An Epistemic Role for Opinion Journalism

A. Advise on Policy

One function that is often attributed to opinion journalists is to advise on policy: opinion journalists provide their readers reasons to adopt a specific policy or course of action.²⁵ At least part of the goal of such pieces seems to be to persuade the readers to take the course of action in question and thus, indirectly, to have some social or political influence. This function is thus closely related to what is called the ‘advocacy’ or ‘mobilization’ function of journalism more generally.²⁶ As Michael Schudson argues, these functions are associated with the partisan press that dominated US journalism through the nineteenth century, but were later broadly shunned under the informational model, except when explicitly confined to the opinion sections.²⁷

Editorials provide good examples of opinion journalism that aims at social and political influence. These are pieces that are standardly signed by the editorial board of the newspaper and that thus represent the views of the newspaper (or the opinion section) as an institution.²⁸ Editorial endorsements of a candidate before elections, for instance, are usually pieces that are clearly intended to influence readers’ voting behavior. They contain arguments and other rhetorical devices meant to urge a particular vote, and they often clearly state the advice or desired outcome of the piece. Some research suggests that editorial endorsements can actually influence the tone and amount of coverage the candidates receive in the other parts of the newspaper, as well as, either directly or indirectly, the readers’ votes.²⁹

Editorials, but also opinion columns, are also sometimes directly addressed to politicians and other policy makers, who are known to read these opinion pieces. Jacobs and Townsley and Schudson cite Walter Lippmann as a paradigmatic example of a columnist who functioned as a policy advisor; for instance, in his famous series of articles rejecting the Truman Doctrine in 1947.³⁰ William Safire in the 1990s or Paul Krugman until recently are other famous examples of columnists who often wrote their columns as critical dialogues with policy makers.³¹

Another function that is sometimes attributed to opinion journalism and that I take to be closely related to policy advising and advocacy is the function of providing value (perhaps especially moral) judgments.³² As Robert Audi notes, it is commonly accepted that ‘reporting should be factual, editorializing evaluative’.³³ The idea that opinion journalism has an evaluative function is also implicit in the common assumptions that all value statements are ‘opinions’ in the sense of the fact/opinion distinction and that opinion journalism captures ‘opinions’ in that sense (see Section I). Under these assumptions, value statement in a newspaper belongs in the opinion section.³⁴ Value statements in opinion journalism sometimes serve to persuade and mobilize readers, and thus, indirectly, to affect social or political change. Nicholas Kristof is a well-known (and sometimes controversial) example of a contemporary opinion columnist who frequently provides both moral evaluations and calls to action, such as in his influential series of columns on starvation, human trafficking, or homelessness. Kristof also often provides advice on philanthropic giving, including in a yearly column on philanthropic holiday gift suggestions.³⁵ He is explicit that journalism has a moral purpose and that it borders on advocacy: ‘It’s incumbent on us to […] try to meet that public desire [to improve public policy], not just to illuminate problems but also to illuminate solutions—to advise people on how they can donate effectively’.³⁶ As I see it, then, the evaluative function of opinion journalism is sometimes subsumed under the policy advisor function delineated here.

B. Facilitate Public Deliberation

A second role that is often attributed to opinion journalism is to facilitate public deliberation or debate. On a first pass, deliberation is a process for forming beliefs or making decisions that involves the consideration and weighing of different arguments.³⁷ When done well, deliberation is supposed to have an epistemically valuable outcome; in particular, the deliberator(s) should reach true or otherwise epistemically valuable views as a result of the deliberation. The mechanisms with which group deliberation outputs such epistemic goods arguably include increasing the available information, detecting factual and reasoning errors, increasing the range of possible solutions considered, and making the manipulation of information more difficult.³⁸ Proponents of deliberative theories of democracy have argued that group deliberation can also have non-epistemic benefits, such as increasing tolerance and mutual understanding, helping build community, and increasing public trust.³⁹

The media has long and often been described as having an important role in facilitating public deliberation. To take one example, the influential 1947 Hutchins Commission report, which outlines five functions of a free and socially responsible press, identifies the second function of media institutions—after providing information—as that of being a ‘forum for the exchange of comment and criticism’ and thus ‘common carriers of public discussion’.⁴⁰ Usually (and as seen in Section I), the function of facilitating public deliberation is attributed specifically to opinion journalism.⁴¹ According to Kelling and Thomas’s survey, opinion journalists themselves deem the role of facilitating civic participation and debate to be the most important to their work.⁴² There are broadly two senses in which (opinion) journalism is said to facilitate public deliberation: the first is by facilitating ‘direct’ group deliberation, and the second is through ‘mediated’ deliberation.

Opinion journalism is said to facilitate direct group deliberation in at least three ways. First, opinion journalism is supposed to provide tools that readers can use when they have small group discussions. For instance, opinion journalists provide information and arguments that can be referenced or deployed in group discussion. As Jacobs and Townsley argue, opinion journalists also provide a ‘cultural repertoire’ including ‘discourses and narratives, scripts and performances’ that can facilitate and, in some cases, enable group discussion.⁴³

Second, opinion journalism is supposed to generate small group discussions among readers. For instance, Jacobs and Townsley argue that opinion writers become representatives of a view or argument in popular culture and incite strong emotions of ‘hate’ or ‘attachment’ to get people to rally behind or against them in debating among themselves.⁴⁴

Finally, opinion journalism can also serve as a venue for small-group discussions, for instance, among political elites and experts, or between representatives of civil society and political elites.⁴⁵ Although these discussions in the written media are typically not face-to-face or conducted in real time, they can still share important features with such paradigmatic discussions. For instance, authors may respond to each other’s writings; a piece might be co-authored; or it may present the transcript of live or near-live conversation, possibly conducted online, as in ‘The Conversation’ feature in the NYT.⁴⁶ To the extent that media-facilitated deliberations are ‘deliberative’ to some degree, the participants can arguably gain whatever benefits are standardly attributed to deliberation. Importantly, however, media-facilitated deliberations have been argued to primarily benefit the audience that witnesses the deliberation. ‘Mediated deliberation’ encompasses deliberations that have non-participant observers in this sense, and it is the second type of mechanism with which (opinion) journalism is said to promote its deliberative function.

Mediated deliberation is often understood broadly to encompass anything that looks like a discussion or has some ‘degree of deliberativeness’ and that takes place in the media.⁴⁷ In particular, a mediated deliberation doesn’t need to take the form of an actual panel, forum, conversation, or exchange between opinion writers; it can refer to the ‘pool of arguments’ across an entire media system.⁴⁸ Perhaps the most distinctive feature of mediated deliberation is that it has an audience that doesn’t participate in but merely observes the quasi-deliberation. This audience is supposed to be an important beneficiary of the mediated deliberation; in particular, audience members are ‘expected to learn from the rich pool of arguments about important issues of public concern presented to them in mediated deliberation’.⁴⁹

Scholars have proposed different criteria to evaluate the degree of deliberativeness of a media environment, but they usually agree on the following: use of arguments (as opposed to mere assertions), inclusion (the range of arguments, perspectives, or evidence considered should be appropriately broad, for instance, by including all reasonable arguments or all relevant perspectives), and responsiveness (the arguments should engage with one another, for instance, by rebutting another or refining one in light of another, etc.).⁵⁰ The media has been criticized for failing each of these criteria.⁵¹ In Section III.B, I will argue for an additional criterion, aggregation, and suggest that opinion journalism often fails to meet it as well. In any case, opinion journalism seems to hold itself to some of these criteria of deliberativeness. For instance, guidelines for guest opinion writing often present use of arguments as necessary for publishing an opinion piece: ‘At its core, an Opinion guest essay provides an argument defined and substantiated with evidence’.⁵² Inclusiveness, too, is a guiding principle of many opinion sections; for instance, their stated goal is to offer the most ‘wide-ranging’ collection of arguments and ideas,⁵³ and ‘provide a diversity of voices and perspectives’.⁵⁴ In sum, mediating deliberation appears to be an important self-ascribed function of opinion journalism.⁵⁵

C. Facilitate Appropriate Uncertainty

Finally, opinion journalism is often attributed certain functions that are much more explicitly epistemic. I have in mind the following types of remarks: opinion journalism should ‘make sense of events’,⁵⁶ ‘explain the significance of the glut of events’,⁵⁷ ‘make the news more understandable […] to put it into a better frame of reference to show its significance’,⁵⁸ ‘point citizens to a deeper understanding of what [is] really important’,⁵⁹ ‘provide analysis and interpretation’, ‘provide explanation, background and context of the news’, or ‘trace causes and predict consequences of events’.⁶⁰ WaPo columnist and editor Karen Attiah puts it vividly thus:

A reporter reports—‘Man bites dog’—and that is just the facts, there is no value judgment about it—on the opinion journalism side we might say, ‘Man bites dog again, but here’s why this matters, or here’s why this doesn’t matter, or here’s why the increased rates of men biting dogs is why our society is about to collapse […]’ […] opinion journalists are able to […] build off of that news gathering, and then add judgments, framing, context, illumination, and perhaps push it forward.⁶¹

As I see it, two types of epistemic roles are implicit in these types of remarks.

The first is a broadly educational role: opinion journalists are supposed to give context or background information that is necessary for understanding the news or its significance. Jacobs and Townsley delineate a similar role of ‘the columnist as teacher’ and argue that Lippmann often embodied it, for instance, in a series of columns that educated the public about the different branches of the government.⁶² This educational function captures part of the remarks above, for instance, that opinion journalists should ‘make news more understandable’, and ‘provide background’ or ‘context’. However, this educational role doesn’t seem specific to opinion journalism. On the informational model, any journalist should include background information or context that is required to come to possess (and perhaps understand) the new information she is trying to convey: for, arguably, the ultimate goal of journalism on the informational model is for the audience to come to possess the information, or, at least, for the information to be comprehensible, and not simply for the information to be written down or archived.⁶³ Although news pieces sometimes remind readers of the most important background facts required to understand some new piece of information (such as what some regulation says or how some event unfolded in the past), some have criticized the media for too often failing to convey enough relevant background information in reporting, for instance, about the frequencies of extreme events that are reported.⁶⁴ Since certain background information is required for many news stories (such as information about the functions of government branches), it would make sense for newspapers or news sites to dedicate a section to it.⁶⁵ But opinion sections aren’t standardly conceived as primarily educational in this specific sense: usually, to say that something is an ‘opinion’ suggests that there also are other opinions on the matter, and that at least some of the content discussed isn’t ‘established fact’ or mere ‘background information’.

I therefore think that the epistemic role specific to opinion journalism in the remarks above is to ‘go beyond the news’ and ‘push it forward’, in the sense of deriving some implications of the news, or attempting to ‘predict’, ‘interpret’, ‘explain’, ‘frame’, or ‘evaluate’ information given in news reporting. As I see it, this was also central to Lippmann’s understanding of the function of opinion journalism, which he elaborates in an address to the National Press Club:

[W]e correspondents perform an essential service. In some field of interest, we make it our business to find out what is going on under the surface and beyond the horizon, to infer, to deduce, to imagine, and to guess what is going on inside, what this meant yesterday, and what it could mean tomorrow.⁶⁶

I propose to conceptualize this epistemic role more precisely as the role to facilitate appropriate uncertainty. Here is what I mean. A vast number of questions of public interest are (epistemically) ‘unsettled’ in the sense that the total available evidence and arguments don’t strongly support any particular answer. The evidence and arguments here are high-quality ones typically gathered and developed by relevant experts, and they are ‘available’ in the sense that at least the relevant experts are aware of it.⁶⁷ There are important unsettled scientific or broadly empirical questions (What are the causes of Alzheimer’s disease? Who will win the election? What are the drivers of inflation, and which policies most effectively reduce it?). There are also important unsettled questions about social, political, economic, moral, or philosophical matters (How should foundational political documents or constitutional provisions be interpreted? What principles should guide the allocation of public resources, such as healthcare or education funding? What role does social media play in shaping voter perceptions of political candidates?). Some of these questions will be settled by more evidence. Others might require more careful analysis and examination of the available evidence to draw conclusions from it. Some are extremely complicated, and might only be capable of being ‘conditionally settled’, for instance, given more precisely defined terms or some set of assumptions about preferences, values, or moral principles.⁶⁸ Note that if a question is unsettled in the sense I defined here, this also means, typically, that there is no expert consensus on the question, i.e., it isn’t the case that a majority of relevant experts are confident in a particular answer.⁶⁹ On the other hand, whether there is consensus on an answer in the general population doesn’t typically track whether a question is settled in this sense.⁷⁰

What are epistemically valuable attitudes to have toward (the answers to) unsettled questions? While this question is entangled with some longstanding debates in epistemology, I set these complexities aside as my subsequent arguments are by and large unaffected by them.⁷¹ I only make some broadly accepted and commonsensical assumptions. One such assumption is that it would be epistemically valuable to base one’s attitudes on the total available evidence, and, in particular, to be only as confident in a particular answer to the unsettled question as the total available evidence warrants. One way to put this more precisely is in terms of credences, where a credence is a degree of belief or confidence; credences are usually given a value in [0, 1], where credence 1 in a proposition represents that the agent has maximal confidence that the proposition is true, and credence 0 represents that the agent has maximal confidence that it is false. One can then say that there is epistemic value in ‘proportioning’ one’s credences to the total available evidence concerning an unsettled question. Consider a toy example. Say that it is unsettled whether Alice has a specific gene g, and that the only available evidence is that 6 out of 10 people have g. Then, a credence that Alice has g that is approximately 0.6 is proportional to the evidence and, even more obviously, credences in this proposition that are approximately 0 or approximately 1 aren’t proportional. The idea of proportioning one’s beliefs to the evidence goes back to Locke and Hume.⁷² It is compatible with both internalist and externalist accounts in epistemology.⁷³ Notice that the total available evidence in question here will most often exceed the evidence available to a given reader of opinion pieces. Thus, my claim isn’t that a subject should proportion their credences to the total available evidence, as this might be beyond their capabilities.⁷⁴ Rather, my claim is that possessing this evidence and then proportioning their credences to it is (or would be) epistemically valuable to them. The idea that it is valuable to proportion one’s credences to the total available evidence, rather than to merely one’s own current evidence, is usually taken to go back to Carnap.⁷⁵ This idea is often supported by arguments from expected (epistemic) utility.⁷⁶ It is also in line with the idea that it is epistemically valuable to seek out expert testimony or to gather new evidence.⁷⁷ To sum up, and again leaving many subtleties aside, my main claim here is that it is epistemically valuable to proportion one’s attitudes in the answers to an unsettled question to the total available evidence—in slogan form, it is epistemically valuable to have appropriate uncertainty in the answers to an unsettled question.

The epistemic function for opinion journalism that I am delineating here, then, is to help readers form such appropriate uncertainty toward unsettled questions of public importance, i.e., to facilitate appropriate uncertainty. Opinion journalists can do (and, in fact do) a number of things that are instrumental for facilitating appropriate uncertainty.⁷⁸ For instance, opinion journalists can present facts or information (including ones reported in the main sections of the news) and explain to what extent these support the various answers to an unsettled question.⁷⁹ Opinion journalists can make their readers aware of a potential answer to an unsettled question and help them distribute their credences appropriately.⁸⁰ They can make explicit various clarifications or assumptions needed to settle a question, and help readers form appropriate attitudes in the relevant conditional answers. Opinion journalists can also more directly report on the total distribution of evidence and arguments discussed by the relevant experts and explain what overall uncertainty this warrants.⁸¹

As I see it, attributing the function of facilitating appropriate uncertainty captures many of the remarks from practitioners and scholars above. In particular, as in my examples above, questions of public importance about ‘prediction’, ‘interpretation’, ‘explanation’, ‘framing’, or ‘evaluation’ are usually unsettled questions in my sense here. Lippmann gives further examples of important unsettled questions the columnist is tasked with investigating. As he says, their answers ‘often exist far away and out of sight of any newspaperman’, ‘in places that the reporter cannot visit’, ‘may lie in the past’, ‘may lie inside the head of a public man’, or ‘in the moving tides of mass opinion, for example about the coming elections’.⁸² For Lippmann, the columnist’s jobs is to ‘deduce’, ‘calculate’, and ‘appraise’ such facts, ‘proposing theories or hypotheses which are then tested by trial and error’, given that ‘we know something but not everything, and not nearly enough’ concerning such important questions.⁸³ Many opinion pieces concern unsettled questions, and many important unsettled questions of public importance get discussed in opinion pieces (though some are discussed in more detail in more specialized venues, such as of science journalism, and some are also sometimes discussed in the reporting sections). Thus, whether or not opinion journalism in practice often succeeds in facilitating appropriate uncertainty (I suggest not in Section III), the role of facilitating appropriate uncertainty does seem to make sense of some self-conception of the aims of opinion journalism.

Finally, I claim that facilitating appropriate uncertainty is a normative function of opinion journalism, given the informational model. On the informational model, journalism is supposed to provide the public with important information; in particular, information that we need to make good decisions in our lives and community. But some of the information we need concerns unsettled questions. All the examples above and in Section I are of unsettled questions that are important for many of us to have appropriate attitudes about. Many of us need to base our decisions and actions on such questions—actions such as which policies or politicians to support or trust, how to avoid certain diseases, how to prepare for the future, etc. Many unsettled questions are also simply interesting. Thus, if journalism’s epistemic goal in general is to help us form epistemically valuable attitudes toward those aspects of the world that are important for or interesting to us, then this should also involve forming epistemically valuable attitudes toward unsettled questions (whatever exactly these attitudes are). On this view, facilitating appropriate uncertainty is simply part of the general epistemic goal of journalism, according to the informational model.

Next, I consider the relationships between the three functions that I delineated here in Section II.

A. Facilitating Appropriate Uncertainty and Policy Advising

Consider, first, the relationship between facilitating appropriate uncertainty and policy advising. Assume that the primary aim of policy advising in opinion journalism is to influence the actions of readers (whether policy makers or members of the public) by influencing their viewpoints. On this assumption, it is clear that policy advising can conflict with facilitating appropriate uncertainty. This follows simply because the attitudes that are most appropriate given the total available evidence and arguments might not be the ones that are most likely to result in a certain action. To take a simple example, if people act at least roughly based on their expected utilities, then it could be that having a slightly higher than appropriate credence that some action will be beneficial will make it more likely that one decides to do it.⁸⁴ This can even happen in cases where the policies or actions in question actually promote public good.⁸⁵ There can thus be tradeoffs between the aims of affecting policy change, swaying votes, etc., and the aim of facilitating appropriate uncertainty. Many opinion pieces, perhaps most famously written by politicians, have been criticized precisely for exaggerating or downplaying facts to get readers to support or implement certain policies.⁸⁶

On an alternative (and more plausible) understanding, policy advising in opinion journalism, at least when done well, should only aim to influence readers’ actions in ways that are compatible with appropriate uncertainty. This is more in line with professional and official norms for medical advice, which dictate that all relevant evidence and uncertainties must be taken into account and that the strength of the advice should be proportional to the level of certainty in the evidence.⁸⁷ Some have defended the point of editorial endorsements by describing their function in a similar manner. On this view, endorsements are supposed to provide the information relevant to making a voting decision that the board gets in part through ‘endorsement interviews’.⁸⁸ Some of it includes information about relevant unsettled questions, such as how the candidates ‘would perform in office and whether they have the steadfast ability to execute the duties of the highest office in the land’.⁸⁹ The editorial recommendation is then a kind of aggregate summary of the total relevant evidence and arguments concerning the candidates. On this understanding, editorial endorsements and policy advice more generally would fit within the informational model.

This understanding has to be unpacked further, however. One crucial difference between policy advice in medicine and in journalism is that medical advice is supposed to be ‘patient-centered’, i.e., the relevant costs and benefits considered are the patient’s health outcomes. In contrast, the opinion journalist who gives policy advice about voting, politics, or even the economy might be (and often is) doing so from a background set of values or interests that aren’t universal or the readers’ own. Indeed, opinion journalism is often understood to proceed from a background set of values or principles, and it is at least implicitly understood that (some of) these values or principles may not be universally or even broadly shared. As Chicago Tribune Opinion Page Editor John McCormick explains: ‘Swaying votes is only one reason for endorsing, and arguably not the most important. Every few years, endorsements […] explain to the world what that publication is, what it advocates, how it thinks, what principles it holds dear’.⁹⁰ Besides providing relevant information, then, editorial endorsements outline the news organization’s (or its editorial board’s) values, and partly bases its recommendation on these. This relates to the evaluative function of opinion journalism discussed in Section II.A: policy advice and advocacy is often done relative to a background set of (moral) values. Some of these are assumed to be (and plausibly are) universal, such as the value of human well-being. But others might not be; they might be values a particular news organization ‘holds dear’, but another doesn’t.

For the same reasons given above, policy advice that proceeds from a background set of values should stay consistent with facilitating appropriate uncertainty, for instance, about questions relevant to which action best achieves or maximizes the background values. But an important issue here is whether the background values themselves are up for epistemic evaluation. Although the fact/opinion distinction, maybe together with the news/opinion distinction, often seem to presuppose that value statements aren’t truth-evaluable, this is a widely discussed and contested metaethical thesis.⁹¹ Thus, if an opinion journalist conveys of some epistemically dubious or uncertain value statement (assuming these exist) that it is either epistemically well-established given the total evidence and arguments, or not apt for epistemic evaluation to begin with, then policy advising can once again conflict with facilitating appropriate uncertainty—only this time concerning evaluative questions. This isn’t to say that the informational model precludes opinion journalism from ever assuming epistemically uncertain value statements (again, assuming they exist) and giving advice on their basis: in particular, opinion journalists could give policy advice addressed specifically to those who share the values in question, without conveying that these values are universally held, epistemically well-established, or not up for debate (if they aren’t). But the worry is that opinion journalism doesn’t always clearly convey these important distinctions. For instance, some editorial boards explicitly state their background values, but then go on to include assumptions that arguably aren’t mere value statements but depend on difficult and unsettled factual questions. Take, for instance, the ‘editorial philosophy’ of the WSJ Opinion Section: ‘free markets, free people’.⁹² On a natural understanding, this statement entails answers to complicated and unsettled factual questions about the economy, the consequences of market regulations, taxation, etc. It thus doesn’t seem ‘amazing’, as former WSJ columnist and editor George Melloan states, that ‘those words today are controversial, assailed by a sizable crowd of intellectuals and politicians who, despite American accomplishments, doubt the efficacy of market capitalism’.⁹³ A parallel point can be made about the NYT’s Editorial Board that, among other things, ‘has long supported a liberal order of nations in which freedom and progress advance through democracy and capitalism’.⁹⁴

In sum, giving policy advice from background principles that either involve uncertain factual claims, or that are themselves epistemically uncertain value statements, can conflict with facilitating appropriate uncertainty. This is because it can mislead the audience into thinking that these statements aren’t uncertain but either well-established or not even up for epistemic evaluation. Facilitating appropriate uncertainty requires being clear about which background assumptions are epistemically uncertain, which are epistemically well-established (including about values or morality), and which aren’t even apt for epistemic evaluation—in each case, it also requires being clear whether these assumptions are universally accepted or not. This, of course, is a difficult task. In Section IV, I suggest that transparency about these categories could nonetheless be a useful tool to facilitate appropriate uncertainty.

B. Facilitating Appropriate Uncertainty and Facilitating Public Deliberation

Consider, finally, the relationship between facilitating appropriate uncertainty and facilitating public deliberation. My first claim here is that the ultimate epistemic goal of facilitating public deliberation in opinion journalism should be to facilitate appropriate uncertainty; in other words, opinion journalism should help facilitate (direct or mediated) public deliberation in order to get the general public to have epistemically valuable attitudes toward unsettled questions (in the sense of Section II.C). Consider, first, its implication that the questions about which opinion journalism is supposed to facilitate deliberation are unsettled. There is certainly epistemic value in discussing settled questions, especially ones over which there is no public consensus, for instance, to understand why there is no public consensus.⁹⁵ But, just as the educational function discussed in Section II.C, informing the public about why people hold views in conflict with the total evidence doesn’t seem specific to opinion journalism (unless that question itself is unsettled). Moreover, and as discussed in Section III.A, it would conflict with facilitating appropriate uncertainty to conflate these categories and mislead the public into thinking that, for instance, a question that is epistemically settled isn’t, or vice versa.

Assuming that deliberations facilitated by opinion journalism concern unsettled questions, it does seem plausible that their ultimate epistemic aim is to help the audience form epistemically valuable attitudes toward these questions. As we saw in Section II.B, after (mediated or direct) deliberation, the audience is supposed to learn about the total pool of evidence and (presumably) form attitudes that are appropriate to it. One difference between pursuing the epistemic aim of facilitating public deliberation and the aim of facilitating appropriate uncertainty might be that, for the former, an opinion journalist should also be concerned with advancing the debate and thus with adding new (and, if possible, even conclusive) evidence or arguments to the pool of total available evidence. This is in line with the idea that the ultimate goal of public deliberation is to figure out the truth and that journalism can contribute to this aim; see, for instance: ‘When journalism succeeds in illuminating questions and debates […] it can help society move conversations about these issues toward resolution’.⁹⁶ In the following I argue, first, that opinion journalism often isn’t the right setting to advance or settle debates of public importance, and, second, that common ways of adding (either old or new) arguments or evidence to a public debate in opinion journalism conflict with facilitating appropriate uncertainty in the general public.

There is an important difference between media-facilitated deliberations and deliberation in paradigm cases of interest to epistemologists. Deliberation that can make progress on unsettled questions is generally conducted among relevant experts. Consider any example of unsettled questions above in Sections I or II.C. Predicting, explaining, and interpreting facts about the economy, science, politics, or society is usually done best among people with relevant expertise, training, and evidence. The ‘pool of arguments’ in such expert deliberations standardly consists of peer-reviewed articles, books, or conferences, in academia or in the relevant industries. Some opinion pieces are written by experts in their fields, but these usually don’t introduce new arguments unknown to other experts; instead, they serve to popularize existing arguments, making them accessible to a general audience. There are, of course, exceptions. In some cases, opinion writers should be considered to be the relevant experts, particularly when interpreting specific political events. Columnists who have spent years covering politics often have deep experience analyzing unfolding events and may have access to insider information that is unavailable to experts in related fields, such as political science.⁹⁷ Columnists might also weigh in on issues that haven’t been discussed by any other relevant expert because these issues are too recent or local. Moreover, some opinion pieces have directly influenced scholarly discussions.⁹⁸ All that being said, opinion pieces don’t often significantly advance the total pool of evidence and arguments about important unsettled questions. This is also because opinion pieces are by their nature always in part addressed to a general audience, and thereby can’t have the detail, complexity, or length to contribute novel arguments to existing debates. In sum, opinion journalism isn’t—and shouldn’t be considered as—the place where significant epistemic progress is made on difficult unsettled questions. But it is (and should be) always in part addressed to a non-expert audience that needs information and appropriate uncertainty. This has important implications for how facilitating public debate can conflict with facilitating appropriate uncertainty.

Consider an opinion piece that aims to facilitate public deliberation by adding or repeating an argument in support of a position concerning an unsettled question. Presenting a new argument for a position or presenting new evidence are standard ways of contributing to debate among experts, for instance, in academia. But in media-facilitated deliberations, this can conflict with facilitating appropriate credences. This is because of the difference just discussed: In expert deliberations, the intended audience of a piece containing an argument consists of the relevant experts who can be assumed to have a good sense of the relevant total evidence and uncertainty. Participants in expert deliberations keep track of the total pool of evidence and arguments by studying the relevant literature. But general audiences can’t be assumed to have a good sense of the total available evidence bearing on a question. They needn’t have read all other relevant opinion pieces on the matter (or even any other opinion piece) or have any relevant training to appreciate the uncertainty of the questions discussed. Thus, the general public might be misled in reading an opinion piece if it only gives an incomplete part of the pool of arguments relevant to a question. The public might think, for instance, that the arguments in the piece are the only, or perhaps the strongest, arguments in favor of a position, and thereby form attitudes that are more extreme than appropriate.

Consider an example. In a series of articles in 2021, Krugman gave policy advice in support of Joseph R. Biden’s economic rescue plan, and, in doing so, he argued that the plan wouldn’t raise inflation to a dangerous degree. In these articles Krugman doesn’t mention the uncertainty over whether the rescue plan could be dangerously inflationary. The argument is made in strong and certain terms. Krugman addresses some potential objections (thus partially meeting the criterion of responsiveness from Section II.B), but only to dismiss them quickly and conclude boldly; see, for instance:

We’ll be coming out of the pandemic with inflation still below the Fed’s target, and it would do little harm to overshoot that target and run the economy hot, leading to a bit of excess inflation—and a bit is all that would happen, because inflation responds slowly to economic conditions. If the boom gets big enough and goes on long enough that inflation actually starts to look like a concern, the Fed can always rein it in by modestly raising interest rates. […] So please, don’t nitpick this plan.⁹⁹

From reading these pieces, a member of the general public might infer that these are the only relevant objections to these arguments, and thus that there is no uncertainty in the matter. An expert audience, on the other hand, would know that the arguments presented are only a proper subset of the total available evidence, and that the latter doesn’t warrant extreme credence in the claim that the rescue plan won’t be dangerously inflationary.

Interestingly, in his contribution to the series ‘I was wrong about’, where NYT columnists discuss mistakes they made in past writings, Krugman mentions these pieces and notes that the issue he was discussing in them was very complicated:

Some warned that the package would be dangerously inflationary; others were fairly relaxed. I was Team Relaxed. As it turned out, of course, that was a very bad call. But what, exactly, did I get wrong? Both the initial debate and the way things have played out were more complicated than I suspect most people realize.¹⁰⁰

My point here is that the intended non-expert audience of an opinion piece shouldn’t be expected to know how complicated or uncertain a question is. Part of the problem in these articles was that Krugman was overconfident (he acknowledges that ‘the whole experience has been a lesson in humility’).¹⁰¹ But another part of the problem is that Krugman only provided arguments for the claim that the rescue plan won’t be dangerously inflationary, without conveying what the total available evidence supports or how unsettled the question is.

Here is another way to put this point. Concerning opinion journalism, Sulzberger says that ‘a great opinion section absolutely advances our core journalistic mission of helping people understand the world […]. It just does so in aggregate, rather than at story level’.¹⁰² The problem I am pointing out is that the intended audiences of opinion pieces might never see the ‘aggregate level’, and thereby fail to adopt the attitudes that might be appropriate given the aggregate of the total available evidence.¹⁰³ The problem in practice is probably even worse: not only do audiences fail to see the total pool of evidence, they only have biased ‘selective exposure’ to opinion pieces.¹⁰⁴ Therefore, I suggest that aggregation is another measure of the deliberativeness of a media environment, insofar as deliberation is also supposed to facilitate appropriate credences in the witness to the mediated deliberation: an opinion piece should ‘aggregate’ in the sense that it should also convey that the question discussed is unsettled (and, if possible, how unsettled it is) given the ‘aggregate level’ of the total available evidence, alongside the arguments or evidence that it presents. The value of aggregation is also supported by the idea that it is epistemically valuable to proportion one’s attitudes to the total available evidence, as discussed in Section II.C. In Section IV, I will briefly discuss the difficult question of what opinion journalists should actually say or write to aggregate in this sense.

Let me observe one last conflict that can arise between facilitating appropriate uncertainty and generating direct deliberation in the sense discussed in Section II.B. This is a simple observation about Jacobs and Townsley’s proposal discussed there and others like it: Getting people to debate by making them hate or love a columnist might require oversimplifying one’s positions or putting one’s arguments more strongly than is warranted by the total available evidence. This is an instance of my point in Section III.A: getting people to act—such as getting them to debate in small groups—might require getting them to have more extreme credences than appropriate. Scholars such as Jeffrey M. Berry and Sarah Sobieraj have pointed out how the ‘outrage industry’ in the opinion media is driven by economic incentives and negatively impacts democratic deliberation.¹⁰⁵ The point here is also continuous with those analyses: the outrage industry—which arguably includes ‘compelling narratives’ and ‘characters’ that make passionate pleas in the face of moral outrage¹⁰⁶—can, and probably often does, also conflict with facilitating appropriate uncertainty, even when it is done to facilitate public deliberation.