Engagement algorithms may trap consumers in cultural overfamiliarity

A model-based study of engagement-driven content curation suggests that recommending more of what users engage with in the short term may eventually reduce their long-term satisfaction. The model, developed by marketing professor Samsun Knight of the University of Toronto’s Rotman School of Management, links repeated exposure to a rising and then falling level of engagement, creating a potential feedback loop in which a platform’s own recommendations increasingly shape what appears familiar to users. The results appeared in the peer-reviewed journal Journal of Cultural Economics under the title “Engagement-based curation and the evolution of taste.”

The study builds on a longstanding economic idea that a taste for something can increase with exposure, but its model predicts that this relationship is not unlimited. Engagement grows under moderate exposure, reaches a peak and then declines as exposure continues. According to Knight, a short assessment window is one possible reason: platforms may interpret limited engagement over a few weeks or months as evidence that content or a genre should no longer be recommended, even though some cultural preferences take much longer to develop. Hip hop, for example, took decades to become a top-selling genre, far beyond the period many streaming services use to evaluate audience response. The model therefore illustrates a distinction between immediate engagement and durable taste rather than proving that any particular platform has already caused disengagement.

The analysis also identifies a feedback problem arising when recommendation systems treat current preferences as independent, fixed traits. Past engagement may partly reflect what the algorithm previously exposed users to, yet the system can interpret that engagement as evidence of an authentic, enduring preference. The familiar material continues to be served, continues to perform well and therefore appears to validate the strategy. In Knight’s model, this process can create a self-reinforcing cycle of overexposure, staleness and churn, progressively weakening the user’s relationship with the platform. He summarizes the concern by saying that the optimization maximizing engagement in one quarter can also create staleness and eventual broader disengagement, risking strong short-term performance while gradually degrading the long-term product. This is a modeled mechanism, not a measured effect reported for Spotify, Netflix or another named service.

Knight proposes a more exploratory approach to recommendation. Intentionally showing users less-familiar material beyond what a narrow short-term engagement metric would justify could expand their exposure to new genres, artists or artistic movements, while reducing the frequency with which previously enjoyed content is repeated. Such exploration might give emerging forms time to develop and could, in principle, support the next cultural phenomenon. The model does not establish that automated exploration will necessarily outperform familiar recommendations, and the research release provides no estimates of optimal exposure levels, intervention size or the time required for a user’s preferences to change. Instead, it frames exploration and overexposure as competing considerations that engagement-only systems may neglect. Knight also points to human-curated playlists and staff recommendations as a possible complement because curators may continue presenting difficult or unproven work on the basis of judgment rather than early performance data.

The broader implication is that cultural discovery depends partly on what recommendation systems choose not to reinforce. A system optimized only for immediate response may efficiently keep serving content users already engage with, but it can also narrow the range of material they encounter and obscure the platform’s role in shaping their tastes. Exploration may appear inefficient under a short-term metric because unfamiliar content often attracts less engagement at first, yet it could preserve opportunities for new cultural preferences to emerge. The study is primarily theoretical and does not demonstrate improved long-term satisfaction, lower churn or measurable cultural benefits in human users. Its contribution is to formalize how engagement feedback, repeated exposure and fixed assumptions about taste can combine to produce overfamiliarity, and why a curation strategy that performs well in the short term may carry longer-term risks.

Subject of Research: A model of how engagement-based recommendation systems may shape cultural taste and produce overfamiliarity.

Article Title: Engagement-driven curation can trap culture consumers in 'overfamiliarity,' leading to less long-term satisfaction

Article References: Engagement-driven curation can trap culture consumers in 'overfamiliarity,' leading to less long-term satisfaction. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: recommendation algorithms, engagement optimization, cultural taste, overfamiliarity, streaming platforms, exploration, consumer churn, cultural economics, Engagement-driven, curation, trap, culture