3.1. Who committed the fallacy?

Notice that, over the course of the discussion so far in this section, audiences (recipients of balanced reports) have gradually become more visible. Bothsiderist reasoning, in both virtuous and vicious instances, begins with framing, and framing is typically something that one agent (speaker/originator) does with at least one other agent (audience/recipient) in mind. This is important because audiences are themselves active participants of bothsiderist reasoning. Once the framing step is completed, audiences may reject the speaker’s characterisation of the state of the evidence; if they don’t, uptake is secured, and the fact of the existence of a controversy over $p$ will enter the common ground (Step 2).Footnote ⁷ And, once the fact of controversy over $p$ enters the common ground, it is again audiences who make the crucial bothsiderist move towards epistemic modesty (Step 3).Footnote ⁸

Technically, then, it is audiences who commit the fallacy of bothsiderism (where a fallacy is indeed committed):

One thing that jumps to the eye in this description of the division of bothsiderist labour is that it remains neutral on the distribution of responsibility for the fallacy among the various participants, thus suggesting (ironically enough) that it is shared equally by all parties. This evaluative neutrality makes sense in the context of a first-time exploratory analysis of a “heretofore untheorized” fallacy type, as is Aikin and Casey’s. But when we turn our attention to concrete exemplifications of the fallacy, it leaves much to be desired.

Returning once more to our initial illustration, imagine a radio listener – call him Pat – who, having tuned in to the infamous BBC Radio broadcast, surmises that the debate over RCC is not yet settled. “Both discussants seemed sure of what they were saying, and the points they made seemed mostly reasonable…”, he might think; “There’s probably some truth to what each said, but I suppose we’ll have to wait and see what’s what”, he might conclude. In drawing this epistemically modest conclusion, Pat partakes in, and strictu sensu commits, the bothsiderist fallacy: he improperly or mistakenly concludes, from the fact that he has witnessed two parties disagreeing over RCC, that RCC is not conclusively settled by the evidence; accordingly, the better (epistemically and practically prudent) option is to suspend judgment on the issue, at least for the time being.

Epistemically modest conclusions are not warranted with respect to RCC: Pat’s reasoning instantiates the fallacious version of bothsiderism. And whose fault is that?, a cynic or a provocateur might ask. Nobody twisted Pat’s arm and made him engage in faulty reasoning; there was nothing stopping him from doing some extra research into the matter, even prior to listening to the broadcast; at the very least, he could have researched the invited discussants’ credentials. If he didn’t, and he came to believe that RCC is not settled after listening to a third-party disagreement broadcast by the BBC, that’s on him.

Rhetorical moves of this sort are simple enough – and notoriously effective (Ceccarelli Reference Ceccarelli2011; Oreskes and Conway Reference Oreskes and Conway2011; Dixon and Clarke Reference Dixon and Clarke2013a). Aikin and Casey’s analysis of the meta-argumentative fallacy of bothsiderism is descriptively illuminating, but lacks the evaluative depth required to resist such moves, and so to vindicate the prevailing intuition that responsibility is (mostly) borne by news sources. As a first step in this direction, the next section takes a closer look at the kind of evidence with which recipients of balanced journalism are presented.

4.1. Who produced the evidence?

The BBC Cambridgeshire broadcast was vitiated by a series of bad framing choices, we saw (Section 3 ). The BBC made a poor choice of segment opener. And, this poor choice was compounded by the decision to platform a ‘balanced’ disagreement on the question as to the reality of climate change (RCC). For, in so doing, the BBC presented its audience with two pieces of misleading higher-order evidence.

First: the provision of a platform to an advocate of the ‘unreasonable minority’ view that climate change isn’t real ( $\neg $ RCC) constitutes evidence that the view, and its advocate, are credible. In general, the strength of the credibility boost conferred to a speaker and his propounded view will plausibly co-vary with the epistemic ‘clout’ of the platform. In this case, given the undisputed prestige enjoyed by the BBC, it seems plausible that the corresponding credibility boost would be especially strong. Therefore, the BBC’s offer of a platform to an advocate of $\neg $ RCC produces HOE ${{\rm{}}_1}$ : higher-order evidence in favour of $\neg $ RCC.

Second: the provision of a platform to a disagreement over the epistemic standing of RCC, between disputants portrayed as sufficiently credible (to warrant an invitation from a reputable news organisation), constitutes evidence that the disagreement is genuine. That is, the disagreement has merit, and so presumably cannot be dismissed out of hand. But the most plausible explanation, when two sufficiently credible disputants genuinely disagree over the epistemic standing of some $p$ , is that the first-order evidence does not conclusively establish the superiority of $p$ over $\neg p$ , and viceversa. Therefore, provision of a platform to a disagreement over RCC between purportedly credible disputants produces HOE ${{\rm{}}_2}$ : higher-order evidence that the first-order evidence is divided between RCC and $\neg $ RCC.Footnote ¹² Since the first-order evidence actually overwhelmingly supports RCC, HOE ${{\rm{}}_2}$ ultimately favours $\neg $ RCC.

HOE ${{\rm{}}_1}$ and HOE ${{\rm{}}_2}$ are individually and jointly misleading with respect to the epistemic standing of RCC.Footnote ¹³ More precisely, HOE ${{\rm{}}_{1 - 2}}$ misrepresent the respective credibility of RCC and $\neg $ RCC, and so are unreliable indicators of the confidence one should entertain about these claims. Since the first-order evidence is not (even close to being) evenly balanced between RCC and $\neg $ RCC, the best explanation of, and the appropriate conclusion to draw from, HOE ${{\rm{}}_{1 - 2}}$ is that one or more parties involved in the disagreement should be downgraded (or outright disqualified) in their capacity of competent and/or sincere evaluators of the first-order evidence (Lackey Reference Lackey, Haddock and Pritchard2008).

But, crucially, this explanation may not be available to all recipients of HOE ${{\rm{}}_{1 - 2}}$ – here, the radio broadcast audience. For, the possibility of correctly assessing HOE ${{\rm{}}_{1 - 2}}$ is conditional on the availability of independently reliable (first-order or higher-order) evidence speaking to the epistemic standing of RCC. Since RCC is a subject of scientific expertise, however, most recipients (both actual and intended) of HOE ${{\rm{}}_{1 - 2}}$ will have limited to no access to such evidence. More precisely, most recipients of HOE ${{\rm{}}_{1 - 2}}$ will have no access to independently reliable first-order evidence about RCC, since the latter presupposes pertinent scientific expertise, and expertise is in short supply. And so most recipients of HOE ${{\rm{}}_{1 - 2}}$ will depend on the availability of reliable higher-order evidence about RCC in order to correctly assess the import of HOE ${{\rm{}}_{1 - 2}}$ .

Again, however, it is not obvious that such reliable higher-order evidence will be easily available; nor that it will be sufficient, quantitatively and qualitatively speaking, to counter the deleterious effects of HOE ${{\rm{}}_{1 - 2}}$ . In part, this is owing to the fact that higher-order evidence is notoriously difficult to rebut (Levy Reference Levy2019: 499): “While an argument can demonstrate that the evidence you cite for $p$ does not in fact support $p$ , as long as the platform provided to you to say it is reputable, it just does confer credibility on you, no matter what anyone says” (see also Lewandowsky et al. Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012, Reference Lewandowsky, Ballard, Oberauer and Benestad2016). And in part, it is because the source of HOE ${{\rm{}}_{1 - 2}}$ – a major news network – is likely to be one of the main, if not the main, source of higher-order evidence about RCC accessible to the audience.

This makes the task of defeating HOE ${{\rm{}}_{1 - 2}}$ significantly harder, to put it mildly, and more costly. Recall that HOE ${{\rm{}}_{1 - 2}}$ , respectively, support the propositions that $\neg $ RCC is credible (HOE ${{\rm{}}_1}$ ), and that the strength of the evidentiary support enjoyed by RCC and $\neg $ RCC is not decisive for either side (HOE ${{\rm{}}_2}$ ). What sort of additional content would need to be introduced in order to successfully rebut these claims?Footnote ¹⁴ Simply denying the truth of HOE ${{\rm{}}_{1 - 2}}$ won’t help, as indicated above. The surest path, in order to successfully defeat HOE ${{\rm{}}_{1 - 2}}$ , would be to undercut the credibility of its source. This is practically difficult however, given the robustness of the BBC’s reputation. Technically, though, Philip Foster (the climate change-denying guest) also had a hand in producing the higher-order evidence in question. Wouldn’t it be possible, at least in principle, to undercut his credibility as a way of neutralising HOE ${{\rm{}}_{1 - 2}}$ ?

Yes and no. Imagine, first, that the second guest on the show had devoted his allotted speaking time to introducing undercutting defeaters – denouncing Foster’s lack of scientific credentials, covert interests, etc. Even if we ignored the implausibility of this scenario (it seems unlikely that the show would have been allowed to air), it seems doubtful that an intervention of this sort would impeach the credibility boost afforded by the BBC’s offer of a platform (HOE ${{\rm{}}_1}$ ).

Alternatively, imagine – even more implausibly – that Foster’s lack of testimonial competence and/or sincerity had been exposed by the broadcast host. Theoretically speaking, this could be a more promising option than the previous. However, there are two problems related to this hypothetical scenario. First, if the point of introducing defeaters is to ensure the factual accuracy of the resulting higher-order evidence, then nothing short of fully neutralising HOE ${{\rm{}}_{1 - 2}}$ will do (see also Kortenkamp and Basten Reference Kortenkamp and Basten2015). But generating HOE ${{\rm{}}_{1 - 2}}$ was the BBC’s own doing; proposing that they turn around and nullify that same evidence is not so much a solution as it is absurd. Secondly, engaging in what would essentially amount to a practical contradiction would presumably damage the credibility of the BBC in the eyes of its audience. For all of the BBC’s flaws, this would be an undesirable outcome: if neutralising a particular token of misleading higher-order evidence comes at the cost of depriving audiences of a key resource, when it comes to bridging knowledge gaps, then the price of such a strategy is much too high (De Cruz Reference De Cruz2020; Goldman Reference Goldman2001).

Let’s recap the train of thought so far. Indiscriminate application of the journalistic balance norm will sometimes foreseeably result in false balance. Falsely balanced reports of third-party disagreements, in turn, produce misleading higher-order evidence about the subjects of those disagreements. Once the misleading higher-order evidence has been generated – once it’s ‘out there’ – it is difficult, and costly, to dismiss it, or make it go away. Airing further testimony that sets the record straight with respect to the first-order evidence may help but likely won’t suffice for the reasons just described. Undermining the credibility of the source of the misleading evidence, in turn, may be more effective. But it carries substantial risks of its own, most prominently that of widening existing epistemic gaps, and further eroding public trust in genuinely trustworthy testifiers.

Thus, it falls upon those at the receiving end of BR to manage the good higher-order evidence alongside the bad: “Third parties witnessing the dispute, and who might otherwise have acquired knowledge via testimony from the participants, also have higher-order evidence in the light of which they should adjust their beliefs” (Levy Reference Levy2019: 493). Crucially, the source of this ‘should’ is epistemic: adjusting one’s beliefs in light of higher-order evidence is the rational thing to do. This means that recipients of BR who have limited (or no) reliable means of sorting the wheat from the chaff may have little choice but to adjust in the direction of epistemic modesty (Section 3 ). If this is right, then it looks as though committing the bothsiderist fallacy, in response to false balance, is (at least sometimes) the rational thing for audiences to do.

This result offers a first theoretically backed indication that the burden of responsibility is not equally shouldered by all those who ‘contribute’ to instances of the bothsiderist fallacy. This is promising, but it’s not enough. Attributions of responsibility require showing that the agent did the deed, did so freely, and did so knowingly. So far, we can lay claim to two out of three. To gain at least some purchase on the third, let us now focus more squarely on the pragmatic dimension of BR.

5.1. What gets passed along, and what gets through?

Journalists are in the business of informing, i.e. of providing audiences with ‘truth claims’. Generally speaking, uptake of an informative speech act requires that the intended recipient believe that the speaker/informer believes that the uttered locution is true. He-said/she-said reporting is somewhat more complex. Here, uptake presupposes that audiences recognise that originators are committed: neither to the truth of each of the locutions jointly constitutive of a token he-said/she-said utterance; nor to the truth of the propositional content of one assertion over the other.Footnote ¹⁵ With respect to surface or on-record content conveyed via he-said/she-said reporting, rather, audiences are required to think that originators are committed to the attribution of the reported locutionary contents to the respective parties in the debate.

On their part, reporters are entitled to assume that the foregoing is indeed what audiences believe – that audiences understand the nature of informative journalism, and the reporter’s role, at least to this extent.Footnote ¹⁶ In turn, audiences are entitled to assume that reporters are reliable and that they comply, not only with Grice’s Cooperative Principle but, more stringently, with the conversational maxims (Quantity, Quality, Relation, and Manner). Importantly, this means that as a rule, audiences can legitimately expect that any false implicatures generated via he-said/she-said reporting are accidental, and not borne out of a deliberate attempt to mislead.

With this in mind, the first order of business is to establish which kind (or kinds) of implicature may be generated by he-said/she-said reporting. I first argue that he-said/she-said reporting qua format (speech act type) produces generalised conversational implicatures: “default inference[s], […] that [capture] our intuitions about a preferred or normal interpretation” (Levinson Reference Levinson2000: 11). I then argue that tokens of he-said/she-said reporting are liable to generate what Jennifer Saul has labelled audience-implicatures (Saul Reference Saul2002). In the next subsection, these verdicts are used to sketch a more fine-grained assessment of the distribution of responsibility for fallacious bothsiderist reasoning triggered by balanced frames.

Recall from Section 4 that disagreements among presumptive epistemic peers may provide additional evidentiary cues about either the evidentiary status of the subject of the disagreement, or the relative epistemic standing of the disputants. These additional pieces of (higher-order) evidence are typically inferred vs articulated explicitly, and may be inferred by audiences, if present, as well as the disputants themselves. When disagreements are presented via he-said/she-said reporting, these evidentiary cues are pragmatically conveyed to audiences via implicature (see e.g. Simion Reference Simion2017). Importantly, the implicatures generated are not specific to he-said/she-said reporting vs other BR formats; they are therefore conversational (because non-detachable) rather than conventional.Footnote ¹⁷ Furthermore, the implicatures generated by he-said/she-said reporting “need not be ‘guessed’ by the hearer, as they are generated by the institutional context itself” (Simion Reference Simion2017: 416, emphasis added). This strongly suggests that the inferences invited by BR qua format are sufficiently stable that they qualify as generalised conversational implicatures (Grice Reference Grice1989; Zufferey et al. Reference Zufferey, Moeschler and Reboul2019; Levinson Reference Levinson2000).

Conversational implicatures require a relatively tight coordination between speakers and audiences.Footnote ¹⁸ But in the context of informative journalism, where speakers paradigmatically address largely unknown and heterogeneous audiences, such coordination is not plausibly guaranteed on a local, case-by-case basis. In fact, coordination isn’t guaranteed even in contexts that are not as pragmatically complex and layered as this: even in relatively ordinary communicative contexts, and with the best of intentions, speakers “can’t fully control what they conversationally implicate” (Saul Reference Saul2002: 241).

More fully, it is possible for a speaker to fail to conversationally implicate $q$ , by uttering $p$ , because $q$ is not required in order to uphold the presumption that the speaker is being cooperative, by the audience’s lights – though it is so required by the speaker’s lights. The same audience, in turn, may think that preserving the cooperativeness presumption requires supposing that the speaker intended to convey (not $q$ but) $q{\rm{*}}$ . That is, “the audience’s state of mind can impose constraints on what is conversationally implicated” (Saul Reference Saul2002: 232).

Thus, an attempted conversational implicature may result in the production of an utterer-implicature on the one hand, and an audience-implicature on the other.Footnote ¹⁹ The notions of audience- and utterer-implicature are helpful, I think, for diagnosing what goes on, and what can go wrong, when audiences are presented with balanced reports of third-party disagreements. To see this, let us examine a slightly fictionalised and simplified version of our initial illustration.

Imagine that network N reports that $p$ : speaker A said RCC, and speaker B said $\neg $ RCC. Let’s suppose that N is a reputable, well-established news outlet both nationally and internationally; on any given day, millions of people from various corners of the globe read one or more pieces on N’s website, or at least check its headlines. And let’s focus our attention on two among the numerous recipients of N’s report that $p$ , Pat and Qat.

As noted earlier, Pat and Qat are entitled to assume that N’s report obeys not only the Cooperative Principle but also the maxims, in particular Quantity and Relation (Simion Reference Simion2017) – “if not at the level of what is said, at least at the level of what is implicated” (Grice Reference Grice1989: 86).Footnote ²⁰

Take, in particular, the second Quantity maxim (Q2).Footnote ²¹ It seems clear that N has violated Q2 in reporting that $p$ : A said RCC, and B said $\neg $ RCC. For, on the understanding that the proposition over which A and B disagree (RCC) is both truth-apt and factual (vs, say, expressing an aesthetic preference), and that A and B’s assertions about RCC are mutually incompatible, it cannot be the case that these assertions are both true. Thus, N’s contribution is strictly “more informative than is required.”

Pat and Qat may therefore reasonably think that “there is some particular point” (at the level of what is implicated) to N’s over-informativeness (at the level of what is said). For instance, Pat may reasonably infer, from the fact that N has quoted both A’s assertion that RCC and B’s assertion that $\neg $ RCC, that N also thought (meant) ${q_1}$ : Both sides bring to the table worthwhile contributions, which should be taken into account when thinking about RCC. In turn, Qat may reasonably think: “N would not have quoted both A’s assertion that RCC and B’s assertion that $\neg $ RCC unless N also thought that ${q_2}$ : Opinions on RCC, including expert opinions, remain divided.”

Pat and Qat may further consider, respectively: N knows (and knows that I know that N knows) that I can see that the supposition that N thinks that ${q_1}$ (resp. ${q_2}$ ) is required; N has done nothing to stop me thinking that ${q_1}$ (resp. ${q_2}$ ); N intends me to think, or is at least willing to allow me to think, that ${q_1}$ (resp. ${q_2}$ ); and so N has implicated that ${q_1}$ (resp. ${q_2}$ ).Footnote ²²

Note that while Pat and Qat reach different epistemically modest conclusions, their inferential paths nonetheless converge insofar as both take the ‘point’ of N’s report to be that of conveying information about the epistemic status of RCC. Since this is among the stable inferences triggered by he-said/she-said reporting, Pat and Qat’s interpretation of N’s communicative intention is both reasonable and appropriate.

If it were also the only possible such interpretation, then ${q_1}$ and ${q_2}$ might qualify as full-blown conversational implicatures of N’s report that $p$ ; moreover, N could be charged with having deliberately misled its audience by reporting that $p$ . But disagreement reports carry, and may be used to convey, more than one (potential) evidentiary moral (see also Section 4). Specifically, N might be operating under the assumption that speaker B “is a person of interest for the audience in a way that turns data about the quality of B’s profile into an interesting piece of news,” and, furthermore that “the audience either already has (shared background) or, alternatively, is offered independent information as to the truth or falsity of the content of that person’s testimony, so as to be able to judge her moral/social/political [or epistemic] profile” (Simion Reference Simion2017: 417). And so, in reporting that A said RCC and B said $\neg $ RCC, N may have intended to convey information, not about the epistemic status of RCC but about the epistemic standing of the parties involved in the debate.

Say this is the case. Under this supposition, the speaker’s intention constitutes a defeater of the success conditions of conversational implicature. For, N may have intended to convey a different proposition ${q_3}$ , but failed to conversationally implicate it if ${q_3}$ is not required to preserve the presumption of cooperativeness, by the audience’s lights. Under this supposition, N will have utterer-implicated, but not conversationally implicated, ${q_3}$ (Saul Reference Saul2002).

To see how this might play out in a bit more detail let us focus again on the second Quantity maxim. N knows that $p$ : A said RCC and B said $\neg $ RCC constitutes an over-informative contribution. N also knows that the audience knows this. In addition, N knows that $s$ : RCC is settled by the evidence. N may also believe that the audience has (prior, independent) epistemic access to $s$ . Given these assumptions, N may therefore think that the most salient (perhaps only) explanation, in order for the over-informative contribution $p$ to count as cooperative (and not breach expectations of impartiality), is that N intended to convey ${q_3}$ : Speaker B is an incompetent, bad faith testifier in the matter of RCC.

Notice that it is perfectly possible for some portion of N’s audience to correctly recognise that N’s intention, in reporting $p$ , was to convey ${q_3}$ . The problem is that ${q_3}$ is not required in order for N’s contribution to be recognised as cooperative by the audience at large. Pat and Qat, in particular, did not recognise N’s intention to convey ${q_3}$ . This is not because they were oblivious to there being something that N intended to convey; but because they incorrectly surmised what that something was. In particular, Pat and Qat took N to have generated epistemically modest implicatures about RCC (respectively, ${q_1}$ and ${q_2}$ ). But, again, neither ${q_1}$ or ${q_2}$ are required in order to satisfy a presumption of cooperativeness by N’s lights: therefore, ${q_1}$ and ${q_2}$ also do not qualify as conversational implicatures. Rather, they are recognisable as audience-implicatures.

5.2. Whose responsibility?

Let’s quickly recap. We have seen that he-said/she-said reporting reliably triggers inferences to epistemically modest conclusions with respect to the object of the reported disagreement. Strictly speaking, these may be merely audience-implicatures: they may be generated by “the audience’s state of mind” in the absence of a corresponding intention, on the originator’s part, to convey these particular conclusions.

In the simplified scenario from Section 5.1, we saw that ${q_1}$ - ${q_2}$ may be generated even though it is not the case that N thinks that “the supposition that [N] is aware that, or thinks that, [ ${q_1}$ - ${q_2}$ ] is required to make [N’s] saying or making as if to say $p$ consistent with [the cooperative] presumption”, or “that it is within the competence of the hearer to work out” that this is the case (adapted from Saul Reference Saul2002: 242).

Descriptively speaking, N’s contribution clearly qualifies as a misleading utterance (Saul Reference Saul2012b). Acts of misleading are typically theorised as admitting of two varieties: speakers may mislead audiences accidentally, in which case they may be blameless for their act; or deliberately, in which case they may be blameworthy.Footnote ²³

Neither category seems to fit the cases under examination, however. The ‘accidental’ reading is clearly inadequate. In turn, the charge of deliberate misleading seems much too strong: it would entail ascribing to news outlets such as the BBC a level of calculated maliciousness that seems, quite simply, far-fetched. Put differently: the former reading would presuppose that, in our original scenario (Section 1), the BBC had absolutely no inkling of the possible inferential ramifications of platforming a ‘balanced’ debate on RCC; the latter would presuppose that they wilfully set up their audience to draw antecedently calculated false conclusions. The cases of interest fall somewhere between these extremes. Bearing in mind that proper and full assessment of responsibility would require saying much more than is possible here, in what follows I sketch how I see the situation.

Recall from Section 2 that by prescribing that reporters “present the views of legitimate spokespersons of the conflicting sides in any significant dispute”, balanced journalism is “presumed to [support] each citizen’s ability to reason to informed conclusions [about the object of a debate] based on all the relevant evidence” (Figdor Reference Figdor2019: 69). Crucially, this presupposes that third-party disagreements represented via he-said/she-said reporting are significant disputes. But if a third-party dispute over $p$ is significant – to the point of being deemed newsworthy by a reputable public service outlet – then it is a remarkably short step to the conclusion that $p$ is not yet settled: more evidence may need to be gathered, more discussion should take place, and judgment had best be suspended in the meantime.

Epistemically modest conclusions such as these, and such as ${q_1}$ - ${q_2}$ , are typically drawn by audiences. They are also among the stably salient implicatures of he-said/she-said reporting, we’ve seen (5.1). So, ${q_1}$ - ${q_2}$ are stably salient explanations of N’s utterance that $p$ : A said RCC, and B said $\neg $ RCC. Therefore, Pat and Qat’s inferences to unwarranted epistemically modest conclusions about RCC were foreseeable from N’s standpoint.

So, N was in a position to foresee that reporting that $p$ carried a significant risk of (epistemic) harm. Plausibly, N was also free to choose whether or not to report that $p$ .Footnote ²⁴ And so N meets the conditions for moral responsibility for action. Since the action in question is both wrong (framing $p$ as a genuine disagreement over RCC is at odds with the evidence) and harmful (unwarranted balanced framing triggers fallacious inferences), N is blameworthy for acting as it did. Depending on whether or not one takes N’s doxastic situation to have been one of “conscious advertence to the risk of wrong”, the verdict of blameworthiness may fall on the side of recklessness or of negligence (Smith Reference Smith1983, Reference Smith2011). Even in the latter case, however, a straightforward case can be made – and we made such a case earlier in this section – for regarding N’s lack of in-the-moment awareness as an instance of tracing culpable ignorance: “ignorance [that] is traceable to some past epistemic dereliction on [N’s] part” (Smith Reference Smith and Peels2016: 95).

Moral philosophers disagree as to whether wrongdoers who act from (i.e. as a result of) culpable ignorance are blameworthy for their (unwitting) wrongful act as well as for their ignorance (Peels Reference Peels2011), or only for the latter (Smith Reference Smith and Peels2016). We won’t enter this debate here. Nor do we need to, I suggest: networks such as the BBC have had multiple opportunities to reflect upon and revise their editorial conduct, and have persistently stuck to the practice of balancing all sorts of disagreements, largely irrespective of the epistemic merit of the ‘sides’ involved. This consideration suffices, I submit, to disqualify any putative lack of “conscious advertence to the risk of wrong” as an excuse for wrongdoing. In reporting that $p$ , N either recklessly or negligently misled its audience. So did BBC Cambridgeshire, in hosting a ‘balanced’ debate over the reality of climate change. So did The New York Times, in describing a congressional hearing on greenhouse gas regulation as a discussion in which “Both Sides Claim Science”.Footnote ²⁵ And for this, they are deserving of blame – and “should be ashamed”.Footnote ²⁶