There are decisions we can make from direct experience. We can taste the food, read the contract, inspect the workmanship or compare the price. Even then, our judgement may be incomplete, but at least some of the relevant evidence is available to us.

Other decisions are different. A patient cannot independently verify every clinical claim. A citizen cannot reproduce the modelling behind a public policy. An investor cannot personally audit every assumption in a company forecast. A student cannot master an entire field before deciding which account of it to accept. In these situations, judgement depends on testimony: on the words, methods and institutions of people who claim to know more than we do.

This dependence is not a failure of intelligence. It is a condition of living in a complex society.

The problem is that the person who needs expertise is often least equipped to recognise it. A genuine expert and a persuasive imitator may use the same vocabulary. Both may display qualifications, cite evidence and speak with confidence. Both may attract followers and institutional approval. The novice sees the performance of knowledge before they can assess its substance.

The question is therefore not simply whether experts should be trusted. Modern life makes some form of reliance unavoidable. The harder question is how claims, experts and institutions should be judged when the knowledge required to assess them is precisely what the non-expert lacks.

The problem comes before trust itself. Before deciding whether reliance is deserved, the non-expert must decide which claims, speakers and institutions merit serious attention.

How do we know what to believe?

Part I - The Necessity of Epistemic Dependence

The ideal of the independent thinker is powerful. It suggests that rational people should examine the evidence for themselves, resist deference and refuse to accept claims merely because an authority has made them.

This ideal protects against manipulation. It reminds us that credentials do not make a claim true and that institutions can preserve error as well as knowledge. But complete intellectual independence is impossible.

John Hardwig described this condition as epistemic dependence. Much of what any individual knows rests on work they have not personally inspected and often could not competently evaluate. Scientific knowledge depends on specialised instruments, distributed teams, technical methods and accumulated findings. Professional knowledge depends on training, standards, case experience and communities of practice. Even experts must rely on other experts outside their own domain.

We do not escape dependence by refusing expertise. We merely transfer it.

A person who rejects an established institution may defer instead to a public commentator, an online community or a charismatic dissenter. A consumer who distrusts professional advice may rely on testimonials. A manager who dismisses academic evidence may accept the authority of personal experience. The language of independence can conceal a change in whom we trust rather than an end to trust itself.

This matters because every source of knowledge has limits. Institutions can become defensive. Professionals can protect status. Communities can reinforce shared errors. Experience can be vivid but unrepresentative. Expertise can be genuine within one domain and carelessly extended into another.

Public judgement begins by accepting a difficult truth: we cannot verify everything ourselves, but neither can we responsibly surrender judgement to those who claim expertise.

The task is not to eliminate dependence. It is to make dependence more discriminating.

Part II - The Performance of Credibility

Expertise is usually encountered through signals.

Four ideas need to be separated. Expertise is competence within a domain. Credibility is the judgement that a person, institution or claim is worthy of belief. Authority is the socially recognised permission to guide judgement or action. Trust is the willingness to rely when independent verification is incomplete. They can reinforce one another, but none guarantees the others.

We see titles, institutional affiliations, awards, publications, follower counts, professional language and confident explanations. These signals matter because knowledge is often invisible. A qualification can indicate sustained study. Professional registration may indicate that minimum standards and accountability mechanisms apply. A publication record may demonstrate engagement with a field. Institutional affiliation can provide access to resources, peer scrutiny and established norms.

But signals are not the same as substance.

A signal becomes useful when it is difficult to obtain without the quality it claims to represent. It becomes weak when it can be copied, purchased, exaggerated or presented without context. The more public life is organised through profiles, rankings and promotional communication, the easier it becomes to confuse the appearance of expertise with expertise itself.

Visibility intensifies this problem. Digital platforms reward forms of communication that can be recognised quickly: certainty, novelty, simplicity and emotional clarity. Genuine expertise often moves differently. It qualifies. It distinguishes what is known from what is inferred. It notices exceptions. It can appear hesitant because the expert understands how much the conclusion depends on context.

The confident non-expert may therefore look more authoritative than the careful expert.

This does not mean that clarity is suspicious or that experts should communicate badly. The ability to explain is part of public responsibility. But communication becomes dangerous when uncertainty is treated as weakness and confidence as evidence.

Alvin Goldman asks how non-experts might distinguish between competing experts. His answer does not require the novice to become an expert. It directs attention towards indirect but meaningful evidence: the quality of arguments, agreement among relevant peers, possible conflicts of interest, comparative track records and the assessments of other competent judges.

These are not perfect tests. Consensus can be wrong. Track records may be selectively presented. Peer communities can exclude dissent or reproduce shared assumptions. Yet the alternative is not judgement free from signals. It is judgement based on signals that are easier to manipulate.

A stronger credibility signal is connected to process. It reveals how a conclusion was reached, what standards governed the work, what evidence might change the view and what consequences follow when the advice is wrong.

Credibility is not established by appearing certain. It is established by being answerable.

Part III - Expertise, Power and the Distribution of Credibility

The question of who gets believed is not settled by competence alone.

Credibility is distributed socially. Some speakers receive an immediate presumption of authority; others must repeatedly prove that they are entitled to speak. Accent, status, profession, gender, race, age, institutional prestige and familiarity can influence how testimony is received before its content is considered.

Miranda Fricker describes this as testimonial injustice: a person may be given less credibility because prejudice affects how they are perceived as a knower. This is not only a matter of hurt or disrespect. It can distort collective judgement by excluding relevant knowledge.

Professional settings offer many examples. A junior colleague may notice a risk but be dismissed because seniority is mistaken for accuracy. A patient may understand their own symptoms but be treated as unreliable because experiential knowledge is valued less than technical language. A practitioner from outside a prestigious institution may offer a stronger account but receive less attention than a weaker claim carrying a more familiar name.

The reverse problem also occurs. Status can produce a credibility surplus. A person recognised as an expert in one field may be treated as authoritative in another. Institutional prestige can shield poor reasoning. Public recognition can become self-reinforcing: the person is invited because they are visible, and their visibility is then used as evidence that they should be invited again.

This is why expertise and authority must be separated conceptually.

Expertise concerns a person’s knowledge or competence within a domain. Authority concerns the social permission to influence judgement or action. Expertise may justify authority, but authority can exist without expertise, and expertise can exist without public authority.

Institutions play a crucial role in connecting the two. At their best, they create procedures through which claims are tested, competence is developed and misconduct has consequences. At their worst, they turn status into insulation. They ask the public to trust the institution while giving the public few reasons to believe that errors will be acknowledged.

Calls for transparency often emerge in response. Yet Onora O’Neill warns that more disclosure does not automatically create more trust. An abundance of information can obscure rather than illuminate. Organisations may become skilled at producing evidence of compliance while remaining difficult to judge.

The relevant question is not whether everything has been disclosed. It is whether people can understand what matters, identify who is responsible and see what happens when standards are not met.

A credible authority does not merely possess information. It occupies a relationship of accountability.

Part IV - Authority Worthy of Trust

What, then, should the public look for when expertise cannot be assessed directly?

No single indicator is sufficient. Credentials matter, but only within their scope. Consensus matters, but only when the relevant community is competent and open to correction. Experience matters, but only when it has been interpreted rather than merely accumulated. Independence matters, but complete freedom from interest is rare. Transparency matters, but only when it improves understanding rather than increasing noise.

When qualified experts disagree, disagreement should not be treated as proof that expertise has failed or that every position is equally credible. The relevant questions include whether the dispute concerns evidence, interpretation, values or acceptable risk; how expertise is distributed across the competing parties; and whether each position is taken seriously within the relevant field. Uncertainty can be real without making judgement arbitrary.

Authority worthy of trust is better recognised through a pattern.

First, it respects the boundaries of competence. The trustworthy expert distinguishes what they know from what lies outside their expertise. They do not use authority earned in one domain as a universal licence to speak.

Second, it makes reasoning inspectable. This does not mean that every technical detail must be presented to every audience. It means that the path from evidence to conclusion can be examined by those capable of examining it, and explained honestly to those who are not.

Third, it remains open to correction. Expertise is not demonstrated by never changing one’s mind. It is demonstrated by having disciplined reasons for changing it. A correction should not be treated as a collapse of authority when it reflects stronger evidence or improved understanding.

Fourth, it discloses interests and incentives that may shape judgement. A conflict of interest does not automatically invalidate a claim, but an undisclosed conflict weakens the relationship through which the claim is offered.

Fifth, it accepts consequences. Advice that affects other people should be connected to professional standards, review, redress or some meaningful form of accountability. Authority without consequence invites performance without responsibility.

These principles are especially important in complex and semi-regulated markets. Consumers may face treatments, technologies or professional claims that are difficult to evaluate before purchase and sometimes even after the outcome is known. The visible provider may not be the most competent. The best-marketed innovation may not be the most appropriate. In such settings, verification mechanisms should not promise certainty. Their purpose is to improve the quality of the signals on which judgement depends.

This is also a responsibility of platforms. A platform does not become neutral merely because it allows many voices to speak. It selects categories, sequences information, ranks claims and determines which signals are made prominent. Repetition and visibility can make a claim feel familiar before it has been assessed. Verification standards and commercial relationships also influence who becomes visible and credible. The design of information is therefore part of the distribution of authority.

The goal should not be a world in which people trust nothing until they have personally verified it. Such a world would be paralysed. Nor should it be a world in which institutional status ends the need for questioning.

A mature public culture combines warranted trust with the possibility of challenge.

Conclusion - Judging What and Whom to Believe

We cannot live without relying on claims, methods and knowledge held by others. The question is not whether to depend, but how to judge what deserves reliance.

Poor judgement treats trust as a choice between obedience and suspicion. It either accepts authority because it appears legitimate or rejects authority because legitimacy has sometimes failed. Better judgement asks more demanding questions.

Is the claim within the speaker’s competence? What process produced it? Is the reasoning open to appropriate scrutiny? Are relevant peers broadly persuaded? What interests may be present? What is the track record? What happens if the advice causes harm? Is the speaker capable of correction?

These questions do not remove uncertainty. They make uncertainty more governable.

They also remind experts that public trust is not owed merely because expertise is real. Knowledge creates responsibility. The greater the asymmetry between expert and non-expert, the greater the obligation to communicate honestly, stay within scope and remain accountable for consequence.

What gets believed and who is believed should never be determined only by visibility, confidence or status.

The answer should emerge from a relationship between competence, integrity and accountability.

Authority becomes credible when it gives people reasons not simply to accept a conclusion, but to judge why acceptance is warranted.