An inferential house of cards – serious financial loss under section 1(2) Defamation Act 2013: Burki v Seventy Thirty Ltd & Ors  EWHC 2151 (QB)
Section 1 of the Defamation Act 2013 had two aims. First, it aimed to raise the test for establishing that a statement is defamatory above the pre-existing common law’s threshold. In order to do so, section 1(1) 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.’ Second, the section aimed to raise the bar to claims still further for commercial claimants. This aim was reflected in section 1(2), which provides that ‘harm to the reputation of a body that trades for profit is not “serious harm” unless it has caused or is likely to cause the body serious financial loss.’ These provisions were justified by the coalition government of the day as necessary reforms designed to give greater prominence to free speech interests within the law of defamation, and to protect ordinary people from being crushed by libel claims brought by big businesses.
Recent judicial decisions, however, have frustrated these aims. Earlier this year, in Lachaux v Independent Print Media Ltd, the Court of Appeal overturned the High Court decision of Warby J, holding that section 1(1) did not fundamentally alter the presumption of damage in libel. Key to this decision was the Court of Appeal’s reasoning that the words ‘likely to’ in section 1(1) simply meant having a ‘tendency’ to cause serious reputational harm. This negated any need for the claimant to provide tangible evidence of such harm, or a likelihood of it occurring, on a balance of probabilities. A ‘tendency’ to cause serious harm could be inferred, in an appropriate case, from the seriousness of the allegation itself, and potentially (though not necessarily) from circumstantial evidence of the nature and extent of its publication. The only actual change rendered by section 1(1), according to the court, was to replace the common law threshold of ‘substantial’ harm (identified in Thornton v Telegraph Media Group Limited) with ‘serious’ harm. Moreover, this change is purely semantic; in Thornton, Tugendhat J refers to the threshold using the term ‘seriousness’ far more frequently than he does the word ‘substantial’.
As a result, in Lachaux, section 1(1) was interpreted in such a way that it did not actually raise the threshold above the level it had already reached at common law. This was in spite of the court’s recognition that Parliament had intended to ‘raise the bar’. This problematic nature of Lachaux – explored by the author in detail elsewhere – has now been laid out all the more plainly by the High Court’s ruling in Burki v Seventy Thirty Ltd, which extends Lachaux in a manner that further frustrates Parliament’s intention in passing section 1(2). Burki was a case in which both a claim in misrepresentation and a counter-claim in defamation ultimately succeeded. (An additional counter-claim in malicious falsehood failed.) It is the defamation claim which is of particular significance.
Tereza Burki had paid a significant sum of money to procure the services of a matchmaking company (Seventy Thirty Ltd, also known as ‘70/30’) in the hopes of meeting a suitable romantic partner. Unsatisfied with the service she received, Ms Burki posted unfavourable reviews of the company on two webpages, one of which – the ‘Google review’ – was ultimately found to attract liability. The judge, HHJ Richard Parkes QC, found the ordinary and natural meaning of the Google review to be that Seventy Thirty Ltd did not have the means to operate an effective matchmaking service, and was engaged in a fraudulent scheme to extract money from its clients.
The question was whether this meaning cleared the twin hurdles set up by sections 1(1) and 1(2). The judge had no difficulty in coming to the conclusion that the allegation of fraud was serious enough that it would tend to cause serious reputational harm. No evidence was needed in support of this, since – following Lachaux – such an inference could be drawn simply from the seriousness of the allegation. He therefore proceeded to consider section 1(2). Agreeing – perhaps defensibly, on coherence grounds – with Warby J’s observation in Pirtek (Uk) Ltd v Jackson that the words ‘likely to’ must bear the same meaning in consecutive subsections of the same statute, the judge determined that section 1(2) should be interpreted as meaning that serious reputational harm (to a body that trades for profit) is that which causes or has a tendency to cause that body serious financial loss. This also fitted with the slightly earlier decision of Nicol J in Euroeco Fuels (Poland) Ltd v Szczecin and Swinoujscie Seaports Authority SA that there was no indication on the face of the statute to limit ‘serious financial loss’ to special damage, nor to preclude courts from inferring such loss from the evidence.
Just as with the section 1(1) hurdle, the claimant does not need – in the judge’s view – to provide evidence of serious financial loss, nor of a likelihood of such loss occurring, in order to satisfy section 1(2), albeit Seventy Thirty Ltd did adduce some such evidence. To do so would (wrongly, in the judge’s view) confuse ‘serious financial loss’ with special damage. Instead, the court could infer a tendency to cause serious financial loss from broader circumstantial facts. In this case, the court did precisely that, inferring such a tendency from evidence that publication took place on the internet, and from the nature of the claimant company’s business being such that the loss of even one client’s business would entail serious loss. The number of readers of the review was ‘unquantifiable’, but the judge inferred from its online location that the readership would have been ‘substantial’.
Thus, the judge concluded that the Google review was ‘liable to cause 70/30 Ltd serious financial loss’. It is entirely understandable that the judiciary should instinctively incline to the view that the words ‘likely to’ should bear the same meaning in concurrent subsections of the same statute. To construe them otherwise would seem, prima facie, absurd. But giving them the same meaning in this instance gives rise to a problem Parliament most likely did not foresee. The problem is that it builds a claim for a commercial claimant by stacking inferences on top of one another, creating an inferential house of cards.
That which is inferred is not proved in quite the same sense as that which is directly evidenced. An inference is given a status akin to evidential proof, but it is fundamentally a reflection of a likelihood – a probability. To infer X is to say that, from the evidence before us, we think the odds are sufficiently in favour of X occurring that we think it likely that X will occur (or has occurred). When one says that a statement will probably cause harm, and that harm will probably cause financial loss, the combined probability of the two is less than that of the first. To put some crude numerical values on it, assume that a certainty of both serious reputational harm and serious financial loss consequent upon that harm is represented by the number 100. The likelihood of serious reputational harm is less than a certainty and so we might say it amounts to 75 per cent of that original certainty. The further likelihood of serious financial loss consequent upon that reputational harm occurring can now only be a percentage of the 75 per cent likelihood of reputational harm; it might be 75 per cent of 75 per cent. Thus, whilst the words ‘likely to’ appear to bear the same semantic meaning, they do not actually do so; the second likelihood is necessarily built upon the first and is thus a quantitatively lesser likelihood than the first.
Where an inference is drawn from available evidence (which evidence does not itself point directly to X occurring), what we are really saying is: ‘it is more likely than not that it is more likely than not that X occurred.’ Thus by drawing the two consecutive inferences in sections 1(1) and 1(2) we are saying something rather precarious. What we are saying is this: (A) given the seriousness of the accusation it is more likely than not that it is likely that the claimant’s reputation will be seriously harmed, and (B) that given (A), and given the nature of the claimant’s business and the forum within which the allegation was made, it is more likely than not that it is likely that the reputational harm will cause the claimant serious financial loss. It is a percentage of a percentage of certainty. But there is a significant risk that, when deciding (B), courts may lose sight of the fact that (A) is itself merely a likelihood of something being likely. In stating that the words ‘likely to’ bear the same meaning in sections 1(1) and 1(2), the decision in Burki, which follows the problematic path that the courts started down in Lachaux, is actively courting precisely that risk.
To this criticism, it might be retorted that all proof – whether evidential or inferential – is taken merely on a balance of probabilities (in civil cases) or beyond a reasonable doubt (in criminal cases). Thus, even if the inferences were excluded, decisions made on the evidence would also represent a mere fraction of a fraction of certainty. Such a retort would imply that inferences and tangible evidence are of the same or similar evidential value. This is an implication that drew support from the Court of Appeal in Lachaux, which pointed out that juries are routinely instructed that they may draw inferences against defendants in criminal trials. But this observation fails to provide an answer to the quantitative point. For in a criminal trial the prosecution may pray in aid an inference to be drawn from evidence that does not itself directly point to the defendant’s guilt, such as the defendant’s refusal to testify. This inference, however, could not stand alone as the sole ‘proof’ of guilt. A criminal case would always contain more evidence than merely the silence of the defendant, and the jury must still be satisfied to of guilt – on all the evidence, which may include an inference – to the criminal standard. Thus, if it draws an inference, it must be satisfied beyond reasonable doubt both that the defendant’s silence indicates that the defendant has something to hide, and that, based on other evidence, what the defendant has to hide is her guilt. If a jury is not sure that the evidence alone indicates guilt, nor that the silence connotes guilt, it must acquit. It is one thing to draw an inference. It is quite another to substitute that inference for any other evidence that points directly to the defendant’s liability, even where the standard of proof is the lower, civil standard.
When interpreting statutory provisions, the starting point is – naturally – the text of the legislation. There are occasions, however, when the legislative drafting is far from perfect and anomalies arise from attributing ostensibly sensible meanings to individual words. In Burki, the decision to give the same interpretation to the words ‘likely to’ in consecutive subsections of section 1 opens the door to claims succeeding where the cumulative likelihood of both serious reputational harm and serious financial loss occurring may actually fail – quantitatively – to reach the civil standard of proof. Most worryingly, the manner in which the capacity to rely on inferences is set out by the court suggests that such a failure to reach the civil standard might go unnoticed – mistaken, quite innocently, for the successful reaching of that standard.
Lachaux gave rise to a situation in defamation law where the interest that the law now protects is a right not only to one’s good reputation, but to a freedom from criticism of a sort that would have a tendency to cause serious reputational harm – even if none ever actually occurs. The problem with this is that it flies in the face of Parliament’s intention when enacting section 1 of the Defamation Act 2013. This problem has now been exacerbated by the ruling in Burki. For it interprets the statute in such a way that a corporate claimant can successfully sue a dissatisfied customer over a single unfavourable review without having to provide any direct or tangible evidence either that its reputation was seriously harmed or that it has suffered, or is likely to suffer, serious financial loss.
It might be suggested that the manner in which the court reached its interpretation of section 1(2) in Burki indicates that it suffered from a lack of attentiveness to context (in much the same way as the Court of Appeal in Lachaux). Alternatively, and more charitably, the judge in Burki might be cleared of such a charge on the basis that following Lachaux – the really problematic decision in this field – ensured a basic, logical coherence between sections 1(1) and 1(2). Nonetheless, the first instance decision of Warby J in Lachaux, which was subsequently overturned on appeal, showed considerably greater sensitivity to the context within which the Act was passed and in which it falls to be interpreted than either the Court of Appeal’s judgment or the judgment in Burki. Warby J had held that, other than in obvious cases involving ‘national media publication of a grave imputation, such as conspiracy to murder or serious sexual crime’ where an inference may readily be drawn as to the likelihood that the statement would cause serious reputational harm, a claimant may need ‘to prove some facts beyond the words themselves and the fact and extent of their publication’ in order to satisfy section 1(1). In other words, serious reputational harm would ordinarily need to be proved as a fact, on the balance of probabilities; only in exceptional, ‘grave’ cases would it be legitimate to infer serious harm. Interpreting section 1(1) in this fashion would have resulted in an ostensibly significant change to the common law, since it ‘may well be that a claim will no longer succeed where the meaning is a serious one but the claimant’s reputation in the eyes of those who read the words complained of is not in fact harmed seriously, if at all’. Such a change would properly reflect Parliament’s intention to ‘raise the bar’ to defamation claims.
However, by a combination of the Court of Appeal’s rejection of Warby J’s approach in Lachaux, and the High Court decision in Burki, we are left with a situation where claimants in defamation actions continue to benefit from an approach that locates a likelihood of serious reputational harm in statements even where there is no evidence that directly ‘proves’ that such a level of harm has, or is likely to occur. All that a claimant need provide is sufficient circumstantial evidence for the court to infer from it a tendency to cause serious reputational harm. Once the court infers such a tendency, the very same limited, circumstantial evidence can be used to infer a likelihood of serious financial loss. This enables a corporate claimant to squash a critical review like a bug, even where it has not and never does actually suffer any serious reputational harm or financial loss. Troublingly, the common law’s deficiencies thus continue to plague the post-Defamation Act era, despite Parliament’s attempt to cure them. The 2013 Act was intended to protect vocally critical members of the public from being squashed like bugs, not to facilitate more efficient squashing.
Thomas D C Bennett
City Law School
  EWCA Civ 1334,  QB 594. A further appeal in Lachaux was heard by the Supreme Court on 13 and 14 November 2019.
  EWHC 1414 (QB),  1 WLR 1985.
 Note 1, .
 See Thomas D C Bennett, ‘Why so serious? Lachaux and the threshold of ‘serious harm’ in section 1 Defamation Act 2013’ (2018) 10(1) Journal of Media Law 1.
  EWHC 2151 (QB).
  EWHC 2834 (QB) at .
  EWHC 1081 (QB).
 Note 1, .
 Ibid, .
 Ibid, .
 This matter is addressed in greater detail in Bennett, n 4, at 16.
  EWHC 2242 (QB),  QB 402.
 Ibid, [57-58].
 Ibid, .