Algorithmic Gatekeeping And Consensus Fragmentation: How Platform Curation Shapes Public Discourse On Social Media News

Abstract

 In the era of algorithm-driven platform communication, social media platforms have become the core channel for the public to access news and information, and the algorithmic gate-keeping mechanism embedded in them is reshaping the logic of public discourse production and the basis of consensus building in an unprecedented way. Taking ‘algorithmic public sphere’ as the core analytical framework, this study adopts a mixed-methods research design, integrating systematic literature synthesis, computational text analysis, and critical discourse analysis, to systematically examine the discourse corpus (N≈1.2 million items) on three major platforms, namely Twitter/X, Facebook, and YouTube, on three types of issues, namely, political elections, public health, and climate change, between 2020 and 2024. 1.2 million) is systematically examined with the aim of revealing how platform curation mechanisms lead to the fragmentation of public consensus. The study finds that: first, algorithmic high push content is significantly higher in affective intensity than low push content (Cohen's d = 0.47-0.61), and conflicting discursive frames systematically dominate algorithmically prioritised distribution of content (43.7% on average), revealing the structural exclusion of engagement-oriented algorithmic logics from rational public discourse; second. Second, the discourse dissemination networks of the three platforms show a highly modular clustering structure (Q = 0.63-0.71), with the proportion of cross-community bridging connections ranging from 8.3% to 11.7%, indicating that the algorithmic gate-keeping mechanism objectively strengthens the tendency of the discourse community's closure; third, the ‘factual definitions’ of the same public issue in different discourse communities have a 4- to 6-fold difference in frame frequency, and the fragmentation of consensus has already occurred. Thirdly, there is a four to six times higher frequency of disagreement between different discourse communities on the ‘factual definition’ of the same public issue, and the fragmentation of consensus has gone beyond the level of emotional polarisation to the epistemological dimension. At the theoretical level, on the basis of critically inheriting Habermas's theory of the public sphere, this study defines the algorithmic gatekeeping mechanism as the ‘level of platform technology design’ not covered by traditional gatekeeping theories, and preliminarily constructs the analytical dimensions of the ‘differentiated platform discourse theory’. At the policy level, the findings of the study are of direct reference value to the regulation of algorithmic transparency, the reconstruction of journalistic professionalism, and the cultivation of digital citizens' algorithmic literacy under the framework of the EU's Digital Services Act