Authoritarianism in the Making: State–Public Narrative Dynamics as Performative Power in Digital Environments

Digital authoritarianism is commonly defined as the instrumental use of digital technologies by states to surveil, repress, and manipulate publics (e.g., Polyakova & Meserole, 2019; Pearson, 2024; Roberts & Oosterom, 2025), particularly in China and Russia (Lamensch, 2021; Shahbaz, 2018). Yet this framing treats technologies as neutral tools and authoritarianism as a fixed structural attribute, obscuring its dynamic and adaptive character. Drawing on political theory and governmentality studies that understand authoritarianism as emergent and continuously enacted (Glasius, 2018; Dean, 2025), we argue—using China as our case—that digital authoritarianism is performatively generated through iterative interactions among state discourse, platform architectures, and patterned public engagement. Although recent work shows that digital infrastructures can produce authoritarian effects even without coercion and within democracies (Feldstein, 2021; Yayboke & Brannen, 2020), prevailing approaches still rely on an instrumental logic that presumes pre-existing authoritarian intent. We instead conceptualize authoritarianism as a patterned practice and, drawing on performativity theory (Butler, 2004), as something brought into being through repeated platformed performances. Extending research on soft authoritarianism (Zhang, 2025; forthcoming), we reinterpret visibility governance (Brady, 2009), participatory persuasion (Repnikova & Fang, 2018), and affective alignment (Zhu & Fu, 2023) as performative processes through which legibility, authority, and emotional intelligibility are iteratively constituted within digitally structured environments. To examine these dynamics empirically, we analyzed 82,034 Weibo posts referencing Japanese politician Sanae Takaichi during a period of geopolitical tension, employing topic modelling, temporal and semantic analysis, and MRQAP to compare how state actors and ordinary users construct political narratives. State actors produced fewer, tightly focused topics, whereas users generated more expansive and fragmented themes. Temporally, user-initiated narratives often appeared concurrently with—or earlier than—state-led ones, showing that narrative initiation is not monopolized by the state. Semantic network analysis revealed clustered and lexically constrained state discourse alongside dispersed and diverse user discourse; MRQAP detected no significant structural similarity between the two, indicating that users do not simply replicate state discursive architectures. Rather than demonstrating top-down diffusion, these patterns expose a fundamentally heterogeneous, polyvocal communicative environment in which state and public discourse operate with substantial autonomy. These findings advance a reconceptualization of digital authoritarianism as environmental conditioning—the subtle calibration of plausibility, salience, and semantic possibility within platformed infrastructures. Authoritarian power emerges not through persuasion or mimicry but through contested performances of interpretive authority shaped by algorithms, platform design, visibility hierarchies, and discursive conventions. This reframing carries three implications: (1) authoritarianism should be understood as a performative dynamic rather than a regime attribute, and thus can materialize within democratic settings; (2) platforms function as generative sites where authority relations are continuously enacted and contested, as reflected in Weibo’s polyvocal ecology; and (3) digital authoritarianism is inherently unstable and contingent, as demonstrated by the heterogeneity of user discourse, the temporal autonomy of narrative initiation, and the weak semantic alignment between state and public narratives.