Disciplining the digital public: Platform mechanisms and the dynamics of emotional polarization Authors: Changrong Lu, Dan Qin, Jiali Huang, Bing Pang, Andrea Cioffi, Andrea Cioffi, Andrea Cioffi, Andrea Cioffi Published: 2026 ABSTRACT Digital platforms increasingly mediate political communication and public opinion formation, raising urgent questions about how Bigtech firms are associated with behavioral norm formation and affective dynamics. This study develops the concept of platform disciplinary mechanisms to examine how interface design, algorithmic recommendation, and interaction feedback correspond to behavioral standardization, cognitive dependency, and emotionally structured group differentiation across online communities. Using a platform-end user-community framework, we investigate whether technological conditions associated with platform governance correspond to systematic changes in emotional polarization within the platform-mediated information environment. We conduct a time-series analysis linking a daily emotional polarization indicator derived from aggregated news sentiment (GDELT) to technology-category intensity measures extracted from Bigtech research outputs indexed in the Web of Science. Because the polarization indicator is based on news sentiment rather than user traces, it captures emotional polarization in the information environment rather than individual psychological states. Treating 2020 as a theory-informed analytic breakpoint, we find that emotional polarization becomes more persistent after 2020 and is more strongly associated with higher levels of technological simplification and emotion-oriented interaction dynamics. These findings are consistent with the view that major platforms are linked to changes in public-opinion dynamics in platform-mediated environments characterized by affective governance. They also suggest that regulation should pay greater attention to structural mechanisms of behavioral normalization and affective manipulation beyond content moderation alone. 1 Introduction As the world enters the era of Internet 3.0, the Internet has profoundly reshaped the patterns of social interaction and development, bringing diverse forms of information and exchanges into a unified medium [1]. Bigtech firms, exemplified by Apple, Amazon, Microsoft, Google, and Facebook, have become important loci of power in information distribution, user guidance, and social organization, owing to their deep mastery of platforms, data, and algorithms [2]. Since the 2016 U.S. election, these companies have become increasingly involved in political agenda‑setting and emotional mobilization through personalized recommendations, precision targeting, and topic manipulation, becoming pivotal institutional actors shaping democratic public opinion [3,4]. A growing body of research has systematically revealed how Bigtech transforms “users” and “engagement” into commercial assets, influencing attention allocation, behavioral choices, and even value identification through algorithmic optimization and interface design [5–7]. Such practices have not only raised global concerns over user privacy but, due to their escalating political influence, have also sparked widespread worries about platform regulation, civil liberties, and democratic stability [8]. The January 6, 2021, Capitol insurrection underscored this trend: protesters relied heavily on social media platforms to disseminate false information, exacerbating social polarization and political extremism [9,10]. This phenomenon is closely linked to emotionally driven algorithms designed to boost engagement by amplifying emotional reactions, thereby facilitating the spread of extremist views and posing a latent threat to democratic institutions [11,12]. Notably, the dual embedding of technology into political structures and social interactions has not consistently delivered the anticipated benefits of information pluralism and public rationality. Recent research on digital transformation highlights that such transformations often involve fundamental changes to organizational strategies, value chains, and structural mechanisms, beyond incremental technological upgrades. In particular, studies investigating the adoption challenges of Industry 4.0 for sustainable digital transformation argue that digital technologies reshape business models and organizational structures in profound ways, affecting resource flows, collaboration, and socio‑technical interactions [13]. This literature underscores that digital transformation should be viewed not merely as an efficiency tool but as a structural process influencing governance, coordination, and institutional arrangements. Instead, while ostensibly supporting diverse expression, digital platforms have in practice intensified information filtering and echo‑chamber effects [14,15]. This paradox raises a set of critical theoretical questions that have become central to political-social inquiry in the digital age. Digital transformation should not be understood solely as an efficiency-enhancing technical shift. Recent scholarship emphasizes that the digital transition should not be viewed merely as an Industry 4.0 efficiency upgrade, but rather as a fundamental reorganization of the social order that directly impacts social cohesion and human-centric governance, mirroring the concerns raised in the transition toward Society 5.0 [16]. In this broader sense, Bigtech platforms are not merely channels for information delivery but part of a wider structural transformation that links technological change to social coordination, public order, and sustainability-related concerns. Why do platforms explicitly designed to expand connectivity and exposure to diverse viewpoints increasingly coincide with opinion homogenization, emotional polarization, and social fragmentation? How do Bigtech firms, while actively encouraging participation and interaction, shape the boundaries of public discourse and the emotional conditions under which political communication unfolds? These questions challenge prevailing assumptions about technological openness and require closer examination of the mechanisms through which platforms structure collective attention and affective alignment. Much of the existing literature approaches these issues from a technology-deterministic perspective, attributing Bigtech’s expanding socio-political influence largely to its control over rapidly advancing technologies, including internet infrastructure, algorithmic systems, smart devices, and the Internet of Things [17]. Within this framework, users are largely conceptualized as passive adaptors whose behavioral patterns and cognitive orientations are shaped by technological architectures that operate beyond individual agency. At the individual level, a substantial body of research has examined how digital technologies intervene in everyday behavior. Lupton [18] first argued that the widespread adoption of smart devices created the material conditions for “self-tracking,” enabling users to record, quantify, and adjust daily routines through wearable technologies, thereby producing increasingly standardized behavioral trajectories [19]. Similarly, IoT systems connect sensors and interfaces to facilitate real-time monitoring and data collection of user activities; through advanced analytics, platforms construct predictive behavioral models that subtly guide user decision-making [20,21]. While such technology-driven and often unconscious forms of self-monitoring may enhance efficiency and convenience, they can gradually undermine individual autonomy and pose latent risks to trust-based mechanisms within democratic societies [22]. At the collective level, scholars have focused on how algorithmic systems reshape patterns of opinion formation and expression by identifying user preferences, assigning classificatory labels, and mapping social trajectories at scale. Balkin [23] conceptualizes this transformation as the emergence of an “algorithmic society,” in which social valuation and behavioral space are increasingly encoded into platform architectures, enabling the organization of group behavior. By tracing digital footprints, platforms deliver highly customized content streams that narrow informational exposure, encourage homogeneous consumption patterns, and influence political cognition and behavioral choice [24]. Pariser’s [25] notion of the “filter bubble” captures how algorithmic personalization reinforces informational closure and resistance to heterogeneous perspectives, increasing susceptibility to misinformation. Building on this insight, DiFonzo et al. [26] advance the “group echo-chamber” framework, showing how prolonged exposure to consonant viewpoints cultivates exclusionary cognitive biases and intensifies emotional distance between social groups. Recent studies further demonstrate that within algorithmically driven attention economies, emotionally charged and extreme content gains disproportionate visibility, amplifying adversarial sentiment and deepening affective divisions in public discourse [27,28]. Moreover, in the domain of political participation, the diffusion of platform-based technological logic has transformed the Internet into a central arena for political mobilization and collective contestation. Kreiss and McGregor’s [3] theory of “instrumental transformation” highlights how network technologies have shifted from passive channels of communication to active mechanisms that shape political behavior, thereby restructuring both modes of participation and processes of decision-making. In this view, digital infrastructures do not merely facilitate political engagement but help reshape the strategic environment in which political actors operate. Woolley and Howard [29], as well as Vaccari, Chadwick, and O’Loughlin [30], further demonstrate that political bots and precision‑targeted advertising systems have become pervasive in campaigning and agenda‑setting, positioning Bigtech platforms as important architects of political discourse. At the same time, by leveraging their control over users’ data, these firms increasingly operate in domains traditionally associated with state governance. As a result, platforms have begun to challenge the state’s monopoly over social coordination and political order, emerging as influential agents in social governance [31,32]. In summary, scholarship on Bigtech grounded in technological-logic paradigms has systematically indicated how platforms guide individual behavior and cultivate group-level opinion homogeneity through algorithmic governance and large-scale data manipulation. This body of work has been instrumental in clarifying platforms’ influence over behavioral formation, cognitive orientation, and information filtering, thereby offering an important theoretical foundation for understanding how platform power intersects with democratic processes and public discourse. Despite these advances, however, three interrelated limitations remain unresolved. First, existing studies tend to privilege technological architectures and their external structuring effects while under-theorizing platform agency and strategic intent. In particular, insufficient attention has been paid to how algorithmic design, interaction rules, and content presentation are deliberately institutionalized to encourage users to internalize platform-imposed norms, values, and behavioral expectations [33]. Second, much of the literature remains analytically fragmented across levels of analysis. Research frequently focuses on individual-level preferences, cognitive biases, or emotional expression, yet lacks a systemic account of how interactions among platforms, users, and communities evolve over time. As a result, the dynamic coupling of behavioral dependency, cognitive convergence, and emotional resonance, and its transformation into identity-based dependence and group polarization—has not been fully theorized. Third, although platforms are widely recognized for connecting like-minded individuals and enabling diverse forms of debate [9], empirical evidence increasingly shows that high-frequency interaction combined with algorithmic curation generates discourse environments characterized by emotional intensity and internal homogeneity [34]. What remains insufficiently explained is the synergistic mechanism linking technological logic, emotional incentives, and community belonging—particularly how users gradually construct dependency-based identities at both cognitive and affective levels, thereby deepening inter-group divisions. To address these gaps and bridge platform studies with communication research on media effects, this paper integrates five closely related strands of scholarship. Agenda-setting research establishes how news coverage structures issue salience and affective tone in public attention [35]. Framing studies further show how evaluative and emotional cues embedded in news narratives shape interpretation and emotional response [36]. Work on affective contagion and affective publics demonstrates how emotionally charged content diffuses through networks and binds communities together [10,28,37]. Research on echo chambers and filter bubbles explains how algorithmic curation and homophily concentrate emotionally consonant content within communities, intensifying within-group cohesion and between-group polarization [25,37]. Finally, studies of recommendation systems reveal how algorithmic ranking and personalization amplify engagement-salient signals often emotional content, thereby reshaping exposure patterns and accelerating affective dynamics [24,27]. Integrating these strands clarifies the mechanism examined in this study. Affective signals embedded in news content (agenda-setting and framing) enter platform information flows, where algorithmic recommendation systems and interface affordances selectively amplify emotionally salient items. Through social contagion and homophilic sharing, these affective cues can be consolidated into community-level emotional identities, which in turn reinforce emotional polarization. This multi-level synthesis provides the conceptual and empirical justification for linking aggregated news-sentiment indicators to platform-mediated emotional dynamics, thereby grounding the GDELT-based empirical strategy within a coherent communication-theoretic framework. Methodologically, this study adopts a two-step empirical strategy. First, we construct an NLP-based technology categorization pipeline to extract platform-facing technological signals from Bigtech research outputs indexed in the Web of Science, capturing the evolving intensity of key technology domains relevant to platform design and governance. Second, we implement a time-series analytical framework that links these technology signals to an emotional polarization proxy derived from aggregated news sentiment in the GDELT database. To assess whether the relationship between technological development and emotional polarization varies across different platform governance logics, we estimate a phase-sensitive dynamic model incorporating a structural breakpoint. This design allows us to test whether shifts in disciplinary regimes are associated with changes in the persistence and transmission of emotional polarization over time. The remainder of the paper is organized as follows. The Literature Review section reviews the relevant scholarship and develops the theoretical framework, covering technical architecture and behavior shaping, disciplinary practices from panopticon to synopticon, dependency-based identity formation, and emotion-driven dynamics and polarization. The Methodology and Analysis section outlines the methodology and analytical strategy, including the research design, data sources from GDELT and Web of Science, the NLP-based classification pipeline, descriptive statistics, and unit root tests. The Results section presents the empirical findings, including breakpoint identification and VAR estimation. The Discussion section examines the implications of the findings, with particular attention to the interpretation of the observed patterns, the role of non-Western platforms, the methodological transparency of ChatGPT-assisted labeling, and the study’s limitations. The Conclusions, Implications, and Future Works section summarizes the study’s main contributions and discussing their implications for platform governance and future research. 2 Literature review 2.1 Technical architecture and behavior shaping 2.1.1 Division of front‑end and back‑end technologies in bigtech: User interaction and system support. Since 2008, Bigtech firms have constructed highly interactive digital ecosystems by integrating a wide array of digital technologies and informational tools [38]. Within these ecosystems, users occupy a dual position as both data generators and algorithmically mediated interaction subjects. Front-end technologies are directly user-facing and primarily responsible for shaping how users perceive, access, and engage with information. Through personalized recommendation interfaces, emotionally responsive interaction designs, and streamlined navigational structures, these technologies substantially reduce the cognitive costs of information search and decision-making, while simultaneously increasing user engagement, convenience, and affective attachment to the platform [39]. In contrast, back‑end technologies operate behind the scenes, often imperceptible to users—providing critical system support. Key components include big‑data analytics, behavioral‑prediction engines, distributed computing frameworks, and deep‑learning models [40]. Their primary functions are to process user data in real time, enhance algorithmic accuracy, and ensure the platform’s operational efficiency and security [41,42]. This clear division of labor between front‑end and back‑end systems underpins simultaneous improvements in user experience and platform performance (see Table 1). Consequently, these integrated platform architectures have evolved into foundational structural conditions that coordinate both micro-level behavioral trajectories and macro-level socio-economic transformations within the digital industry [43]. 10.1371/journal.pone.0342143.t001Table 1Comparative overview of front‑end and back‑end technologies.Technology TypeRepresentative TechnologiesFunctions and ObjectivesUser Impact and Community Interaction Front‑End Technologies Personalized recommendation; interface optimization; affective design; mobile applications; virtual assistantsEnhance user experience and interaction efficiency; strengthen dependency and emotional identificationIntensify in‑group interaction frequency and emotional resonance, making users more receptive to unified viewpoints Back‑End Technologies Big data analytics; behavior prediction; distributed computing; deep learning; database managementProcess user data in real time; optimize recommendation algorithms; support platform operationsIncrease precision of algorithmic control and content delivery within communities, reinforcing viewpoint homogenizationSource: Compiled by the authors based on Bhatt et al. [44], Miraz [45], and Sharma [46]. Comparative overview of front‑end and back‑end technologies. Source: Compiled by the authors based on Bhatt et al. [44], Miraz [45], and Sharma [46]. 2.1.2 Technological simplification: Standardizing user behaviors and patterning community interaction. Unlike traditional information recipients or independent content creators, users within this ecosystem act simultaneously as data consumers and producers. This dual role means that user behavior is no longer driven solely by personal volition but is continually shaped and optimized within the front‑end technological interaction framework. Front‑end technologies not only determine how end users access and consume information but also, through specific interaction designs, influence how they generate and express content. As back‑end systems’ capacity to analyze and regulate user data improves, behaviors become increasingly predictable and patterned, thus magnifying the effects of front‑end–driven technological simplification. Technological simplification refers to the extensive deployment of streamlined, intuitive interaction tools, such as “like” buttons and one‑click sharing, that markedly reduce users’ cognitive load in navigating operations, making decisions, and expressing emotions (e.g., emojis, fixed phrases), enabling complex tasks to be executed in fewer steps and enhancing both interaction efficiency and user experience [47,48]. However, despite apparent gains in convenience, simplification embeds users’ actions within standardized, modular, and automated structures, promoting behavioral homogeneity and diminishing expressive diversity [49]. Under this regime, platforms gain greater predictive power over user conduct, while users grow increasingly reliant on platform‑prescribed interaction patterns for content dissemination, curtailing opportunities for exploratory or diverse expression [31]. Consequently, the flow of information and modes of interaction within communities are profoundly altered. Communities such as interest groups or fan bases on social media serve as key sites of user interaction, and their internal dynamics and affective bonds heavily depend on simplification designs. Under technological simplification, emotional expression within these circles becomes more efficient, formulaic, and convergent. Recent research shows that, compared to complex forms of articulation, users tend to favor simplified emotional tools (e.g., emojis, set phrases, or like/report buttons) to convey sentiment, a mechanism that particularly amplifies negative expressions for their directness and efficiency [50]. When a topic elicits a shared emotion among community members (such as anger or discontent), simplification features further intensify “taking sides” behaviors [51]. Users are more inclined to adopt uniform labels, slogans, or memes to signal their stance rather than engage in nuanced debate. This “take‑sides” dynamic not only accelerates emotionally driven information spread but also transforms expression into a symbolic ritual detached from substantive discussion [52]. At its core, this phenomenon reflects an invisible logic of discipline. 2.2 Disciplinary practices: Technological governance from the panopticon to the synopticon 2.2.1 The transition from panopticon to synopticon. The concept of discipline was first articulated by Foucault to describe the systematic control and shaping of individual behavior within social institutions—such as law, education, and the military—so that conduct aligns with societal norms and ultimately yields widely accepted values and codes of conduct [53,54]. Disciplinary practices have traditionally been understood to operate in two modes: the panopticon and the synopticon, which some scholars regard as successive stages in the evolution of disciplinary regimes [55]. The panopticon model, derived from Bentham’s “Panopticon,” emphasizes how a small number of power holders continuously monitor a larger population, generating a felt sense of surveillance that compels individuals to self‑regulate in accordance with institutional rules [56,57]. In contrast, the synopticon model relies on mutual surveillance among many: individuals are visible not only to the platform but also to one another, experiencing an invisible but pervasive peer‑driven oversight [58]. With the rise of Bigtech, disciplinary mechanisms have become more pervasive and covert within online communities, shifting from panoptic to synoptic modalities. Online communities formed on the basis of shared interests and extended through social media and other digital channels exhibit hierarchical structures and tightly knit network characteristics [59,60]. In its early stages, Bigtech deployed front‑end algorithmic monitoring and personalized recommendations to shape user behavior unilaterally, manifesting a classic panoptic effect: users internalize platform rules as default norms when algorithms deliver highly targeted content, and they consciously conform to avoid reduced visibility or engagement penalties [61]. However, as user populations swelled and interactions diversified, the unilateral discipline of the panopticon proved less effective in managing complex, emotion‑driven group dynamics [62]. Consequently, Bigtech has increasingly adopted a synoptic approach, leveraging peer‑to‑peer mechanisms to enhance mutual supervision and self‑discipline. By embedding features such as “likes,” comments, and shares into the user interface, platforms enlist users themselves as enforcers of community norms. Kingsbury and Hong [63] observe that public feedback and collective pressure prompt individuals to adjust their behavior to align with the prevailing values and viewpoints of their peer group. Posts that deviate from community expectations often incur negative reactions, downgrading of content, or even exclusion from the circle. Although front‑end advances appear to decentralize oversight, they in fact strengthen emotional cohesion and behavioral conformity among users. In essence, whether operating as a panopticon or a synopticon, disciplinary practices deploy technology to subtly regulate user behavior, fostering unconscious acceptance of platform rules and the gradual formation of uniform social norms. The deeper impetus underlying these practices—namely, the cultivation of dependency‑based identities and affective control—provides a critical entry point for further analysis of disciplinary mechanisms (see Table 2). 10.1371/journal.pone.0342143.t002Table 2Comparative analysis of the panopticon and synopticon models.DimensionPanopticon ModelSynopticon Model Design & Concept A centralized watchtower surveils surrounding individualsA decentralized network of mutual public surveillance Power Structure Authority is concentrated in a central observerPower appears dispersed but remains centralized within the platform Mode of Surveillance Unidirectional, passive monitoringMultidirectional, active peer-to-peer monitoring Manifestation Few observe the manyMany observe the few Psychological Mechanism Individuals self-regulate out of fear of being watchedGroups supervise each other, prompting voluntary behavior adjustment Applicable Contexts Traditional institutions (prisons, schools, factories) and early Internet environmentsContemporary social media and online communities requiring public participation Means of Identity Formation Individuals perceive external authority and constrain themselvesMutual peer pressure and public opinion within the group shape identity Emotional Impact Extreme emotional responses are rareEmotional polarization is pronounced Fundamental Logic Cultivate dependency-based identity through behavior and thought control to achieve social order and managementSource: Compiled by the authors. Comparative analysis of the panopticon and synopticon models. Source: Compiled by the authors. 2.2.2 The deep motivation of discipline: Shaping dependency-based identity. Dependency-based identity refers to an individual’s or group’s sense of belonging that becomes highly bound and homogenized through continuous interaction and cognitive internalization within a particular environment. This identity consists of three core dimensions: behavioral dependence, cognitive dependence, and emotional dependence. These dimensions evolve sequentially and reinforce each other [64,65]. At the behavioral level, users initially adapt to platform rules through standardized interactions such as liking, sharing, and other simplified actions. This process leads to a uniform behavior pattern that diminishes individual differences and strengthens users’ dependence and loyalty toward the platform. Subsequently, based on this behavioral dependence, users progressively adopt the mainstream values of the platform, increasingly rejecting alternative information channels [25]. This cognitive homogenization enhances the platform’s control over the flow of information, allowing it to dominate the direction of social discourse. Emotional dependence, however, is the core of the disciplinary practice. Under the influence of front-end technologies’ emotional engagement mechanisms, users form a strong sense of community belonging, seeing the platform as an essential medium for emotional expression and social connection [66]. Notably, compared to the traditional Panopticon model, emotional drivers are more pronounced in the Synopticon model. As Bigtech shifts its disciplinary mode from the unidirectional surveillance of the Panopticon to the more group-interactive and peer-supervised Synopticon model, users’ emotional ties and collective emotional consensus are effectively strengthened [58]. Users’ emotional identification with the platform’s mainstream viewpoints makes them more likely to consciously support and propagate the topics defined by the platform, further consolidating the platform’s discursive power in public opinion [67]. Thus, emotional drive has become a critical byproduct of Bigtech’s shift in disciplinary practice. It extends the platform’s control beyond technological tools to emotional and psychological levels, making the regulation not only a matter of behavior but also of internalized emotional alignment. 2.3 Emotion-driven dynamics: The double-edged sword of unified public opinion and emotional polarization The analysis above highlights that the core motivation of Bigtech’s disciplinary practices lies in shaping users’ dependency-based identity, turning them into strong supporters of the platform’s discursive power [68–70]. This unified public opinion is heavily reliant on an emotion-driven logic. Front-end technologies, through emotional feedback mechanisms, enhance users’ emotional satisfaction and sense of community belonging, making emotion-driven interactions the norm within these groups. In this context, users increasingly neglect the diversity of information, and cognitive and emotional homogeneity within these groups grows, leading to irrational group psychology and behavior patterns [71,72]. This emotion-driven interaction logic explains the decline of rational expression and the rise of emotional dissemination in online spaces. Compared to traditional communication models, online platforms increasingly depend on emotional resonance and interaction to spread information. Negative emotions, such as anger, fear, and dissatisfaction, because of their high arousal and attentiveness, are more readily accepted and shared by users, thus exerting a more direct and powerful influence on individual attitudes and behaviors [37]. Negative emotions, such as anger and fear, trigger stronger physiological and psychological reactions and, therefore, gain more widespread attention and higher levels of participation on platforms [10]. Research has shown that the spread of anger, in particular, is rapid and widespread, contributing to what is referred to as the “digital anger” phenomenon, where users, due to the emotional stimulus, are more likely to engage in the propagation of anger [73]. The platform’s dynamic recommendation algorithms further amplify emotional interactions among users by adjusting content push strategies in real-time, thereby increasing the rapid spread of negative emotions on social media [74]. Under the Synopticon model, the spread of negative emotions is influenced by multiple interrelated mechanisms [75,76], such as multi-directional information flow and the “echo chamber” effect. These effects cause emotions to accumulate rapidly within the group, ultimately leading to emotional polarization [77,78]. Furthermore, the spread of anger has been shown to promote homogeneity effects, intensifying emotional division between groups [79]. This mechanism not only strengthens emotional homogenization within groups but also fosters cognitive biases and hostility between opposing groups [80]. Therefore, the preferential spread of negative emotions on social platforms is not merely the result of algorithmic biases but reflects human psychological structures’ inherent sensitivity to threat information, ultimately driving the spread of emotionally charged, polarized, and adversarial content. As the disciplinary model shifts from the Panopticon to the Synopticon, users develop stronger internal cohesion through group interaction and emotional resonance, but this also exacerbates emotional conflict between groups and shrinks the space for rational discussion. This transformation not only reshapes the structure of social public opinion but also poses a challenge to the long-term sustainability of Bigtech’s disciplinary mechanisms. In conclusion, Bigtech’s use of technological simplification and disciplinary practices has significantly contributed to the standardization of user behavior patterns and reinforced emotion-driven interactions, resulting in the emotional polarization in the platform-mediated public sphere (Fig 1). In sum, the theoretical framework specifies a set of directional and time-varying implications that cannot be settled by conceptual reasoning alone. In particular, the model suggests that (a) platform-facing technologies that lower expressive cost and heighten engagement-salient feedback should be associated with stronger affective divergence, and (b) once platform governance shifts from one-way monitoring to interaction-amplified peer surveillance, affective dynamics may become more self-reinforcing over time. These implications are empirically contestable: their magnitude, timing, and even sign may vary across technological domains and institutional phases. We therefore move to an empirical strategy that tests the model’s observable implications by linking technology-category signals to a time-series measure of emotional polarization, and by examining whether the estimated relationships differ before and after a structural transition in platform disciplining mechanisms. Theoretical framework of Bigtech’s disciplinary mechanisms. Source: Created by the authors. 3 Methodology and analysis This section tests the observable implications of the platform disciplining mechanisms framework in a time-series setting. Rather than restating the research questions developed earlier, we focus on two empirical expectations derived from the theory. First, if platform-facing technological development strengthens disciplining capacity, shifts in technology-category intensity should co-move with changes in emotional polarization in the platform-mediated information environment. Second, if disciplining regimes differ, the technology-polarization relationship should vary across phases, with stronger associations after the hypothesized regime transition. To evaluate these expectations, we construct a matched daily time-series that links (i) an emotional polarization proxy derived from aggregated news sentiment signals and (ii) platform-facing technology-category intensity measures derived from Bigtech research outputs. We then estimate a dynamic model with a structural breakpoint and report baseline estimates alongside diagnostic checks. 3.1 Data collection and processing Building on the analytical framework established above, this section operationalizes emotional polarization and technology-category intensity in a manner consistent with the paper’s theoretical claims. The proxy for emotional polarization is derived from the GDELT Global Event Database, which provides large-scale, time-stamped sentiment information for news events worldwide. Specifically, we extract daily sentiment scores associated with reported events and compute the cross-event dispersion of sentiment on each day. The standard deviation of daily sentiment scores is used to capture the degree of affective divergence in the information environment, with higher dispersion indicating greater emotional polarization across contemporaneous news coverage. We adopt the cross-event standard deviation as the dispersion measure for three reasons. First, SD is the canonical second-moment summary of a distribution and aligns directly with the variance-based notion of affective divergence used in prior aggregate sentiment work [81]. Second, GDELT’s sentiment scores are continuous and approximately symmetric around zero, conditions under which SD is informative and not dominated by skewness artifacts that would otherwise motivate Gini- or IQR-based alternatives. Third, the first-differencing operation we apply (yielding dstd) and the additional winsorized and spike-excluded specifications reported in the robustness checks below already absorb the principal robustness concerns that Gini/IQR would address—namely, sensitivity to extreme values. We acknowledge that fully replicating the analysis with rank-based dispersion (Gini, IQR) would provide additional reassurance and identify it as an avenue for future extension. During data preprocessing, observations with abnormal timestamps, duplicated records, or insufficient daily event counts were excluded to ensure the reliability of the dispersion measure. Specifically, we define “insufficient daily event counts” as days with fewer than 500 events, which are excluded to stabilize the dispersion-based sentiment measure and avoid volatility driven by extremely small samples. After cleaning, the final dataset consists of 2,475 daily observations, covering the period from February 2015 to December 2021. The date-indexed daily series used to construct the empirical variables are provided in S1 Data; the first-difference polarization proxy (dstd) and model-ready inputs are reproducibly generated from these series using S1 Code. This time span captures both the expansion phase of platform-centered political communication and the subsequent period in which emotionally driven interaction dynamics became increasingly salient, thereby providing an appropriate temporal window for assessing the structural impact of platform disciplinary mechanisms on emotional polarization. Although GDELT captures event-level and media-level sentiment rather than the emotional states of individual social-media users, there are well-established theoretical and empirical pathways that justify its use as an indirect indicator of platform-amplified affective dynamics. Classical agenda-setting and framing research demonstrates that the emotional valence embedded in news coverage shapes issue salience and the interpretive frames available to audiences, thereby influencing subsequent perception and evaluation [35,36]. In the platformized media environment, digital platforms no longer merely coexist with news media but function as algorithmic aggregators and redistributors of news content: headlines, frames, and affective cues originating in the news ecosystem are selectively prioritized, reformatted, and recirculated through personalized feeds and recommendation systems [3,30]. A growing body of experimental and large-scale observational research further shows that emotionally valenced content seeded in news flows readily enters social media streams, where it affects user attention, emotional response, and engagement patterns [10,28]. These findings support a mediated transmission pathway in which affect travels from news production to platform curation and subsequently to user interaction. Taken together, the literature implies a sequential mechanism—news emotion → platform aggregation and recommendation → user exposure → affective contagion and community consolidation, through which media-level affective signals become socially consequential. Under this mediated pathway, systematic changes in aggregated news sentiment (as captured by GDELT) can be interpreted as a co-moving indicator of the affective inputs that platforms ingest, amplify, and circulate. Accordingly, changes in aggregated news affect (as captured by GDELT) are interpreted here as shifts in the affective inputs circulating in the platform-mediated information environment. Importantly, this proxy does not measure individual users’ psychological states; it captures variation in affective tone at the media level that platforms can ingest and amplify through ranking and redistribution. Technological categorization data were retrieved from the Web of Science platform, limiting the search scope to papers published by five tech giants—Apple, Amazon, Facebook, Google, and Microsoft—between 1975 and 2023. To match the time span of the emotional polarization data, only papers published from 2015 to 2021 were retained, with precise publication dates supplemented through web scraping. To align publication outputs with the daily GDELT timeline, we aggregate category-level publications into daily counts and construct daily technology-intensity series. Because publication events are sparse at the daily level, we apply a k-day moving-average smoothing (k = 7) to reduce day-to-day zero-inflation noise in the intensity series, rather than imputing publication dates. This procedure yields a complete dataset containing 49,775 paper abstracts. Because WoS-indexed publications are publicly visible research outputs rather than direct records of internal technology deployment, the technology-intensity series should be interpreted as research-output signals of broader technological conditions. Published research may lag internal R&D or deployment decisions by one to three years; therefore, the observed post-2020 associations may partly reflect R&D trajectories initiated earlier and subsequently becoming visible in the publication record. We thus interpret these variables as indicators of evolving platform-facing technological emphasis, rather than as immediate measures of feature deployment. We further note that prior work using platform-trace emotion data has consistently identified emotional amplification and polarization as platform-level dynamics on social media. For instance, Brady et al. [82] showed that moralized emotional language on Twitter predicts higher diffusion and engagement, indicating how emotional content shapes network propagation dynamics. Bail et al. [83] found that exposure to opposing views on social media platforms such as Twitter or Facebook can paradoxically increase affective polarization rather than reduce it. Rathje et al. [84] demonstrated that content expressing out-group animosity on Facebook and Twitter strongly predicts engagement metrics, suggesting that platform architectures inherently amplify certain emotional expressions. Compared with these proprietary Twitter or Facebook emotion datasets, GDELT’s news-sentiment dispersion proxy operates upstream of within-platform engagement traces: it captures the affective tone of news inputs that platforms ingest and redistribute rather than the user‑level engagement signals themselves. Relative to proprietary trace-level emotion indicators, GDELT offers public availability, longitudinal continuity, and replicability, at the cost of greater distance from individual user affective states. We view this as a deliberate analytical trade-off rather than a substitute for trace-based studies. Ethics statement. This study used aggregated media-level data from GDELT and bibliographic research records obtained through Web of Science. It did not involve the recruitment of or interaction with human participants, the use of human biological materials, or access to identifiable private personal data. Accordingly, institutional ethics approval and informed consent were not required. 3.2 Technology categorization and time-series construction We identified Bigtech-affiliated publications in Web of Science using a predefined firm list and affiliation-based search strings. Records were then mapped into ten technology categories using a two-stage NLP pipeline that combines unsupervised clustering with active-learning-assisted labeling. Category assignments were subsequently aggregated into a time-aligned technology-intensity series matched to the GDELT daily timeline. This construction is intended to capture temporal variation in the prominence of platform-facing technology domains in Bigtech firms’ research portfolios. The NLP procedure treats each article abstract as one text unit. After text preprocessing, abstracts were converted into SentenceTransformer-based semantic embeddings. UMAP dimensionality reduction was then applied before clustering. The initial technology categories were generated using MiniBatchKMeans, with the final number of clusters set to 10 (k = 10). This cluster number was selected to balance thematic granularity, category interpretability, class size, and alignment with the theoretical distinction between user-facing and infrastructural platform functions. TF-IDF or class-based TF-IDF keywords were used to interpret the clusters rather than as the primary clustering features. To refine the initial categories, we implemented an active-learning-assisted labeling procedure. In each round, a small set of low-confidence abstracts, defined by the lowest predicted category probabilities or ambiguous category membership, was selected for review. The active-learning process was conducted over six rounds, with approximately 30 low-confidence abstracts reviewed per round. ChatGPT was used as an auxiliary labeling and category-interpretation tool in the later rounds, following a pre-specified labeling guideline and prompt template. The authors reviewed the category interpretations and used the reviewed labels to improve the classifier. ChatGPT was not treated as an independent human coder; rather, it functioned as a model-assisted review tool for boundary cases and category interpretation. The final categories were interpreted using representative keywords, sample abstracts, and the active-learning review process. For example, Type0 captures security, privacy-preserving computation, and trust infrastructure; Type3 captures runtime systems and dynamic software infrastructure; Type4 captures social platforms, search, and user activity; and Type5 captures image processing and visual-content labeling. These categories were then mapped onto platform-disciplinary functions as an interpretive bridge between Bigtech research-output signals and the theoretical framework. The full Web of Science retrieval strategy, screening protocol, feature-construction procedure, clustering settings, prompt template, validation notes, and technology-category mapping are provided in S1 Text. The date-indexed daily series used to construct the empirical variables are provided in S1 Data. All code used for data processing, technology categorization, time-series construction, and statistical analysis is provided in S1 Code to ensure reproducibility. 4 Results 4.1 Descriptive analysis and stationarity test Based on the keywords generated from clustering (see Table 3) and Rodríguez and Ortún’s [38] front-back technology framework, the 10 technology categories are provisionally interpreted as two major dimensions: 10.1371/journal.pone.0342143.t003Table 3Ten technology categories: Representative keywords, front-/back-end orientation, and primary links to platform disciplinary mechanisms.Technology categoryRepresentative keywords (TF-IDF)OrientationPrimary disciplinary mechanismType 0 — Privacy, security & trust infrastructuresecurity, integrity, malicious, secure inference, MPC, protocolFront-endCognitive dependencyType 3 — Runtime & dynamic recommendation systemstype system, polymorphic, runtime, dynamic, parameterizedFront-endBehavioral standardizationType 4 — Social platform, search & user activitysocial, platform, user, activity, search, information-seekingFront-endCognitive dependencyType 5 — Image processing & visual labelingimage, pixel, labeling, visual, CRF, neighborhoodFront-endEmotional differentiationType 1 — Query optimization & workload infrastructurebig data, workload, optimizer, complex query, decisionBack-endBehavioral standardization (supporting)Type 2 — Neural-network modeling & tuningDNN, activation, hidden layer, neural, parameterBack-endCognitive dependency (supporting)Type 6 — Access control & policy managementsecurity, user, transaction, control, policy, accessBack-endBehavioral standardization (supporting)Type 7 — Code & computational-method engineeringcode, pairwise, geometry, experiment, boundaryBack-endBehavioral standardization (supporting)Type 8 — Motion & spatiotemporal signal processingmotion, velocity, spatiotemporal, phase, signalBack-endEmotional differentiation (supporting)Type 9 — Ad auctions & attention allocationauction, slot, allocation, position, revenue, clickBack-endEmotional differentiation (supporting)Note. Categories are derived from a SentenceTransformer–UMAP–MiniBatchKMeans (k = 10) pipeline applied to 49,775 Bigtech-affiliated Web of Science abstracts; keywords are top TF-IDF/ class-based TF-IDF terms per cluster (S1 Text). For readability, each category is mapped to a single primary mechanism among behavioral standardization, cognitive dependency, and emotionally structured differentiation; “(supporting)” denotes back-end categories that enable rather than directly enact a mechanism. The full keyword lists, representative abstracts, and the complete category-to-mechanism mapping are provided in S1 Text. Ten technology categories: Representative keywords, front-/back-end orientation, and primary links to platform disciplinary mechanisms. Note. Categories are derived from a SentenceTransformer–UMAP–MiniBatchKMeans (k = 10) pipeline applied to 49,775 Bigtech-affiliated Web of Science abstracts; keywords are top TF-IDF/ class-based TF-IDF terms per cluster (S1 Text). For readability, each category is mapped to a single primary mechanism among behavioral standardization, cognitive dependency, and emotionally structured differentiation; “(supporting)” denotes back-end categories that enable rather than directly enact a mechanism. The full keyword lists, representative abstracts, and the complete category-to-mechanism mapping are provided in S1 Text. (1) Front-End Technologies (Type 0, Type 3, Type 4, Type 5): These technologies are more directly related to user-facing interaction and content presentation. For instance: (1) Front-End Technologies (Type 0, Type 3, Type 4, Type 5): These technologies are more directly related to user-facing interaction and content presentation. For instance: Type 0 is associated with data privacy and encryption-related functions that may enhance user trust; Type 0 is associated with data privacy and encryption-related functions that may enhance user trust; Type 3 and Type 4 are associated with recommendation- and platform-mediated interaction patterns that shape emotional expression; Type 3 and Type 4 are associated with recommendation- and platform-mediated interaction patterns that shape emotional expression; Type 5 is associated with image-processing functions that may strengthen the emotional rendering of content. Type 5 is associated with image-processing functions that may strengthen the emotional rendering of content. (2) Back-End Technologies (Type 1-2, Type 6-9): These technologies are more closely related to the platform’s underlying computational logic and support functions. (2) Back-End Technologies (Type 1-2, Type 6-9): These technologies are more closely related to the platform’s underlying computational logic and support functions. Type 1 is associated with big data optimization Type 1 is associated with big data optimization Type 2 with neural network modeling and parameter tuning Type 2 with neural network modeling and parameter tuning Type 6 with user behavior and analysis; and Type 6 with user behavior and analysis; and Type 9 with advertising and allocation algorithms. Collectively, these technologies improve data-processing efficiency and predictive accuracy, and may indirectly shape how emotional content is selected, prioritized, and circulated. Type 9 with advertising and allocation algorithms. Collectively, these technologies improve data-processing efficiency and predictive accuracy, and may indirectly shape how emotional content is selected, prioritized, and circulated. The keyword clusters in Table 3 provide the empirical basis for this interpretive grouping, while the front-end/back-end framework offers the conceptual rationale for distinguishing user-facing functions from system-support functions. The classification is therefore intended as an analytically useful and theory-informed grouping, rather than a claim of exhaustive functional equivalence across all categories. To improve readability, Table 3 provides a compact overview of the ten technology categories, representative keywords, and front-end/back-end orientation; the full mapping to the three disciplinary mechanisms of standardization, dependency, and differentiation is documented in S1 Text. Based on the above analysis, we obtained the descriptive statistics table for the relevant variables, as shown in Table 4. 10.1371/journal.pone.0342143.t004Table 4Descriptive statistics of variables.VariableObsMeanStd. Dev.MinMax dstd 24750.007−.071.22 type0 2475.522.27.061.924 type3 2475.555.292.1612.003 type4 2475.575.243.0751.847 type5 2475.275.22701.214Source: Compiled by the authors. Descriptive statistics of variables. Source: Compiled by the authors. Descriptive statistics show that the mean of the emotional polarization indicator (dstd) is zero, with relatively stable fluctuations (standard deviation = 0.007), but the range of extreme values is large (−0.071 to 0.22), implying localized emotional disturbances. Among front-end technologies, Type 4 (social platforms) has the highest and most stable application intensity (mean = 0.575), while Type 3 (dynamic recommendation) shows significant fluctuations in activity (standard deviation = 0.292, maximum value = 2.003). Type 5 (image processing) experiences periods with zero values (mean = 0.275), which may be related to the cycle of its privacy functions. The extreme value differences in back-end technologies (e.g., Type 0 maximum value = 1.924) reflect phase-specific deployment peaks. Through the ADF unit root test (see Table 5), it was found that the original emotional polarization indicator (std) is non-stationary (ADF = −1.168, p = 0.687), while its first-order difference series (dstd) passes the stationarity test (ADF = −40.195, p = 0.000). All technical classification variables (type0-type9) are stationary (p < 0.05), meeting the prerequisites for time series modeling. 10.1371/journal.pone.0342143.t005Table 5ADF unit root test results.VariableTest statistic5% critical valuep-valueStationary Std −1.168−2.8600.6872Not stationary dstd −40.195−2.8600.0000stationary Type0 −4.064−2.8600.0011stationary Type1 −4.373−2.8600.0003stationary Type2 −4.554−2.8600.0002stationary Type3 −4.692−2.8600.0001stationary Type4 −4.646−2.8600.0001stationary Type5 −4.041−2.8600.0012stationary Type6 −4.418−2.8600.0003stationary Type7 −4.420−2.8600.0003stationary Type8 −4.738−2.8600.0001stationary Type9 −4.509−2.8600.0002stationarySource: Compiled by the authors. ADF unit root test results. Source: Compiled by the authors. 4.2 Breakpoint identification and mechanism transition Analysis of user growth trends on major social platforms from 2012 to 2024 indicates that 2020 marked a key turning point: prior to this, platforms like Facebook, YouTube, and Reddit relied on a “panopticon” disciplinary mechanism, which monitored user behavior through centralized algorithms and implemented standardized content recommendations, rapidly attracting new users (see Fig 2). After this point, the user growth rate slowed to nearly zero, and platforms shifted to a “synopticon” model, relying on user interactions and group supervision to shape emotional resonance. User growth trends of major social platforms (2012-2024). Data Source: DataReportal (accessed 15 Sep, 2025), series compiled and visualized by the authors based on the reported platform user statistics. From a theoretical perspective, this shift can be attributed to the differences between the two disciplinary mechanisms: the Panopticon relies on one-way algorithmic control to lower the threshold for user participation but struggles to maintain long-term engagement; the Synopticon, on the other hand, requires users to actively adapt to the emotional rules within their circles (e.g., emotional expression and content preferences), making it harder for new users to integrate. Empirical analysis further reveals a significant increase in the persistence of emotional polarization after 2020 (as indicated by the rise in the Dstd L1 coefficient in the subsequent VAR model), suggesting that the Synopticon model amplifies the inertia of emotional transmission through the coupling of technological logic and user interaction. We treat 2020 as an analytic breakpoint motivated by a visible structural change in platform growth dynamics and the broader platform environment. This structural shift is not merely an empirical artifact but aligns with broader observations that digital transitions and algorithmic governance strategies are fundamentally reconfigured when confronting macro-level contexts of profound global uncertainty [85]. Our aim is not to attribute the breakpoint to a single historical event, but to evaluate whether the technology–polarization relationship exhibits phase sensitivity consistent with a regime-dependent disciplining process. We therefore estimate models separately across the pre-2020 and post-2020 phases and interpret differences as regime-contingent associations rather than causal effects. Substantively, 2020 coincides with at least four platform-relevant developments that jointly motivate its use as an analytic breakpoint: (i) the COVID-19 pandemic shifted public discourse heavily online and increased platform reliance for political information; (ii) major platforms expanded emotional-reaction affordances and short-form video recommendation surfaces during 2019–2020 (e.g., the rapid scaling of TikTok and Reels, the broader rollout of Facebook reaction buttons beyond the like); (iii) the U.S. 2020 election cycle and the January 6, 2021 events brought renewed scrutiny of platform recommendation logics; and (iv) user-growth saturation among major platforms plausibly shifted incentives from acquisition to engagement intensification. We do not claim any single event is dispositive, but the convergence of these structural shifts is what the Chow-type test in Table 6 captures empirically. 10.1371/journal.pone.0342143.t006Table 6Chow-type structural break test for the 2020 analytical breakpoint.BreakpointObservations usedPre-break observationsPost-break observationsTested null hypothesisTest statisticp-valueConclusion2020-01-012,4731,742731Parameter stability across pre- and post-2020 periodsF(6, 2461) = 12.507< 0.001RejectedNote. The Chow-type known-break test examines whether the intercept and lagged coefficients in the dynamic model are stable across the pre-2020 and post-2020 periods. The tested model regresses dstdₜ on dstdₜ-₁, lagged technology-category variables (type0, type3, type4, type5), a post-2020 dummy, and their interaction terms. Robust standard errors are used. Rejection of parameter stability supports the use of phase-specific estimation but should not be interpreted as causal evidence that 2020 itself produced affective polarization. The formal test rejects the null hypothesis of parameter stability. Chow-type structural break test for the 2020 analytical breakpoint. Note. The Chow-type known-break test examines whether the intercept and lagged coefficients in the dynamic model are stable across the pre-2020 and post-2020 periods. The tested model regresses dstdₜ on dstdₜ-₁, lagged technology-category variables (type0, type3, type4, type5), a post-2020 dummy, and their interaction terms. Robust standard errors are used. Rejection of parameter stability supports the use of phase-specific estimation but should not be interpreted as causal evidence that 2020 itself produced affective polarization. The formal test rejects the null hypothesis of parameter stability. To empirically validate the 2020 breakpoint, we conducted a Chow-type known-break test specifying January 1, 2020, as the analytical breakpoint. The test is based on a dynamic model specification consistent with our VAR framework, regressing dstdₜ on its first lag, the lagged technology-category variables, a post-2020 dummy, and their interaction terms. The robust joint test rejects the null hypothesis of parameter stability across the pre- and post-2020 periods (F(6, 2461) = 12.507, p < 0.001). Substantively, the persistence term (lagged dstd) exhibits the most pronounced shift, rising from 0.149 before 2020 to an implied coefficient of 0.664 after 2020 (detailed interaction coefficients are provided in S1 Appendix). This formal structural break test provides strong empirical support for our phase-specific estimation strategy, though it underscores that 2020 serves as a proxy for a broader structural transition rather than a single causal event. 4.3 Empirical analysis based on the VAR model To ensure comparability across the pre-2020 and post-2020 subsamples, we estimate the same VAR specification in both phases and select the lag order using the same information-criterion procedure (reported in S1 Code). To examine the dynamic impact of technological categorization on emotional polarization, this study constructs a phase-based VAR model with 2020 as the breakpoint. The key results are as follows: (1) Persistence of Emotional Polarization: (1) Persistence of Emotional Polarization: The coefficient of the lag term (Dstd L1) increased significantly from 0.149 before 2020 to 0.664 (p < 0.001) after 2020, indicating a stronger self-reinforcing effect of emotional polarization under the shared surveillance model. (2) Differentiated Impact of Technological Categorization: (2) Differentiated Impact of Technological Categorization: Frontend Technologies: Type0 (Data Security) and Type3 (Dynamic Recommendation) showed a significant increase in their positive effects on emotional polarization after 2020, with coefficients rising from 0.000015** and 0.00004 to 0.0021 and 0.0033** respectively. Type4 (Social Platforms) transitioned from an insignificant impact to a significant positive effect (0.0022**), reflecting the synergistic amplification between algorithmic content push and user interaction. Type5 (Image Processing): The coefficient reverses from −0.0008 to a statistically significant negative effect (−0.0035***), which may reflect that certain image-processing and privacy-adjacent design choices reduce the platform’s ability to capture, infer, or amplify users’ emotional signals, thereby dampening affective escalation within platform-mediated interactions. (3) Granger Causality Test: (3) Granger Causality Test: Before 2020, none of the technological variables passed the significance test (p > 0.05). After 2020, Type0 to Type5 show significant Granger predictability for emotional polarization (p < 0.05), with Type5 exhibiting the strongest test statistic (Chi2 = 10.672). This pattern suggests that, in the post-breakpoint phase, technology-category signals are not merely contemporaneous correlates but become informative predictors within the modeled dynamics. 4.4 Robustness checks for the emotional polarization proxy To address potential concerns regarding the measurement validity of the emotional polarization proxy derived from news sentiment, we performed a series of sensitivity checks. As reported in Table 7, the core finding that the persistence of emotional polarization increases significantly post-2020 remains robust across multiple proxy specifications. 10.1371/journal.pone.0342143.t007Table 7Summary of emotional-polarization persistence and selected technology cross-lag associations, pre- versus post-2020.Pre-2020Post-2020Δ (Post–Pre) Panel A. Emotional-polarization persistence (Dstd L1) Baseline (dstd)0.149***0.664***+0.515Standardized (dstd)0.1490.664+0.515Winsorized (dstd)0.3500.662+0.312Spike-excluded (dstd)0.3270.635+0.308Level (std)0.9980.993−0.005Winsorized level (std)0.9980.993−0.005 Panel B. Selected technology cross-lag coefficients (baseline VAR, lag 1) Type0 — security/ trust infrastructure0.000015**0.0021**+0.0021Type3 — dynamic recommendation0.000040**0.0033**+0.0033Type4 — social-platform technologies−0.0001370.0022**+0.0024Type5 — image processing/ privacy−0.000799−0.0035***−0.0027Note. Entries are coefficients from phase-specific dynamic (VAR) models estimated separately for the pre-2020 (n = 1,742) and post-2020 (n = 731) subsamples around the 2020-01-01 analytic breakpoint. Panel A reports the persistence coefficient on the lagged dependent variable (Dstd L1) for the baseline first-difference proxy and five robustness specifications; Δ is the post–pre change. Panel B reports first-lag coefficients of selected technology-category intensities on Dstd in the baseline VAR. * p < 0.05, ** p < 0.01, *** p < 0.001. Summary of emotional-polarization persistence and selected technology cross-lag associations, pre- versus post-2020. Note. Entries are coefficients from phase-specific dynamic (VAR) models estimated separately for the pre-2020 (n = 1,742) and post-2020 (n = 731) subsamples around the 2020-01-01 analytic breakpoint. Panel A reports the persistence coefficient on the lagged dependent variable (Dstd L1) for the baseline first-difference proxy and five robustness specifications; Δ is the post–pre change. Panel B reports first-lag coefficients of selected technology-category intensities on Dstd in the baseline VAR. * p < 0.05, ** p < 0.01, *** p < 0.001. The persistence coefficient rises from 0.149 to 0.664 in the baseline first-difference model. This pattern holds when using a standardized dstd measure, a dstd measure winsorized at the 1st and 99th percentiles (rising from 0.350 to 0.662), and a specification excluding the top 1% absolute daily spikes (rising from 0.327 to 0.635). These results confirm that the increased persistence is not an artifact of rescaling, extreme first-difference observations, or isolated spike days. Additionally, level-based specifications (raw and winsorized std) demonstrate high persistence near 1.0 in both phases, confirming the strong memory of the underlying baseline series. Visualizations of these proxy dynamics are provided in the Supporting Information (S1 Appendix). 5 Discussion 5.1 Findings by time phase Table 7 summarizes the core phase-specific dynamic results for the two periods separated by the 2020 analytical breakpoint. The first pattern is a clear increase in polarization persistence after 2020. Specifically, the lagged term of polarization (Dstd L1) rises from 0.149 in the pre-2020 phase to 0.664 post-2020 (both statistically significant), indicating that affective divergence becomes substantially more self-reinforcing under the hypothesized synoptic, interaction-amplified regime. Substantively, this increased persistence suggests that under the synoptic regime, emotional contagion is no longer episodic but has formed a self-reinforcing feedback loop, making affective divergence more entrenched within community interactions. More strikingly, translating these coefficients into temporal dynamics implies that the half-life of an emotional shock to the information environment lengthens from approximately 0.4 days to roughly 1.8 days under the post-2020 regime, which is an over fourfold increase in the temporal footprint of any given affective disturbance. 5.2 Findings by technology type The second pattern concerns the differentiated effects of platform-facing technology domains. After 2020, the coefficients for Type0 (data security-related platform infrastructure) and Type3 (dynamic recommendation) increase markedly and remain positive and significant, consistent with the model’s expectation that scalable data processing and recommendation logic can strengthen engagement-salient exposure patterns. To contextualize these magnitudes substantively: post-2020, a one-standard-deviation increase in Type3 (dynamic recommendation) intensity is associated with a daily change in the dispersion proxy that represents roughly 14% of the overall average daily fluctuation (0.007). Type4 (social platform technologies) shifts from statistically indistinguishable effects before 2020 to a significant positive association afterward, suggesting a stronger coupling between platform interaction design and affective dynamics in the later regime. In contrast, Type5 (image processing/ privacy-related functions) becomes significantly negative after 2020. This pattern may reflect that certain image-processing and privacy-adjacent design choices (e.g., reduced granularity of user profiling or targeting precision, and increased friction in content circulation) can dampen affective escalation within platform-mediated interactions. We interpret this result as suggestive rather than definitive, given that Type5 captures a bundle of visual-content and privacy-adjacent platform capabilities. Overall, these technology-specific findings imply that front-end interaction designs serve as direct amplifiers of affective escalation, whereas reducing the granularity of data capture or affective targeting (as suggested by the Type 5 estimates) represents a potential structural intervention to dampen such polarization. Third, the Granger predictability tests are consistent with a regime-sensitive interpretation. Before 2020, technology categories do not exhibit systematic predictive power; after 2020, key front-end categories (Type0–Type5) display significant Granger predictability, suggesting that technology signals become more informative for polarization dynamics in the later phase. These results support a mechanism-consistent reading of the theoretical framework: once disciplining shifts toward synoptic dynamics, emotional polarization shows stronger inertia, and platform-facing technological domains (especially recommendation- and interaction-related categories) are more strongly associated with polarization dynamics. At the same time, the negative association for Type5 highlights the model’s “double-edged” implication: not all technology deepens polarization, and design choices that limit data extraction or targeting may partially offset escalation pressures. These findings have implications for emerging platform-governance frameworks. For example, the EU Digital Services Act requires very large online platforms and search engines to assess systemic risks linked to the design, functioning, and use of their services, including algorithmic systems, recommender systems, advertising systems, and data-related practices [86]. Our results suggest that such regulatory debates may benefit from looking beyond content moderation alone and paying closer attention to structural platform mechanisms, including algorithmic recommendation, interaction-feedback loops, and affective engagement infrastructures. This policy implication should be read cautiously: the present study provides observational and proxy-based evidence, but it highlights why systemic-risk assessment may need to consider the design conditions under which affective polarization becomes more persistent. 5.3 Generalizability and broader implications While our empirical proxy captures generalized media sentiment, the disciplinary mechanisms identified likely operate across diverse digital contexts, though their manifestations may vary. For instance, non-Western platforms like WeChat, which rely heavily on strong-tie social networks, may exhibit different affective contagion dynamics compared to algorithmic-driven short-video platforms like TikTok. The rapid scaling of emotion-reaction affordances and infinite-scroll interfaces on platforms such as TikTok during and after 2018–2020 may form part of the broader platform ecology surrounding the observed post-2020 structural shift. This broader ecology also cautions against treating the Big Five ecosystem as fully representative of global platform governance, since platforms such as WeChat and ByteDance/TikTok may organize interaction, recommendation, and affective feedback through different affordance configurations and institutional logics. Furthermore, the methodological use of AI tools such as ChatGPT for category labeling in our study highlights a broader implication: as algorithms increasingly participate in defining and categorizing knowledge, establishing clear labeling guidelines and human-in-the-loop verification processes becomes essential to prevent the uncritical institutionalization of algorithmic biases. 5.4 Limitations Despite its contributions, this study has limitations that point toward future research directions. First, the empirical analysis focuses on the relationship between platform-facing technological development and an aggregated emotional polarization proxy derived from news sentiment, rather than on individual psychological polarization or platform-internal exposure data. Accordingly, our findings speak to affective dispersion in the platform-mediated information environment (digital public sphere) rather than to users’ internal emotional states; future work could triangulate our proxy with user-trace, survey-based measures, or platform transparency disclosures to better test the proposed micro-to-meso transmission mechanisms. Furthermore, as the GDELT proxy aggregates sentiment signals at a global scale, there is a risk that country-specific effects, particularly those within the U.S. political landscape which is often the focal point of polarization research, may be partially diluted. Due to the lack of access to raw event-level metadata required for localized filtering at this stage, we were unable to perform a formal U.S.-centric robustness check. We acknowledge this limitation and encourage future studies to utilize geographically filtered datasets to evaluate whether these platform disciplinary mechanisms manifest differently across specific national media systems. Second, while this paper emphasizes the effects of platform disciplining mechanisms, subsequent research could explore user heterogeneity in greater depth, examining how different groups accept, negotiate, or resist behavioral and emotional structuring. Third, our phase-based analysis treats 2020 as an analytic breakpoint to evaluate whether the technology-polarization relationship is regime-sensitive; however, this design does not isolate the regime transition from contemporaneous macro shocks, and future work could incorporate alternative breakpoint tests and additional exogenous controls. Additionally, the complex macro-environment inevitably introduces risks of omitted variable bias, and we must acknowledge the potential for reverse causality, where heightened societal polarization may simultaneously drive platforms to develop specific technology categories to manage or capitalize on user engagement. Finally, extending this framework to non-Western or comparative digital contexts would help assess the generalizability of platform disciplining mechanisms and contribute to the development of more inclusive theories of global digital governance. 6 Conclusions, implications, and future works In the context of digital platforms becoming an increasingly central arena for political communication and the reconfiguration of social structures, this study examined how Bigtech platforms are associated with user behavior and emotional polarization through technology-driven mechanisms. By introducing the theoretical framework of platform disciplining mechanisms, this paper developed an integrated analytical model encompassing individual behavior guidance, group identity formation, and the structural patterning of public opinion. Empirically, we observe that emotional polarization becomes more persistent after 2020, and that front-end technology domains related to recommendation and interaction feedback exhibit stronger post-breakpoint associations with polarization dynamics. These patterns are consistent with a regime-sensitive disciplining process operating through platform-mediated affective inputs. Given that the polarization indicator is media-level, our findings speak to affective dynamics in the platform-mediated information environment rather than individual psychological states. Building on these empirical results, the central finding of this paper is that Bigtech platforms do not merely influence user behavior at the level of technical facilitation. Rather, the coordinated operation of interface design, algorithmic distribution, and interaction feedback is consistent with a broader disciplining process that operates across behavioral, cognitive, and emotional dimensions. At the individual level, this mechanism is associated with gradual dependency on platform logics through standardized interaction paths, simplified decision-making, and the reward of emotionally resonant engagement. At the group level, these mechanisms are consistent with patterns of identity alignment within communities and cognitive segmentation between groups through affiliation signals and affective feedback loops. This dual process suggests that platforms should not be understood only as neutral intermediaries in information circulation. Instead, they increasingly operate as institutional environments in which social order and identity dynamics are coordinated through interface architecture, recommendation logic, and user feedback. The rise of platform disciplining mechanisms is therefore associated with reconfigurations in the ecology of public opinion in the digital age, including the formation, stabilization, and fragmentation of social consensus. Importantly, the disciplining process identified in this study does not reduce to content filtering or isolated algorithmic control. It reflects a broader governance system in which interface architecture, recommendation logic, and user feedback are jointly associated with patterns of behavioral motivations, cognitive orientations, and emotional structures. As a result, individuals embedded in highly personalized platform environments may increasingly express collective emotions and identity-based alignments, even as their experiences appear individualized at the interface level. These findings challenge prevailing assumptions about technological neutrality and platform openness that continue to inform much of the policy and public discourse on digital platforms. The power of platforms is not confined to the moderation or amplification of specific content but lies more fundamentally in their systemic alignment with emotional dynamics and identity logics at scale. This systemic alignment helps account for the observed contraction of deliberative spaces, the persistence of emotional polarization, and the deep coupling between platform design and democratic participation. From this perspective, emotional governance emerges as a critical, yet often overlooked, dimension of contemporary platform power. The implications of these findings extend directly to platform governance. Existing regulatory frameworks predominantly target content moderation, misinformation control, and data privacy protection. While these measures address important risks, they remain insufficient for confronting the deeper behavioral and emotional interventions enabled by platform technologies. The empirical evidence presented here suggests that governance approaches should expand from content-level oversight toward structural accountability for how platforms design interaction mechanisms and emotional feedback systems. In particular, emotional polarization risks should be incorporated into platform responsibility assessments, and platforms should be expected to evaluate the downstream social effects of recommendation and interaction architectures. In this context, governance strategies may benefit from encouraging the introduction of mechanisms that disrupt excessive behavioral convergence and emotional homogenization, such as diversified exposure pathways or friction in interaction feedback. Moreover, institutional oversight should not rely solely on platform self-regulation. The establishment of neutral and professional supervisory bodies—comprising governmental agencies, academic experts, and technical specialists—could provide independent evaluation of platform disciplining capacities. Such bodies would be positioned to assess how platform technologies co-evolve with behavior, identity construction, and community segmentation, and to integrate these assessments into broader public governance frameworks. By institutionalizing external monitoring of platform disciplining mechanisms, it becomes possible to balance technological innovation and commercial freedom with safeguards against large-scale emotional manipulation and social fragmentation. These findings point to several extensions that follow directly from our analytical design. Because the proxy operates at the media level, future work could triangulate the GDELT-based indicator with platform-trace data and survey-based measures to test the proposed micro-to-meso transmission pathway more directly. Because it aggregates globally, geographically filtered analyses—particularly U.S.-centric ones—would help recover country-specific dynamics that the global signal may dilute. Beyond the institutional-level account developed here, user-level studies could probe heterogeneity in how different populations accept, negotiate, or resist platform-imposed structuring. Finally, the regime-sensitive pattern we document around 2020 invites tests with alternative breakpoints and extensions to non-Western ecologies such as WeChat or short-video platforms like TikTok, where comparable disciplining mechanisms may operate through different affordance configurations. Rapid scaling of emotion-reaction features and infinite-scroll interfaces on platforms such as TikTok after 2018–2020 may form part of the broader platform ecology surrounding the observed post-2020 structural shift, highlighting that findings for the Big Five ecosystem may not fully generalize to other global platforms, including WeChat and other non-Western platforms. Taken together, these directions would strengthen the empirical foundation for institutional approaches to affective governance and inform comparative theories of digital platform regulation. Addressing these structural mechanisms of affective governance is therefore a necessary step toward ensuring the broader sustainability of digital transformation, where technological advancement aligns with social resilience and inclusive development rather than undermining them [87]. Supporting information Daily time-series dataset used to construct the empirical variables. (XLSX) Supplementary methodological documentation for web of science retrieval, screening, clustering, and AI-assisted labeling. (DOCX) Reproducible scripts for data processing and statistical analysis. (ZIP) Supplementary results and figures for the breakpoint and proxy-sensitivity analyses. (DOCX) 22 Mar 2026 Dear Dr. LU, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. We note that some of the comments of the reviewers refer to specific articles for you to cite. Please note that it is not mandatory that you cite these specific articles and you are welcome to seek alternatives manuscripts in the literature that are relevant to your manuscript’s content. Please submit your revised manuscript by May 06 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Andrea Cioffi Academic Editor PLOS One Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Please note that PLOS One has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. 3. Thank you for stating in your Funding Statement: [The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Grants-in-Aid for Scientific Research (KAKENHI C)Political Identify and Practices of Bigtech in the Digital Political Space (24K04704)]. Please provide an amended statement that declares *all* the funding or sources of support (whether external or internal to your organization) received during this study, as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now. Please also include the statement “There was no additional external funding received for this study.” in your updated Funding Statement. Please include your amended Funding Statement within your cover letter. We will change the online submission form on your behalf. 4. Please ensure that you include a title page within your main document. You should list all authors and all affiliations as per our author instructions and clearly indicate the corresponding author. 5. Please upload a copy of your Supporting Information files, to which you refer in your text on pages 26, 27, 42, and 43. If the figure is no longer to be included as part of the submission please remove all reference to it within the text. Please ensure that each Supporting Information file has a legend listed in the manuscript after the references list. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Partly ********** 2. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Yes ********** 3. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Yes ********** 4. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: No Reviewer #3: Yes ********** Reviewer #1: I appreciate the opportunity to review this manuscript. The study tackles an important and timely question regarding how platform governance mechanisms intersect with affective polarization in digital public spheres. While technically sound, several aspects could benefit from further clarification: Breakpoint Justification: The selection of 2020 as a structural breakpoint is theoretically motivated (platform growth transition), but a formal structural break test (e.g., Bai–Perron or Chow test) would strengthen empirical justification. Causal Interpretation: The manuscript appropriately avoids strong causal claims; however, given the observational time-series design, this limitation should be emphasized more clearly in the discussion. Measurement Validity of Affective Polarization: The use of media-level sentiment dispersion as a proxy for platform-mediated affective dynamics is theoretically defended. Still, additional sensitivity checks (e.g., alternative dispersion metrics) would enhance robustness. Magnitude Interpretation: While statistical significance is reported, effect sizes should be contextualized substantively to help readers understand real-world implications. Reviewer #2: 1.The formal terminology of main headings has to change in order to comply with the following 6 main sections/titles: 1.Introduction, 2.Literature Review, 3.Methodology and Analysis, 4.Results, 5.Discussion, 6.Conclusions Implications and Future Works. All other-typed headings can change to aforementioned, or to be fixed as subsections underneath the aforementioned 6 main sections. 2.The legends/captions of all Figures can be placed at the bottom of their graph areas, not on the top of them. 3.The narrative of the study contains many bold-typed, but non-numbered, subheadings. For this reason, all non-numbered subheadings have to be numbered in a sequence, jointly with the aforementioned 6 main headings/sections of the revised manuscript. 4.Regarding the digital platforms and their role of regulating them in order to better address “structural mechanisms of behavioral normalization and affective manipulation beyond content moderation alone”, there should be a broader and updated international theoretical coverage, since the half of citations are dating back more than a decade ago of publishing, whereas the topic of digital technologies in social context has attracted plentiful research interest in the last 5 years of literature production. Indicative studies (there is also a literature blossom of similar ones) that can be considered in the revised manuscript are the following: DOI: 10.1007/978-3-031-66801-2_2 , DOI: 10.3389/fenvs.2024.1371047 , DOI: 10.1007/978-3-031-30351-7_9 , DOI: 10.1007/978-3-031-30351-7_1 ,DOI: 10.3390/su13179577. 5.Almost all subsections 5.2-5.5 contain non cross-citing information, thus, checking and citing them in a more systematic manner can better validate the discussion on findings. 6.In a separate and autonomous Discussion section authors are recommended to structure 2 subsections in which the a) per type-findings, b) per time-findings in decades’ intervals among the whole examined period, to be conveyed. In such a way a more comprehensive understanding of findings can be obtained when reading the study. Up to 2 extra and cross-cited text pages can be devoted in structuring this Discussion section. Reviewer #3: Strengths This is a timely and ambitious contribution at the intersection of platform studies, computational social science, and affective-polarization research. The introduction of “platform disciplinary mechanisms” as a tripartite process (behavioral standardization via interface design, cognitive dependency via algorithmic recommendation, and emotionally structured group differentiation via interaction feedback) is genuinely novel and successfully bridges Foucaultian governmentality with contemporary digital-media theory. The empirical strategy—linking academic-publication intensity in Big Tech research domains to a media-level affective-polarization proxy—is creative and avoids the common reliance on proprietary platform trace data, thereby sidestepping many privacy and replicability barriers. The use of K-means clustering plus ChatGPT-assisted active learning for categorizing technology types (security, recommendation systems, emotion-oriented interaction, etc.) is methodologically innovative and transparently documented. Findings that affective polarization becomes more persistent post-2020 and more tightly coupled to front-end, emotion-oriented technology categories align with broader public and scholarly concerns about algorithmic amplification of outrage. The policy implications—advocating regulation that targets structural mechanisms rather than content moderation alone—are thoughtful and forward-looking. Major Concerns (requiring revision) Causal language vs. correlational evidence. The abstract and conclusion repeatedly employ verbs such as “organize,” “reorganize,” “shape,” and “function as institutional actors that … through affective governance.” The VAR/Granger analysis demonstrates temporal precedence and regime-dependent associations but cannot establish that platform technologies cause polarization (or vice versa). Endogeneity is plausible: rising polarization may itself stimulate Big Tech R&D in recommendation and moderation technologies. I recommend toning down causal claims to “coincide with,” “are associated with,” or “co-evolve with,” and adding a dedicated paragraph in the Discussion on reverse causality, omitted-variable bias, and the limitations of publication proxies as measures of deployed platform mechanisms. Validity of the affective-polarization proxy. Using the standard deviation of GDELT news-sentiment scores as a daily indicator of “affective polarization in the information environment” is clever and publicly replicable, yet it remains one step removed from platform users. GDELT aggregates global news tone; platforms operate in closed ecosystems with their own amplification logics. The manuscript justifies the proxy well but should more explicitly discuss (a) why news-tone dispersion is a valid upstream indicator of platform-mediated affective dynamics, (b) potential mismatches between journalistic sentiment and social-media user emotion, and (c) comparisons with alternative proxies (e.g., Twitter or Facebook emotion scores from prior studies). An edge-case consideration: GDELT’s global scope may dilute country-specific effects; a robustness check restricting to U.S.-centric events would strengthen claims. Choice and justification of the 2020 breakpoint. The breakpoint is described as “theory-informed,” yet the exact theoretical rationale (algorithmic shifts, COVID-19, regulatory pressure, or the Capitol insurrection) is not fully elaborated. Contemporaneous macro-shocks could confound results. Future readers will ask whether findings are robust to alternative breakpoints (e.g., 2018 Cambridge Analytica aftermath or 2022 Musk-Twitter acquisition). Sensitivity analyses and a clearer narrative link between 2020 events and specific platform disciplinary mechanisms (e.g., increased emotional-reaction buttons, recommendation of short-form video) would address this. Technology-categorization methodology. The clustering procedure (K-means + ChatGPT active learning) is promising but requires expanded reporting: exact number of abstracts, feature vector construction (TF-IDF? embeddings?), number of clusters chosen and validation metrics (silhouette score, human inter-rater reliability on a hold-out set), and how the 10 final categories map onto “front-end” vs. “back-end” disciplinary functions. The manuscript mentions “Type3 recs” and “emotion-oriented interaction dynamics” but does not provide the full mapping table in the main text (only implied in Supporting Information). Readers need this transparency to evaluate construct validity. Omitted robustness and alternative specifications. Beyond the diagnostics noted in point 2, consider: (i) different sentiment-dispersion metrics (Gini, inter-quartile range), (ii) log-transformation or differencing of technology-intensity series, (iii) inclusion of news-volume controls, and (iv) subsample analyses excluding major news spikes. These would rule out artefactual results. Minor Points English polishing: break up some 50–60-word sentences; ensure consistent terminology (“affective polarization” vs. “emotional polarization” in title/abstract). Visuals: add a conceptual diagram of the platform–end-user–community framework and a summary table of VAR results. Discussion: expand on generalizability (non-Western platforms, TikTok, WeChat), user heterogeneity, and ethical implications of using ChatGPT for labeling. No issues of dual publication, research ethics (no human subjects), or competing interests apparent. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #1: No Reviewer #2: Yes: Grigorios L. Kyriakopoulos Reviewer #3: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation. NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications. 9 May 2026 Response to the Editor and Reviewers Manuscript ID: PONE-D-26-02998 Dear Dr. Cioffi and Reviewers, Thank you very much for the opportunity to revise our manuscript, "Disciplining the Digital Public: Platform Mechanisms and the Dynamics of Emotional Polarization" (PONE-D-26-02998). We would like to express our deepest gratitude to you and the three anonymous reviewers for the highly constructive criticisms and insightful comments. We have carefully studied all the comments and addressed them through targeted revisions, additional analyses, expanded methodological documentation, and clearer discussion of limitations. We believe that these revisions have substantially improved the clarity, rigor, and theoretical depth of our paper. First of all, we have carefully reviewed all five Journal Requirements and revised the manuscript and supporting materials accordingly to comply with PLOS ONE's style, file-naming, code-sharing, and submission standards. In specific, regarding Requirement 3, the amended Funding Statement (declaring all sources of support and explicitly noting that no additional external funding was received) is included in our cover letter as requested. Regarding Requirement 5, a complete legend for each Supporting Information file (S1 Text, S1 Appendix, S1 Data, and S1 Code) has been added at the end of the revised manuscript, immediately following the References list. Authorship update. During the revision process, we added Bing Pang (Graduate School of Asia-Pacific Studies, Waseda University, Tokyo 169-8050, Japan) as a co-author. Mr. Pang made substantive contributions to the revised manuscript by assisting with data processing, code organization, empirical analysis, robustness checks, and the preparation of reproducible supporting materials. All authors, including the newly added co-author, have reviewed and approved the revised manuscript and agree with the updated authorship and author order. We also updated the author information, affiliations, and author-contribution details in the submission system as required. Below, please find our point-by-point responses to the reviewers' comments. Responses to Reviewer 1 1. Comment: Breakpoint Justification: The selection of 2020 as a structural breakpoint is theoretically motivated (platform growth transition), but a formal structural break test (e.g., Bai–Perron or Chow test) would strengthen empirical justification. Response: We thank the reviewer for this highly constructive comment. We completely agree that a formal structural break test is necessary to strengthen our empirical justification. In the revised manuscript, we have now conducted a formal Chow-type known-break test specifying January 1, 2020, as the analytical breakpoint. The robust joint test strongly rejects the null hypothesis of parameter stability across the pre- and post-2020 periods (F(6, 2461) = 12.507, p < 0.001), providing strong empirical support for our phase-specific estimation strategy. These formal test results and interpretations have been added to the Results section. (Please see Section 4.2 and Table 6, Page 35-39). 2. Comment: Causal Interpretation: The manuscript appropriately avoids strong causal claims; however, given the observational time-series design, this limitation should be emphasized more clearly in the discussion. Response: We thank the reviewer for this important observation. We agree and have further toned down causal wording throughout the manuscript, especially in the Abstract, Introduction, Results, and Conclusion. The revised text now consistently frames the findings as associations, co-movements, or regime-sensitive relationships rather than direct causal effects. We also expanded the Discussion and Limitations sections to state explicitly that the observational design does not permit strong causal inference and to acknowledge possible reverse causality and omitted-variable bias. (Please see Section 5.4 Limitations, Page 46). 3. Comment: Measurement Validity of Affective Polarization: The use of media-level sentiment dispersion as a proxy for platform-mediated affective dynamics is theoretically defended. Still, additional sensitivity checks (e.g., alternative dispersion metrics) would enhance robustness. Response: Thank you for this valuable methodological suggestion. We completely agree with this change. To enhance robustness, we have conducted additional sensitivity checks across multiple specifications of the dispersion proxy (standardized, winsorized, spike-excluded, level, and winsorized-level versions), reported in Table 8 and visualized in S1 Appendix. These specifications collectively address concerns about scaling, extreme observations, and isolated spike days. We retain the standard deviation as the principal dispersion measure for the methodological reasons outlined in Section 3.1; rank-based alternatives such as Gini or IQR are noted as a useful direction for future extension. (Please see Section 4.4 and Table 8, Page 41-43, as well as Supporting Information S1 Appendix). 4. Comment: Magnitude Interpretation: While statistical significance is reported, effect sizes should be contextualized substantively to help readers understand real-world implications. Response: Thank you for this insightful comment. We completely agree with this suggestion. We have revised the discussion to substantively contextualize the effect sizes, explicitly explaining what the increased persistence and specific coefficients mean for real-world platform governance and community interactions. (Please see Section 5.1 and 5.2, Page 43-46). Responses to Reviewer 2 1. Comment: The formal terminology of main headings has to change in order to comply with the following 6 main sections/titles: 1. Introduction, 2. Literature Review, 3. Methodology and Analysis, 4. Results, 5. Discussion, 6. Conclusions Implications and Future Works. All other-typed headings can change to aforementioned, or to be fixed as subsections underneath the aforementioned 6 main sections. Response: We thank the reviewer for this clear guidance. We agree and have restructured the manuscript so that it now follows the six required main sections. All former major sections have been reorganized accordingly, and the remaining headings have been renumbered as subsections under the new structure. (Please see the revised manuscript structure throughout). 2. Comment: The legends/captions of all Figures can be placed at the bottom of their graph areas, not on the top of them. Response: We thank the reviewer for this helpful formatting suggestion. We agree and have standardized the figure captions so that all figure titles and notes are now placed directly at the bottom of their respective graph areas. Specifically, this formatting update has been applied to Fig 1. Theoretical Framework of Bigtech’s Disciplinary Mechanisms in Section 2.3, and Fig 2. User Growth Trends of Major Social Platforms (2012-2024) in Section 4.2. (Please see Pages 23 and 36). 3. Comment: The narrative of the study contains many bold-typed, but non-numbered, subheadings. For this reason, all non-numbered subheadings have to be numbered in a sequence, jointly with the aforementioned 6 main headings/sections of the revised manuscript. Response: Thank you for your careful reading and suggestion. We completely agree with this change. All non-numbered subheadings have now been sequentially numbered to align with the 6 main sections. (Please see Section 2.1.1, 2.1.2, 2.2.1, etc., throughout the manuscript). 4. Comment: Regarding the digital platforms and their role of regulating them in order to better address “structural mechanisms of behavioral normalization and affective manipulation beyond content moderation alone”, there should be a broader and updated international theoretical coverage, since the half of citations are dating back more than a decade ago of publishing, whereas the topic of digital technologies in social context has attracted plentiful research interest in the last 5 years of literature production. Response: Thank you for recommending these highly relevant and updated studies. We completely agree with this suggestion. We have carefully read these excellent works and integrated them into our manuscript to broaden our theoretical coverage on digital transformation, macro/micro-economic impacts, uncertainty contexts, and environmental/social sustainability. (Please see Introduction, pp. 3-4; Section 2.1.1, pp. 12-13; Section 4.2, pp. 37-38; Section 6, pp. 52-53). 5. Comment: Almost all subsections 5.2-5.5 contain non cross-citing information, thus, checking and citing them in a more systematic manner can better validate the discussion on findings. Response: We thank the reviewer for this helpful comment. We agree and have revised the Results and Discussion so that the empirical findings are now interpreted in a more systematic manner, with clearer links to the theoretical framework and the existing literature. The discussion of each major result now connects the observed pattern to the relevant mechanism, rather than leaving the subsections isolated. (Please see Section 5, Pages 43-48). 6. Comment: In a separate and autonomous Discussion section authors are recommended to structure 2 subsections in which the a) per type-findings, b) per time-findings in decades’ intervals among the whole examined period, to be conveyed. In such a way a more comprehensive understanding of findings can be obtained when reading the study. Up to 2 extra and cross-cited text pages can be devoted in structuring this Discussion section. Response: We thank the reviewer for this constructive recommendation. We agree and have created a separate, autonomous Discussion section (Section 5) with clear subsections specifically dedicated to '5.1 Findings by Time Phase' and '5.2 Findings by Technology Type' to comprehensively convey our results. (Please see Section 5.1 and Section 5.2, Pages 43-46). Responses to Reviewer 3 1. Comment: Causal language vs. correlational evidence. The abstract and conclusion repeatedly employ verbs such as “organize,” “reorganize,” “shape,” and “function as institutional actors that … through affective governance.” The VAR/Granger analysis demonstrates temporal precedence and regime-dependent associations but cannot establish that platform technologies cause polarization (or vice versa). Endogeneity is plausible: rising polarization may itself stimulate Big Tech R&D in recommendation and moderation technologies. I recommend toning down causal claims to “coincide with,” “are associated with,” or “co-evolve with,” and adding a dedicated paragraph in the Discussion on reverse causality, omitted-variable bias, and the limitations of publication proxies as measures of deployed platform mechanisms. Response: We thank the reviewer for this important and well-taken point. We agree and have toned down causal language throughout the manuscript, especially in the Abstract, Introduction, Results, Discussion, and Conclusion. The revised text now consistently frames the findings as associations and phase-sensitive relationships rather than direct causal effects. Additionally, we have added a dedicated paragraph in the Discussion section to explicitly address reverse causality, omitted variable bias, and proxy limitations. (Please see Section 5.4 Limitations, Page 46). 2. Comment: Validity of the affective-polarization proxy. Using the standard deviation of GDELT news-sentiment scores as a daily indicator of “affective polarization in the information environment” is clever and publicly replicable, yet it remains one step removed from platform users. GDELT aggregates global news tone; platforms operate in closed ecosystems with their own amplification logics. The manuscript justifies the proxy well but should more explicitly discuss (a) why news-tone dispersion is a valid upstream indicator of platform-mediated affective dynamics, (b) potential mismatches between journalistic sentiment and social-media user emotion, and (c) comparisons with alternative proxies (e.g., Twitter or Facebook emotion scores from prior studies). An edge-case consideration: GDELT’s global scope may dilute country-specific effects; a robustness check restricting to U.S.-centric events would strengthen claims. Response: We thank the reviewer for this helpful suggestion. We agree and have strengthened the explanation of the proxy in the Methods and Discussion. The revised manuscript now clarifies that GDELT-based sentiment dispersion is used as a media-level indicator of affective dynamics in the platform-mediated information environment, not as a direct measure of individual psychological states. Specifically, in revised Section 3.1, we now explicitly compare our GDELT-based proxy with prior platform-trace emotion data (Brady et al. 2017 [Ref 82]; Bail et al. 2018 [Ref 83]; Rathje et al. 2021 [Ref 84]) and clarify that GDELT operates upstream of platform-internal engagement traces, providing public availability and longitudinal continuity at the cost of distance from individual user states. We also discuss the conceptual gap between journalistic sentiment and user emotion, and we completely agree that a global aggregate may inadvertently dilute specific national dynamics, such as those in the U.S. context. However, due to the lack of access to the raw, event-level metadata required for precise localized filtering in our current database, we were unable to conduct a separate U.S.-centric robustness check. We have followed your suggestion by explicitly adding this as a methodological limitation in the Discussion section and highlighting the need for localized filtering in future research. (See Section 5.4, Page 46). 3. Comment: Choice and justification of the 2020 breakpoint. The breakpoint is described as “theory-informed,” yet the exact theoretical rationale (algorithmic shifts, COVID-19, regulatory pressure, or the Capitol insurrection) is not fully elaborated. Contemporaneous macro-shocks could confound results. Future readers will ask whether findings are robust to alternative breakpoints (e.g., 2018 Cambridge Analytica aftermath or 2022 Musk-Twitter acquisition). Sensitivity analyses and a clearer narrative link between 2020 events and specific platform disciplinary mechanisms (e.g., increased emotional-reaction buttons, recommendation of short-form video) would address this. Response: We thank the reviewer for this important observation. We completely agree and have substantially strengthened the theoretical and empirical justification for treating 2020 as an analytic breakpoint. In the revised manuscript, we now explicitly outline four converging structural shifts around 2020: (i) the COVID-19 pandemic's impact on online discourse, (ii) the rapid scaling of short-form video and emotion-reaction affordances (e.g., TikTok, Reels), (iii) the scrutiny following the 2020 U.S. election/Jan 6 events, and (iv) user-growth saturation. Furthermore, we implemented a formal Chow-type structural break test, which statistically confirms that the dynamic relationship shifted significantly post-2020. This ensures the breakpoint is both contextually grounded and empirically validated. (Please see Section 4.2 and Table 6, Page 35-39). 4. Comment: Technology-categorization methodology. The clustering procedure (K-means + ChatGPT active learning) is promising but requires expanded reporting: exact number of abstracts, feature vector construction (TF-IDF? embeddings?), number of clusters chosen and validation metrics (silhouette score, human inter-rater reliability on a hold-out set), and how the 10 final categories map onto “front-end” vs. “back-end” disciplinary functions. The manuscript mentions “Type3 recs” and “emotion-oriented interaction dynamics” but does not provide the full mapping table in the main text (only implied in Supporting Information). Readers need this transparency to evaluate construct validity. Response: We thank the reviewer for this detailed and constructive suggestion. We agree that the original manuscript did not provide sufficient transparency regarding the technology-categorization procedure. In the revised manuscript, we expanded Section 4.1 to explain more explicitly that the ten technology categories are interpreted as two broad dimensions on the basis of the clusteri Submitted filename: Response to Reviewers.docx 8 Jun 2026 Dear Dr. LU, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Jul 23 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors. We look forward to receiving your revised manuscript. Kind regards, Andrea Cioffi Academic Editor PLOS One Journal Requirements: If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author Reviewer #1: All comments have been addressed Reviewer #2: All comments have been addressed Reviewer #3: All comments have been addressed ********** 2. Is the manuscript technically sound, and do the data support the conclusions??> Reviewer #1: Yes Reviewer #2: (No Response) Reviewer #3: Yes ********** 3. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: (No Response) Reviewer #3: Yes ********** 4. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #1: Yes Reviewer #2: (No Response) Reviewer #3: Yes ********** 5. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: (No Response) Reviewer #3: Yes ********** Reviewer #1: I have reviewed the revised manuscript and the authors’ point-by-point responses. In my opinion, the authors have adequately addressed the major concerns raised during the previous round of review. The justification for the 2020 breakpoint has been strengthened through both theoretical explanation and a formal Chow-type structural break test. The causal language has been appropriately toned down, and the manuscript now presents the findings as associations rather than causal effects. The authors have also added useful robustness checks for the emotional polarization proxy, clarified the limitations of using GDELT-based media-level sentiment dispersion, and expanded the methodological description of the technology-categorization procedure. The revised manuscript is more clearly structured, the discussion is better connected to the empirical findings, and the limitations are now stated more transparently. The supporting information, including data, code, methodological documentation, and robustness results, also improves the reproducibility and transparency of the study. Overall, the manuscript now satisfies the major criteria for publication in PLOS ONE. The study presents original research, the analyses are described in sufficient detail, the conclusions are appropriately supported by the data, and the manuscript is intelligible and substantially improved. I therefore recommend that the manuscript be accepted for publication. Reviewer #2: (No Response) Reviewer #3: This revised manuscript represents a meaningful contribution to the growing literature on platform governance, affective dynamics, and the structural mechanisms through which digital platforms may shape public opinion environments. The core theoretical innovation—conceptualizing “platform disciplinary mechanisms” (interface design, algorithmic recommendation, and interaction feedback loops) as processes that foster behavioral standardization, cognitive dependency, and emotionally structured group differentiation—is timely and well-aligned with contemporary debates surrounding algorithmic amplification, short-form video affordances, and the limits of content-moderation-centric regulation. The empirical strategy, while necessarily proxy-based and observational, is thoughtfully designed and now bolstered by formal structural-break testing, extensive robustness checks, and unusually transparent documentation of the NLP pipeline used to derive technology-intensity measures from nearly 50,000 Big Tech-affiliated abstracts. The central finding—that emotional polarization in the information environment (operationalized via GDELT news-sentiment dispersion) exhibits greater persistence after 2020 and stronger associations with technological simplification and emotion-oriented interaction dynamics—offers plausible correlational evidence consistent with the affective-governance thesis. The authors are to be commended for the care with which they have tempered causal language and for the expanded Limitations section that explicitly discusses reverse causality, omitted-variable bias, and the distance between news-tone dispersion and individual psychological states. Strengths of the Revision • Formal breakpoint validation: The Chow test result (F(6, 2461) = 12.507, p < 0.001) provides rigorous statistical grounding for treating 2020 as an analytic regime shift, complementing the rich contextual narrative around COVID-19 discourse migration, short-form video scaling (TikTok/Reels), post-election scrutiny, and user-growth saturation. • Robustness of the polarization proxy: Testing standardized, winsorized, spike-excluded, level, and winsorized-level versions of the dispersion measure, and demonstrating stability of the post-2020 persistence increase across first-difference specifications, directly addresses earlier concerns about scaling artifacts and extreme observations. • Methodological transparency (S1 Text): The new supporting text detailing SentenceTransformer embeddings, UMAP reduction, MiniBatchKMeans (k=10), WCSS/silhouette diagnostics, six-round active-learning protocol (~30 abstracts/round), ChatGPT-assisted labeling with pre-specified guidelines, and the front-end/back-end interpretive heuristic is a model of open science practice. This significantly strengthens construct validity claims for the ten technology categories. • Restructured Discussion: The creation of dedicated subsections 5.1 (Findings by Time Phase) and 5.2 (Findings by Technology Type), together with systematic cross-referencing to the theoretical framework and prior literature, eliminates the earlier impression of isolated subsections and improves interpretive coherence. • Cautious interpretation and limitations: The consistent reframing from causal verbs (“organize,” “shape,” “function as institutional actors”) to associational language (“coincide with,” “are associated with,” “co-evolve with”), plus the explicit paragraph on endogeneity and proxy limitations, aligns the rhetorical claims with the observational design. Minor Suggestions for Further Improvement While the revisions have addressed the major concerns, the following presentational and interpretive refinements would further strengthen the manuscript: 1. Main-text technology category overview. Although S1 Text contains the full mapping, a compact summary table in the main text (perhaps as Table 3 or integrated into Section 4.1) listing the ten final categories, exemplar keywords/TF-IDF terms from each cluster, approximate front-end vs. back-end orientation, and linkage to the three disciplinary mechanisms (standardization, dependency, differentiation) would improve immediate accessibility for readers who do not consult the supplement. This need not duplicate every detail—two or three rows per dimension would suffice. 2. Publication-lag and proxy interpretation. The use of WoS-indexed research outputs as a proxy for “technological conditions associated with platform governance” is reasonable but introduces a temporal lag (research often published 1–3 years after internal deployment decisions). A brief acknowledgment in Section 3.2 or 5.4 that the observed associations may partly reflect earlier (pre-2020) R&D trajectories surfacing in the publication record would add nuance, especially given the sharp post-2020 regime shift detected. 3. Generalizability to non-Western platforms. The Discussion (5.3) and Future Works paragraph appropriately flag the need for extension to WeChat, TikTok (ByteDance), and other platforms whose affordances and recommendation logics may differ from the Apple–Amazon–Meta–Google–Microsoft ecosystem. A short additional sentence noting that TikTok’s rapid global scaling of emotion-reaction and infinite-scroll features after 2018–2020 may itself be a partial driver of the breakpoint (captured in the aggregate “emotion-oriented interaction dynamics” category) would help readers situate the Big-Five focus within a broader platform ecology. 4. VAR results summary table. A single, reader-friendly table (or small set of panels) early in the Results section that reports the key persistence coefficients and selected cross-lag associations for the baseline and preferred robustness specifications, pre- and post-2020, would allow quick appraisal of effect magnitudes without requiring readers to navigate between Table 6, Table 8, and the appendix figures. 5. Policy implications paragraph. The Conclusion and the final paragraph of Section 5.2 already gesture toward regulatory attention to structural mechanisms beyond content moderation. Expanding this into one concise, evidence-linked sentence or short paragraph—e.g., referencing the EU Digital Services Act’s systemic-risk assessment obligations or analogous debates in other jurisdictions—would increase the manuscript’s relevance to ongoing policy conversations without overstating causal claims. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #1: No Reviewer #2: Yes: Grigorios L. Kyriakopoulos Reviewer #3: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation. NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications. 21 Jun 2026 Response to the Editor and Reviewers Manuscript ID: PONE-D-26-02998 Dear Dr. Cioffi and Reviewers, Thank you very much for the opportunity to revise our manuscript, “Disciplining the Digital Public: Platform Mechanisms and the Dynamics of Emotional Polarization” (PONE-D-26-02998). We sincerely appreciate the editor’s and reviewers’ thoughtful comments and constructive feedback. We have carefully considered all remarks and revised the manuscript accordingly. We believe these revisions have improved the clarity, rigor, and overall presentation of the paper. Below, we provide our point-by-point responses to the reviewers’ comments. Responses to Reviewer 1 1. Comment: All comments have been addressed. The reviewer states that the major concerns raised in the previous round have been adequately addressed and recommends acceptance. Response: We sincerely thank Reviewer 1 for this positive evaluation and for acknowledging the substantial improvements made in the revised manuscript. We are very pleased that the revisions have satisfactorily addressed the major concerns raised in the previous round, including the justification of the 2020 breakpoint, the more cautious use of causal language, the robustness checks for the emotional polarization proxy, and the expanded methodological description. We greatly appreciate the reviewer’s supportive assessment. Responses to Reviewer 3 Reviewer 3: This revised manuscript represents a meaningful contribution to the growing literature on platform governance, affective dynamics, and the structural mechanisms through which digital platforms may shape public opinion environments. The core theoretical innovation—conceptualizing “platform disciplinary mechanisms” (interface design, algorithmic recommendation, and interaction feedback loops) as processes that foster behavioral standardization, cognitive dependency, and emotionally structured group differentiation—is timely and well-aligned with contemporary debates surrounding algorithmic amplification, short-form video affordances, and the limits of content-moderation-centric regulation. The empirical strategy, while necessarily proxy-based and observational, is thoughtfully designed and now bolstered by formal structural-break testing, extensive robustness checks, and unusually transparent documentation of the NLP pipeline used to derive technology-intensity measures from nearly 50,000 Big Tech-affiliated abstracts. The central finding—that emotional polarization in the information environment (operationalized via GDELT news-sentiment dispersion) exhibits greater persistence after 2020 and stronger associations with technological simplification and emotion-oriented interaction dynamics—offers plausible correlational evidence consistent with the affective-governance thesis. The authors are to be commended for the care with which they have tempered causal language and for the expanded Limitations section that explicitly discusses reverse causality, omitted-variable bias, and the distance between news-tone dispersion and individual psychological states. Strengths of the Revision • Formal breakpoint validation: The Chow test result (F(6, 2461) = 12.507, p < 0.001) provides rigorous statistical grounding for treating 2020 as an analytic regime shift, complementing the rich contextual narrative around COVID-19 discourse migration, short-form video scaling (TikTok/Reels), post-election scrutiny, and user-growth saturation. • Robustness of the polarization proxy: Testing standardized, winsorized, spike-excluded, level, and winsorized-level versions of the dispersion measure, and demonstrating stability of the post-2020 persistence increase across first-difference specifications, directly addresses earlier concerns about scaling artifacts and extreme observations. • Methodological transparency (S1 Text): The new supporting text detailing SentenceTransformer embeddings, UMAP reduction, MiniBatchKMeans (k=10), WCSS/silhouette diagnostics, six-round active-learning protocol (~30 abstracts/round), ChatGPT-assisted labeling with pre-specified guidelines, and the front-end/back-end interpretive heuristic is a model of open science practice. This significantly strengthens construct validity claims for the ten technology categories. • Restructured Discussion: The creation of dedicated subsections 5.1 (Findings by Time Phase) and 5.2 (Findings by Technology Type), together with systematic cross-referencing to the theoretical framework and prior literature, eliminates the earlier impression of isolated subsections and improves interpretive coherence. • Cautious interpretation and limitations: The consistent reframing from causal verbs (“organize,” “shape,” “function as institutional actors”) to associational language (“coincide with,” “are associated with,” “co-evolve with”), plus the explicit paragraph on endogeneity and proxy limitations, aligns the rhetorical claims with the observational design. Minor Suggestions for Further Improvement While the revisions have addressed the major concerns, the following presentational and interpretive refinements would further strengthen the manuscript: Response: We sincerely thank Reviewer 3 for the thoughtful and generous assessment of our revised manuscript, as well as for the helpful minor suggestions. We are very pleased that the revision was recognized as a meaningful contribution with improved theoretical framing, empirical rigor, and transparency. We also appreciate the reviewer’s detailed recognition of the improvements made in the breakpoint analysis, proxy robustness checks, methodological documentation, discussion structure, and caution in interpretation. In the revised manuscript, we have further refined the presentation according to the five minor suggestions, as detailed below. 1. Comment: Main-text technology category overview. Although S1 Text contains the full mapping, a compact summary table in the main text (perhaps as Table 3 or integrated into Section 4.1) listing the ten final categories, exemplar keywords/TF-IDF terms from each cluster, approximate front-end vs. back-end orientation, and linkage to the three disciplinary mechanisms (standardization, dependency, differentiation) would improve immediate accessibility for readers who do not consult the supplement. This need not duplicate every detail-two or three rows per dimension would suffice. Response: We thank the reviewer for this helpful suggestion. We agree that the main text should provide a more immediately accessible overview of how the technology categories relate to the disciplinary-mechanism framework. In response, we revised Table 3 by adding a compact mapping between the ten technology categories, their front-/back-end orientation, and their primary linkage to the three disciplinary mechanisms: behavioral standardization, cognitive dependency, and emotionally structured differentiation. This revision allows readers to understand the interpretive bridge in the main text, while the fuller methodological details and validation notes remain available in S1 Text. (Please see Section 4.1 and Table 3, Pages 33~34) 2. Comment: Publication-lag and proxy interpretation. The use of WoS-indexed research outputs as a proxy for “technological conditions associated with platform governance” is reasonable but introduces a temporal lag (research often published 1-3 years after internal deployment decisions). A brief acknowledgment in Section 3.2 or 5.4 that the observed associations may partly reflect earlier (pre-2020) R&D trajectories surfacing in the publication record would add nuance, especially given the sharp post-2020 regime shift detected. Response: We thank the reviewer for this valuable methodological observation. We agree that WoS-indexed research outputs should be interpreted carefully because they may involve a publication lag relative to internal R&D or deployment decisions. We have therefore added a clarification in Section 3.1 explaining that the technology-intensity series should be understood as publicly visible research-output signals of broader technological conditions, rather than as direct measures of immediate platform feature deployment. The revised text also notes that post-2020 associations may partly reflect R&D trajectories initiated earlier and subsequently appearing in the publication record. (Please see Section 3.1, Page 28) 3. Comment: Generalizability to non-Western platforms. The Discussion (5.3) and Future Works paragraph appropriately flag the need for extension to WeChat, TikTok (ByteDance), and other platforms whose affordances and recommendation logics may differ from the Apple-Amazon-Meta-Google-Microsoft ecosystem. A short additional sentence noting that TikTok’s rapid global scaling of emotion-reaction and infinite-scroll features after 2018-2020 may itself be a partial driver of the breakpoint (captured in the aggregate “emotion-oriented interaction dynamics” category) would help readers situate the Big-Five focus within a broader platform ecology. Response: We thank the reviewer for this important suggestion. We have revised the discussion of generalizability to place the Big Five ecosystem within a broader global platform ecology. The revised text now notes that the rapid scaling of emotion-reaction affordances and infinite-scroll interfaces on platforms such as TikTok during and after 2018–2020 may form part of the broader platform ecology surrounding the observed post-2020 structural shift. We also clarify that the findings should be extended cautiously to other global and non-Western platforms, including WeChat and ByteDance/TikTok, where affordance configurations and governance logics may differ. (Please see Section 5.3 and Section 6, Pages 46~47 and 53) 4. Comment: VAR results summary table. A single, reader-friendly table (or small set of panels) early in the Results section that reports the key persistence coefficients and selected cross-lag associations for the baseline and preferred robustness specifications, pre- and post-2020, would allow quick appraisal of effect magnitudes without requiring readers to navigate between Table 6, Table 8, and the appendix figures. Response: We thank the reviewer for this suggestion, which helps improve the readability of our results. To allow readers to appraise effect magnitudes quickly without having to navigate back and forth between multiple tables, we have added a consolidated summary table near the beginning of the Results section (new Table 7). This table combines the core phase-specific VAR results and proxy-sensitivity results in a panel format: Panel A reports the emotional-polarization persistence coefficient (Dstd, lag 1) under the baseline specification and five robustness specifications, with its pre- and post-2020 values and their difference; Panel B reports the lag-1 coefficients of the selected technology categories (Type0, Type3, Type4, Type5) on emotional polarization in the baseline VAR, together with their pre- and post-2020 changes. All values in the table are drawn directly from the existing VAR estimates and proxy-sensitivity checks; the table introduces no new analysis and serves only as a consolidated quick reference. The full coefficient estimates, standard errors, and supplementary robustness results remain available in S1 Appendix. We have also added a brief signpost at the beginning of Section 4.3 directing readers to this summary table for a rapid comparison of effect magnitudes across phases and specifications. (Please see Section 4.3 and the new Table 7, Pages 40~41) 5. Comment: Policy implications paragraph. The Conclusion and the final paragraph of Section 5.2 already gesture toward regulatory attention to structural mechanisms beyond content moderation. Expanding this into one concise, evidence-linked sentence or short paragraph-e.g., referencing the EU Digital Services Act’s systemic-risk assessment obligations or analogous debates in other jurisdictions-would increase the manuscript’s relevance to ongoing policy conversations without overstating causal claims. Response: We thank the reviewer for this constructive recommendation. We agree that the manuscript can more clearly connect the findings to ongoing policy debates without overstating causal claims. We have therefore expanded the policy implications paragraph by referring to the EU Digital Services Act’s systemic-risk assessment framework and by explaining how our findings speak to structural platform mechanisms beyond content moderation, including algorithmic recommendation, interaction-feedback loops, and affective engagement infrastructures. We have kept the interpretation cautious by emphasizing that the study provides observational and proxy-based evidence rather than direct causal proof. (Please see Section 5.2, Pages 45~46) Once again, we sincerely thank the editor and reviewers for their constructive feedback. Their comments have substantially improved the manuscript. Submitted filename: Response_to_Reviewers_auresp_2.docx 9 Jul 2026

Disciplining the Digital Public Platform Mechanisms and the Dynamics of Emotional Polarization PONE-D-26-02998R2 Dear Dr. LU, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Andrea Cioffi Academic Editor PLOS One Additional Editor Comments (optional): Reviewers' comments: Reviewer's Responses to Questions Comments to the Author Reviewer #3: All comments have been addressed ********** 2. Is the manuscript technically sound, and do the data support the conclusions??> Reviewer #3: Yes ********** 3. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #3: Yes ********** 4. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #3: Yes ********** 5. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #3: Yes ********** Reviewer #3: This revised manuscript represents a meaningful and timely contribution to the literature on platform governance, affective dynamics, and the structural mechanisms through which digital platforms shape public-opinion environments. The core theoretical innovation—platform disciplinary mechanisms (interface design, algorithmic recommendation, interaction feedback loops) fostering behavioral standardization, cognitive dependency, and emotionally structured group differentiation—is conceptually sharp and well-aligned with contemporary debates on algorithmic amplification, short-form video affordances, and the limits of content-moderation-centric regulation. The empirical strategy, while necessarily proxy-based and observational, is thoughtfully designed and now further strengthened by formal structural-break testing (Chow test), extensive robustness checks across multiple specifications of the polarization proxy, and unusually transparent documentation of the NLP pipeline (S1 Text). The central finding—that emotional polarization in the information environment exhibits greater persistence after 2020 and stronger associations with technological simplification and emotion-oriented interaction dynamics—offers plausible correlational evidence consistent with the affective-governance thesis. The authors have consistently tempered causal language and expanded the Limitations section to address reverse causality, omitted-variable bias, proxy distance from individual psychology, and publication lag. The R2 revisions directly and effectively address all five minor suggestions from Reviewer 3, resulting in a more accessible, interpretable, and policy-relevant manuscript. The addition of the consolidated VAR summary Table 7 is particularly helpful for readers. The manuscript is now ready for acceptance. Strengths of the Revised Manuscript (R2) 1. Theoretical contribution & framework clarity: The platform disciplinary mechanisms concept, linked to three concrete processes (standardization, dependency, differentiation), provides a coherent bridge between platform affordances/governance and affective polarization dynamics. Table 3 now makes this mapping immediately accessible in the main text. 2. Methodological transparency & open science: S1 Text’s detailed documentation of embeddings, dimensionality reduction, clustering validation, active-learning protocol (~30 abstracts/round × 6 rounds), and ChatGPT-assisted labeling with pre-specified guidelines is a model of reproducible qualitative-to-quantitative text analysis. This significantly bolsters construct validity for the ten technology categories. 3. Empirical rigor & robustness: Chow test provides statistical grounding for the 2020 breakpoint. Multiple proxy specifications (standardized, winsorized, spike-excluded, first-differenced) and the new consolidated Table 7 demonstrate stability of the post-2020 persistence increase. Publication-lag nuance is now explicitly acknowledged. 4. Cautious, evidence-aligned interpretation: Consistent shift to associational language, explicit discussion of endogeneity/proxy limitations, and careful framing of policy implications (observational evidence informing structural-mechanism attention under frameworks such as the EU DSA) align rhetorical claims with the observational design. 5. Improved readability & coherence: Dedicated subsections 5.1 (Findings by Time Phase) and 5.2 (Findings by Technology Type) with systematic cross-referencing to theory, plus the early Results summary Table 7, eliminate earlier impressions of isolated findings and enable quick appraisal of effect magnitudes. Remaining Minor / Optional Suggestions No substantive issues remain. The following are very minor presentational or forward-looking suggestions for the authors’ optional consideration prior to final proofing: • Table/figure numbering & cross-references: Confirm that all in-text references to the new Table 7 and revised Table 3 are correctly numbered and that supplementary tables in S1 Appendix retain clear, non-conflicting numbering. • Policy implications brevity: The expanded paragraph referencing the EU DSA is appropriately cautious; if the journal imposes strict word limits, the authors could consider moving one supporting clause to a footnote while retaining the core linkage to systemic-risk assessment of structural mechanisms. • Future-work specificity: The call for extension to WeChat, TikTok/ByteDance, etc., is well-placed. A single additional sentence sketching one concrete comparative dimension (e.g., differing infinite-scroll vs. algorithmic curation emphases) could further sharpen the agenda without lengthening the section materially. • Final proofread: A light final pass to ensure uniform spacing, quotation formatting, and reference style (Vancouver/NLM) is recommended, though the manuscript already appears polished. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #3: Yes: Prof Nirmal Kumar Ganguly ********** Submitted filename: PLOS_ONE_Reviewer Prof N K Ganguly_Report.docx PONE-D-26-02998R2 PLOS One Dear Dr. LU, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS One. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing. If we can help with anything else, please email us at customercare@plos.org. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Dr. Andrea Cioffi Academic Editor PLOS One