious harm to the reputation of the claimant” or is likely to do so.

The touchstone here is not the nature of the statement but the impact of “its publication”. Those two words are plainly critical. It would be impossible to construe them as implicitly importing the single publication rule that applies in some other countries. It is presumed that Parliament does not intend to alter the common law unless (and to the extent that) such an intention is expressed or is necessarily implicit in the wording used. There is nothing in the 2013 Act or in the legislative history to show or to suggest that in enacting section 1 Parliament intended to adopt the single publication rule or to depart in any other way from the common law meaning of the word “publication”. Quite the contrary. Section 15 of the 2013 Act expressly provides that in the Act “‘publish’ and ‘publication’ have the meaning they have for the purposes of the law of defamation generally”. And section 8 of the Act, which enacts the “single publication rule” for the purposes of the law of limitation (and for that purpose only) expressly achieves that aim by using the word “publication” in its common law sense.

Once it is accepted, as it must be, that “publication” for the purposes of the 2013 Act bears its established common law meaning save where the Act expressly says otherwise, there are only two available readings of section 1(1) of the 2013 Act. Either it means that a statement is defamatory if any publication of that statement causes serious harm to the claimant’s reputation; or it means the statement is defamatory only if and to the extent that the publication of it causes serious harm.

The first reading is the one for which the claimant contends. It must be rejected for all manner of reasons. Perhaps the most obvious and pertinent objection is that this reading would permit a claim for libel to succeed in respect of publication of a statement that did not cause serious harm just because there had been some other publication of the statement that did. The seriously harmful publication might have been made in circumstances that make it not actionable for reasons such as limitation, foreign law or absolute immunity. Or it might have been lawful because there was a defence of privilege or – as here – a public interest defence. Indeed, the other publication might not have been sued upon at all. Take this case. On the first reading of section 1 that I have identified, the claimant would have won if he had confined his claim on the TED Talk to publication after 29 April 2020 even though (on the judge’s findings) that publication did not cause him any serious reputational harm. He would have won because he could show that serious harm was caused by earlier publication of which, on this hypothesis, no complaint was made. Such an outcome would be contrary to the clear legislative purpose behind the provision and the Act as a whole.

The correct reading of section 1(1) of the 2013 Act must therefore be the second of the two interpretations I have identified. A statement is defamatory only if and to the extent that its publication causes serious harm to reputation or is likely to do so, and not otherwise. This is entirely consistent with the language of section 1(1) which does not contain an exhaustive definition of what is defamatory but an exclusionary criterion. It does not say that a statement is defamatory if its publication causes serious harm. It says a statement is not defamatory unless its publication has that effect.

There is nothing incoherent about this approach. Indeed, in my view it is the approach advocated by the claimant that is open to the charge of incoherence. I have given some reasons for that already, but there are others. The common law approach to the concept of “publication” accounts for the principle that a public interest defence, once established, does not necessarily endure for ever. A defendant has to show that at the time of each and every publication, he or she continued to believe that it was in the public interest to publish the statement and that this belief remained reasonable when those particular publications were made. The approach is a granular one. The public interest defence is not available in respect of publications made after the defendant ceases to hold the belief in question or such a belief ceases to be reasonable. By parity of reasoning, and for the purposes of consistency and fairness, a claimant cannot succeed in establishing liability in respect of publications which do not cause serious harm, because there is some other publication that does, or because serious harm is caused by the “publication” taken as a whole. Likewise, it would be unprincipled to treat serious harm caused by conduct which is not actionable because a defence has been made out as a sufficient reason to grant a remedy in respect of other conduct which, viewed in isolation, is not harmful enough to justify this.

The quality of being defamatory at common law is an abstract one that depends exclusively on the nature and content of the statement and its inherently harmful tendency. And I accept it may seem odd, to those steeped in that common law approach, that the same statement published over a period of time by the same defendant may be defamatory for only some of that time. This however is an inevitable consequence of the change enacted by section 1(1) which made it an essential ingredient of a statement being defamatory, and therefore actionable, that “its publication has caused serious harm” as a matter of fact (or is likely to do so).

I do not agree that this analysis requires proof that each individual publication caused serious reputational harm. There will doubtless be cases in which an individualised approach is both possible and necessary. That might be so, for instance, in a case of publication to a small number of identified individuals only one of whom turns out to have believed the allegation complained of. But the statutory words “its publication” are flexible enough to embrace other kinds of case, including the typical case of media or online publication involving a mass of individual publications to numerous unknown individuals. There is no principled objection to the application in such a case of a modified version of the traditional pragmatic approach of the common law, which is to consider the relevant publications collectively when assessing reputational harm. The modification is that in a case of this kind the size and shape of the relevant group of publications will not be known for certain until after the public interest defence has been assessed.

If this approach gives rise to any practical difficulties they will, I apprehend, arise in relatively few cases (noting as I do, that though section 1 of the 2013 Act came into force on 1 January 2014, this is the first occasion on which the court has been called on to consider this issue). The paradigm if not the only case giving rise to difficulties will be one where the claimant contends that the public interest defence, if available, falls away at some identified point(s) in time. Claimants should be able to address the evidential implications of such contentions for their own case on serious harm. I consider any potential problems are capable of sensible case management. Be that as it may, the court is bound to give effect to the ordinary meaning and purpose of the words of the legislation which Parliament has enacted.

For these reasons, which are in large part developments of those given by the judge, I would dismiss the appeal on Ground 1.

It is common ground that a relevant and potentially significant factor when deciding whether publication has caused serious harm to reputation is the scale of publication or, putting it another way, the total number of publications. The judge noted this in her statement of the relevant legal principles. It was one of the points on which the parties focused their arguments on the issue of serious harm. The claimant’s first criticism of the judge’s analysis of the matter is that she went astray: instead of making an assessment of the absolute numbers of relevant publishees or the overall scale of the relevant publication she based her conclusions instead on assessments of relative scale. The first “crucial difference” which the judge identified between the initial publication of the TED Talk and the publication of the Tweet was that the latter was “likely to be only a fraction” of the former. A comparison of this kind tells one nothing of importance, submits Mr Williams. A similar error is apparent, he argues, in the judge’s approach to the Phase two publication of the TED Talk which she said was “close to a tenth” of the figure from 15 April 2019.

In my judgment this criticism is misplaced. The judge concluded that the initial publication of the TED Talk was probably on a scale “at least equivalent to a high circulation domestic newspaper”. That would imply a number in the order of 1 million. The judge later found that “the number of people within the jurisdiction to whom the Tweet was published” was likely to be only a fraction of this [my emphasis]. In other words, the judge was making a finding about the absolute scale of publication, using a fraction as a method of calculation or estimation. Although she did not specify the fraction she had in mind it is clear from the context that it was a small one. As everyone in this case accepts, precision in these matters is impossible. Likewise, when assessing the scale of Phase Two publication of the TED Talk the judge began with her earlier finding about the initial publication of that talk and used a fraction – in this instance a tenth – only as a way of arriving at an estimate of the absolute scale of later publication. She then applied the same “small” fraction she had mentioned earlier to arrive at an estimate of the scale of publication in this jurisdiction. I am satisfied that the broad conclusions at which the judge arrived on these issues were open to her on the evidence.

Mr Williams has advanced his criticism of the judge’s observations on this topic on the footing that what she meant was that most of the defendant’s followers “generally share her broad views” so that many of those who read the Tweet and viewed the TED Talk “had already formed a view of [the claimant] that was not particularly positive (because they strongly opposed his views on Brexit)”. He argues that it cannot be assumed that Twitter is an “echo chamber” in this sense; that there was no proper basis for the judge’s conclusions that the defendant’s Tweet was published in such an “echo chamber” and still less for the conclusion that this was true of the TED Talk. He submits that in any event there is “a world of difference” between thinking that a public figure is wrong on an important political issue and alleging that he is “the law-breaking stooge of a hostile … foreign government”. The allegations in the Tweet and the TED Talk were new and particular allegations which would make any reader, even one with an existing adverse view of the claimant, think substantially the worse of him. If Mr Williams’ analysis of the judge’s reasoning is correct, then I would agree that it involved an error of principle.

Where a defendant publishes a specific allegation of a seriously damaging kind in circumstances which would ordinarily lead to an inference of serious reputational harm the fact, if it be so, that those to whom that allegation is published are politically opposed to the claimant or dislike him or have a generally low opinion of him for some other reason is not a proper basis on which to reject that inference. Such an approach would be at odds with well-established and salutary principles. A person’s reputation is not a simple question of whether they are liked or disliked. Nor is reputation a single indivisible whole, it is composed of sectors. At common law it is clear that evidence of bad reputation must be confined to the sector of the claimant’s character relevant to the libel: see Gatley on Libel & Slander 13^th ed (2022) at 34-083 and cases there cited. If it were otherwise a person who was widely disliked or had a bad reputation in one sector of their reputation would find it hard to succeed in a defamation claim whatever grave falsehood was published about them. That would be unprincipled and contrary to the public interest.

It is for this reason, I believe, that judges of the Media and Communications List have consistently, and in my view correctly, rejected arguments to the effect that a serious allegation of specific wrongdoing does not cause serious harm if the audience dislikes the claimant for some other reason. In Barron v Collins [2017] EWHC 162 (QB) at [56] I held that “it does not follow from the fact that a publishee is a political opponent of the claimant that they will think no worse of the claimant if told that he or she has covered up sexual abuse”. In Monroe v Hopkins [2017] EWHC 433 (QB), [2017] 4 WLR 68 at [71(8)] I accepted the submission that “If someone is hated for their sexuality or their left-wing views, that does not mean they cannot be libelled by being accused of condoning the vandalisation of a war memorial”. In Turley v Unite the Union [2019] EWHC 3547 (QB) the allegation was one of dishonest submission of an application to join the union at a concessionary rate (see [100]). At [114(iii)], Nicklin J observed that proof that readers of the offending words “held some pre-existing hostility towards the claimant” on political grounds “cannot lead to the conclusion that such a reader would discount what they read about the claimant”. The point is familiar to judges and practitioners in this field. It was made to Steyn J by the claimant’s Counsel at the trial and it does not appear to have been in dispute. In Riley v Sivier [2022] EWHC 2391 (KB), [2022] EMLR 6 at [114]-[115], a judgment handed down the month after her decision in this case, Steyn J herself cited and applied the reasoning in Monroe at [71(8)].

In the circumstances, I doubt that the judge made the error attributed to her by Mr Williams. In my opinion what she meant by her reference to an “echo chamber” is a closed environment in which the information people receive is merely repetition of the same things that they have heard or said before and already believe. That is consistent with the literal meaning of “echo”. I agree that it cannot be assumed that Twitter itself is an echo chamber in this sense, but the judge did not do this. What she said was specific to the defendant and her followers on Twitter. Nor do I think that what the judge said about those matters can fairly be described as an “assumption”. What emerges from a fair reading of the judge’s language is that she considered it reasonable to infer that the majority of the publications complained of were sent into an “echo chamber” of this kind. That inference fed into the judge’s overall conclusions that serious reputational harm was not established in respect of the Tweet or Phase Two publication of the TED Talk.

This is a legally permissible line of reasoning, up to a point. Proof that the relevant sector of the claimant’s reputation is bad among those to whom the statement complained of was published can reduce damages, perhaps very substantially. A claimant is only entitled to recover compensation for injury to the reputation he actually has. By the same token proof of an existing bad reputation in the relevant sector must be relevant to the question of whether the publication of a statement caused serious harm to the claimant’s reputation.

However, the authorities set clear limits on the means by which a relevant bad reputation can be proved. Other publications to the same effect as the words complained of, or relating to the same incident as referred to in those words, are inadmissible for this purpose, and this rule covers previous publications by the same defendant: Gatley (op. cit.) at 34-086 citing Dingle and Lachaux SC at [22]. The judge had applied “the Dingle rule” earlier in her judgment, when rejecting as “impermissible” the defendant’s attempt to prove a pre-existing bad reputation by reliance on earlier publications of the allegation that the claimant “had lied about his contact with the Russian government” and reliance on the investigations of the ICO and the Electoral Commission and media reporting of those investigations: see [87]-[90]. At trial the defendant did not advance any other basis for concluding that the claimant had a relevant pre-existing bad reputation. Nor did the defendant rely on any proposition about an “echo chamber”. This raises the question of what evidential basis there was for the inference drawn by the judge.

On this appeal Mr Millar KC has submitted that we should be extremely hesitant to interfere with the judge’s conclusion. He has emphasised that an appeal is not a re-hearing but a review, citing authoritative explanations of why it is generally inappropriate for an appellate court to second-guess the trial judge on questions of fact (Fage UK Ltd v Chobani UK Ltd [2014] EWCA Civ 5, [2014] ETMR 26 at [114] (Lewison LJ) and Biogen Inc v Madeva Plc [1997] RPC 1 at [45] (Lord Hoffmann)). I am very mindful of the general principle of appellate restraint and the sound reasons that underlie it. Mr Millar has not, however, descended into any detail. He has not identified for us any “island” or item of evidence in the case and submitted that this is something from which the judge was entitled to draw the inference she did. For the claimant, Mr Williams submits that the reason we have not been shown any such item is that, although there was undoubtedly a sea of evidence, it was all about other matters and contained nothing relevant to this issue. Having carefully examined the judgment and all the other material that has been placed before us I find myself driven to the conclusion that this is correct. In my judgment, there was no legally admissible evidential basis for the judge’s inference that the TED Talk and Tweet did not cause serious harm because most of those to whom they were published already believed in the truth of the allegation which they contained.

It is not crystal clear what the judge meant by her findings that most of the relevant publishees would fit this description. I am however satisfied that this ground of decision is separate from the judge’s reasoning about an “echo chamber”. It seems to me that the judge’s language can only mean either (a) that these were people whose opinions of the claimant were a matter of indifference to him or (b) that a bad reputation in the eyes of these people could not have any practical consequences for the claimant. The first interpretation would be contrary to principle. The issue to be decided was whether the claimant’s reputation had suffered harm that was serious. The claimant’s feelings about that harm are a different matter. They fall for consideration only after the claimant has established the serious harm to reputation which is nowadays the gist of the tort. These points are fundamental and well understood by practitioners and judges in this field. The claimant gave evidence of his upset at the publications complained of. It was not put to him that he did not care what the defendant’s followers thought of him. For all these reasons, I do not think this is what the judge meant.

Further, the origin of the term “no consequence” seems to be my judgment in Ames v Spamhaus Project Ltd [2015] EWHC 127 (QB), [2015] 1 WLR 3409, a case cited by the judge in her summary of the relevant legal principles. In Ames, the concept of people whose opinion was “of no consequence to the claimant” was deployed in the second sense I have identified, to mean people whose negative opinion was incapable of being consequential for the claimant, and therefore did not cause him serious harm.

In Ames the claimants sued in respect of the publication to at most a few thousand people in England and Wales of the allegation that they were ‘spammers’. The defendants applied to strike out the claim as an abuse of process on the footing that the claimants had no realistic prospect of satisfying the serious harm requirement. The claimants were foreign citizens resident and in business in California. It was argued that they had no significant personal or business connections here so that any negative views of them that might have been formed by the small number of publishees were inconsequential and incapable of having any significance for the claimants. It is in that context that I referred at [92] to the question of whether publication in this jurisdiction was “clearly inconsequential” and agreed that “seriously harmful allegations about a person may not cause ‘serious harm’ to reputation if they are made to people whose opinion of the claimant is of no consequence to the claimant.” I held however that on the evidence the claimants’ prospects of proving serious harm were not fanciful.

The notion of inherently harmful allegations being inconsequential or of “no consequence” as a matter of fact was therefore a narrow and specific one. It arose in an unusual context where the principal question was the prospect of injury to business reputation. It was not a ground of decision. The merits of this line of reasoning have not been debated before us but whatever they may be I am satisfied that it cannot be applied to the very different facts of this case. Again, this is not a conclusion for which the defendant argued at the trial. Nor is it one for which Mr Millar has identified any basis in the evidence that was before the judge. In the absence of an explanation in the judgment itself, although I bear in mind all the warnings against appellate interference to which I have already referred, I can only conclude that there was no sufficient basis for this aspect of the judge’s findings.

Mr Williams has a fourth criticism of the judge’s approach to the assessment of serious harm. He complains that the judge failed to refer to online comments which were in the papers before her even though these “can be evidence of reputational harm to the extent they can be said to be a natural and probable consequence of the publication complained of” (Economou v de Freitas [2016] EWHC 1853 (QB), [2017] EMLR 4 at [129]). The potential relevance of such comments is not in doubt. The claimant’s closing argument relied on comments about the TED Talk. The judge expressly addressed these, concluding that there were only four that provided “minimal” support for the claimant’s case. The judgment did not deal explicitly with comments on the Tweet but nor did the claimant’s closing. A judgment need not address every single aspect of the evidence and argument. I therefore see no merit in this point.

My conclusion that the trial judge erred in the ways I have identified is not enough in itself to justify the reversal of her decision. A decision on serious harm is an evaluative one, akin to the exercise of a discretion. To succeed in an appeal against such a decision the appellant has to identify a “flaw in the judge’s treatment of the question to be decided … which undermines the cogency of the conclusion”: Re Sprintroom Ltd [2019] EWCA Civ 932, [2019] BCC 1031 at [76]. Hence, a judicial determination of whether the serious harm requirement is satisfied will “rarely” be disturbed in the absence of an “error of principle potentially critical to the outcome”: Lachaux SC at [21]. These are high hurdles for an appellant to overcome.

Nonetheless, so far as the TED Talk is concerned, I have concluded that the judge’s errors do fatally undermine her conclusion. In my judgment, if those errors are put to one side it was an inevitable inference from the evidence before the judge that publication of the TED Talk after 29 April 2020 caused serious harm to the reputation of the claimant. There was little direct evidence of harm, but as the judge held this was “unsurprising” given the well-known difficulties of obtaining such evidence: see Economou v De Freitas [2018] EWCA Civ 2591, [2019] EMLR 7 at [28] and [31]. This was, as the judge also held, “a serious talk on a serious subject” given by “an award-winning journalist” on an “authoritative and credible international platform”. In these circumstances the inference that serious harm was caused flows from the inherent gravity of the allegation and its natural tendency to cause serious reputational harm, coupled with the judge’s own findings as to the scale of publication in this phase, taken at its lowest.

As for the gravity of the allegation, I would endorse the judge’s summary. It was one of “serious, repeated dishonesty … about a secret relationship he had with a potentially hostile foreign power.” Beyond this, the claimant was accused of acting in breach of the law on electoral funding, which is a significant additional element. As I have indicated, the judge implicitly found that initial publication of the TED Talk in this jurisdiction was in the order of 1 million views. She estimated that the scale of publication after 29 April 2020 was about a tenth of what went before. The implicit finding is therefore that the TED Talk was viewed in this jurisdiction at least 100,000 times in Phase Two. That is broadly equivalent to the circulation of a broadsheet national daily newspaper. Those are weighty considerations. Given the claimant’s prominent role in public life and business a conclusion that there was serious harm to his reputation must follow. There is nothing about the circumstances of publication or the identity or characteristics of the publishees that could undermine such a conclusion.

I am not persuaded, however, that the judge was wrong to dismiss the claim in respect of the Tweet. The natural and ordinary meaning conveyed by the Tweet was the same as that of the TED Talk. There clearly was substantial publication at around the time the Tweet was posted. However, for the reasons I have given the question that is critical to the outcome of this claim is not whether publication of the Tweet caused serious harm between 24 June 2019 and 29 April 2020 (when it was protected by the public interest defence) but whether serious harm was caused by publication in Phase Two. For that purpose, the judge’s key findings are that access to the Tweet will have peaked “at or shortly after the time of first publication” and that it will have moved “a long way down the defendant’s timeline”.

These propositions are not in dispute, which is unsurprising as they are trite observations about how Twitter works. They make it inherently improbable that anyone in this jurisdiction read the Tweet after 29 April 2020, which is ten months after it was posted. It is all the more unlikely that any such person both read the Tweet and followed the link to the TED Talk. There was no direct evidence that this happened. I have been unable to identify a basis for inferring that it did. The agreed figures for retweets and likes of the Tweet to which the judge referred in her paragraph [67] show that between the end of June 2019 and 17 November 2021 the total historic number of retweets and likes of the Tweet reduced by roughly 10%. We are told that this is likely to reflect the closure of accounts by some of those who had retweeted or liked the Tweet in the past. However that may be, the figures do nothing to indicate an increase in publication during the relevant period. They tend to support the conclusion that the harm was done at the start and that in the relevant period there was no publication, or no harm, or none that was serious.

For the reasons I have given I would dismiss the first and second grounds of appeal and uphold the judge’s decision to dismiss the claim in respect of the Tweet, but I would allow the appeal on ground three and set aside against the judge’s order dismissing the claim in respect of the TED Talk. In its place I would substitute an order that the claim in respect of the TED Talk be dismissed so far as it relates to publication up to 29 April 2020 but that there be judgment for the claimant for damages to be assessed in respect of publication of the TED Talk in this jurisdiction between 29 April 2020 and the date of judgment.

If my Lord and my Lady agree, it would remain to deal with consequential matters. These include the appropriate procedure for assessing damages and for resolving whether any other remedies should be granted. The additional pleaded claims that would be relevant in the light of what I have said are for an injunction to prohibit the defendant from publishing or causing further publication and prohibitory orders against third parties pursuant to s 13 of the 2013 Act. The claimant does not seek to prevent publication of the entire TED Talk but only the sentence that is complained of (see paragraph [12] above). Although the defendant has admitted responsibility for the publication of the words complained of up to trial, it is common ground that she is not able to control what the TED organisation does. There is an issue about the extent to which she should seek to persuade it to edit the TED Talk or cease publication of the talk in its current form.

I agree.

I also agree.