Expert Authority and Objectivity: Why the Public is Not Equipped to Adjudicate Expert Disagreement

Introduction

Alberto Giubilini, Rachel Gur-Arie, and Euzebiusz Jamrozik¹ (henceforth GGJ) argue that the non-expert public’s appraisal of someone as an expert is necessary for whether they have expert authority. They contend that expertise is contingent on whether someone possesses some “set of epistemic features that warrant trusting” them “as an expert,”² and whether someone has these features is a function of whether the public believes that person has authority to speak on those matters. The need for the public’s perspective on expert authority is derived primarily from the fact that the public is vested in domains that affect their interests and, so they are also vested in whether putative experts satisfy the responsibility to fulfill those interests. This suggests to GGJ that whatever criteria experts normally use to train and identify other experts—what GGJ call “internal legitimization,” for example, expertise conferred through education, dissertation defenses, certification, licensure, peer review, etc.—is insufficient for determining whether someone is an expert in domains of public interest, and by implication, whether that person speaks with expert authority.

GGJ further argue that people who want to be attributed expert authority by the non-expert public can improve their trustworthiness among the non-expert public if, as part of their communication with the public, putative experts acknowledge uncertainty regarding their knowledge claims and admit that disagreement among putative experts is common when there are uncertainties or differences in value judgment.

Here, I begin by clarifying what I understand to be the domains of expertise GGJ are addressing as well as their conceptions of expertise and expert authority. Then I sketch the basic structure of their argument. In the remainder of the paper, I offer three objections. First, I argue that their concerns about two common accounts of expertise—which I identify as veritism and the Studies of Expertise and Experience (SEE) account of Collins and Evans³—and their implications for expert authority are too weak to undermine the viability of these approaches. Second, I accept that some form of trust is necessary for establishing expert authority but argue that it is located with experts rather than the lay public. Third, I argue that, even if non-experts did stand in an epistemic position sufficient to evaluate expert authority, transparency would not necessarily or obviously facilitate trust in experts. I close by acknowledging the precarious epistemic position of the non-expert, but I point to some alternative accounts of expert authority that address some of GGJ’s concerns while also preserving the objectivity of epistemic authority.

Two Points of Clarification

Before sketching and critiquing their argument, two points of clarification are in order. First, GGJ stipulate that they are concerned only with “expertise about matters that affect the interests of the general public or relevant portions of it.”⁴ As such, I take it they are not interested in accounting for the kind of expertise in niche scientific domains, such as astrobiology or geomorphology—domains that do not directly impact public interest, if at all—or domains with only aesthetic interest, such as Olympic competition and conceptual art—domains which, while accessible to the public, exist primarily for the public’s passive entertainment. Though I worry that distinguishing types of expertise according to whether they are “relevant to specific goals in which we have some stakes”⁵ risks damaging the concept of expertise,⁶ I will, nonetheless, follow GGJ and restrict my comments to domains that seem to meet this requirement, and which, I will presume, include domains such as public health, climatology, medicine, and law.

A second point of clarification regards their definitions of expert and expert authority. In several places, GGJ define expert authority not in terms of objective epistemic traits of trustworthiness, but in terms of whether someone is (descriptively) trusted as an expert by some group. They infer from this public trust that the person should be trusted. For example, as noted above, they write that “trust is constitutive at least of a certain type of expertise” such that “[o]ne cannot be that type of expert without being trusted as an expert.”⁷ Later they say that “it is impossible to separate expertise from being trusted as having expert authority.”⁸ Still later (and here is where they introduce the normative “should”), they say: “A lack of trust in public health experts can be particularly serious insofar as public health experts need to enjoy what Matthew Bennett calls ‘recommendation trust.’ In other words, trust that ‘I should do something because they have told me I should.’”⁹

On a straightforward reading of these claims, public trust in someone as an expert is not simply “valuable” (as GGJ say in their abstract), but logically necessary for expert authority, and they say as much: “The very notion of expertise and of expert authority would be undermined by the erosion of trust.”¹⁰ However, this reading of trust as a necessary condition of expertise renders the quote from page 16 conceptually problematic: “A lack of trust in public health experts” would not, on their account, be (strictly speaking) possible. This is because if the public doesn’t trust someone as an expert, then they are not an expert in whom the public lacks trust, and therefore, are not someone the public “should” come to trust. If that is correct, there is nothing “particularly serious” about such a lack of trust because, without trust, the putative public health official is no more entitled to trust than any other non-expert. This straightforward reading also leads to unanswerable questions such as: How many people among the lay public must trust someone as an expert in order for them to be an expert and thus have epistemic authority? And, if large portions of the population trust, for example, Creationist scholars and Darwinist scholars equally, would this mean that both are authoritative despite holding contradictory views?

A more charitable reading of “trusted as” can be rendered in light of their claim on page 13, that “expertise is the possession of any set of epistemic features that warrant trusting someone as an expert” (italics mine), and that expertise is “not separable from an assessment of whether a certain relationship of epistemic trust can be justified between that individual and the relevant community.”¹¹ So, for the purposes of this paper, I will stipulate that an expert has expert authority when the lay public perceives reasons to believe the expert should be trusted.

This reading is consistent with their descriptions of trusting someone as an expert because that person is “considered to possess knowledge or skills that are relevant to specific goals in which we have some stakes”¹² and that “trust is a form of reliance on others’ skills, knowledge, and moral traits.”¹³ Since knowledge, skills, and moral traits are aptitudes that exist independently of public perception, then what is relative is simply the non-expert public’s perception of whether someone has these aptitudes. This reading is also more consistent with what I take to be their overall intent, which is to increase public trust in people who in fact stand in a better epistemic position than the general public on issues such as vaccine safety. In virtue of their epistemic position, some people are (objectively) experts and, therefore, should be taken seriously by the public.

This reinterpretation renders “expertise” objective by making it contingent on features inherent to a domain while allowing “expert authority” to be contingent, to some extent, on non-experts’ perception of someone as authoritative without opening the whole account to critiques of purely subjectivist accounts of expertise, as discussed in Collins¹⁴ and Nichols.¹⁵

But, of course, authority and expertise come apart in important ways, and I agree with GGJ that “[e]xperts need to be trusted not just because they are credible… They also need to be trusted because they are trustworthy….”¹⁶: an expert must use their expertise responsibly, that is, within the scope of their domain and not fraudulently or manipulatively. Someone who misuses their competence may continue to meet the criteria for expertise in a given domain but fail to meet epistemic conditions sufficient for being an authority in that domain (such as Olympic sports physician Larry Nassar). And it is reasonable for a non-expert to be uncertain about whether an expert’s public behavior should be interpreted as evidence for or against their authority (see the recent controversy over podcast star Andrew Huberman¹⁷).¹⁸ For the purposes of my response to GGJ, I will focus on those places where expertise and expert authority overlap—that is, where extenuating circumstances are such that someone possesses expertise and they are acting responsibly so the person is, prima facie, an epistemic authority—and I will aim my critique at the concept of expert authority.

Putting my cards on the table, I agree that a member of the non-expert public may be sufficiently ignorant of a domain of expertise such that, even if someone meets the relevant internal legitimization criteria for expertise, the layperson may withhold the attribution of expertise when it is deserved. This would certainly mean the expert doesn’t have authority for that person in the sense of what the authors cite as “recommendation trust.”¹⁹ Where I diverge is that I do not think recommendation trust is either necessary or sufficient for expertise or expert authority. I think there is an objective sense of expert authority such that, once the requisite evidence is made available and accessible to the non-expert public, and the public has accessed it, it would be epistemically irresponsible for them not to trust the expert. Similarly, an internally legitimized expert may choose to use his expertise fraudulently and deceitfully, such that all non-experts might perceive him to be authoritative when he is not. Such an expert may have authority for that layperson insofar as the layperson acts (however mistakenly) on that expert’s testimony. Nevertheless, once the evidence of fraud is brought to light, the layperson should reject the belief that the expert’s advice or testimony is authoritative, even if they continue to perceive the expert as, on the whole, humble, honest, and transparent. The central problem with GGJ’s account is that they put the entire question of whether an internally legitimized expert is authoritative in the hands of the non-expert public who may accept or reject expert authority. They do not make room for normative expert authority outside of public trust in domains that affect public interest.

One final clarification is in order regarding having authority and exercising authority. It is not unreasonable to think that if experts cannot exercise authority, they cannot rightly be said to have authority.²⁰ If one interprets GGJ as saying that public trust in experts is necessary for experts to exercise their authority, then, by contraposition, in withholding trust, laypersons effectively denude experts of authority. But this reading would conflate epistemic authority with political or “administrative” authority.²¹ Scholars like Robert Paul Wolff²² interpret authority as inherently implying power over someone else, as when a ruler has authority over their subjects. If a ruler cannot exercise power, then it makes little sense to say they have power. But with respect to expertise, the relevant notion of “authority over” is epistemic: it refers to what non-authorities should believe, not whether authorities are able to compel belief. For example, if someone, A, has a better view of something than someone else, B, and there is a question as to whose perspective is better regarding that view, the objective answer is A’s. A has more epistemic authority than B to speak to those issues. Note that this is true even if B doesn’t accept A’s authority (and also true if no one accepts A’s authority). A doesn’t need B’s (nor anyone’s) approval, trust, or agreement to stand in a better epistemic position than B. Therefore, even if B doesn’t trust A, A can still exercise epistemic authority simply by testifying to what they see. The unfortunate prophet, Cassandra, was no less correct for not being believed.

This does not, of course, mean that Cassandra’s contemporaries were epistemically blameworthy for not believing her. If your neighbor is a physician, but you have no reason to know that, you would not perceive their casual medical claims made at, say, a neighborhood barbeque to be authoritative, even if they are.²³ This is where “recommendation trust” becomes important. The question at issue in my response to GGJ is whether there is a sense of expert authority independent of the lay public’s perception.

The Central Argument

The motivating context for GGJ’s argument is an inconsistency in recommendations regarding COVID-19 vaccines for children 12 to 15 years old. The UK’s Joint Committee on Vaccination and Immunisation (JCVI) recommended against the vaccine due to “considerable uncertainty” regarding its safety.²⁴ On the other side of the Atlantic, the US Centers for Disease Control and Prevention (CDC) did recommend the vaccine, claiming the benefits outweighed the risks. And a few days after the JVCI’s report, the UK government authorized the vaccine for use in 12-15-year-olds, citing a broader range of risks (including social risks to mental health and school disruption due to isolation requirements) of not authorizing the vaccine.

It is worth noting there are two kinds of disagreement involved in this case. The first is inter-disciplinary disagreement among specialists in a field over the facts in a case because a question is new and evidence is evolving.²⁵ The second is extra-disciplinary disagreement among specialists in different fields working on the same problem who disagree because the aims of their domains differ. An expert climate scientist may be right to claim that anthropogenic global warming is happening but wrong to claim support for a public policy that, according to public policy experts, would not successfully mitigate climate change. The disagreement between the JCVI and the CDC is inter-disciplinary. The disagreement between the JCVI and the UK government is extra-disciplinary.

Epistemically, this matters because, in cases of the former, non-experts should suspend judgment regarding the claim at hand. This is not to say that the conflicting experts lack authority in general, but simply that there is no settled expert opinion regarding the safety of vaccines.²⁶ To be sure, there may be no authoritative view regarding the specific question of the vaccine’s safety in children—and I will discuss different accounts of authority below. The point is that, until there is settled expert opinion, non-experts must admit that the expert authorities cancel one another out for the moment, and GGJ admit as much: “one might conclude that the two conflicting expert views would, so to speak, cancel each other out.”²⁷ To see why I think the experts retain general authority, note that it would not be appropriate for a non-expert, when faced with conflicting expert testimony, to then turn to a group of non-experts—say a vocal anti-vaxxer group—to “break the tie.” Only genuine experts can cancel-out other experts on a question intrinsic to their domain.

Extra-disciplinary disagreement is less epistemically problematic, though no less frustrating for non-experts. The two groups of experts are commenting on different claims. Consider a common medical situation: A physician recommends a treatment that, while likely to be successful, will take a great toll on the patient (for example, a surgery that would reduce chronic pain but that would prevent the patient from seeing their grandchildren for six months due to travel limitations). It is obvious why the patient might value seeing their grandchildren more than having less chronic pain, especially during the child’s formative development or if the parents need support during a difficult financial period. Neither the surgeon’s expertise nor authority regarding medical benefit is in question; there are simply other matters of importance to be balanced against medical benefit, and which the patient is in a better position to assess. In this case, the JCVI is concerned with medical risks and benefits, while the UK government is concerned with how those benefits trade off with other matters of importance—matters that lay within their distinct purview. I take it that the heart of GGJ’s claim about the importance of trust for expert authority lies with the inter-disciplinary disagreement among experts.

GGJ begin their argument by defining expertise in contrast to two common views of expertise: the veritistic—or, truth-based—view of scholars like Alvin Goldman²⁸ and the Studies of Expertise and Experience (SEE) view of Harry Collins and Robert Evans.²⁹ Veritistic accounts are inadequate, according to GGJ, because there is no impartial stance from which to assess the truth of an expert’s claims. They acknowledge that an expert’s claims should be “significantly” more likely to be true than a non-expert’s and that “significant disagreement” among experts creates a problem for expert authority,³⁰ but there is no objective way to determine that sort of “significance.” A common proposal in the literature is for non-experts to vet expert testimony in terms of “the extent to which other experts agree,” which GGJ call “a kind of expert majority rule.”³¹ But GGJ reject that proposal, noting that “‘majority rules’ is a fragile basis for expert authority, as majorities can be formed out of bad incentives.”³² Polarized, biased, financial, and reputational motives can lead experts to collude over standards in a domain. They point out that the history of science offers numerous examples where the minority view was proved correct, thus some experts might make bad recommendations based on consensus views.³³

The SEE account of Collins and Evans³⁴ offers a slightly different approach to establishing objective expertise, namely, as “the condition where one can contribute to domain-specific conversations with one’s peers, after one has learned to master the language and methodologies specific to a certain domain.”³⁵ GGJ acknowledge that knowledge “could be built this way and it often will,”³⁶ but they are concerned that, even if that were a necessary condition for expertise, it could not be sufficient because it doesn’t account for the public’s interest in the type of expertise under consideration. “At a minimum,” they write, “there must also be some level of societal relevance of a certain field which warrants attribution of expertise.”³⁷ They explain that, while linguistic immersion of the sort Collins and Evans describe can be effective when experts “investigate a new phenomenon,” expertise “encompasses more aspects than merely meeting epistemic conditions…. It includes, in particular, the capacity to recognize relevant aspects of decision-making processes, such as how individuals’ and society’s interests will be affected…”³⁸

What criteria do GGJ suggest instead of objective criteria derived from the domain itself (veritism) or by the way that expertise is acquired in a domain (SEE)? They claim that someone’s claim is reliable to the degree that the non-expert public believes it is reliable. They explain: “Indeed, it is the fact that someone we believe to be an expert makes a certain claim that provides good reason to believe that claim is true.”³⁹ Objective reliability in terms of “knowing the truth typically contributes to [trusting someone as an expert], but it is not a necessary condition for it.”⁴⁰

From this account of expertise, they conclude that expert authority is also, by implication, contingent on public perception. Since the authority of experts is not necessarily (and certainly not always) derived from truth (given the inability to determine when a putative expert is “significantly likely” to be correct) and given that “majority rules” is too fragile, the most plausible candidate source for expert authority, according to GGJ, is the public perception that someone is reliable.⁴¹ “If this is correct,” they conclude, “then it is impossible to separate expertise from being trusted as having expert authority.”⁴² Those with expert authority, then, are simply people “we [the non-expert public] have stronger, or sufficiently strong reasons to trust.”⁴³

Nowhere does GGJ say that public trust in experts is sufficient for expert authority. They only say it is necessary, and they do not offer additional comment on what other condition that, when added to public trust, might be sufficient for expert authority. It is conceivable that they would accept some disjunctive account, according to which truth or immersion in expert domains or some other minimally adequate explanation of expert competence would suffice. But that is speculation on my part. The claim GGJ present is that public trust is a necessary condition for expert authority.

If they are right about expert authority, then the conditions for correctly determining whether an expert is trustworthy become the criteria for correctly accepting their expertise as authoritative, and these, according to GGJ, include “humility, honesty, and transparency.”⁴⁴ GGJ then focus on the ways that transparency can facilitate increased trust in putative experts. Returning to the question of the safety of COVID-19 vaccines for children, a relevant assessment would go beyond empirical assessment, to include the vulnerability of particular children, the unknown rate and severity of the vaccine’s side-effects, the psychological and economic impact of children continuing to stay home. Thus, the decision of whether to recommend the vaccine is as much about risk-tolerance and values assessment as current empirical standards. If the public had been made aware of these additional elements of expert judgment, they might not have been so skeptical of the vaccine. Thus, GGJ argue that experts should have been more transparent with the public about their own degree of uncertainty and the nature of the seeming disagreement among experts on the topics. Doing so, according to GGJ, would have given the lay public stronger reason to believe they were reliable, and therefore, stronger reason to attribute them expert authority.

Objection 1: The Case Against Objective Expert Authority is Insufficient

The motivation for GGJ’s subjectivist account of expert authority is derived from their skepticism about the ability of objective accounts to sufficiently account for the normative authority of expertise. They worry that consensus can be biased and that even well-meaning expert can rally around false claims. With respect to SEE, they worry that linguistic immersion leaves out public interests.

I share many of their concerns about veritism, and I have expressed these elsewhere.⁴⁵ However, to interpret the claim about “the extent to which experts agree” as “mere agreement” does a disservice to the power of longitudinal, independent expert consensus on a topic. Even if experts do not necessarily or always arrive at the truth, their considered judgments after extended study, experiment, and collaboration amount to more than “majority rules.” This is because scientific claims emerge in the context of theoretical frameworks and a history of empirical findings in their respective domains. Acquiring this sort of epistemic standing occurs through a number of processes that would seem to confer epistemic authority: rigorous training, independent review, the scrutiny of peers (at conferences and in journals), and linguistic immersion.⁴⁶ Further, commitments to various claims in a given domain are held in varying degrees. But if everyone in a domain accepts a claim to at least some minimum threshold of belief, that’s evidence that it is the best available judgment on the issue. GGJ do not elaborate on the epistemic implications of this process, but if objective epistemic authority can be established by longstanding, well-respected processes internal to expert domains, their claim that standard accounts of expertise are inadequate has not yet been established.

There is no question that research funding is often biased toward certain projects and political interests or that research questions are framed in accordance with researchers’ prejudices and personalities.⁴⁷ And the examples that GGJ offer on pages 12 and 13—e.g., of behavior like “self-silencing” in case of uncertainty or peer pressure to follow the “majority”— must be taken seriously. And concerns about disregarding the public’s perspective on the adequacy of expert “solutions” are well-founded, as demonstrated by complaints from the disability rights community about how their interests are often overlooked by the medical establishment for not asking people with disabilities whether their initiatives were actually helpful.⁴⁸

But the important lesson about COVID-19 is that, for the most part, the experts were right: The virus was dangerous, distancing was necessary until a vaccine was ready, hard risk/benefit choices would have to be made, and the vaccine works. While there was much uncertainty and much disagreement to be rectified, that’s true for any novel problem that requires a specialized, multi-disciplinary solution. And in all such cases, only people trained in the vast terminology and methodology of a domain, as well as its current literature, can begin to judge whether any particular claim in a scientific domain merits respect.

When, on the specific question of the safety of vaccines for children 12 to 15 years old, experts found themselves in a state of disagreement, then it was true that, as GGJ put it, “[p]eople would still not know which experts to trust.”⁴⁹ But in that case, it was also true for scientists and other experts working on problems related to COVID-19. When the people who are in the best position to know whether the vaccine is safe discover reasonable disagreement among themselves, then everyone should be uncertain, and that is the epistemically responsible conclusion. But this responsibility is not grounded in whether the public trusts the experts. Imagine that the CDC exemplified all of the virtues of trustworthiness GGJ recommend and that the JCVI exhibited the opposite, but at the end of the day the CDC stills says the vaccine is safe for minors and the JCVI says it isn’t. Does that mean the CDC, in exhibiting those virtues, breaks the epistemic tie and becomes authoritative for the public over the JCVI? Not obviously. Epistemically, the public—including other experts, such as pediatricians—should suspend judgment.⁵⁰

It is also important to note that, despite GGJ’s criticisms of alternative accounts of expert authority, these accounts are also equipped to address public concerns about transparency and disagreement in the way GGJ prescribe. Even if expert authority is objective, experts can acknowledge that, as with COVID-19, no single field of expertise has the authoritative perspective on a question, such as the efficacy of a vaccine, and that there is a measure of disagreement on the conclusions. Experts from epidemiology, virology, public health, infectious disease, sociology, ethics, and economics can be brought together (and in many states were brought together during the pandemic) to discuss various possible impacts of any particular recommendation regarding vaccines. To avoid concerns about excluding the public, people with relevant, on-the-ground experience of the effects of expert recommendations can be included in discussions regarding the adequacy of certain interventions (see Collins’s discussion of “experience-based experts”⁵¹). The public could have just as easily been made aware of this uncertainty, disagreement, and the different possible conclusions based on different value-assessments. Whether the public would have trusted experts on the basis of this transparency is a question I will address in Objection 3.

Objection 2: Trust is Critical for Expert Authority, But Not Public Trust

A second reason to reject the idea that public trust is necessary for expert authority, is that public trust adds no normative element to the processes that establish expert consensus. Both the terms of debate on a domain-specific question and the methods used to assess those terms in light of the question are determined solely by experts. And while these terms and methods were indeed the process of extensive collaborative projects over a significant period of time and required trust among the expert participants, the fact remains that the public’s approval or disapproval of the process, outcomes, or character of those involved was not necessary for establishing the normative force of those outcomes for non-experts.

Note first that the terms of the debate that non-experts must choose from are already set by experts in specialized domains. For example, in order for a conspiracist to make a claim such as, “SARS CoV-2 was created in a lab in China for purposes of political warfare,” they already accept certain claims grounded in expertise: that viruses are transmitted via a certain mechanism, SARS CoV-2 is a virus, that it is harmful, and that its impact is global. The only reason the conspiracy has meaning is because experts have given meaning to the terms of the debate. But those same experts do not then make conspiratorial claims. The lay public has to trust experts to some degree even to reject their claims. And note second that the means by which these terms are evaluated are solely the purview of those working within the field—e.g., how experts distinguish viruses from one another, how they explain why viruses make people sick, and how they measure infection rates (R-values) and dispersion parameters (K-values).

And, of course, with more than two centuries of insights on these issues, not even experts can acquire an understanding of their domain first-hand. They have to place a great deal of trust in the myriad of experts that established the domain as it currently stands. Sociologist Gil Eyal explains:

A great deal of “normal science” depends on trust between scientists. Scientists cannot check for themselves each and every detail within another’s work upon which they may be building. If they did, the pace of scientific advance would grind to a halt. They rely on informed trust, on reputation, on collective assessments of the skills, meticulousness, and integrity of other scientists within their “core-set.” This is called “virtual witnessing.”⁵²

If this is right, then presumably the virtues GGJ are concerned with (humility, honesty, transparency) are built into the scientific process. Experts must trust that their forebears provided them with good information on which they can improve. The question, then, is whether there is room for public trust in that process.

If even experts are dependent on the history of expertise in their domain and currently accepted claims and methods in their domain, then any particular expert’s claim is, by definition, a consensus statement—a considered judgment that reflects, to greater or lesser degrees, the state of a domain to which the non-expert has no access except through this expert’s testimony. Thus, when experts turn to present information about their domain to the public, they are not doing so in a vacuum. Stephen John explains that, “Many claims about climate change, encountered in the mass media or textbooks, for example, do not concern the climate. Rather, they are claims about consensus among climate scientists about the climate.”⁵³ This implies that every time a putative expert authority states a social problem or proposes a solution, both the problem and solution have been framed by shared understanding and interests of the specialists in that domain. To even understand a problem in an expert domain and considerations of which specialists to trust, the non-expert public has accepted “as experts” those who have crafted the conversation, and their choice among experts has already been constrained by those experts.⁵⁴

Consider the historical debate over the nature of disease transmission—between miasma theory (according to which diseases were “bad air” that came from decomposing matter) and germ theory (which says many illnesses are caused by microorganisms). Both theories were proposed by experts, and both were consistent with a variety of observations, but miasma theory is considered by contemporary experts to be antiquated, and versions of the germ theory persist. If the lay public read old manuscripts from proponents of miasma theory—experts, to be sure—and found them to be more humble, honest, and transparent than proponents of germ theory, GGJ would seem committed to the claim that, if all else was equal between the groups,⁵⁵ proponents of miasma theory had expert authority, and the lay public should believe them, while they should dismiss proponents of germ theory.

But if trust is already baked into the interactions among experts, then contemporary theories of disease are the best available perspective on disease, and the process determined that that view is authoritative, irrespectively of whether the public perceives proponents of that view as trustworthy. The public’s perception of whether one group of experts has authority does not add to nor detract from what they should believe, that is, whom they should find authoritative.

Of course, there are many cases where the public doubts expert authority because they are concerned that experts have a conflict of interest with something the public finds more valuable. And when that happens, the public should certainly speak up. But if the public’s values do not align with experts’, the remaining question is: Who has the best epistemic position on the appropriate course of action? And we do not have to imagine the implications of allowing the public to influence expert judgment about the appropriate course of action. Here’s an example from Harry Collins:

In Britain…the North Sea oil rig Brent Spar came to the end of its life in 1995. The best option seemed that of sinking it to the seabed, but environmental groups opposed this move and were supported by academics. Shell bowed to the pressure and cut up the rig on land. Later analysis, accepted by all parties, showed that disposal at sea would have been the most environmentally friendly option….⁵⁶

In this case, everyone agreed about the stakes for the general public. And everyone agreed about what is in the public’s interests. But laypeople disagreed with the experts about the appropriate action to take. Unfortunately, in hindsight, “all parties” agreed that the lay response was suboptimal compared with the proposed expert response.

Objection 3: Indicators of Trustworthiness are No Less Fraught than any Other

Even if GGJ’s case for the necessity of public attribution for expert authority were strong, it is a further question whether transparency would promote the requisite trust needed for that attribution. There are three reasons to think it would not.

First, in some cases, transparency has undermined public trust in expertise. We saw this vividly when emails from the University of East Anglia’s Climatic Research Unit were released to the public in an scandal now known as “Climategate.”⁵⁷ Stephen John explains that the lay public believed “these emails showed that the climate scientists at UEA were engaged in ‘non-’ or ‘anti-’ scientific practices, for example by confusing correlation with causation, by refusing to include certain data sets in their analyses, by refusing to publish papers by certain authors and so on.”⁵⁸ Harry Collins writes, “even the scientific standards of honesty and openness, it seemed, were being sacrificed for the sake of victory in the case of scientists’ opinions on the climate.”⁵⁹

This is not the sort of transparency that GGJ have in mind, of course. If the researchers had known their emails would be public, they wouldn’t have sent them. Nonetheless, the point stands that if their honest perspective on how information should be handled in the face of a doubting public had been made transparent, the loss of trust would have been no less substantial. Some of the emails sound as if they are actively withholding information from the public. And if the non-expert public had been correct in interpreting them this way, that would certainly have struck a damaging blow to the objectivity of expert authority. Of course, what was really happening was considered “normal and respectable” scientific behavior.⁶⁰ It was simply that the non-expert public could not grasp the details relevant even to their own interests in the science. We saw similar examples of this sort of misunderstanding of science when the lay public complained about the rapidity with which the COVID-19 vaccine was developed. Despite extensive public statements regarding the process for vaccine development, skeptics perseverated on the speed of the process rather than whether the globally collaborative efforts were sufficient.

Second, we know that trust in experts on some issues is better predicted by political affiliation than objective markers of trustworthiness. Sociologist Gil Eyal explains that in the US, conservatives have shifted their stance on science irrespective of any evidence regarding actual scientific practices:

In the early 1970s, Americans self-identifying as “conservative” were more likely than “moderates” or “liberals” to answer that they had “a great deal” of confidence in science. By 2010, however, they were the least likely, due to a precipitous decline of about 12 percent. … This decline was especially pronounced among educated conservatives, with college or graduate degrees.⁶¹

And although the specific numbers change depending on what specific questions are asked,⁶² it turns out that mistrust in “government science” transcends both politics and nationality. When “simply told about a GMO food product, British consumers made more or less reasonable estimates of its safety, ‘but when told that the British government had stated that it was satisfied that the product was safe, the levels of confidence in the safety of that product fell sharply.’”⁶³ This suggests that, while we may reasonably worry that scientists influence one another for the sake of consensus, even more concerning is that many members of the non-expert public make decisions about whom to trust on ideological grounds or motivated misunderstandings.

And finally, there are voluminous examples demonstrating how any particular marker of trustworthiness can be manipulated against real experts and in favor of people who lack expertise. Note that, in the US, all research articles funded with public money are, by law, made freely available to the public, through platforms such as NCBI and PubMed. This means that a substantial measure of transparency already exists for many scientific questions related to public interest. And yet, many non-expert groups “have used email hacks, Freedom of Information Requests and so on, to undermine climate science.”⁶⁴ It seems that whatever markers might be used by honest experts to demonstrate honest attempts to understand a complex issue can be manipulated by those who already disagree with the experts to confuse or undermine trust among the broader public.

It is worth noting that GGJ reject criteria like those offered by Goldman,⁶⁵ for example, the reasons conflicting experts offer, the degree to which they agree with other experts, etc., on the grounds that any particular criteria could be irrelevant to an issue, inscrutable to a member of the lay public, or not distinguishable for two people who claim to be experts (that is, they may both have the same credentials, etc.). But the same criticisms apply to indicators of trustworthiness, such as transparency.

Of course, if GGJ aren’t concerned about who the non-expert public trusts, then this third objection is irrelevant. The public trusts the people and groups they trust, and when they do, those people have expert authority. But if the attribution of expert authority comes apart from objective expert authority, and if it is possible for laypeople to be mistaken in their attributions—if there are people the lay public should trust—then the arguments above suggest that indicators of trustworthiness, by themselves, cannot serve the function that GGJ need them to.

Conclusions: What’s a Non-expert to Do?

Perhaps the most disturbing thing about contemporary technology is the ease with which it allows people to take advantage of us. It allows non-experts to mimic experts. It allows deceivers to undermine genuine experts. It allows contemporary sophists, as Plato paraphrases Aristophanes, to “make the weaker argument the stronger.”⁶⁶ I fully agree with GGJ that indirect indicators of expertise—credentials, track records, confidence, consensus statements, plain, accessible language, admitting uncertainty and failures, and so on—are not sufficient for the lay public to discern experts from charlatans, or for deciding conclusively who is authoritative and who is not. While useful for identifying some kinds of experts, such as accountants, lawyers, carpenters, and so on, the stakes in those cases are usually low. And even then, non-experts sometimes choose frauds or fakes or trespassers. As the stakes get higher and our need for experts grows, these indirect criteria become increasingly unreliable.⁶⁷

GGJ suggest this gap can be closed if experts cultivate a set of virtues related to trust—humility, honesty, transparency—and extend them to the non-expert public in their communication. Happily, we have seen that these virtues are already built into the scientific process and that this is part of what makes science so successful. But I have shown why extending these virtues to communication with the public would not likely close that gap and certainly would not contribute to the authority of experts.

Does this mean that the non-expert is forever relegated to the losing end of the epistemic power dynamic with digital manipulators? I think the answer is no, but that the situation is more complicated than GGJ make it out to be.⁶⁸ The balances will not be leveled with a series of lists that non-experts can follow (as suggested by Goldman) or the series of virtues aimed at trust suggested by GGJ, which might be well-meaning but can also be faked to further manipulate the public.

In my view, the solution is inherently social but in a sense that allows for the objectivity of expert authority. Expert domains are related in complex ways,⁶⁹ and experts in closely related domains have insights into one another’s domains. When non-experts form trusting relationships with experts in domains that are related to the one they’re interested in, they have a means by which to calibrate their trust. This is how team-based medicine works, when different specialists, nurses, social workers, trainees, etc. come together to create a care plan for a patient. No one on that team goes through a checklist before the meeting starts to make sure everyone has the relevant expertise or demonstrates the requisite traits for expert authority. The checks and balances that ensure the right people are in the right roles are voluminous—far more than any one person could even perform, much less in the moments before a meeting. All that’s left in those moments is to trust the structures that brought them there, also to trust that there are processes in place for catching mistakes. For other scholars who adopt a social, trust-based but objective account of expert authority, see Almassi and Brennan.⁷⁰

To be sure, most domains do not have the level of checks and balances found in medicine, and non-experts should adjust their trust levels accordingly.⁷¹ And if a non-expert does not have a trust-relationship with an expert in a closely enough related domain, they may simply have to suspend judgment on what to believe about any particular issue. Nevertheless, rejecting objective authority altogether and rendering expert authority subject to whether a non-expert trusts someone as an expert effectively undermines all strategies for real experts to come together and solve hard problems for the benefit of non-experts.