The anniversary moment: why so many retrospectives now
A striking number of these papers are explicitly occasioned by anniversaries — Social Media + Society’s twentieth-anniversary issue produces Boyd2026-op, Marwick2026-ss, Baym2026-tr, and Tornberg2026-lc in close conversation, while Helmond2026-ll revisits “platformization” a decade after coining it and Bastos2025-ya eulogizes Twitter’s open-API era as a closed chapter of research infrastructure. This clustering is itself a symptom: a generation of scholars who built the vocabulary of social media studies — networked publics, participatory culture, platformization, surveillance capitalism — is now watching that vocabulary come apart from the objects it was built to describe. The shared move across these essays is diagnostic rather than celebratory: each asks not “what is social media doing to us” but “has social media, as a category, ceased to exist in the form we theorized it.”
From “social” to “parasocial” to “post-social”
The most direct expression of this rupture is terminological. Boyd2026-op argues that the “social” in social media is now a misnomer: feeds reward scrolling over posting and one-sided attention to influencers over reciprocal exchange, and she proposes retiring the term in favor of “parasocial media.” Tornberg2026-lc generalizes this further, arguing that the entire social-media paradigm — built on user-generated content circulating through a social graph — is dissolving into three distinct “post-social” formations: algorithmic broadcasting, semi-private micro-communities, and AI-mediated chatbot communication, each requiring different analytic tools. Marwick2026-ss supplies the autoethnographic counterpoint, using her own LiveJournal years to show what reciprocity and intimacy actually felt like before platforms optimized for scale, and insisting that today’s alienation is a contingent political-economic outcome, not an inevitability. Helmond2026-ll triangulates these claims infrastructurally: Meta’s retirement of the Like button is read as the symbolic end of the social-graph paradigm and the pivot toward AI-driven inference, a shift she folds into a redefinition of platformization itself as a transcoding process extending now into “Big AI” and even space-based compute (Google’s Project Suncatcher). Together these four essays stage the same argument from four angles — terminological, infrastructural, affective, and methodological — that the platform configuration theorized circa 2010–2015 has given way to something else.
Political economy, unfinished business
A second cluster insists that what has changed is less the technology than the intensification of political-economic dynamics already visible a decade ago. Baym2026-tr explicitly re-audits her own 2015 diagnosis — wealth inequality, algorithmic opacity, precarious labor, data extraction — and finds all four worse, crystallized in the “broligarchy” and in generative AI’s appropriation of user-generated content as training fodder. Bruns2026-yv narrates Twitter’s specific arc from prosocial, flat-network openness to Musk-era “enshittification,” using Doctorow’s framework (also invoked by Boyd2026-op) to show how mismanagement and commercial pressure, not technological necessity, hollowed out a once-valuable public-debate infrastructure — and finds that would-be successors (Mastodon, Threads, Bluesky) inherit the same governance failures rather than solving them. Bastos2025-ya narrows the lens to research infrastructure itself: Twitter’s suite of APIs enabled an entire “Twitter studies” subfield, and its closure is a rupture for computational social science comparable to the platform ruptures these other essays describe for users. Petre2019-zh, though earlier, supplies the conceptual hinge connecting this cluster to platform governance studies proper: “platform paternalism” names how platforms moralize the boundary between legitimate and illegitimate visibility-seeking in ways that mask their own material interests — a dynamic that anticipates and informs De2026-ld’s account of “changecraft,” the discursive strategies (aestheticized infrastructure, continuity-framing, patchworking) through which Meta, TikTok, YouTube, and X narrate constant change as benign and participatory while consolidating power.
Scams, extraction, and the blurring of legitimacy
Swartz2026-zb pushes the political-economy argument into new territory by arguing that scams are not peripheral to platform economies but constitutive of them: the same anti-fraud infrastructures that govern “scam” also authorize surveillance and legitimate certain forms of capitalism (hustle culture, crypto, MLMs) over others. This essay reframes the “decline” narrative common to Marwick2026-ss and Boyd2026-op in economic rather than affective terms — what has eroded is not just reciprocity but the line between legitimate and illegitimate value extraction, with “network scams” replacing the dyadic con. Read alongside Petre2019-zh’s account of platforms policing “gaming” while gaming the system themselves, and De2026-ld’s changecraft, a consistent picture emerges: platforms have always needed a vocabulary of moral boundary-work (organic vs. artificial, scam vs. hustle, continuity vs. rupture) to legitimate extraction, and that vocabulary has only grown more elaborate as the underlying economy has become more exploitative.
Theoretical anchors: platforms before the reckoning
Two foundational texts, chronologically prior to the retrospective wave but filed here as its necessary baseline, show how far the conversation has traveled. Gillespie2010-as established that “platform” itself was always a piece of discursive positioning — a term whose computational, architectural, figurative, and political connotations let YouTube claim neutrality while making substantive editorial interventions. van-Dijck2018-up scaled this insight to “the platform society,” arguing platforms had penetrated the core of Western institutions (news, transport, health, education) and converted public-value questions into contests with corporate infrastructure. Both texts anticipated almost every move the 2026 retrospectives make: Gillespie’s skepticism about platform neutrality rhymes with De2026-ld and Petre2019-zh; van Dijck’s public-vs-private-value framing underwrites Marwick2026-ss and Baym2026-tr’s calls for infrastructures of care. What the newer essays add is the sense of an ending — that the platform-society configuration van Dijck described is itself now being superseded.
The AI society and the political economy of hype
That supersession is theorized most directly by Tornberg2026-im, which posits a tendential shift “from the platform society to the AI society”: network-effect monopolies give way to “cognitive lock-in,” predictive logics give way to generative ones, and platform-era struggles over visibility give way to “alignment politics.” This essay functions as connective tissue between the retrospective cluster and a set of papers specifically diagnosing AI’s political economy and discourse. Vertesi2026-lv argues that AI critique is systematically misdirected into five “decoys” (ontological, inevitability, disruption, safety, regulatory) that distract from the underlying “Project of AI” — financiers and corporations assembling networks of capital and infrastructure. Hepp2026-oi complements this with ethnographic texture, distinguishing Silicon Valley’s “loud futuring” (CEO hype) from the “quiet futuring” of Bay Area communities (Effective Altruism, Rationalism) that supplies its ideological reservoir. Stanusch2026-ec and Dodds2026-df examine how these imaginaries circulate discursively — the former mapping industry “premediation” and “preclusion” strategies during the Altman controversy, the latter theorizing AI hype in journalism as a systemic infrastructure rather than mere narrative, with journalists caught playing both hype-maker and hype-watcher. Weinbrand2026-sf grounds these dynamics in a concrete case, showing how Google’s AI Overviews are discursively fought over by Google, journalists, and SEO marketers, with users’ and environmental concerns conspicuously absent — a finding that echoes Dodds’s point about hype’s erasure of hidden costs and labor. Li2026-zv closes this arc from the audience side, finding that ordinary users experience “AI slop” as a fuzzy, contested, only partly negative category, complicating both platform-decay narratives and techno-hype accounts alike.
Democracy, disinformation, and trust at platform scale
A parallel strand extends the political-economy argument into democratic institutions. Bennett2018-ys established the “disinformation order” thesis — that strategic disinformation reflects institutional legitimacy crisis more than media technology — and Bennett2025-xs now updates it, arguing that platforms enable new “digital surrogate organizations” that pull conservative parties toward illiberalism, using Trump’s MAGA transformation as the central case. Kristensen2025-ni asks a related, narrower empirical question about whether alternative platforms themselves drive discursive polarization. Fletcher2026-lv provides the long-run empirical backdrop against which these claims should be read: trust in news has been declining since the 1980s — before the internet — and accelerated only around 2000, with the US a global outlier rather than a representative case, and with trust actually rising in unfree-press countries. This essay usefully complicates the declinist instincts of the anniversary essays by insisting on description over narrative, a methodological caution that resonates with Hegarty2026-qr’s argument, discussed below, that decline itself is a story that needs historicizing rather than simply accepting.
The problem of knowing what counts as evidence
Underwriting all of this is a quieter methodological crisis: how can scholarship make any of these claims when platforms control access to the data needed to test them? Heiss2026-qv names this directly, arguing that platform-academic entanglement structurally resembles tobacco- and pharma-industry capture of research, and that regulatory mechanisms like the EU’s DSA Article 40 are so far inadequate. Bastos2025-ya’s elegy for Twitter’s API ecosystem is the concrete историческое case of this abstract problem — the loss of a specific, unusually open data infrastructure that had made an entire subfield possible.
Decline as genre, not just diagnosis
Finally, Hegarty2026-qr stands slightly outside — and above — this whole cluster by treating “internet decline narratives” as an object of study in their own right. Its argument lands as a useful, even uncomfortable, gloss on nearly every other paper filed here: IDNs periodize a threshold of decline, assign blame to Big Tech, and imagine a lost golden age, but in doing so they risk reproducing corporational determinism, erasing user agency, and romanticizing an internet that was always exclusive, Western, and male-dominated. Read against this frame, Marwick2026-ss’s LiveJournal nostalgia, Boyd2026-op’s parasocial renaming, Bruns2026-yv’s Twitter elegy, and even Baym2026-tr’s “it’s bad, what now” all become instances of a recognizable cultural form — legitimate critiques, certainly, but also situated performances of generational loss that this topic cluster, taken as a whole, cannot simply narrate from the inside. The collective arc of these twenty-five papers is thus double: a genuine diagnosis of platformization’s exhaustion and reconfiguration around AI, scam logics, and democratic erosion — and, running alongside it, an emerging self-awareness that the retrospective genre itself demands the same critical scrutiny its authors direct at the platforms they mourn.