More papers, fewer ideas: How evaluation asymmetries drive scientific convergence

Despite sustained growth in scientific output, contemporary research displays marked convergence in topics, claims, and communicative form. This Perspective offers a behavioural explanation for this paradox. We argue that disjoint evaluation regimes operating at different levels of the scientific system generate a stable but epistemically inefficient equilibrium. Author-level evaluation is largely neutral with respect to publication volume, whereas journal-level evaluation is sensitive to output volume through denominator-based performance indicators.This structural asymmetry reshapes selection behaviour. Using a minimal illustrative model, we show how variance-sensitive editorial evaluation predictably suppresses high-uncertainty contributions while favouring low-risk, centrally aligned work. The resulting equilibrium gives rise to a constellation of behavioural adaptations, including upstream fragmentation, editorial conservatism, suppression of epistemic outliers, and convergence in both scientific content and narrative form.Unlike accounts centred on funding scarcity or individual risk aversion, this framework explains selective convergence as an outcome of routine evaluation dynamics rather than exceptional pressures or misconduct. By focusing on selection geometry rather than individual motivation, the Perspective provides a unifying interpretation of persistent patterns in contemporary scientific production and communication.