Towards a Post-Social Media Studies

Summary

This conceptual paper argues that the “social media” paradigm — built on user-generated content circulating through social graphs to form networked publics — has been hollowed out, and that digital media studies needs a corresponding paradigm shift. Törnberg and Rogers identify three interlocking dynamics dissolving the old configuration: a move from social-graph to interest-based algorithmic recommendation, generative AI flooding platforms with synthetic content, and user retreat into semi-private spaces. In its place they map three distinct “post-social” formations — algorithmic broadcasting platforms, semi-private spheres/micro-communities, and AI-mediated communication — each requiring its own concepts, methods, and normative apparatus. The intervention is diagnostic and programmatic rather than empirical, calling for a wholesale reorientation of the field.

Key Contributions

• A “post-social media” framework that names a rupture in the organizing assumptions of two decades of digital communication scholarship.
• A typology of three coexisting post-social formations, each analytically distinct: algorithmic broadcasting, semi-private spheres, and AI-mediated chatbot communication.
• Conceptual vocabulary updates: replacing “networked publics” with notions like “algorithmic publics,” “imitation publics,” and the “user-spectator.”
• A method agenda matched to each formation: flow analysis and rhythmanalysis for broadcasting platforms; ethnographic and bounded-place approaches for micro-communities; auditing and experimental designs for chatbot media.
• A normative framing centered on epistemic fragmentation, platform power, and the erosion of shared democratic infrastructure.

Methods

The paper is a theoretical synthesis. It reconstructs the foundational social-media paradigm (networked publics, participatory culture, platformization, surveillance capitalism), then diagnoses contemporary transformations through secondary evidence — including Meta’s own FTC antitrust filings, Pew and Gallup survey data, Elon University AI usage figures, and prior empirical work on declining posting rates — and develops a comparative analysis across the three emergent formations to derive methodological implications.

Findings

• Meta’s court-submitted data: only 7% of Instagram time and 17% of Facebook time involves content from friends/followed accounts — major platforms no longer function primarily as social networks.
• Recommendation systems privilege passive engagement signals (watch time, dwell, completion) over active social signals (likes, shares, follows).
• TikTok’s “For You Page” logic has driven industry-wide convergence toward algorithmic, interest-based feeds.
• Posting and viewing on Facebook and Twitter/X fell roughly 50% between 2020 and 2024.
• AI chatbot use (52% of US adults in 2025) now exceeds active posting on Facebook (35%), Instagram (20%), and Twitter/X (11%).
• Synthetic content scales effectively within engagement-optimized systems, with weak platform incentives to distinguish human from AI authorship.
• Established methods — network analysis, content analysis, API-based collection — are increasingly misaligned with post-social structures.

Connections

This piece serves as a programmatic umbrella for several adjacent strands in the register: it shares its diagnosis of declining posting and platform decay with Tornberg2025-ir and Bruns2026-yv, and its concern about AI saturating platform content with DeVerna2025-dl, Bouchaud2026-lr, and Schroeder2026-im. Its claim that chatbots constitute a new media form lacking shared texts or publics resonates with experimental and auditing work on LLM influence such as Hackenburg2025-dj and Le-Mens2025-qz, while its methodological pessimism about API-based research connects to ongoing data-access debates exemplified by Freelon2024-sc, Ohme2026-nv, and Rieder2025-ju.

Podcast

A research-radio episode discusses this paper: Listen