Affective, defective, and infective narratives on social media about nuclear energy and atomic conflict during the 2022 Italian electoral campaign

The study examines how disinformation spreads across social media during the 2022 Italian electoral campaign, focusing on the nuclear narrative in two strands: civilian nuclear energy and fears of atomic escalation in the Russo-Ukrainian war. In an environment where anyone can produce political content at scale and social media are pivotal for political communication, polarization and hyper-partisan media have fostered fertile ground for misinformation. Italy’s historical sensitivity to nuclear power (post-Chernobyl and post-Fukushima) and the concurrent geopolitical tensions increased the salience of the term “nuclear” during the campaign. Prior work shows Twitter’s outsized role in shaping perceptions despite representativeness concerns; insights can often transfer across platforms with similar affordances. This study adopts a cross-platform lens using the concept of “bridges” (from transmedia theory) to describe hyperlink-driven narrative connections that move audiences between platforms. Research questions: RQ1) What content is conveyed by bridges in the nuclear debate and for what purpose? RQ2) What role do bridges play in disinformation dissemination dynamics across the social media ecosystem? The study anticipates polarized, fragmented discourse, with bridges connecting to mainstream and below-the-radar platforms that may foster echo chambers and limit serendipity.

The paper situates its inquiry within several strands of literature: (1) Disinformation, polarization, and the role of social media in political communication and public opinion (e.g., Tucker et al. 2018; Stier et al. 2018; Wu et al. 2019). (2) Italian nuclear energy debates historically framed by major nuclear accidents and largely studied through traditional media or energy/environmental lenses (e.g., Contu et al. 2016; Bersano et al. 2020; Bernardi et al. 2018; Santarossa 1990; Standish 2009; Butler et al. 2011). (3) Social media research on post-Fukushima discourse (Japan-focused) often using text analysis, with Twitter as a key site for political discourse despite representational issues (Kim and Kim 2014; Tsubokura et al. 2018; Jungherr 2016; McGregor 2019; van Klingeren et al. 2021; Ellisona and Boyd 2013). (4) Cross-platform dynamics via hyperlinks and user account linkages that create interconnected media environments (Ryfe et al. 2016; Fu and Shumate 2017; Singer and Franklin 2019; Martin and Dwyer 2019; Park and Kaye 2020; Del Vicario et al. 2017; Gruzd et al. 2018; Amara et al. 2023), and digital methods approaches (Rogers 2019, 2021, 2023). (5) The transmedia concept of bridging (Hayes 2006; Bolter and Grusin 2000) reframed here to capture narrative devices that draw audiences across platforms. The study also draws on literature categorizing disinformation types and their effects, the role of fringe platforms, and echo-chamber/trench-warfare dynamics (Karlsen et al. 2017; Born and Edgington 2017; Sanovich et al. 2018; Douglas et al. 2019; Rogers 2021).

Design: Explanatory sequential mixed-methods. Quantitative detection of cross-platform hyperlinks (“bridges”), followed by qualitative content exploration and labeling to capture narrative nuances. Conceptual frame: Trench Warfare Dynamic for platform-specific interactions and affordances; “bridges” conceptualized as narrative-driven hyperlinks that intentionally move audiences between platforms. Data collection: Two Boolean queries were constructed—Q1 (nuclear references: nucleare/nucleari/atomico/atomica/atomici/atomiche) and Q2 (Italian elections terms, hashtags, and key political actors). The final query Q = Q1 ∩ Q2 to isolate nuclear-related discourse within the electoral context. Platforms and tools: Twitter data via 4CAT using Twitter Academic API; Facebook and Instagram via CrowdTangle (public pages, public groups, verified profiles only). Facebook/Instagram datasets were uploaded to 4CAT for unified preprocessing. Pre-processing (bridge detection): Filtered datasets to retain only posts containing URLs; exported to spreadsheets; manually flagged only URLs that linked to other social platforms. Destination categories included: mainstream social networking sites (e.g., LinkedIn, Reddit, TikTok, Facebook, Twitter, Instagram), non-mainstream social networks (e.g., Mastodon, Gab, 4chan, VK), instant messaging (Telegram), video-sharing (YouTube, Twitch, Bitchute, Rumble), and landing pages (e.g., Linktree). Qualitative data enrichment: Manually explored bridged content to add labels: content type; destination platform and account; user category on the source platform (politic, news, alternative news, activist, generic, comedian, music); topic and position (e.g., Nuclear Energy pro-nuke vs no-nuke; atomic conflict positions such as Anti-NATO, Pro-Russia, Anti-USA). Categorization was iterative based on emergent information. Visualization: Labeled data imported into Cortext (Network Mapping) to map source users/channels to destination users/channels; exported GEXF and visualized in Gephi. Instagram-specific handling: Because Instagram does not allow clickable links in captions, bridges were primarily identified via “link in bio” patterns, often pointing to landing pages or profile hubs that create star-topology networks across platforms.

- Dataset and bridge counts: 660 bridges analyzed across Twitter, Facebook, and Instagram, pointing to secondary platforms (YouTube, LinkedIn, Mastodon, Telegram, etc.). A platform can be primary or secondary depending on the source of the bridge. - Twitter bridges: Main destinations were Facebook (56.88%), YouTube (26.67%), and Mastodon instances (sociale.network and mastodon.uno) at 8.86%. Remaining bridges led to mainstream (LinkedIn, Instagram) and niche platforms (e.g., Friendica). Topic split: nuclear energy (81%), atomic conflict (13.1%), mixed (5.9%). Disinformation on Twitter skewed toward hyper-partisan and unverified information; conspiracy theories were less prevalent than expected. - Facebook bridges: Destinations were more diverse, led mainly to YouTube (59.78%), with Instagram, Telegram, and Twitter each around 11.75%. Minor destinations included TikTok, Twitch, and fringe platforms such as Mastodon (sociale.network), Rumble, and Sfero. Disinformation included conspiracy content alongside hyper-partisan and unverified information, with a more varied ecosystem oriented to below-the-radar platforms. - Instagram bridges: Identified mostly via “link in bio,” often to landing pages that aggregate multiple destinations (star topology). Cross-posted accounts largely maintained consistent editorial lines across platforms, limiting exposure to qualitatively different content in the crossing step. - Nuclear energy debate: Strong polarization among politicians and activists. Pro-nuke (e.g., Lega Salvini Premier, Noi con Salvini) framed EU taxonomy as a victory, used affective rhetoric (e.g., comparing environmentalists to “Taliban”), and included factually inaccurate claims (e.g., fourth-generation nuclear as a near-term bill-lowering solution). No-nuke actors (e.g., M5S, Alternative, PD, Angelo Bonelli) countered with affective messaging (e.g., risks to “children’s future”) and at times shared inaccurate content (e.g., a false map of potential plant sites, later debunked by PagellaPolitica). Activists amplified both sides. - Bridges to Mastodon: Though not numerous, they enabled cross-posting to instances like sociale.network, opening parallel discussions with highly polarized no-nuke sentiment, creating echo-chamber-like environments by design. - Video-centric bridging: Facebook and YouTube were key conduits. Content included political interviews, satirical and ironic videos, and channels promoting pro- or anti-nuclear stances. - Conspiracy and atomic conflict narratives: Many YouTube-oriented bridges led to channels expressing pro-Russia and anti-NATO positions, hyper-partisan commentary on events (e.g., Zaporizhzhia), or broader conspiracies (e.g., freemasonry/NWO). Some content mimicked TV news formats to enhance credibility. When videos were removed from YouTube for policy violations, they were often reuploaded to Rumble or Bitchute, described as “free from censorship.” - Messaging platforms: Telegram channels (e.g., Lantidiplomatico ~90,000 subscribers) spread unverified rumors rapidly; posts showed heavily negative reaction distributions (e.g., 93% vomit/anger reactions). Bridges often pointed to entire channels rather than specific posts, fostering echo-chamber dynamics and anonymity of content creators. - Ecosystem complexity of alt-news: Profiles interconnected across multiple mainstream and fringe platforms (e.g., YouTube, Rumble, Bitchute, Gab, MeWe, Telegram), seemingly to maximize virality (infectivity) and persist despite moderation on mainstream platforms. - Overall characterization: Bridges commonly conveyed affective (emotion/fear-based), defective (partisan, unverified, or false), and potentially infective (viral-prone) content. Approximately half of destination platforms were under-the-radar (e.g., Sfero) or video/messaging platforms (YouTube, Rumble, Telegram), which have comparatively looser content controls.

Findings address RQ1 by showing that bridges predominantly transmit affective rhetoric (emotional and fear-based appeals), defective information (partisan, unverified, or false claims), and potentially infective content designed for virality via multiplatform dissemination. Examples include political hyperbole regarding nuclear energy, debunked maps of potential plant sites, and alarmist atomic-conflict narratives. For RQ2, bridges function as communicative instruments that: (a) disseminate preferred narratives (including propaganda, hyper-partisanship, conspiracy); (b) augment messages using auxiliary content (e.g., music) and TV-like formats; and (c) shepherd audiences into favorable environments (echo chambers) with platform affordances that reinforce the message (e.g., Mastodon instances or Telegram channels with polarized reactions). Bridges also facilitate migration to fringe platforms with lower moderation, enabling persistence and amplification despite takedowns on mainstream platforms. This cross-platform ecology undermines institutional trust and may contribute to political disengagement and extremism. The results underscore that platform affordances and inter-platform linkages shape exposure and engagement, that below-the-radar platforms play outsized roles in sustaining disinformative ecosystems, and that bridging can reduce serendipity by directing users into ideologically homogeneous spaces.

The study maps cross-platform “bridges” in the 2022 Italian electoral discourse around nuclear energy and atomic conflict, revealing polarized, affective, defective, and potentially infective narratives propagated through hyperlinks across mainstream and fringe platforms. Methodologically, it contributes a hybrid, elastic cross-platform technique grounded in digital methods and a transmedia perspective that conceptualizes bridges as narrative devices. Substantively, it highlights the prominence of video-sharing and messaging platforms and the strategic role of fringe environments in sustaining disinformation. Policy-wise, it argues for attention to below-the-radar platforms, noting gaps in current regulatory frameworks (e.g., DSA’s limited coverage) and suggesting early-stage scrutiny of emergent fringe spaces. Future research could extend multimodal analyses (especially audiovisual content), incorporate private/closed-group data where ethical and feasible, and test interventions at inflection points where cross-platform spread can be reduced.

- Data access: Facebook and Instagram data limited to public pages, public groups, and verified profiles per Meta’s policies; private user content not included. - Platform affordances: Instagram’s audiovisual focus limits textual analysis; post–July 31, 2022, videos under 15 minutes were classified as Reels and not supported by CrowdTangle APIs, reducing available Instagram video data. - Generalizability: The specific discourse and period may limit generalizability, though observed discursive strategies likely transfer to other contexts. - Qualitative labeling: Iterative manual classification may introduce coder subjectivity despite systematic procedures.