Rachel Riley v Laura Murray

At the trial of this libel action the claimant was awarded damages of £10,000 in respect of a tweet posted by the defendant which alleged that the claimant had stated that Jeremy Corbyn deserved to be violently attacked, denounced her for doing so, and suggested that others should never engage with her. The Judge rejected the defences of truth, honest opinion, and publication on matter of public interest that are provided for by sections 2, 3 and 4 of the Defamation Act 2013. This appeal raises issues about the interpretation and application of each of those defences in the somewhat unusual circumstances of this case.

The claimant was a well-known television presenter. She had a Twitter account with some 625,000 followers. She had spoken out publicly to condemn what she regarded as the fostering of anti-Semitism in the Labour Party under its then Leader Jeremy Corbyn. The defendant was Mr Corbyn’s Stakeholder Manager. She too was on Twitter, with some 7,252 followers. So also was Owen Jones, a well-known journalist and prominent Labour supporter whose Twitter identity featured the Labour red rose. He had about 1 million followers.

On 10 January 2019, Owen Jones posted a message on Twitter making reference to an incident in which an egg had been thrown at the former leader of the British National Party, Nick Griffin:

“Oh: I think an egg was thrown at him actually. I think sound life advice is, if you don’t want eggs thrown at you, don’t be a Nazi. Seems fair to me.”

On the afternoon of 3 March 2019, someone assaulted Mr Corbyn by throwing an egg at him when he was visiting a mosque. There were media reports that a man had been arrested, but that Mr Corbyn had not been injured and had continued his visit following the incident. A series of tweets followed during the late afternoon and evening of 3 March.

First, at 18:16, the claimant tweeted a screenshot of Owen Jones’ January 2019 tweet with the comment “Good advice”, accompanied by pictures of a red rose and an egg. This became known as “the Good Advice Tweet” or “GAT”. The GAT was a response to the attack on Mr Corbyn, as would be evident to any reader who knew of that attack. Because it is central to the issues, a copy of the GAT is annexed to this judgment.

The GAT received some 1.5m impressions, which is to say it was seen that many times. Different people responded to it differently. The evidence at trial included a large number of immediate responses. These showed that some thought it was “Rachel Riley calling Jeremy Corbyn who’d just been attacked a Nazi” and condemned her for “celebrating a physical assault on him”, “applauding his assault” and “condoning violence”. At least one called for her to face “criminal charges for incitement”. Others said, “I didn’t read it as that but as tongue-in-cheek highlighting [Owen Jones’] sanctimony over this”, “she didn’t label him a Nazi … she highlighted a clear hypocrisy”, she was “pointing out the hypocrisy of Owen’s position”, “If it’s wrong to throw an egg at someone you agree with its wrong to do it to someone you disagree with – simple.” These are just samples of what was itself a selection, but they give the flavour of the two kinds of response which the GAT provoked.

One of those who responded to the GAT in a way that tended towards the first of these two lines was the defendant. At 20:10 on 3 March 2019 she posted a reply to the Good Advice Tweet (“the Reply Tweet”) in these terms:

“You are publicly encouraging violent attacks against a man who is already a target for death threats. Please think for a second about what a dangerous and unhealthy role you are now choosing to play in public life.”

Because it was a reply, the Reply Tweet made the text of the GAT available to its readers. The claimant did not respond. At trial she said she did not recall whether she had seen it. She has never brought a claim in libel in respect of the Reply Tweet.

Later still, at 21.03, the defendant posted the Defendant’s Tweet, which was in these terms:

“Today Jeremy Corbyn went to his local mosque for Visit My Mosque Day, and was attacked by a Brexiteer.

Rachel Riley tweets that Corbyn deserves to be violently attacked because he is a Nazi.

This woman is as dangerous as she is stupid. Nobody should engage with her. Ever.”

The Defendant’s Tweet did not reply to or quote the GAT or include a screenshot of it nor did it otherwise provide the reader with access to the text of the GAT or any information about its content other than this account of what it said.

On 4 March 2019, the claimant and defendant staked out their positions on Twitter. At 00.17, the claimant responded by quote tweeting the Defendant’s Tweet, referring to it as an “appalling distortion of the truth” and thanking all those who had “checked the facts of this to call [it] out”. The claimant used raised eyebrow and head in hands emojis to express dismay at “those calling for my arrest, urgh”. At 07:38 the defendant replied to this, addressing the claimant in these terms: “Your tweet said ‘good idea’ to the words ‘if you don’t want to get egged, don’t be a Nazi’. The obvious interpretation of that is that you’re saying Corbyn is a Nazi and it’s a good idea to punch him. If you meant something different, please clarify it?” The claimant did not respond on Twitter.

The claimant sued for libel, complaining of the Defendant’s Tweet.

As is now standard practice, there was a trial of preliminary issues at which the court determined the natural and ordinary meaning of the words, to what extent the words were a statement of fact or an expression of opinion, and to what extent they were defamatory of the claimant. Nicklin J applied the well-established and uncontroversial principles he had summarised in Koutsogiannis v The Random House Group Ltd [2019] EWHC 48 (QB), [2020] 4 WLR 25 [11]-[16]. An important aspect of these for present purposes is the “single meaning rule” explained in Koutsogiannis [11]:-

“The Court’s task is to determine the single natural and ordinary meaning of the words complained of, which is the meaning that the hypothetical reasonable reader would understand the words bear. It is well recognised that there is an artificiality in this process because individual readers may understand words in different ways: Slim v Daily Telegraph Ltd [1968] 2 QB 157, 173D–E, per Lord Diplock.”

The process of identifying the single natural and ordinary meaning of a statement is an objective one. The intention of the publisher is irrelevant and no evidence is admissible other than the publication complained of itself: Koutsogiannis [12(ii), (x)].

Applying these principles Nicklin J determined that:

“(i)

The natural and ordinary meaning of the [Defendant’s Tweet] is:

(1)

Jeremy Corbyn had been attacked when he visited a mosque.

(2)

The Claimant had publicly stated in a tweet that he deserved to be violently attacked.

(3)

By so doing, the Claimant has shown herself to be a dangerous and stupid person who risked inciting unlawful violence. People should not engage with her.

(ii)

Paragraphs (1) and (2) are statements of fact. Paragraph (3) is an expression of opinion.

(iii)

Paragraphs (2) and (3) are defamatory at common law.”

Meanings (2) and (3) have come to be known as “the Factual Allegation” and “the Opinion” respectively.

At the trial of the action the defendant denied that the publication of those imputations met the threshold requirement imposed by s 1 of the 2013 Act, that they had caused or were likely to cause serious harm to the claimant’s reputation. On that question, evidence of actual responses was relevant. Having examined the evidence Nicklin J found for the claimant on that issue. In the alternative, the defendant asserted that the Factual Allegation was true, that the Opinion was honest opinion, and that the Defendant’s Tweet as a whole was a statement on a matter of public interest, the publication of which she reasonably believed to be in the public interest. The onus lay on her to establish these defences. Nicklin J accepted that the Defendant’s Tweet was honestly published on a matter of public interest but found that none of the statutory defences had been established. He therefore entered judgment for the claimant.

Damages were reduced by two important findings. The first was that the Good Advice Tweet was ambiguous, and the claimant was aware of this, so that posting it was “provocative, even mischievous” conduct which posed an obvious risk of misunderstanding and hostile reaction. This was held to be conduct that mitigated the damage, or “directly relevant background context” which had the same effect in accordance with Burstein v Times Newspapers Ltd [2000] EWCA Civ 338, [2001] 1 WLR 579, or both. Secondly, the Judge held that this was “not a case in which the damages award has an important role to play in vindicating the claimant’s reputation.” That was because the judgment and reporting of it would make clear the vindication to which the claimant was entitled.

This court refused the defendant permission to challenge the damages assessment but granted her permission to appeal against the Judge’s rejection of the statutory defences. By a respondent’s notice the claimant asks us to uphold the Judge’s decision on additional or alternative grounds. These include an invitation to reverse findings which the Judge made in favour of the defendant on elements of the statutory defences.

I shall consider each of the three defences in turn, whilst keeping in mind throughout that these are interlocking and to some extent potentially overlapping defences. It is important that we interpret and apply the statute in a way that is not only faithful to Parliament’s intention as expressed by the language of the 2013 Act and the Human Rights Act 1998 but is also, as far as possible, coherent.

Section 2 of the 2013 Act abolished and replaced the common law defence of justification. Section 2(1) provides “It is a defence to an action for defamation for the defendant to show that the imputation conveyed by the statement complained of is substantially true”. The relevant “imputation” for these purposes was the Factual Allegation which the Judge had identified at the preliminary trial: that “the claimant had publicly stated in a tweet that [Mr Corbyn] deserved to be violently attacked.” The Judge held that this was not substantially true because it was not an accurate account of the Good Advice Tweet.

Rejecting the defendant’s primary pleaded case the Judge said: “The Good Advice Tweet, taken at face value, plainly does not state that Jeremy Corbyn deserved to be violently attacked.” The Judge went on to reject the defendant’s alternative case, that the Factual Allegation was true because it was one that “could reasonably” be taken from the Good Advice Tweet and/or a meaning that “some or all of the people who read it” had in fact drawn from the Good Advice Tweet. The Judge said that the Good Advice Tweet was “open to more than one interpretation and was therefore ambiguous”. It could be read as suggesting that “there was an element of hypocrisy or inconsistency in Owen Jones’ original tweet” (on the footing that Jones’ attitude to the acceptability of throwing eggs at politicians varied according to his attitude to the politician’s views) (“the hypocrisy meaning”). Or it “could be (and was) read in a similar way to the defendant’s interpretation that Jeremy Corbyn was a Nazi and he too deserved to be similarly attacked”. The Judge said that this ambiguity was “obvious”.

The Judge rejected a submission on behalf of the claimant that he should apply the single meaning rule to the Good Advice Tweet. To do so would, he said, “stifle the very important fact that it was ambiguous”. But the Judge went on:-

“77.

Nevertheless, the Defendant’s defence of truth fails. What the Defendant stated, as a matter of fact, in the Defendant’s Tweet is not substantially true; it was at best half the story, presented to readers of the Defendant’s Tweet as if it was the full story. Critically, it took away the important fact that what the Good Advice Tweet said was a matter of interpretation or opinion, upon which reasonable views could differ, and replaced it with the Defendant’s unequivocal statement of what it meant as a matter of fact. In doing so, the Defendant’s Tweet was a misrepresentation of what the Claimant had said in the Good Advice Tweet.

78.

The position in which the Defendant finds herself could easily have been avoided. If she had said, in the Defendant’s Tweet, for example, that the Claimant had posted a Tweet which was capable of suggesting, or implied, that Jeremy Corbyn deserved to be violently attacked then she may well have had a viable defence of truth (or honest opinion). But she did not do this. She took upon herself the burden of describing, as a matter of fact, what the Claimant had said and failed because she removed the element of ambiguity. Worse, she added the two elements that the Claimant had stated that Mr Corbyn “deserved to be violently attacked”. By doing so, the Defendant put forward the very worst construction that could be put upon the Good Advice Tweet and stated, as a fact, that this was what the Claimant had said.

79.

These are not trivial differences, or ones that could be excused as small errors of detail, or exaggeration, within the permitted parameters of a defence of truth. There is a significant and material difference, not least in terms of likely harm to reputation, between offering an interpretation of what someone has said, and pronouncing unequivocally the interpretation …”

So the Judge found that the Defendant’s Tweet was not substantially true because it failed to reflect the ambiguity of the GAT. The Defendant’s Tweet told readers that the claimant had made an unequivocal statement that Jeremy Corbyn deserved to be violently attacked when in truth the GAT was ambiguous and no more than capable of bearing such an interpretation, as well as another quite different interpretation.

The defendant challenges this conclusion on two grounds. First, it is said that the Judge’s approach was wrong in law because the question for the court in a case such as this is “whether a section of the audience could reasonably have understood the GAT to contain the meaning stated” in the Defendant’s Tweet. In support of this contention the defendant relies, as she did below, on Begg v BBC [2016] EWHC 2688 (QB) (“Begg”). The defendant argues that on this approach she was entitled to succeed because on the Judge’s own findings of fact a section of the audience could (and did) reasonably interpret the GAT in this way.

Secondly, it is said that the Judge’s approach was wrong in fact because, on his own findings, “the GAT did contain the meaning stated” in the Defendant’s Tweet, which was enough for a finding of substantial truth (the emphasis is mine). Elaborating this argument before us Mr McCormick QC submitted that where a person makes a public statement that contains two meanings, (a) and (b), a publication that alleges that she made statement (a) is true. Meaning (b) is irrelevant. Mr McCormick argued that here, each of the meanings identified by the Judge was “a public statement in the GAT”, and that was a conclusive answer to the claim.

By her respondent’s notice the claimant seeks to meet these points by contending that the claimant “did not state anything in [the GAT], whether expressly or by implication, which can reasonably be interpreted or characterised as a public statement by the respondent that Jeremy Corbyn deserved to be violently attacked.” Logically, this point must come first.

At first blush the claimant appeared to be challenging the Judge’s findings about the two meanings the GAT could bear, as summarised at [20] above. That would be bold. A reviewing court will be slow to interfere with a judicial determination of the single meaning conveyed by words that are complained of as defamatory: Stocker v Stocker [2019] UKSC 17, [2020] AC 593 [58]-[59]. I would take the same approach to findings of fact about the range of meanings that could reasonably be drawn from a published statement, made by an experienced specialist judge. But the argument that Mr Bennett QC advanced to us was subtly different. He pointed to the meaning identified in paragraph [74] of the judgment: that “Jeremy Corbyn … deserved to be similarly attacked” (emphasis added). He submitted that, as is common ground, this meant “attacked with an egg”. Mr Bennett said that the Judge went no further than this, and that there is “a gulf” between this and the Factual Allegation. In simple terms, the argument is that “violently assaulted” suggests something much more serious than “being egged”; so the Judge’s finding as to meaning cannot support the defence of truth.

I see some force in the point. To a lawyer the words “violent assault” can cover a wide range of conduct from an armed attack causing really serious physical harm to comparatively trivial conduct. It would include an “egging”. But to the ordinary reader “violent assault”, stripped of context, would probably suggest something towards the more serious end of the scale. There is a difference. Mr Bennett’s problem is that Nicklin J expressly described the meaning he outlined as “similar … to the defendant’s interpretation”. Reading his judgment as a whole, I see no indication that he considered there was a significant difference between the two. I had not myself seen the point until it was raised on behalf of the claimant. I do not think we are entitled to substitute our view for that of the trial Judge. We should proceed on the footing that the GAT was ambiguous in the way that he identified and that the Factual Allegation was in substance one of the reasonable meanings of the GAT.

This way of putting the question encapsulates the thrust of the defendant’s point of law and her point of fact. In my judgment, the Judge was right for the reasons he gave.

I think the key to this issue is to keep firmly in mind throughout that what had to be proved true was the single defamatory Factual Allegation conveyed by the Defendant’s Tweet. To repeat, for convenience, this was that the claimant had “publicly stated in a tweet that [Jeremy Corbyn] deserved to be violently attacked”. I do not think the Judge’s finding that the GAT did not make such a statement can reasonably be gainsaid. The defendant’s argument to the contrary depends on reading the Factual Allegation identified at the preliminary trial as if it covered the case of a second, implicit meaning that could be and was read in by some reasonable readers.

This kind of argument about “the meaning of meaning” does arise from time to time in libel cases. It is rarely productive. Judges are generally very well aware of the need for clarity and precision in their determinations of meaning. Few will be more aware of that than Nicklin J, as Judge in Charge of the Media and Communications List. Here I see no ambiguity or room for debate about the matter. What is more, the Judge had made the meaning determination himself and he evidently had no doubt about what he had meant by it. As I read it, his judgment as a whole makes clear that the Factual Allegation was an imputation that the claimant had made an express statement; it did not encompass a possible implicit meaning to similar effect. In my view it is clear that the Factual Allegation was not literally true.

In this light the defendant’s submissions on this point can be seen to rest on a false equivalence. The Judge’s finding, that the GAT could reasonably be interpreted as conveying two meanings, one of which was substantially similar to the Factual Allegation, does not establish the truth of the proposition that “the claimant publicly stated in a tweet that Jeremy Corbyn deserved to be violently attacked.”

That is not the end of the matter because, as Nicklin J observed at [51], a defence of truth can succeed if the defendant proves the “essential” or “substantial” truth of the “sting” of the libel; proof of every detail is not required; peripheral inaccuracies, even extensive ones, should not distract the court. The defence that was open to the defendant was that it was substantially true to say (a) that the claimant had “publicly stated in a tweet that Jeremy Corbyn deserved to be violently attacked” because (b) the claimant had publicly tweeted something that could reasonably be, and had been, interpreted as conveying that suggestion – as well as another - by way of implication. It seems to me self-evident that these two things are not identical. As a matter of law the question for the Judge was whether, when it came to the defamatory sting of the Defendant’s Tweet, the true position, (b), was substantially or essentially equivalent to the alleged position, (a).

To answer that question the Judge had to and did make an assessment of the relative gravity of (a) and (b). He found that they were substantially different in gravity. He concluded that the Defendant’s Tweet “took away the important fact that the meaning of the GAT was a matter of interpretation or opinion … and replaced it with the defendant’s own unequivocal statement of what it meant as a matter of fact.” That was “a misrepresentation of what the claimant had said”. The defendant had put forward “the very worst construction” of the GAT and “stated, as a fact, that this was what the claimant had said”. These, said the Judge, were “not trivial differences, or ones that could be excused as small errors of detail, or exaggeration, within the permitted parameters of a defence of truth”. There was “a significant and material difference, not least in terms of likely harm to reputation” between “offering an interpretation of what someone has said, and pronouncing unequivocally the interpretation”. The Judge said this was “more than demonstrated” by the review of the responses to the Defendant’s Tweet which he had conducted for the purpose of deciding the issue of serious harm.

In my judgment these passages correctly reflect the applicable legal principles and the Judge was entitled to conclude that, for these reasons, the Defendant’s Tweet was not shown to be substantially true.

I do not think that Begg assists on this issue. That was a libel action in which the BBC successfully proved the truth of two defamatory imputations identified by the trial judge, Haddon-Cave J (as he then was). These were “(1) The claimant is an extremist Islamic speaker who espouses extremist Islamic positions. (2) The claimant had recently promoted and encouraged religious violence by telling Muslims that violence in support of Islam would constitute a man’s greatest deed.” To prove the truth of those imputations the BBC relied on various speeches given by the claimant. The defendant in the present case relies on a section of the judgment in which Haddon-Cave J explained his approach to the truth defence. Particular emphasis is placed on paragraph [61], where the judge said that at this stage of the analysis the court was not constrained by the single meaning rule, but had far more flexibility.

“The reason is that the Court is concerned with a quite different exercise, namely simply deciding whether the defendant has proved the ‘sting’ (i.e. of the ‘single’ meaning established in the first exercise) to be ‘substantially true’. In so doing, the Court does not have to find a ‘single’ meaning or even a range of reasonable meanings in relation to every disputed passage. The Court simply has to decide whether a section of the audience would reasonably take the words spoken to convey a particular message. Thus, if the Court were to conclude that at least a section of the audience would reasonably take the Claimant’s words to carry a particular message, that would be sufficient to support a finding that his words conveyed that message, even if it could not be said with certainty that the words were understood or conveyed the same message to everyone present.”

It is not clear that this passage was essential to the decision in the case. As Mr McCormick for the defendant accepted in the course of argument, the claimant’s speeches and postings were held to represent “an overwhelming case of justification for the BBC”; they showed “... a consistent pattern of behaviour on the part of the claimant of fomenting extremist ideas and ideology before engaged and receptive Muslim audiences”: see [368]-[369]. In any event, in the passage I have cited Haddon-Cave J was not purporting to lay down a rule of law of universal application; he was identifying the approach he would follow in the particular case before him. It must be right that, as Nicklin J observed at [71], “Whether a finding that at least a section of the publishees reasonably understood the relevant publication to convey a particular meaning will be sufficient to prove the substantial truth of an imputation will be case-specific”. It will depend, among other things, on the meaning to be proved true.

What Haddon-Cave J said in Begg was that a meaning reasonably taken from words by a section of the audience would be “sufficient to support a finding that the claimant’s words conveyed that message” (my emphasis). That is easy to understand in context; the central issue for decision in Begg was whether the claimant had “promoted and encouraged extremism”. The present case is different. Here, the Factual Allegation was not that the GAT “conveyed a message”. It was an allegation that the claimant had made an express and unequivocal public statement of a particular kind. Hence, as I have explained, the literal truth of the Factual Allegation cannot be established by showing that the GAT could be and was reasonably read by some as containing an implied message to the same or similar effect. When deciding whether the Factual Allegation is substantially true, the fact that the GAT could be taken to imply such a meaning is relevant but not conclusive.

Section 3 of the 2013 Act provides, so far as relevant:-

“3 Honest opinion

(1)

It is a defence to an action for defamation for the defendant to show that the following conditions are met.

(2)

The first condition is that the statement complained of was a statement of opinion.

(3)

The second condition is that the statement complained of indicated, whether in general or specific terms, the basis of the opinion.

(4)

The third condition is that an honest person could have held the opinion on the basis of (a) any fact which existed at the time the statement complained of was published; (b) anything asserted to be a fact in a privileged statement published before the statement complained of.

(5)

The defence is defeated if the claimant shows that the defendant did not hold the opinion.

…

(8)

The common law defence of fair comment is abolished and, accordingly, section 6 ofthe Defamation Act 1952 (fair comment) is repealed.”

Subsections (2), (3), (4) and (5) reflect the previous common law to this extent at least: the statement must be one of opinion rather than fact; it must indicate its basis in some way (I shall call this “the Basis Condition”); and there must be some fact on the basis of which an honest person could have held the opinion (“the Objective Honesty Condition”); but satisfaction of these conditions will not be enough if the claimant proves as a fact that the defendant did not hold the opinion she expressed (I will call this “the Dishonesty Disqualification”).

The defendant had to satisfy the Judge that each of the three conditions in s 3(2)-(4) was met in relation to the imputation that “By so doing, the Claimant has shown herself to be a dangerous and stupid person who risked inciting unlawful violence. People should not engage with her.” The Judge had decided at the preliminary trial that this was an expression of opinion, so the first condition was met. At the main trial, the Judge held that the Basis Condition was also met. But he held that the Objective Honesty Condition was not met because the Opinion was “expressly premised” on the truth of the Factual Statement and that, as he had already held, was untrue.

The Judge rejected the defendant’s submission that on the true construction of s 3(4)(a) it was enough for her to prove that an honest person could have expressed the Opinion based on the terms of the GAT, and that “the reaction to the Good Advice Tweet more than demonstrates that an honest person could have expressed the Opinion”. The Judge noted that the well settled position of the common law was that “if the alleged facts relied on as the basis for comment turn out to be untrue, a plea of fair comment avails the defendant nothing”:London Artists Ltd v Littler Grade Organisation Ltd [1969] 2 QB 375, 395 (Edmund Davies LJ) approved in Tse Wai Chun v Cheng [2001] EMLR [31] (Lord Nicholls).Having considered commentary in Duncan & Neill on Defamation and Blackstone’s Guide to the Defamation Act 2013, and his own previous obiter observations in Morgan v Associated Newspapers Ltd [2018] EWHC 3690 (QB) [63]-[64], the Judge concluded that the literal reading of s 3(4)(a) relied on by the defendant would involve a “fundamental (and radical) departure from the settled position at common law” with surprising consequences which Parliament would not have intended.

In these circumstances it was not necessary for the Judge to consider the claimant’s reliance on the Dishonesty Disqualification but “in fairness to the defendant” he stated his conclusion that she did honestly hold the Opinion.

The defendant appeals on the ground that the Judge was wrong “to rely on the common law to give s 3(4)(a) of the 2013 Act a construction which its words do not bear”. Alternatively, if reference to the common law was permissible, it is said that the Judge “failed to take account of s 6 of the Defamation Act 1952.” The claimant supports the Judge’s approach to the Objective Honesty Condition but also, by her respondent’s notice, seeks to challenge his decision on the Basis Condition. The argument is one that was not advanced below but no objection is taken on that ground and we allowed it to be advanced. The argument is that the Basis Condition was not met “because the basis of the opinion indicated in the Defendant’s Tweet, namely that the [claimant] had made a public statement that Jeremy Corbyn deserved to be violently attacked is a basis that is factual in character, and that (purported) factual basis was (and is) untrue.”

In my opinion the proposition quoted at [43] above is misconceived. The only question raised by s 3(3) of the 2013 Act is whether the statement complained of indicated the basis of the opinion which it contained. That is a question of analysis or assessment which turns exclusively on the intrinsic qualities of the statement complained of. If the statement did not indicate the basis for the opinion the analysis stops there and the defence fails. If it did, the condition is met and the analysis moves on to the next stage. The extraneous question of whether the matters indicated as the basis for the opinion are true or false is immaterial at this stage of the analysis. As Nicklin J held at [92], “The issue (at this stage) is not whether the factual premise is right, but whether it was sufficiently indicated”.

That is how the common law operated. The five ingredients of the common law defence of fair comment were identified by the Supreme Court in Joseph v Spiller [2010] UKSC 53, [2011] 1 AC 852, approving with qualifications the judgment of the Court of Final Appeal of Hong Kong in Tse Wai Chun v Cheng (above) [16]-[21]. The first condition was that the statement was on a matter of public interest. The second was that it was “recognisable as comment, as distinct from an imputation of fact”. The third was, so far as relevant, that “the comment must be based on facts which are true” so that “if the facts on which the comment purports to be founded are not proved to be true … the defence of fair comment is not available”. The fourth condition was that “the comment must explicitly or implicitly indicate, at least in general terms, what are the facts on which the comment is being made”. The fifth was that “the comment must be one which could have been made by an honest person…” The questions of whether the statement indicated the basis for the opinion and whether the basis for the opinion was true or privileged were therefore treated as separate and distinct matters.

The Explanatory Notes to the 2013 Act explain (at para 19) that Section 3 removes the common law condition that the statement must be on a matter of public interest but otherwise “broadly reflects the current law while simplifying and clarifying certain elements.” It can readily be seen that the statutory conditions in s 3 of the 2013 Act reflect or at least contain echoes of the second to fifth common law conditions, albeit with some changes of order and wording. The language of s 3(3) is very close indeed to that of Spiller, and it is no surprise that the Notes say at para 22 that “Condition 2 (in subsection (3)) reflects the test approved by the Supreme Court in Joseph v Spiller that ‘the comment must explicitly or implicitly indicate, at least in general terms, the facts on which it is based.’” It is s 3(4) (Condition 3) that calls for an examination of whether there was “a fact which existed at the time” on the basis of which the opinion could have been honestly expressed. Thus, the statutory structure corresponds to that of the common law: the questions of whether the statement indicated the basis for the opinion and whether there was a true factual basis for that opinion are dealt with separately. It is conceded by the claimant, indeed it is inherent in her argument, that the Defendant’s Tweet did indicate the basis on which it was made. Accordingly the Basis Condition is satisfied.

The claimant’s contentions as to the Basis Condition rest on two main foundations. The first is a further passage in the same paragraph of the Explanatory Notes that I have quoted. Paragraph 22 goes on to say this:

“Condition 2 and Condition 3 (in subsection (4))aim to simplify the law by providing a clear and straightforward test. This is intended to retain the broad principles of the current common law defence as to the necessary basis for the opinion expressed but avoid the complexities which have arisen in case law, in particular over the extent to which the opinion must be based on facts which are sufficiently true and as to the extent to which the statement must explicitly or implicitly indicate the facts on which the opinion is based.”

The claimant’s submission is that this wording indicates that in order to “retain the broad principles of the current common law defence” ss 3(3) and 3(4) are to be “read in tandem”. That may be so, but the claimant’s approach would require us to add words into s 3(3) and I can see no sufficient basis for doing so. The subsection seems to me to be clear and unambiguous. To read into the second condition an implicit requirement that the basis indicated in the statement must be shown to be substantially true would depart from the common law structure. It would create an obvious overlap and tension with the express requirement in the third condition. That would introduce obscurity and complexity which is the opposite of the intention indicated in the Notes.

The second main foundation for the claimant’s argument on this issue is a decision not cited to Nicklin J, namely that of HHJ Parkes QC in Burki v SeventyThirty Ltd [2018] EWHC 2151 (QB). The case has persuasive authority as Judge Parkes, a defamation specialist as a practitioner and as a judge, undoubtedly did hold that a s 3 defence of honest opinion could not succeed if the stated basis for the opinion was factual, and false. That is clear from his decision on the facts: see [243]. His reasons for reaching that view of the law were explained in [230]:

“In the light of the firm restatement in Joseph v Spiller of the common law requirement that comment be based on facts that are true, it would have been very radical departure from the common law (rather than a ‘broad reflection’ of it) for Parliament to intend that it should be enough for a defendant to show a basis for his comment – when that basis consisted of facts – that was plainly untrue”.

Burki therefore supports the claimant’s overall case that s 3 should not be interpreted as allowing a defendant who expresses an opinion on a false factual premise to succeed in a defence of honest opinion. But in my opinion, the claimant’s reliance on Burki as authority for a radical re-interpretation of s 3(3) is misplaced. Judge Parkes gave no clear indication of where in the statute he located the requirement for a true factual basis. To the extent he gave any indication, it seems to me that his preferred analysis may have been that the common law principle he referred to was reflected in s 3(4)(a) of the 2013 Act, as Nicklin J held in the present case: see [231].

Read literally, all that s 3(4)(a) requires is that there should at the time of publication be some fact, of whatever nature, that would allow an honest person to express the opinion complained of. The statutory wording does not link the Basis Condition and the Objective Honesty Condition. On the face of it, therefore, there need be no connection between the facts indicated in the statement complained of and the fact(s) that can support the opinion. The defence could succeed even if the facts indicated were false, as long as there was some other fact on the basis of which an honest person could have held the opinion.

To judge by the materials before us, the unanimous opinion of commentators and judges so far has been that this was not Parliament’s intention. Two related reasons have been identified for that conclusion. The first is the one identified by Judge Parkes in Burki: that it would involve a “radical” departure from the common law rather than retaining the “broad principles” of the common law, as the Explanatory Notes suggest was the intention. Nicklin J agreed with that view in his obiter dicta in Morgan and in his decision in the present case, as does the commentary in Blackstone’s Guide. That commentary points to other features of the Parliamentary history of the provision.

The second reason, with which Nicklin J also agreed, is that this interpretation would lead to outcomes that seem odd, and unlikely to have been intended. This point is made forcefully in this extract from paragraph 4.50 of Blackstone’s Guide which the Judge set out in his [95]:

“It cannot have been intended that an opinion expressed on wholly false facts can be supported on an entirely different basis. Otherwise, for example, a person could be accused of dishonesty, or of being a danger to the public, on the basis of some recent alleged, but entirely false, conduct in his or her public capacity, and the comment could be defended as one which could be held by an honest, but prejudiced, obstinate, etc. person, on the basis of some conduct in a wholly different and private capacity, years previously. …”

To make this example concrete, and take it further, suppose a defendant who told the world that the claimant had stolen his car and was dishonest. If the allegation of theft was false the common law defence of justification would have failed as would a defence of truth under s 2 of the 2013 Act. At common law, the defence of fair comment would also have failed, because the indicated basis for the comment was false. Yet on a literal reading of s 3(4)(a) the defendant could now successfully defend the expression of opinion if there was some entirely different factual basis on which the opinion could honestly have been held, for instance, that the claimant had cheated in an exam at the age of 16. Mr McCormick was unable to identify any logical reason why Parliament would have required the basis for the opinion to be indicated if there were no link to the facts which could support it.

There are other odd consequences of this interpretation. In this example the allegation of dishonesty would only strike the reader as an expression of opinion at all because of the stated (but false) factual basis; a bare allegation of dishonesty will normally be analysed as an allegation of fact. If the defendant held the opinion based on an honest belief in the allegation of theft the Dishonesty Disqualification would not apply. So the defendant would be given a defence of honest opinion on facts honestly but falsely stated, provided some other fact could be found to support the same opinion. The defence would be available adventitiously. The position would be at least as strange if the defendant could rely on a fact previously reported on a privileged occasion by someone else, as would be the case under s 3(4)(b) on the defendant’s interpretation.

Against these considerations, the defendant can and does rely on the “golden rule” that the Court should give statutory language its ordinary and natural meaning. Mr McCormick submits that s 3(4)(a) is clear and unambiguous and the effect it would have on its literal interpretation is not absurd. Mr McCormick submits that there is no justification for resort to the Explanatory Notes or, if there is, the Notes support his client’s case because they make clear an intent to “simplify” the law. He says the view favoured by commentators and by the Judge requires a lot of words to be read into s 3(4)(a). And he argues that in a case such as the one I have described the claimant would not be unduly disadvantaged. The claimant would win damages for the publication of the false defamatory factual allegation. The damage consequent on the publication of the opinion would be “the price the claimant has to pay for freedom of expression”.

There is no doubt that some of these are attractive points. We would normally strive for a construction that gives the statutory words their normal meaning. For my part, however, I am more attracted to the view adopted by the Judge. I believe we are entitled to look at the Explanatory Notes. As Lord Hodge DPSC observed in R (Project for the Registration of Children as British Citizens) v Secretary of State for the Home Department [2022] UKSC 3, [2022] 2 WLR 343 [30]: “The context disclosed by such materials is relevant to assist the court to ascertain the meaning of the statute, whether or not there is ambiguity and uncertainty …” I note that the Supreme Court took that approach when construing s 4 of the 2013 Act in Serafin v Malkiewicz [2020] UKSC 23, [2020] 1 WLR 2455 [58], [66] (Lord Wilson). The Notes in this case militate against Mr McCormick’s argument for the reasons given in Burki and by the Judge in this case.

I do not think it likely that Parliament intended to break the link, firmly established in the common law, between the facts indicated in the statement and the fact(s) that could support the opinion. We have been shown nothing in the legislative history to suggest as much. No logical explanation for doing so has been offered, nor can I see any plausible policy justification for the strange consequences I have identified. The more likely explanation for the statutory wording, in my view, is that the draftsman was attempting to establish an Objective Honesty Condition that would be clear and simple to operate by comparison with the previous law, and the defendant’s interpretation is available as an unintended by-product.

By 1948, there was a problem with “the rigidity with which the rule is applied that the plea of ‘fair comment’ must fail unless all the defamatory facts contained in the matter complained of and on which the comment is based are truly stated”: see para 89 of the Report of the Committee on the Law of Defamation (aka “the Porter Committee”) Cmd. 7536. In section 6 of the Defamation Act 1952, Parliament used the following convoluted language to address this problem:

“In an action for libel or slander in respect of words consisting partly of allegations of fact and partly of expression of opinion, a defence of fair comment shall not fail by reason only that the truth of every allegation of fact is not proved if the expression of opinion is fair comment having regard to such of the facts alleged or referred to in the words complained of as are proved.”

The much simpler wording of s 3(4)(a) means that if a fact exists that could support the opinion that will be enough, and it will not matter how many false assertions were made. As paragraph 28 the Explanatory Notes say, “Subsection (8) repeals section 6 of the 1952 Act … This provision is no longer necessary in light of the new approach set out in subsection (4).”

In this case, however, it is not necessary to resolve the issue. The appeal on this aspect of the case fails on the short and simple basis that the Opinion that had to be defended was that “by so doing the claimant has shown herself to be a dangerous and stupid person who risked inciting unlawful violence” and “people should not engage with her” (emphasis added). “By so doing” is shorthand for “by publicly stating in a tweet that Jeremy Corbyn deserved to be violently attacked.” As the Judge held, the fact that the Opinion was expressly (and I would add exclusively) premised on the truth of the Factual Allegation means that it cannot survive the failure of the defendant’s case on the issue of truth.

If other facts existed on the basis of which an honest person could hold the opinion that the claimant was “a dangerous and stupid person” or that “people should not engage with her” those facts could not come to the defendant’s rescue. That is because they could not support the view that the claimant had “shown herself to be” such a person “by publicly stating in a tweet that Jeremy Corbyn deserved to be violently attacked”. The defendant’s contention, that the Opinion can be supported by the existence of the GAT coupled with the defendant’s understanding of the meaning and/or effect of the GAT and/or the reactions of others, fails for this reason. It also fails because for these purposes “the defendant’s interpretation of the GAT” was an opinion not a fact.

Mr McCormick rightly points out that this approach places something of a premium on the words “by so doing”. It also puts significant weight on the other words I have emphasised above. I agree that there are wider implications. As it seems to me at present, where a defendant expresses a defamatory opinion that is expressly based on a single factual assertion the s 3 defence is likely to fail if that assertion is false. That is because the court is likely to frame the meaning in similar terms to those used by Nicklin J here. But there is nothing new or wrong in this. This is a common way of formulating a meaning, which properly reflects the words complained of. It has long been the law that a defamatory opinion cannot be defended if it expressly stated a basis which was wholly untrue. It is hard to see any injustice in that.

Besides, this will not be the position in every case. The basis for a statement often includes more than one alleged fact, sometimes a range. The basis for an opinion may be indicated implicitly and very broadly. A well-known example is Kemsley v Foot [1952] AC 345, where the words complained of were “Lower than Kemsley”and the name of the claimant, an active newspaper proprietor, was considered a sufficient allusion to the entire journalistic output for which he was responsible. A more modern example is provided by Lowe v Associated Newspapers Ltd [2006] EWHC 320 (QB), [2007] QB 580. Eady J held it was enough for the subject-matter of the opinion to be indicated, and that a commentator may (within certain limits) seek to support the opinion on the basis of relevant extraneous facts. In cases of these kinds s 3(4)(a) is likely to have a role to play, whatever its true construction. There may be cases where a multiplicity of facts is indicated in one of these ways, of which just one would be enough to support the s 3 defence. Trial judges will need to be alert to attempts artificially to force such cases into the category we are dealing with here, with a view to imposing a straitjacket on the honest opinion defence. But I am confident that can be done.

Section 4 of the 2013 Act abolished and replaced the common law defence known as Reynolds privilege. It provides, so far as relevant:-

“4 - Publication on matter of public interest

(1)

It 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.

(2)

Subject to subsections (3) and (4), in determining whether the defendant has shown the matters mentioned in subsection (1), the court must have regard to all the circumstances of the case.

…

(4)

In determining whether it was reasonable for the defendant to believe that publishing the statement complained of was in the public interest, the court must make such allowance for editorial judgement as it considers appropriate.

(5)

For the avoidance of doubt, the defence under this section may be relied upon irrespective of whether the statement complained of is a statement of fact or a statement of opinion.”

Section 4(1) requires the court to determine three issues: (i) was the statement complained of, or did it form part of, a statement on a matter of public interest? If so (ii) did the defendant believe that publishing the statement complained of was in the public interest? If so (iii) was that belief reasonable? See Economou v de Freitas [2018] EWCA Civ 2591, [2019] EMLR 7 [87].

The Judge found for the defendant on the first issue, explaining:

“120…. The Claimant’s Good Advice Tweet was published on Twitter, on a public platform, and was intended by the Claimant to be read both by her large number of followers and more widely. The Claimant was well aware that her words, in the Good Advice Tweet, were likely to provoke public comment and engagement. Comment upon, or response to, the Good Advice Tweet was a matter of public interest. It was certainly not a “personal and private” matter.

The Judge was also “satisfied that the defendant had demonstrated” the second element of the s 4 defence. He accepted her evidence. This was that on reading the GAT she thought that the claimant was “endorsing and encouraging” the attack on Jeremy Corbyn; that this was “reckless and irresponsible”, “deliberately provocative” and “sent a dangerous message to the wider world and was a dangerous and stupid thing to do”. She had monitored reaction to the Good Advice Tweet during the evening and saw several replies which made points similar to the ones she later expressed in the Defendant’s Tweet. She believed that her Tweet was in the public interest. The written submissions for the claimant before trial did not appear to the Judge to attack these contentions, and in cross-examination the defendant was not challenged about her belief that posting the Defendant’s Tweet was in the public interest.

The Judge held however that the defence failed at the third stage: the defendant had failed to demonstrate that her belief was reasonable. The Factual Allegation was, he said, one of some seriousness the publication of which had caused serious harm to the claimant’s reputation by misleading those who relied on the Defendant’s Tweet as a description of what the claimant had said. He went on:-

“124.

The most significant factor, in my assessment, is the failure of the Defendant to include the Good Advice Tweet in the Defendant’s Tweet. I accept the Defendant’s evidence that she did not do so because she did not want to drive further traffic to it, but this cannot be a good enough reason for depriving readers of the Defendant’s Tweet of accurate information about, and the proper context of, the Good Advice Tweet. In Turley, I observed that “it can never be in the public interest for a journalist to misrepresent in an article the information or evidence that s/he has obtained”: [153]. The Defendant is not a journalist, but this fundamental principle applies equally to her or anyone else seeking to avail themselves of a s.4 defence in answer to a defamation claim. A person who misrepresents the material they have is likely to find it difficult to establish that s/he reasonably believed that the resulting inaccurate publication was in the public interest.”

The defendant challenges this aspect of the Judge’s decision on four grounds. It is said that the Judge’s approach (a) wrongly required the defendant to give a true description of the meaning of the GAT, rather than a reasonable and honest description, (b) failed to take into account the series of steps which the defendant took to arrive at her belief that publication was in the public interest, instead regarding her conduct as impetuous, (c) failed to make proper allowance for editorial judgment, and (d) failed to take account of the provocative nature of the claimant’s conduct.

By her respondent’s notice the claimant contends that the Judge’s rejection of the s 4 defence should be upheld on the additional or alternative grounds that (1) he should have held that the defendant failed on the first issue because the propriety or otherwise of the claimant’s conduct in posting the Good Advice Tweet “is not capable in law of being ‘a matter of public interest’ for the purposes of s 4(1)(a)”; and/or (2) the defendant’s belief was not reasonable because she “could not reasonably have believed that what was being imputed factually to [the claimant] in the Defendant’s Tweet was true”.

Mr Bennett advanced an elaborate argument based on the reasoning of the Court of Appeal in Serafin v Malikiewicz[2019] EWCA Civ 852, [2019] EMLR 21 and a passage in Gatley on Libel and Slander, the gist of which was that matters of public interest are and should be confined to such matters as “[T]he business of government and political conduct … the protection of public health and safety; the fair and proper administration of justice; … the conduct of the police; …. breach of charitable fiduciary rules; involvement in serious crimes; corporate malpractice.” (The passages cited are extracted from a list in para 16-006 of Gatley 13^th ed.). Mr Bennett submitted that the Judge had failed to address the issue in appropriate detail. On a true analysis, he submitted, the Defendant’s Tweet was “on” the GAT, which was a witty, mildly mischievous, throwaway, two-word remark, turning Owen Jones’ words back on him. It represented the personal reaction of the claimant, who is not a politician, to Jones’ earlier tweet in the light of reported events on 3 March 2019. The Defendant’s Tweet amounted to nothing more than one tweet purportedly summarising and commenting on another.

I would uphold the Judge’s decision and reject the claimant’s case on this issue. It would in my opinion be wrong to take a narrow view of what can count as a matter of public interest. It is clear that the concept does not extend to matters which are purely “personal and private such that there is no public interest in their disclosure” (Reynolds v Times Newspapers Ltd [2001] 2 AC 127, 176-177 (Lord Bingham CJ), approved in Flood v Times Newspapers Ltd [2012] UKSC 11, [2012] 2 AC 273[33] (Lord Phillips)). But the overall criterion identified by Lord Bingham was a broad one: “matters relating to the public life of the community and those who take part in it.” I do not think Serafin assists the claimant. Even if Mr Bennett is right to submit that some of what this court said about the public interest in that case has survived the Supreme Court’s decision, which is not obvious, the Court of Appeal’s decision on the point turned on an analysis of the particular words complained of, which were held to be “aimed at the claimant’s conduct in his personal capacity as a contractor and supplier…”

Nor do I think a great deal is to be gained by reference to non-exhaustive lists of topics that have been found to qualify as matters of public interest, such as the one in Gatley, if the point is that the case falls outside the scope of such a list. That does not significantly advance the argument. But as it happens the Gatley list, which draws on decisions from a range of legal contexts, includes another case on s 4 of the 2013 Act in which this court held that criticism of the present claimant for what she had said on Twitter was at least arguably a publication on matter of public interest. In Riley v Sivier [2021] EWCA Civ 713, [2021] 4 WLR 84, the allegation complained of was, in summary, that she had engaged upon, supported and encouraged a campaign of online abuse and harassment of a 16-year-old girl and “by so doing” was a “serial abuser” who had acted hypocritically, recklessly, irresponsibly and obscenely. Defences of truth and honest opinion were struck out. So was a s 4 defence but an appeal was allowed on the footing that this merited a trial.

It was common ground that in this context the public interest is “necessarily a broad concept”. In a judgment with which the President of the Queen’s Bench Division and Henderson LJ agreed I observed at [20] that “In the common law of fair comment, it was well-established that matters of public interest included the public conduct of public figures and any statement which the claimant—whether or not she was a public figure—had put before the public for consideration”. I said that one way the case could be put is that “the matters of public interest which the article was ‘on’, or about, were the public conduct of a prominent public figure and, in particular, statements she had made or caused to be made publicly (a) in a media interview and (b) on Twitter.”

The present case seems to me similar but stronger. I consider the claimant’s arguments on this issue to be wholly artificial. She is not an obscure private individual, nor was this some kind of private joke at Owen Jones’ expense that had nothing to do with the public at large. The claimant was not chatting privately with friends. She chose instead to use a public platform to address a readership that exceeds that of most if not all national newspapers. And she did so on a political topic. She was, as the defendant’s Reply Tweet put it, choosing to play a role in public life. The way I would put it is that the matter of public interest which the Defendant’s Tweet was “on” or about was the conduct of the claimant, a well-known celebrity and prominent political activist, in publishing to her hundreds of thousands of followers a provocative tweet relating to matters of political significance, namely the attacks on Nick Griffin (which had been the subject of an approving publication to over a million people by a prominent journalist in January 2019) and Jeremy Corbyn (which was a newsworthy subject of current interest that was firmly in the public domain).

There are two main questions here. The first is the one raised by the respondent’s notice: could the defendant reasonably have believed that the GAT conveyed the Factual Allegation? I have dealt with this point at [26]-[27] above. The answer is that the Judge found that in substance the Factual Allegation reflected one of the reasonable meanings of the GAT, and we have no grounds to interfere with that conclusion.

The second main question is whether we should reverse the Judge’s further conclusion that, although the defendant reasonably believed the GAT to convey the Factual Allegation, it was nevertheless unreasonable for her to believe that it was in the public interest to say what she did in the Defendant’s Tweet. I have not been persuaded that we should interfere with this conclusion either.

This was not a finding of fact but an evaluative decision. To overturn it on appeal, as Mr McCormick points out, the defendant does not need to show that it was perverse. But she does need to persuade us that the decision was “wrong … by reason of some identifiable flaw in the treatment of the question to be decided, such as a gap in logic, a lack of consistency, or a failure to take account of some material factor, which undermined the cogency of the conclusion”: Re Sprintroom Ltd [2019] EWCA Civ 932, [2019] BCC 1031 [76]. That is still a fairly demanding test. I would add that in applying it the court should bear in mind that a judgment given after a trial does not have to address every argument or point of fact, and “must be read against the background of what was actually in dispute between the parties”: Fage UK Ltd v Chobani UK Ltd [2014] EWCA Civ 5, [2014] FSR 29[116]-[117] (Lewison LJ). Reviewing the judgment on the issue of reasonable belief I see no flaw that undermines its cogency.

For the Judge, the critical feature of the Defendant’s Tweet was that it misrepresented the GAT by suggesting that the claimant had made an unequivocal public statement that Jeremy Corbyn deserved to be violently attacked, when in truth the GAT was ambiguous. This being so, the Defendant’s Tweet misled the public and its publication was not in the public interest. It was unreasonable for the defendant to believe that it was. It is necessarily implicit in this reasoning, as it seems to me, that the misrepresentation and misleading were not the result of some permissible oversight but either conscious or at least careless.

There was evidence of actual knowledge that the GAT was ambiguous in the way the Judge found it to be. The defendant’s own account was that before posting the Defendant’s Tweet she went through Twitter and saw reactions to what the claimant had said in the GAT. Many of those reactions adopted the hypocrisy meaning: see [6] above. They were there to be seen. It was a reasonable interpretation of this evidence that the defendant must have seen both interpretations. The Judge did not spell out such a finding in terms but I think the better reading of his judgment is that he considered the defendant was aware of the ambiguity. He referred to her “decision not to include the Good Advice Tweet, or accurately to describe what it said”, which implies such knowledge. And he found that the tweets posted by the defendant after the GAT and before the Defendant’s Tweet “demonstrate an awareness that she was offering an interpretation” of the GAT. That no doubt reflects the fact that these tweets referred to the claimant “implying” the anti-Corbyn message. The defendant never said in her evidence that she had been ignorant of the ambiguity.

She did however say, when cross-examined about third-party tweets in response to the GAT, that she had seen none that said anything to the effect that “this is a comment on hypocrisy”. The point was not pursued further in questioning, and the Judge did not address this evidence directly. The Judge also referred (at [126]) to “the defendant’s failure to appreciate that the Good Advice Tweet was capable of another interpretation”. In the circumstances, I prefer not to rest my decision on the interpretation of the judgment I have identified in [79] above. I would uphold the Judge’s decision on the basis that he was entitled to conclude and did conclude, at least, that the defendant ought reasonably to have appreciated the ambiguity.

This seems to me to follow inescapably from his clear findings that it was “obvious” that the GAT was ambiguous ([74]) and that “there were two obvious meanings: the hypocrisy meaning or the meaning that suggested that Jeremy Corbyn deserved to be egged because of his political views” ([155]). As I read the judgment, Nicklin J concluded at a minimum that it was unreasonable for the defendant to believe that it was in the public interest to characterise the GAT as she did, and to express the opinion she did, when it was obvious and therefore should have been apparent to her that a different and much less damaging interpretation of the GAT was available. I regard that conclusion as unassailable given his factual findings, which are not impugned.

The significance of ambiguity for a public interest defence in libel has been considered in two cases to which it is helpful to refer: Bonnick v Morris [2002] UKPC 31, [2003] 1 AC 300 and Banks v Cadwalladr [2022] EWHC 1417 (QB). In Bonnick, the court was dealing with the common law Reynolds defence. In both cases, the context was ambiguity in the statement complained of as a libel. The cases are nonetheless instructive. In Bonnick, Lord Nicholls said at [24] that a journalist should not be penalised for making a wrong decision on a question of meaning on which people might reasonably take different views. But he went on at [25] to say that this “should not be pressed too far”.

“In the normal course a responsible journalist can be expected to perceive the meaning an ordinary reasonable reader is likely to give to his article. Moreover, even if the words are highly susceptible of another meaning, a responsible journalist will not disregard a defamatory meaning which is obviously one possible meaning of the article.”

In Banks Steyn J, DBE applied these observations in the context of the s 4 defence. At [123] she summarised her analysis in this way:

“A defamatory meaning should not be ignored by a journalist if it is “obviously one possible meaning” ([25]) or “glaringly obvious” ([27]); to do so would not be reasonable. But if that threshold is not reached, the reasonable belief of a journalist who did not perceive the more damaging meaning falls to be assessed by reference to the less damaging meaning.”

It may be that these principles do not transpose directly into a situation such as the present, but I do not think the defendant can reasonably argue for any more generous test. She has never done so. She has not addressed the issue. The defendant’s argument has always been that her conduct should be assessed exclusively by reference to what she reasonably believed the GAT to mean, and that on that footing it was reasonable for her to believe that it was in the public interest to publish the Factual Allegation and the Opinion. That, in my view, is simplistic and wrong. When assessing the reasonableness of a belief that it is the public interest to denounce a person as dangerous and stupid for what they have said in a public statement, it must be relevant that the statement has an obvious alternative and lesser meaning which is not worthy of such denunciation. Here, the Judge was entitled to conclude that the defendant ought reasonably to have appreciated that the GAT could also be interpreted as conveying the hypocrisy meaning and that it was therefore unreasonable for her to believe that presenting the position unambiguously, as she did, was in the public interest.

This approach seems to me consistent with one strand of the authorities to date, which is that “a belief [is] reasonable for the purposes of s 4 only if it is one arrived after conducting such enquiries and checks as it is reasonable to expect of the particular defendant in all the circumstances of the case”: Economou v de Freitas [2016] EWHC 1853 (QB), [2017] EMLR 4 [241], approved [2018] EWCA Civ 2591, [2019] EMLR 128 [101] and endorsed by the Supreme Court as “no doubt helpful” in Serafin (above) at[67].

I see no inconsistency with my judgment in Yeo v Times Newspapers Ltd [2015] EWHC 3375 (QB), [2017] EMLR 1, aspects of which are relied on by the defendant. In the passages relied on, at [175] and [179], I said that in a case such as that one “it will be ‘fair’ to present readers with factual conclusions honestly and reasonably drawn by journalists who were themselves witnesses to the key events; it is permissible to summarise, and to be selective; … fairness does not require the publisher to present the reader with all the factual material that could support a competing assessment … it is not incumbent on the responsible journalist to lay out for the reader all the pros and cons relevant to a particular conclusion.” Yeo was very different from this case. It was a decision on the application of the Reynolds defence to newspaper reports of an undercover journalistic investigation of a leading politician. But I also said (at [175]) that “if the evidential picture is misrepresented or presented in a wholly unbalanced way, that may well be unfair”. Here, the Judge found that the evidential picture had been unreasonably misrepresented.

This disposes of the first element of this ground of appeal. It also means that I can deal quite shortly with the alleged failure to make proper allowance for editorial judgment. The defendant made clear that the omission of the GAT from the Defendant’s Tweet was not accidental but deliberate. That is the only relevant editorial decision. The defendant never said that she took it because she had decided to disregard the hypocrisy meaning. On her account, she could not have done that, as she did not appreciate the ambiguity. The defendant’s only explanation for excluding the GAT from her tweet was that she did not want to “drive additional traffic” to the GAT. The Judge accepted this was her reason but held that it was not a good enough reason for depriving those who read the Defendant’s Tweet of the GAT or further information about it. He was entitled to reach that conclusion.

As the Judge pointed out at [12], the defendant could have used the simple expedient of a screenshot. And as he pointed out at [131], hundreds of people were able to criticise the claimant for posting the GAT in ways that were on the face of it defensible expressions of opinion. Indeed, the defendant was one of these. A few hours before she posted the Defendant’s Tweet she had posted the Reply Tweet. This gave her followers the text of the GAT coupled with her interpretation of its meaning and her comments upon the claimant’s conduct (“dangerous and unhealthy”). The claimant has never complained about the Reply Tweet, which appears to me to be plainly an expression of opinion that fully and accurately indicates its basis. It is hard to understand why the public interest could reasonably be thought to justify repetition of substantially the same defamatory messages in the form of the Factual Allegation and Opinion, but without the source material.

I see no merit in the argument that the Judge “failed to take account of the steps the defendant took to arrive at her belief that publication was in the public interest”. He had little need to do so, when he accepted that the belief was honest, and the single ground on which he held it to be unreasonable was a failure to reflect an obvious and important ambiguity. On that ground, he had regard to the Reply Tweet and others which revealed the defendant’s thinking about meaning at the material time. The other points relied on do not advance the defendant’s case. To take two examples, the importance of the topic cuts both ways, because the more important the topic the greater the importance of accuracy; it would clearly be contrary to the public interest to publish a false and misleading allegation that a prominent celebrity had publicly approved the use of violence against a leading politician. And if the point is (as in part it is) that the evidence showed a considered approach rather than an impetuous one, it tends to undermine the defendant’s case by making it harder to justify her disregard of the hypocrisy meaning. The “provocative nature” of the claimant’s tweet is a matter of hindsight and an irrelevance for this purpose.

For these reasons I would dismiss the appeal.

I agree that the appeal should be dismissed for the reasons given by Warby LJ.

I accept that, as Mr McCormick QC submitted on behalf of the defendant, the fact that the judge found that the Good Advice Tweet (“GAT”) had, among others, an obvious meaning that “Jeremy Corbyn deserved to be egged” (paragraph 155 of the judgment below), is relevant to the issues of whether: the defendant’s tweet was substantially true, for the defence of “truth”; and the defendant’s belief, though honest, was reasonable, for the defence of “publication on matter of public interest”.

I agree that the answer to the defence of truth, in the particular circumstances of this case, is provided by paragraph 77 of the judgment below, which is set out in paragraph 21 above of Warby LJ’s judgment. In circumstances where the GAT was deliberately “provocative, even mischievous” (so justifying a reduction in damages awarded to the claimant) and ambiguous, it was not substantially true to report only one possible meaning of GAT as having been tweeted by the claimant.

Further, in such circumstances, the defendant’s honest belief that the GAT conveyed the factual allegation was not reasonable. This is because the GAT was obviously ambiguous.

I agree that the appeal should be dismissed for the reasons given by Warby LJ.