Boaz Miller

This chapter investigates how online conversational environments might be designed to promote epistemic health rather than merely reduce incivility. Drawing on the authors’ empirically informed collaboration with an industry partner developing engagement platforms for publishers, it argues that prevailing industry approaches to “healthy conversation” disproportionately prioritize civility norms while neglecting epistemic norms governing truth, evidence, and inquiry. The analysis distinguishes epistemic toxicity from incivility and examines existing moderation tools, reporting systems, and fact-checking practices, showing that most rely on the identification of manifest norm violations—an approach well suited to detecting insults or harassment but ill suited to tracking misinformation or evidential deficiencies. Survey data from platform users further reveal a striking mismatch between industry priorities and user expectations: participants value epistemic quality at least as highly as civility. To explain this gap, the chapter introduces the “violation-identification problem,” according to which breaches of epistemic norms are typically non-manifest and therefore resistant to automated or surface-level detection. The discussion concludes that improving epistemic health online requires a shift away from civility-centered moderation toward design strategies modeled on high-quality offline epistemic practices, especially those foregrounding evidential standards and norms of inquiry.

This article highlights the role of epistemic norms in mitigating the spread of misinformation. The mixed-methods study includes exploratory reconstructions and survey experiments. Two intervention approaches proved efficient in reducing the sharing of misinformation, but only one significantly differentiated between true and false information. This study contributes to the literature on normative countermeasures and is the first to emphasize epistemic norms. Although misinformation is fundamentally an epistemic issue, scholars rarely refer to the rich epistemic literature for conceptualization and theory. The article also draws on focus theory and additional social literature to ensure a holistic framework. In addition to underscoring the role of epistemic norms, we suggest temporal proximity as a new factor to consider in focus theory-based measures. Ethical, empirical, and practical implications are discussed, emphasizing the need to control for countermeasures’ effect on true information sharing, which may threaten to suppress public discourse.

Fostering healthy online conversations is essential to the integrity of public discourse, yet the norms that guide such conversations remain contested and difficult to enforce. This paper develops and empirically grounds a conceptual and empirical framework for understanding and addressing online toxicity. Building on the distinction between epistemic and civil norms, we argue that norm violations are the proper target of moderation. While this paper is primarily conceptual, it is informed by empirical observations drawn from a collaboration with a platform designer. Drawing on a dataset of user comments and responses to moderation scenarios, we identify eight recurring types of norm violations and analyse patterns of user agreements about whether such content should be removed. Our findings reveal that while civil norm violations prompt relatively consistent responses, epistemic violations elicit wide disagreement, raising challenges for universal moderation standards. We conclude by proposing a set of general, evidence-informed principles for platform designers and moderators. These recommendations emphasise context-sensitive moderation, platform affordances that encourage epistemic responsibility, and the integration of civil and epistemic considerations into online infrastructure. Our work bridges theoretical and practical perspectives, offering both conceptual clarity and actionable insights for scholars, designers, and practitioners engaged in shaping healthier digital discourse.

דמיון מתחולל בחלקו מחוץ לראש באמצעות עזרים טכנולוגיים. עובדה זו כשלעצמה אינה חדשה, אולם ‏טכנולוגיות חדשות משנות את טיבו של הדמיון. מחוללי מדיה סינתטית, כגון דָאלִי, ופריטי מדיה סינתטיים, ‏כגון זיופים עמוקים (‏דיפ-פייקס; תמונות וסרטונים ריאליסטיים שנוצרו באופן אלגוריתמי ושמציגים אנשים ‏מבצעים או אומרים משהו שלא עשו או אמרו) מערערים על אמות המידה האפיסטמיות הבסיסיות שלנו. עם ‏זאת, טבעו של האיום האפיסטמי החדש הנשקף מהם נותר חמקמק, שכן ייצוגים בדיוניים או מעוותים של ‏המציאות הם עתיקים לפחות כמו הצילום עצמו. אפיונים קיימים באפיסטמולוגיה חברתית של האיום הגלום ‏בזיופים עמוקים מתקשים לזהות מה בעצם חדש בו. למאמר זה שתי מטרות: לאפיין את האיום האפיסטמי ‏החדש הנשקף ממדיה סינתטית, ולזהות את התנאים שבהם פריט מדיה סינתטית הוא בעייתי מבחינה ‏אפיסטמית. אנו טוענים שהאיום האפיסטמי הגלום בזיופים עמוקים בפרט, ובמדיה סינתטית בכלל, הוא ‏שלראשונה למשתמשי מחשב רגילים זמינה האפשרות המעשית ליצור בזול וללא מאמץ עולמות בדיוניים שלא ‏ניתן להבחין ביניהם לבין העולם האמיתי ולשתפם בתפוצה רחבה. פריט מדיה סינתטי הוא רעיל מבחינה ‏אפיסטמית בהקשר נתון כאשר אדם בעל הכרה כלשהי של ההקשר עלול לבלבל את העולם הבדיוני המתואר בו ‏עם העולם הממשי או עם ייצוגים מהימנים שלו בנושאים בעלי חשיבות מוסרית או אפיסטמית. ‏ Imagination takes place partly outside the mind, through technological aids. This fact is not new in ‎itself, but new technologies are changing the nature of imagination. Synthetic media generators, ‎such as DALL·E, and synthetic media items, such as deepfakes (realistic images and videos ‎algorithmically generated to depict people doing or saying things they never actually did or said), ‎challenge our fundamental epistemic standards. But the nature of the new epistemic threat they ‎pose remains elusive, since fictional or distorted representations of reality are at least as old as ‎photography itself. Existing characterizations in social epistemology of the threat posed by ‎deepfakes struggle to pinpoint what exactly is novel about it. This paper has two aims: to ‎characterize the new epistemic threat posed by synthetic media, and to identify the conditions ‎under which a synthetic media item is epistemically malignant or toxic. We argue that the ‎epistemic threat posed by deepfakes in particular, and by synthetic media more broadly, lies in the ‎unprecedented practicable possibility afforded to ordinary computer users to cheaply and ‎effortlessly create fictional worlds that are indistinguishable from the real world and share them ‎widely. A synthetic media item is epistemically toxic or malignant in a given context when a person ‎acquainted with the context is liable to confuse the fictional world depicted in it with the real ‎world or with credible representations thereof on matters of moral or epistemic significance.‎

Synthetic media generators, such as DALL-E, and synthetic media artifacts, such as deepfakes, undermine our fundamental epistemic standards and practices. Yet, the nature of their epistemic threat remains elusive. After all, fictional or distorted representations of reality are as old as photography. We argue that the novel epistemic threat of synthetic media is that, for the first time, synthetic media tools afford ordinary computer users the practicable possibility to cheaply and effortlessly create and widely share fictional worlds indistinguishable from the real world or credible representations of it. We further argue that a synthetic media artifact is epistemically malignant in a given media context for a person acquainted with the context when the person is misled to confuse the version of the world depicted in it with the real world in an epistemically or morally significant way.

This Element is about the social dimensions of scientific knowledge. The first section asks in what ways scientific knowledge is social. The second section develops a conception of scientific knowledge that accommodates the insights of the first section, and is consonant with mainstream thinking about knowledge in analytic epistemology. The third section asks under what conditions we can tell, in the real world, that a consensus in a scientific community amounts to shared scientific knowledge, as characterized in the second section, and how to deal with scientific dissent. The fourth section reviews the ways epistemic and social elements mutually interact to coproduce scientific knowledge. This Element engages with literature from philosophy of science and social epistemology, especially social epistemology of science, as well as Science, Technology, and Society (STS), and analytic epistemology. The Element focuses on themes and debates that date from the start of the second millennium.

Leading prescriptions for addressing the spread of fake news, misinformation, and other forms of epistemically toxic content online target either the platform or platform users as a single site for intervention. Neither approach attends to the intense feedback between people, posts, and platforms. Leading prescriptions boil down to the suggestion that we make social media more like traditional media, whether by making platforms take active roles as gatekeepers, or by exhorting individuals to behave more like media professionals. Both approaches are impracticable and wrong.

We present a novel model of individual people, online posts, and media platforms to explain the online spread of epistemically toxic content such as fake news and suggest possible responses. We argue that a combination of technical features, such as the algorithmically curated feed structure, and social features, such as the absence of stable social-epistemic norms of posting and sharing in social media, is largely responsible for the unchecked spread of epistemically toxic content online. Sharing constitutes a distinctive communicative act, governed by a dedicated norm and motivated to a large extent by social identity maintenance. But confusion about this norm and its lack of inherent epistemic checks lead readers to misunderstand posts, attribute excess or insufficient credibility to posts, and allow posters to evade epistemic accountability—all contributing to the spread of epistemically toxic content online. This spread can be effectively addressed if people and platforms add significantly more context to shared posts and platforms nudge people to develop and follow recognized epistemic norms of posting and sharing.

Social inequality may obstruct the generation of knowledge, as the rich and powerful may bring about social acceptance of skewed views that suit their interests. Epistemic equality in the context of justification is a means of preventing such obstruction. Drawing on social epistemology and theories of equality and distributive justice, we provide an account of epistemic equality. We regard participation in, and influence over a knowledge-generating discourse in an epistemic community as a limited good that needs to be justly distributed among putative members of the community. We argue that rather than trying to operationally formulate an exact criterion for distributing this good, epistemic equality may be realized by insisting on active participation of members of three groups in addition to credited experts: relevant disempowered groups, relevant uncredited experts, and relevant stakeholders. Meeting these conditions fulfills the political, moral, and epistemic aims of epistemic equality.

Normatively inappropriate scientific dissent prevents warranted closure of scientific controversies and confuses the public about the state of policy-relevant science, such as anthropogenic climate change. Against recent criticism by de Melo-Martín and Intemann of the viability of any conception of normatively inappropriate dissent, I identify three conditions for normatively inappropriate dissent: its generation process is politically illegitimate, it imposes an unjust distribution of inductive risks, and it adopts evidential thresholds outside an accepted range. I supplement these conditions with an inference-to-the-best-explanation account of knowledge-based consensus and dissent to allow policy makers to reliably identify unreliable scientific dissent.

This chapter explores properties that bind individuals, knowledge, and communities, together. Section ‎‎1 introduces Hardwig’s argument from trust in others’ testimonies as entailing that trust is the glue ‎that binds individuals into communities. Section 2 asks “what grounds trust?” by exploring assessment ‎of collaborators’ explanatory responsiveness, formal indicators such as affiliation and credibility, ‎appreciation of peers’ tacit knowledge, game-theoretical considerations, and the role moral character ‎of peers, social biases, and social values play in grounding trust. Section 3 deals with establishing ‎reliability standards for formation and breaching of trust. Different epistemic considerations and their ‎underpinning of inductive risks are examined through various communication routes within a ‎discipline, between disciplines, and to the public. Section 4 examines whether a collective entity can ‎be trusted over and above trust that is given to its individual members. Section 5 deals with the roles ‎technological artifacts play in distributed research and collective knowledge. It presents the common ‎view in which genuine trust cannot, in principle, be accorded to artifacts, so as an opposite view. We ‎show that what counts as a genuine object of trust is relevant to debates about the boundaries of ‎collective agency and as a criterion for extended cognitive systems.‎

According to the Value-Neutrality Thesis, technology is morally and politically neutral, neither good nor bad. A knife may be put to bad use to murder an innocent person or to good use to peel an apple for a starving person, but the knife itself is a mere instrument, not a proper subject for moral or political evaluation. While contemporary philosophers of technology widely reject the VNT, it remains unclear whether claims about values in technology are just a figure of speech or nontrivial empirical claims with genuine factual content and real-world implications. This paper provides the missing argument. I argue that by virtue of their material properties, technological artifacts are part of the normative order rather than external to it. I illustrate how values can be empirically identified in technology. The reason why value-talk is not trivial or metaphorical is that due to the endurance and longevity of technological artifacts, values embedded in them have long-term implications that surpass their designers and builders. I further argue that taking sides in this debate has real-world implications in the form of moral constraints on the development of technology.

There is an existing debate regarding the view that technological instruments, devices, or machines can assert ‎or testify. A standard view in epistemology is that only humans can testify. However, the notion of quasi-‎testimony acknowledges that technological devices can assert or testify under some conditions, without ‎denying that humans and machines are not the same. Indeed, there are four relevant differences between ‎humans and instruments. First, unlike humans, machine assertion is not imaginative or playful. Second, ‎machine assertion is prescripted and context restricted. As such, computers currently cannot easily switch ‎contexts or make meaningful relevant assertions in contexts for which they were not programmed. Third, ‎while both humans and computers make errors, they do so in different ways. Computers are very sensitive to ‎small errors in input, which may cause them to make big errors in output. Moreover, automatic error control ‎is based on finding irregularities in data without trying to establish whether they make sense. Fourth, ‎testimony is produced by a human with moral worth, while quasi-testimony is not. Ultimately, the notion of ‎quasi-testimony can serve as a bridge between different philosophical fields that deal with instruments and ‎testimony as sources of knowledge, allowing them to converse and agree on a shared description of reality, ‎while maintaining their distinct conceptions and ontological commitments about knowledge, humans, and ‎nonhumans.‎

This paper reviews current debates in social epistemology about the relations ‎between ‎knowledge ‎and consensus. These relations are philosophically interesting on their ‎own, but ‎also have ‎practical consequences, as consensus takes an increasingly significant ‎role in ‎informing public ‎decision making. The paper addresses the following questions. ‎When is a ‎consensus attributable to an epistemic community? Under what conditions may ‎we ‎legitimately infer that a consensual view is knowledge-based or otherwise ‎epistemically ‎justified? Should consensus be the aim of scientific inquiry, and if so, what ‎kind of ‎consensus? How should dissent be handled? It is argued that a legitimate inference ‎that a ‎theory is correct from the fact that there is a scientific consensus on it requires taking ‎into ‎consideration both cognitive properties of the theory as well as social properties of ‎the ‎consensus. The last section of the paper reviews computational models of ‎consensus ‎formation.‎

Journalism and media studies lack robust theoretical concepts for studying journalistic knowledge ‎generation. More specifically, conceptual challenges attend the emergence of big data and ‎algorithmic sources of journalistic knowledge. A family of frameworks apt to this challenge is ‎provided by “social epistemology”: a young philosophical field which regards society’s participation ‎in knowledge generation as inevitable. Social epistemology offers the best of both worlds for ‎journalists and media scholars: a thorough familiarity with biases and failures of obtaining ‎knowledge, and a strong orientation toward best practices in the realm of knowledge-acquisition ‎and truth-seeking. This paper articulates the lessons of social epistemology for two central nodes of ‎knowledge-acquisition in contemporary journalism: human-mediated knowledge and technology-‎mediated knowledge. ‎

People increasingly form beliefs based on information gained from automatically filtered Internet ‎sources such as search engines. However, the workings of such sources are often opaque, preventing ‎subjects from knowing whether the information provided is biased or incomplete. Users’ reliance on ‎Internet technologies whose modes of operation are concealed from them raises serious concerns about ‎the justificatory status of the beliefs they end up forming. Yet it is unclear how to address these concerns ‎within standard theories of knowledge and justification. To shed light on the problem, we introduce a ‎novel conceptual framework that clarifies the relations between justified belief, epistemic responsibility, ‎action, and the technological resources available to a subject. We argue that justified belief is subject to ‎certain epistemic responsibilities that accompany the subject’s particular decision-taking circumstances, ‎and that one typical responsibility is to ascertain, so far as one can, whether the information upon which ‎the judgment will rest is biased or incomplete. What this responsibility comprises is partly determined by ‎the inquiry-enabling technologies available to the subject. We argue that a subject’s beliefs that are ‎formed based on Internet-filtered information are less justified than they would be if she either knew how ‎filtering worked or relied on additional sources, and that the subject may have the epistemic ‎responsibility to take measures to enhance the justificatory status of such beliefs.‎.

According to Popper's rationality principle, agents act in the most adequate way according to the objective situation. I propose a new interpretation of the rationality principle as consisting of an idealization and two abstractions. Based on this new interpretation, I critically discuss the privileged status that Popper ascribes to it as an integral part of all social scientific models. I argue that as an idealization, the rationality principle may play an important role in the social sciences, but it also has inherent limitations that inhibit it from having the privileged status that Popper ascribes to it in all cases.

Mainstream analytic epistemology regards knowledge as the property of individuals, rather ‎than groups. Drawing on insights from the reality of knowledge production and dissemination ‎in the sciences, I argue, from within the analytic framework, that this view is wrong. I defend ‎the thesis of ‘knowledge-level justification communalism’, which states that at least some ‎knowledge, typically knowledge obtained from expert testimony, is the property of a ‎community and possibly none of its individual members, in that only the community or some ‎members of it collectively possesses knowledge-level justification for its individual members’ ‎beliefs. I address several objections that individuals, qua individuals, have or are able to ‎acquire knowledge-level justification for all the beliefs they obtain from expert testimony. I ‎argue that the problem I identify with individualism is invariant under any specific account of ‎justification, internalist or externalist. ‎

It is commonly argued that values “fill the logical gap” of underdetermination of theory by evidence, namely, values affect our choice between two or more theories that fit the same evidence. The underdetermination model, however, does not exhaust the roles values play in evidential reasoning. I introduce WAVE – a novel account of the logical relations between values and evidence. WAVE states that values influence evidential reasoning by adjusting evidential weights. I argue that the weight-adjusting role of values is distinct from their underdetermination gap-filling role. Values adjust weights in three ways. First, values affect our trust in the testimony of others. Second, values influence the evidential thresholds required for justified epistemic judgments. Third, values influence the relative weight of a certain type of evidence within a body of multimodal discordant evidence. WAVE explains, from an epistemic perspective, rather than psychological, how smokers, for example, can find the same evidence about the dangers of smoking less persuasive than non-smokers. WAVE allows for a wider effect of values on our accepted scientific theories and beliefs than the effect for which the underdetermination model allows alone; therefore, science studies scholars must consider WAVE in their research and analysis of evidential case studies.

The social sciences today, Lee McIntyre argues, are in the same state in which the natural sciences were in the Dark Ages. In the same way that religion inhibited the progress of science and the growth of knowledge in the Dark Ages, so is political correctness inhibiting progress in the social sciences and the growth of knowledge today. This is why, so he argues, the social sciences do not follow the scientific method like the natural sciences do, and are hence incapable of offering effective solutions to pressing social problems such as crime, famine, and war. The reason why political correctness is able to affect science in this way is our fear of knowledge. Human beings are simply too terrified to discover unpleasant truths about themselves, so they prevent certain hypotheses from being seriously tested in social science research. Rather, they prefer to indulge in comforting pseudo-scientific ideology. These are bold claims, but McIntyre’s argument to support them is thin and weak. In particular, it fails to come close to meeting the standards of proof by empirical evidence that McIntyre requires the social sciences to meet

Philosophers have recently argued, against a prevailing orthodoxy, that standards of knowledge partly depend on a subject’s interests; the more is at stake for the subject, the less she is in a position to know. This view, which is dubbed “Pragmatic Encroachment” has historical and conceptual connections to arguments in philosophy of science against the received model of science as value free. I bring the two debates together. I argue that Pragmatic Encroachment and the model of value-laden science reinforce each other. Drawing on Douglas’ argument about the indispensability of value judgments in science, and psychological evidence about people’s inability to objectively reason about what they care about, I introduce a novel argument for Pragmatic Encroachment.

In August 2002, three Indian computer scientists published a paper, ‘PRIMES is in P’, online. It presents a ‘deterministic algorithm’ which determines in ‘polynomial time’ if a given number is a prime number. The story was quickly picked up by the general press, and by this means spread through the scientific community of complexity theorists, where it was hailed as a major theoretical breakthrough. This is although scientists regarded the media reports as vulgar popularizations. When the paper was published in a peer-reviewed journal only two years later, the three scientists had already received wide recognition for their accomplishment. Current sociological theory challenges the ability to clearly distinguish on independent epistemic grounds between distorted and non-distorted scientific knowledge. It views the demarcation lines between such forms of presentation as contextual and unstable. In my paper, I challenge this view. By systematically surveying the popular press coverage of the ‘PRIMES is in P’ affair, I argue--against the prevailing new orthodoxy--that distorted simplifications of scientific knowledge are distinguishable from non-distorted simplifications on independent epistemic grounds. I argue that in the ‘PRIMES is in P’ affair, the three scientists could ride on the wave of the general press-distorted coverage of their algorithm, while counting on their colleagues’ ability to distinguish genuine accounts from distorted ones. Thus, their scientific reputation was unharmed. This suggests that the possibility of the existence of independent epistemic standards must be incorporated into the new SSK model of popularization.

David Chalmers thinks his iPhone exemplifies the extended mind thesis by meeting the criteria ‎that he and Andy Clark established in their well-known 1998 paper. Andy Clark agrees. We take ‎this proposal seriously, evaluating the case of the GPS-enabled smartphone as a potential mind ‎extender. We argue that the “trust and glue” criteria enumerated by Clark and Chalmers are ‎incompatible with both the epistemic responsibilities that accompany everyday activities and the ‎practices of trust that enable users to discharge them. Prospects for revision of the original ‎criteria are dim. We therefore call for a rejection of the trust criterion and a reevaluation of the ‎extended mind thesis.‎

Information providing and gathering increasingly involve technologies like search ‎engines, which actively shape their epistemic surroundings. Yet, a satisfying account ‎of the epistemic responsibilities associated with them does not exist. We analyze ‎automatically generated search suggestions from the perspective of social ‎epistemology to illustrate how epistemic responsibilities associated with a ‎technology can be derived and assigned. Drawing on our previously developed ‎theoretical framework that connects responsible epistemic behavior to ‎practicability, we address two questions: first, given the different technological ‎possibilities available to searchers, the search technology, and search providers, ‎who should bear which responsibilities? Second, given the technology’s ‎epistemically relevant features and potential harms, how should search terms be ‎autocompleted? Our analysis reveals that epistemic responsibility lies mostly with ‎search providers, which should eliminate three categories of autosuggestions: those ‎that result from organized attacks, those that perpetuate damaging stereotypes, and ‎those that associate negative characteristics with specific individuals.‎.

Scientific consensus is widely deferred to in public debates as a social indicator of the existence of knowledge. However, it is far from clear that such deference to consensus is always justified. The existence of agreement in a community of researchers is a contingent fact, and researchers may reach a consensus for all kinds of reasons, such as fighting a common foe or sharing a common bias. Scientific consensus, by itself, does not necessarily indicate the existence of shared knowledge among the members of the consensus community. I address the question of under what conditions it is likely that a consensus is in fact knowledge based. I argue that a consensus is likely to be knowledge based when knowledge is the best explanation of the consensus, and I identify three conditions—social calibration, apparent consilience of evidence, and social diversity, for knowledge being the best explanation of a consensus