A paradox of deliberative science communication under bounded rationality: insights from audience-initiated science journalism

This study examines how deliberative science communication operates in everyday journalism under conditions of bounded rationality and proposes a framework for understanding satisficing forms of deliberation in mainstream science journalism. Empirically, it draws on a year-long deliberative initiative, Fakt teadust päevas (“A Science Fact a Day”), co-created with the Estonian tabloid Õhtuleht , where one science-related fact was published daily across print, online, and social media based on questions from both scientists and the public. A qualitative content analysis examined how audience questions and expert responses were brought into relation, focusing on topic selection, epistemic framing, contributor roles, and alignment with deliberative principles, complemented by analysis of editorial correspondence, interviews, and the authors’ observation notes. Findings show that audience questions cluster around four orientations: practical uncertainty, cognitive curiosity, collective risk, and identity-related meaning-making. Scientists’ responses do not simply mirror these concerns, but systematically translate them into expert-defined explanatory frames through closure, bounded completeness, diagnostic assessment, or conceptual boundary-setting. While reason-giving is strong across all groups, inclusivity and reciprocity remain weak, resulting in asymmetrical, one-way engagement. Professional role strain, project-based time scarcity, fragile institutional infrastructures, academic autonomy norms, risk aversion, and limited discursive competence are key factors constraining deliberation. Building on these findings, the study proposes a framework of satisficing deliberative science communication that conceptualises deliberation as a situated, resource-dependent practice emerging from the interaction between audience orientations, expert response modes, and infrastructural conditions. Deliberation appears as partial, bounded, and locally optimised rather than fully realised.