In 1947, the Hutchins Commission on Freedom of the Press published a report that briefly unsettled the American newspaper industry before its principal concerns were largely set aside. Its central conclusion was straightforward: a free press was necessary, but not sufficient, to ensure an informed public. The decisive factor was not the availability of information, but its quality: the extent to which media organizations accepted responsibility for accuracy, contextualization, and the faithful representation of social reality. The Commission raised concerns about concentrated ownership and commercial pressures on editorial standards. It did not consider the possibility that the entire incentive structure of mass communication might one day be reorganized around a metric bearing no necessary relationship to truth. At the time, that possibility was not yet within the range of practical concerns.
It is now. The transformation has taken place incrementally and then comprehensively, and its consequences have become embedded in the ordinary operation of contemporary life, making them difficult to identify with precision. Across much of the world, media systems are now organized, at the operational level, around capturing and retaining attention rather than representing reality, transmitting knowledge, or fulfilling the other public functions historically claimed by journalism.
The question is no longer whether this reorientation has occurred. The evidence is extensive and the underlying shift is widely documented. The more consequential question concerns its implications for the relationship between communication and value. What is lost when the production of a message becomes subordinate to the reaction it generates? And who benefits when audience response is treated as a reliable proxy for informational value?
Engineering of audience response
In 1922, Walter Lippmann argued that the mental representations through which people understand the world rarely correspond directly to external reality. His observation was descriptive: the discrepancy between reality and its representation in the public mind was a structural characteristic of mass communication, not simply an error that better reporting could eliminate. Information was necessarily selected, framed, and filtered before reaching its audience. What Lippmann could not have anticipated was a media environment in which these decisions would increasingly be made not by editors exercising professional judgment, but by systems optimizing for a single variable: the probability that a particular item of content would sustain a reader’s attention long enough to produce a measurable interaction.
The distinction is consequential. Editorial judgment, however imperfect and however influenced by the biases of its practitioners, was at least nominally directed toward an assessment of public significance. A story might be selected because it concerned an important development or clarified an issue the audience needed to understand. Algorithmic ranking, by contrast, is not inherently organized around either objective. Its principal criterion is engagement, a measure that correlates imperfectly with importance, inconsistently with intellectual interest, and not necessarily at all with the kind of understanding that requires sustained attention to complex issues.
The empirical consequences are well documented. A 2018 study published in Science by researchers at MIT found that false news spread farther and faster than true news on Twitter. The researchers identified human sharing behavior, rather than automated amplification alone, as a central factor: false stories were more novel and elicited responses that made users more likely to redistribute them. This finding is not simply evidence of human gullibility. It demonstrates what can happen when communication infrastructure rewards novelty and emotional salience rather than factual accuracy. The infrastructure creates conditions conducive to the observed outcome. The resulting behavior is then attributed to human nature, an explanation that is convenient for the owners of these systems but insufficient to account for the role of their design.
The transformation concerns the operational purpose of communication itself. Transmission, understood as the transfer of meaning from sender to receiver, with fidelity as a principal measure of success, has increasingly been subordinated to performance: the generation of reactions, the elicitation of engagement, and the production of measurable responses that a system can record and optimize. A message that conveys information without generating a reaction may register as a failure under this model. A message that generates an intense response while contributing little to the audience’s understanding may register as a success. Engagement metrics do not distinguish reliably between these outcomes because they were not designed to measure that distinction.
From persuasion to behavioral conditioning
The historical precedents for this shift are instructive, and less reassuring than they might initially appear. The twentieth century witnessed large-scale efforts to influence public opinion through organized propaganda. What distinguished the propagandists of the interwar period, from the relatively crude nationalist appeals associated with the First World War to the more sophisticated techniques developed during the 1930s, was their recognition that rational persuasion was slow, unreliable, and vulnerable to counterargument. Emotional conditioning could operate more rapidly, produce more durable dispositions, and prove considerably more difficult to rebut because it did not necessarily depend on explicit claims that could be tested or falsified. Its objective was to shape predispositions.
The contemporary attention economy did not emerge from an explicit attempt to reproduce this logic. Rather, it arrived at comparable mechanisms through the large-scale optimization of engagement metrics. In doing so, it revealed a pattern already recognized by propagandists: emotional responses can be elicited more readily than sustained reasoning; outrage often circulates more rapidly than analysis; and maintaining attention frequently depends less on informing an audience than on stimulating its affective responses.
The distinction between the interwar propagandist and the contemporary media algorithm lies principally in their objectives, not necessarily in every consequence of their methods. The propagandist sought particular political outcomes. The algorithm is designed to maximize measurable outcomes such as time spent on a platform or the probability of further interaction. Both can achieve their respective objectives by directing communication toward emotional responses rather than deliberative judgment. The algorithmic system operates at a scale that may exceed that of previous propaganda infrastructures, although the comparison does not imply that their institutional contexts, intentions, or political consequences are identical.
This is what it means to describe communication as performance rather than transmission. Performance here does not imply theatricality or deliberate deception. It refers to the increasing tendency to design messages around their anticipated reception, including the reactions they are likely to elicit, rather than primarily around the information they convey. Political communication has adapted rapidly to this environment. A political speech intended to support deliberation and a political statement designed to circulate widely are structurally different forms of communication. The latter has increasingly displaced the former in settings where visibility and immediate audience response are rewarded. Corporate communication has undergone a comparable adjustment: narrative coherence and emotional resonance have often taken precedence over operational transparency because the prevailing distribution systems reward the former more readily and do not directly measure the latter.
The decoupling of visibility and substantive value
The consequences extend beyond journalism and politics to the broader processes through which societies produce, distribute, and recognize value. When attention becomes the principal currency of cultural exchange, the relationship between the intrinsic qualities of a work and its visibility, never perfectly correlated even under favorable conditions, becomes increasingly unreliable. Content adapted to the demands of attention capture, delivering its central proposition within the first few seconds and eliciting an immediate emotional response that encourages further distribution, is well positioned to circulate. Works whose value depends on patience, contextual knowledge, or repeated engagement face structural disadvantages that critical recognition alone cannot necessarily overcome.
This tension is not new. The relationship between commercial success and cultural merit has been contested for as long as cultural goods have been exchanged in markets. What has changed is the precision with which the performance of different forms of content can now be measured, and the speed with which those measurements inform subsequent production decisions. A publisher operating in 1970 might have suspected that more demanding material attracted a smaller audience. The publisher had no equivalent of today’s continuous analytics showing the precise scale of the difference, the point at which readers abandoned a publication, or the alternative content they selected. These data are now available, often in real time, to organizations distributing content through digital platforms. The economic incentives created by this information are not difficult to identify.
The result is a growing separation between performance value and substantive value. Content can achieve economic and social success, measured in reach, recognition, and distribution, without necessarily improving understanding or offering substantial aesthetic or intellectual depth. Its success may instead reflect its ability to perform within systems designed to capture attention. The resulting cultural environment should not be understood as evidence of an absolute decline in quality. Rather, it is an environment in which quality and visibility have become less reliably connected, distorting the conditions under which content is produced, recognized, and preserved.
The limits of media literacy
The conventional response to concerns about media quality is to advocate media literacy: the development, through education, of the critical skills required to evaluate sources, recognize manipulation, and assess information with appropriate skepticism. This approach is well intentioned, and its limitations are substantial.
Media literacy addresses the demand side of the problem. It seeks to equip individuals with the skills to resist manipulation and evaluate information independently of the mechanisms that determine its visibility. These skills are valuable, but they cannot resolve the underlying problem on their own. The principal difficulty is not necessarily that audiences are unable to distinguish reliable from unreliable information when they encounter it directly and have sufficient time to assess it. It is that the infrastructure determines which information reaches them, under what conditions it is presented, and how much cognitive capacity remains available for evaluation. A reader who could correctly assess a misleading headline given sufficient time and context may share it before completing that assessment because the system encourages immediate redistribution rather than reflection.
The media literacy framework locates the problem primarily in individual cognition and proposes an individual remedy. The more fundamental problem is architectural. It resides in the design of the systems through which public information is distributed: ranking algorithms, engagement metrics, and autoplay functions that determine, at scale, which material receives exposure and which remains largely unseen. These mechanisms are not neutral channels of transmission. They function as editorial systems with identifiable selection criteria and measurable preferences, operating at a scale that exceeds that of many traditional editorial institutions. Yet governance frameworks have frequently treated digital platforms as neutral infrastructure rather than accounting fully for their functional role in selecting and distributing public information.
In most jurisdictions, regulatory responses have concentrated on content moderation, including the removal or restriction of particular categories of harmful material, rather than on the incentive structures that continually generate and amplify such material. Removing individual instances of misinformation from a system optimized to distribute misleading content is analogous to treating the symptoms of a public-health problem while leaving its underlying cause intact. The comparison is imperfect, but the distinction it identifies is important: interventions targeting individual items of content do not necessarily alter the economic and technical conditions that encourage their production and circulation. The Hutchins Commission was concerned with the effects of commercial pressure on editorial standards. In contemporary digital media, that pressure has become more deeply embedded. It is no longer merely an external influence on the system; it is part of the system’s operating logic.
What the metrics displace
The Hutchins Commission’s central concern was the relationship between a free press and a self-governing public. Democratic deliberation, in its account, depended on citizens having access to accurate, contextualized information about the world they were collectively responsible for governing. That relationship has not disappeared. It has become more difficult to sustain under a media infrastructure whose commercial incentives do not necessarily align with the informational conditions democratic deliberation requires.
The discrepancy between what contemporary media systems optimize and what an informed public requires cannot be resolved through individual behavior alone. A reader who commits to more deliberate consumption, verifies claims before sharing them, and seeks substantive analysis over immediate stimulation is making choices against the prevailing incentives of a system designed to make those choices more difficult and their alternatives more rewarding. This is not simply a failure of individual discipline. It is a predictable response to an environment engineered, through substantial investment and increasingly sophisticated optimization, to encourage particular forms of behavior.
Lippmann’s mental representations of the external world have always been imperfect. The distinctive feature of the current environment is that the processes through which those representations are formed are increasingly mediated by systems that have no direct incentive to maximize their accuracy, but substantial incentives to maximize their emotional salience and capacity to sustain attention. What reaches the audience is therefore not necessarily the message that best represents reality, but the message that performs most effectively under the system’s criteria.
This is not merely a media literacy problem. It is a problem of institutional and technical design. Addressing it requires collective responses capable of changing the conditions under which information is selected, distributed, and monetized, rather than relying exclusively on individuals to compensate for those conditions through more responsible behavior. The persistent emphasis on individual responsibility is consequential precisely because it shifts attention away from the architecture of the system and toward the conduct of the people operating within it. That framing is not only durable; it is also compatible with leaving the underlying incentives largely unchanged.
