In Indonesia, Liquefied Petroleum Gas (LPG) subsidies—specifically the distribution of 3-kilogram cylinders—constitute a critical social policy designed to secure affordable access to household fuel for low-income families, informal workers, and small vendors whose livelihoods depend on cooking and small-scale production. The subsidy program, administered through the Indonesian Ministry of Energy and Mineral Resources, ensures that these cylinders are sold at below-market prices, thereby functioning as a material safety net in everyday life. When the government temporarily suspended retail sales of subsidized LPG in February 2025, restricting distribution to designated agents, the policy disruption immediately triggered widespread unrest: households faced shortages, queues lengthened, and small businesses were unable to operate. This episode illustrates how a technical adjustment in subsidy targeting rapidly escalated into a livelihood crisis, underscoring the socio-economic significance of LPG subsidies as more than a fiscal instrument but as a cornerstone of energy justice and social protection in Indonesia. The sentiment categorization scheme used in this study is presented in Table 1.

The suspension of subsidized LPG distribution rapidly became a focal point of national debate, with mainstream media outlets extensively covering the policy reversal and amplifying public discontent across television, print, and online platforms. This visibility was further intensified through social media, where news clips, citizen protests, and viral commentary circulated widely, transforming a technical energy policy into a matter of collective grievance and civic mobilization. The convergence of traditional and digital media not only magnified the regulatory adjustment but also facilitated broad societal engagement, as audiences actively responded, shared, and debated the issue in real time. This dynamic underscores how policy disruptions in Indonesia are increasingly mediated through hybrid information ecosystems, setting the stage for a deeper examination of how social media reshapes journalism and audience participation.

Furthermore, the adoption of social media in journalism has provided not only a new way to distribute news but also to reach a wider audience beyond geographical boundaries. It has also brought about a fundamental shift in the dynamic between the media and its audience. The relationship can no longer be seen as one of producer (journalism) and consumer (audience); instead, it has transformed the static, one-way communication model into a more interactive and participatory journalism-audience ecosystem (Bossio, 2017; Halse et al., 2018; Kalsnes and Larsson, 2018; Kubin et al., 2024; Van Dijk, 2006). In contrast to audiences in traditional media, in digital media, interaction becomes real-time, and audiences can be said to participate in the news (Schrøder, 2019). The presence of social media then provides an extension of audience involvement (Bossio, 2017). This transformation of digital interaction aligns with cybermedia theory, which conceptualizes online spaces as dynamic arenas shaped by user participation and technological affordances (Nasrullah, 2014). Social media has become an alternative for various groups to get information apart from traditional media sources (Vázquez-Herrero et al., 2022). People can comment, like, or spread content in the form of news uploaded by the media industry on social media, which is a form of audience engagement with the media (Tenenboim, 2022).

The collaboration of journalism and social media activities has created new values among its users, one of which is the use of social media as a medium for online citizen aspirations (Saco, 2002; Wilhelm, 2000). Therefore, an interesting note given by Rycroft (2007) is that the virtual space on the internet encourages the emergence of a political culture? This culture operates in a new public space, which is a virtual space where values are exchanged among members (Habermas, 1974; Jones, 2012; Papacharisi, 2002). Unlike the space provided by traditional mass media, the presence of social media used by the media industry in disseminating news as an alternative media provides a kind of open space that can be used by anyone to convey their aspirations and criticisms (Camp and Chien, 2000; Papacharisi, 2002). There is also no hierarchy in social media, and it is as if there is no longer a distinction between producers and consumers of information that has existed, directly or indirectly, in traditional media. A topic can be created by anyone and discussed as a topic of debate, followed by anyone (Jordan, 1999). A detailed chronology of these policy events is shown in Table 2.

Comment sections in digital journalism illustrate how audiences move beyond passive consumption to active participation. They not only signal loyalty to news outlets (Krebs and Lischka, 2019) but also serve as spaces for fact-checking, providing additional information (Lee, 2012; Liu and McLeod, 2019), expressing opinions (Waddell, 2018, 2019), and engaging in peer discussion (Barnes et al., 2018; Fletcher and Park, 2017; Wardle and Williams, 2010; Zamith and Lewis, 2014). Media organizations have long experimented with such interactivity—for example, editorial blogs launched by the BBC, CBS, The Guardian and The New York Times in the 2000s allowed audiences to question and comment on editorial positions (Sterling, 2009). Whether or not to enable commentary remains at the discretion of the newsroom (Robinson, 2010), yet its presence significantly shapes how news is received and debated.

Social media amplifies this participatory dynamic by extending the public sphere into digital spaces. Audience engagement through likes, shares, and comments has measurable cognitive and behavioral impacts, motivating individuals to influence others and reinforcing their role within communities (Guo and Chen, 2022). Inter-coder reliability statistics are summarized in Table 3. In political contexts, however, these platforms also facilitate the circulation of misinformation, hate speech, and polarization. Reader participation thus enhances the value of news by creating deliberative spaces (Ziegele et al., 2017), but it also introduces challenges that require technological filtering to manage sentiment and abusive language (Yahav and Schwartz, 2018).

At the same time, social media functions as a source of news material for the industry itself. The rise of user-generated content (UGC) has blurred the boundary between professional journalism and citizen reporting. Events such as the Iraq War in 2003, the London bombings in 2005, and the Egyptian reforms in 2011 demonstrated how audience-produced content circulated globally and was incorporated into mainstream coverage (Fuchs, 2014; Gillmor, 2004; Heinrich, 2011; Pavlik, 2001; Tuten, 2008). Audiences once regarded solely as receivers of news have increasingly become sources, commentators, and even reporters, with blogs and social media posts frequently quoted as part of news discourse (Jahng, 2018).

Audiences with the power of mobile phones, connections to the online world and the existence of social media are a combination of flexibility in reporting and disseminating reports of events in the field. Heinrich (2011) offers the concept of “network journalism” to explain how professional journalism workers collaborate with citizens in disseminating information or news. Although similar names, such as participatory journalism, production, and crowd-sourced journalism, are ambiguous in the practice of journalism related to the reporting system (Scott et al., 2014), at least the participation of the audience and news comments on social media cannot be ignored in current journalism practices (Hille and Bakker, 2014) because comments show not only the audience's engagement with the media but also the engagement among fellow audiences (Almgren and Olsson, 2015; Tran and Ngo, 2018).

This study is guided by the theoretical lens of Appraisal Theory (Martin and White, 2005), which provides a framework for analyzing how language encodes attitudes, emotions, and evaluative stances. By situating audience comments within the broader scholarship on the digital public sphere, anonymity, user-generated content, and participatory journalism, the study emphasizes that online discourse is not only informational but also affective and symbolic. Appraisal Theory enables the categorization of sentiments—such as hostility, critique, or support—while highlighting how emotional intensity and evaluative language function as markers of civic frustration, symbolic contestation, and digital activism. This conceptual orientation positions comment sections as arenas of affective politics, where emotional discourse intersects with participatory governance and shapes the dynamics of public engagement in digital spaces.

Furthermore, comment sections can also reflect the emotional aspects of the readers themselves (Graf et al., 2017). The presence of anonymous accounts or profiles in news comment sections is a concern (Barnes et al., 2018). Among them are providing negative comments and reducing the credibility of the news (Waddell, 2018). In Christopherson's (2007) view communication on the internet allows for the presence of anonymous accounts based on technical anonymity and social anonymity. Technical anonymity refers to the absence of any information approaching the account owner, while social anonymity operationalizes anonymity with pseudonyms, nicknames, or by comparing the absence of photos with photos (Graf et al., 2017).

The various comments that appear in the comment fields on the studied social media will be categorized according to the sentiment categories, as adapted from the development of sentiment categories by Amarasekara and Grant (2019).

The open nature of online news comment sections allows for the expression of public opinion, but it is also susceptible to the display of negative sentiments, often conveyed through offensive or abusive language (Sohn et al., 2019). The sentiments found within news commentary columns are a clear illustration of how the audience favors or disfavours media content (Lee, 2012; Waddell and Bailey, 2017). It should be noted that the tone of an audience's comments does not affect perception (Kim and Masullo Chen, 2021). The audience is divided into two camps, namely, supportive of like-minded content and non-supportive of counter-attitudinal content (Guo and Chen, 2022). It has been demonstrated that comments on news articles do more than simply express public opinion. Findings from scholarly research further suggest that these forums are exploited for personal attacks (Chen, 2019; Thorne and Reinhardt, 2008). However, several studies have indicated that news websites are removing comment sections and directing commentary toward social media platforms, with the intention of sidestepping negative sentiment and discourteous language (Liu and McLeod, 2019). A general overview of the YouTube comment corpus is presented in Table 4.

Despite extensive scholarship on audience participation in digital journalism, several gaps remain. Prior studies have largely emphasized engagement metrics, political polarization, or fact-checking practices, while the emotional intensity and ad hominem attacks that dominate online discourse have received limited systematic attention. Moreover, ethnographic approaches to comment analysis are rarely applied to YouTube, leaving unexplored the cultural and symbolic dimensions of audience reactions on this platform. Scholarship on the digital public sphere has also tended to privilege deliberation and information exchange, overlooking the emergence of symbolic figures who embody public frustration and become rallying points for civic sentiment. The distribution of comments across the six sentiment categories appears in Table 5. In addition, Southeast Asian policy controversies—particularly those involving socio-economic issues such as LPG subsidies—are underrepresented in comparative research, and the role of humor, sarcasm, and emojis as shorthand affective markers remains peripheral in existing analyses. This study addresses these gaps by employing ECA to examine 426 YouTube comments on the LPG subsidy regulatory adjustment in Indonesia, foregrounding emotional discourse, symbolic contestation, and affective communication. In doing so, it contributes to a deeper understanding of how digital platforms function as arenas for emotional release, civic engagement, and participatory governance in social media contexts. Descriptive statistics for comments categorized under appearance sentiment are shown in Table 6.

The objective of this study is to investigate how audiences emotionally respond to the LPG subsidy regulatory adjustment on YouTube, categorizing sentiments through ECA and highlighting the role of symbolic contestation in digital civic engagement. This study contributes to the literature by foregrounding emotional discourse and symbolic contestation in digital audience participation, applying ECA to YouTube comments on the LPG subsidy regulatory adjustment in Indonesia. By highlighting how audiences' sentiment intersects with civic engagement, the research expands and underscores the role of digital platforms as arenas for emotional release, symbolic activism, and participatory governance.

The LPG subsidy regulatory adjustment in Indonesia is theoretically significant because it demonstrates how governance decisions are increasingly shaped by public sentiment expressed in digital arenas. The temporary suspension of 3-kilogram LPG cylinders not only disrupted household and small business routines but also triggered extensive coverage across mainstream media outlets throughout the country. Simultaneously, the incident went viral on social media platforms, where discussions of scarcity and citizen protests circulated widely and rapidly. This dual visibility—through both mass media and participatory digital spaces—amplified the regulatory adjustment and compelled policymakers to respond in real time. For digital governance research, the case illustrates how emotional discourse and symbolic contestation in comment sections and viral debates function as forms of affective accountability, highlighting the central role of digital publics in shaping state responsiveness and policy communication. The characteristics of comments categorized as sexual and sexist sentiment are provided in Table 7.

The LPG subsidy regulatory adjustment in Indonesia unfolded within a compressed timeline in February 2025. On 1 February, the Ministry of Energy and Mineral Resources announced a ban on retail sales of subsidized 3-kilogram cylinders, triggering public unrest and extensive media coverage on 2–3 February as households and small businesses faced acute shortages. In response to escalating protests and viral dissemination across social media, President Prabowo Subianto revoked the regulation on 4 February, restoring retail distribution (ANTARA, 2025; Kompas.com, 2025; Kontan.co.id, 2025).

To capture audience sentiment during this critical period, 426 YouTube comments were collected (MetroTV, 2025) between February 5 and March 1, 2025, immediately after the government's policy reversal on February 4, 2025, ensuring that the dataset reflects audience reactions in the aftermath of the suspension and its rapid revocation rather than during the initial announcement phase. A descriptive summary of hostile comments is included in Table 8. The dataset included both primary comments and replies, each treated as an independent unit of analysis, allowing the study to capture the sentiment landscape in full complexity. Replies—while sometimes functioning as affirmations or reinforcements of existing views—could also introduce new perspectives, tones, or discursive strategies, thereby encompassing direct reactions to the news clip as well as dialogic exchanges among users.

After gathering 426 YouTube comments, the sentiment framework was applied independently by two coders. The diagram illustrates the workflow from data collection to reliability outcome.

The comparison of coding outcomes demonstrates that the two evaluators reached near-perfect agreement across all sentiment categories. The observed discrepancies account for < 3% of the total corpus (426 comments), affirming both the clarity of the coding framework and the consistency of its application. The most notable variation appeared in the General/Neutral category, where differing interpretations of ambiguous comments produced minor divergence. Crucially, the overall Krippendorff's Alpha of 0.993 confirms exceptionally strong reliability, ensuring that these small differences do not undermine the dataset's integrity. Rather, they illustrate the interpretive nuance characteristic of qualitative content analysis, while simultaneously reinforcing the study's methodological rigor and robustness. To capture the dynamics of this digital contestation, the study employs the sentiment categories developed by Amarasekara and Grant (2019). These categories are particularly appropriate for Indonesian political discourse, where rational critique often coexists with sarcasm, humor, ad hominem attacks, and symbolic support for ordinary citizens. By applying this framework, the study is able to document both substantive policy critiques and affective expressions, showing how emotional discourse and symbolic contestation in comment sections function as mechanisms of civic engagement and affective accountability in digital governance.

The ECA approach is utilized to document and comprehend how meaning is generated in social media responses following news coverage. Within this study, meaning extends beyond mere textual analysis; it is also derived from the choice of phrasing, message format, employment of capitalization, and the contextual or situational background. The method maintains a dynamic nature, continuously cycling through concept development, sampling, data collection, coding, analysis, and interpretation (Altheide, 1987). Furthermore, this method is fundamentally based on the concept of constant discovery and constant comparison across the range of emerging user comments. The primary goal is to both uncover and contrast the relevant situations, communicative styles, underlying meanings, and nuances. Descriptive statistics for positive sentiment comments are presented in Table 9. These elements are subsequently managed through a process of coding and final categorization based on the commonalities in expressed sentiment.

ECA is utilized not just to convey countable, quantitative data, but to deliver rich descriptive information on sentiments arising from social media news. From a social and media perspective, this ethnographic approach to digital media helps to investigate the social practices, norms, and experiences that ultimately unveil the routines, interactions, processes, and identities that are formed and performed in the digital domain (Bakry et al., 2022; Forberg and Schilt, 2023; Hine, 2015; Shumar and Madison, 2013). The main focus of this research is on the first domain of attitude, subdivided by affect, which signifies emotional reactions in the comments given (Martin and White, 2005; Tran and Ngo, 2018) (Figure 1).

This research began with the systematic collection of 426 YouTube comments related to the LPG subsidy regulatory adjustment, which served as the primary dataset. These comments were then categorized using a sentiment framework adapted from Amarasekara and Grant (2019), encompassing six categories: appearance, sexual/sexist, hostile, positive, critiques/negative, and general/neutral. Two independent coders applied this framework to the corpus, paying close attention not only to textual meaning but also to communicative markers such as phrasing, capitalization, and emoji use. Reliability was tested through Krippendorff's Alpha, yielding a near-perfect score of 0.993, which confirmed the consistency of the coding scheme. Beyond quantitative classification, the ECA approach emphasized interpretive ethnography, examining communicative styles, symbolic figures, and cultural practices embedded in digital discourse. A detailed overview of critique and negative sentiment comments is shown in Table 10. The method followed a dynamic cycle of constant discovery and comparison—concept development, sampling, coding, analysis, and interpretation—ensuring both methodological rigor and qualitative richness. Ultimately, this procedure allowed the study to foreground emotional discourse and symbolic contestation, demonstrating how YouTube comment sections function as arenas for emotional release, civic engagement, and participatory governance.

The analysis in this study focused on all comments posted on YouTube in response to the LPG subsidy regulatory adjustment, including replies to comments. Both primary comments and replies were treated as independent units of analysis. This methodological choice was based on the understanding that replies, while sometimes functioning as affirmations or reinforcements of existing views, may also introduce new perspectives, tones, or discursive strategies that enrich the overall sentiment landscape. By incorporating replies alongside original comments, the study ensured that the analysis captured sentiment as a multifaceted and independent expression of audience reaction.

The temporary chaos in the distribution of 3-kilogram gas cylinders began with a policy announcement from the Indonesian government. On February 1, 2025, Deputy Minister of Energy and Mineral Resources, Yuliot Tanjung, stated that the sale of these subsidized cylinders, commonly known as tabung melon [“melon cylinder”] due to its melon-like shape, would no longer be permitted through retailers. This policy, according to various news reports, was introduced to improve the targeting of subsidized gas, which was originally intended exclusively for low-income households. The regulations stipulated that distribution would be restricted to government-appointed distributors or official agents.

These restrictions, however, quickly led to widespread public unrest. Stocks of 3-kilogram cylinders at retailers and small shops dwindled or vanished, causing significant disruption for households, small traders, and home-based businesses. This resulted in long queues for gas across various Indonesian cities.

In response to the regulatory adjustment, Minister of Energy and Mineral Resources, Bahlil Lahadia, appeared before the House of Representatives (DPR) on February 3, 2025. The meeting concluded with a decision to improve the management of the gas distribution system. The following day, President Prabowo Subianto officially lifted the ban on sales through retailers, effectively reversing the earlier policy.

Despite this reversal, public outcry had already erupted on social and mainstream media platforms. On February 4, 2025, during an unscheduled inspection of a gas distribution center, Minister Bahlil personally witnessed public outrage. A 57-year-old resident and daily wage worker, identified as Effendy, openly criticized the minister and urged the government to implement policies that do not burden home-based producers.

This incident, a rare public outburst directed at a government official, quickly went viral. It was widely covered by national media, including television, print, and online outlets. Google Trends data from February 3–5, 2025, documented “Bahlil” as a top-searched keyword related to the gas distribution turmoil. The confrontation was notably broadcast by MetroTV on its Metro Pagi Primetime program and later uploaded to the official MetroTV YouTube channel.

The content uploaded ultimately received responses from YouTube users. In less than a week, various comments were posted on the news clip about the residents' anger toward a minister. Descriptive statistics for general and neutral sentiment comments are summarized in Table 11.

Comments responding to YouTube content related to the regulatory adjustment over the distribution of 3-kilogram LPG reflect a wide spectrum of audience reactions, each carrying distinct characteristics that represent individual perspectives. The average comment length is 13.69 words, with the shortest consisting of only three words and the longest extending to 120 words. This suggests that, in digital spaces, audiences generally prefer brief and direct expressions, often reflecting immediate emotional responses. This tendency is further illustrated by the frequent use of emojis—😂, 😅, and 🤣—which dominate among the 426 comments analyzed. These emojis highlight how sarcasm and humor emerge as rapid, shorthand forms of engagement, often outpacing verbal expression.

The emotions of the audience expressed through comments, both in text and emojis, can then be categorized based on sentiments as shown in Table 4.

Each YouTube comment was coded once into a single sentiment category. Categories were treated as mutually exclusive, with coders assigning each comment to the dominant sentiment expressed. Even when comments contained overlapping features, such as hostile language combined with appearance-based remarks, the primary sentiment was identified and coded accordingly. This ensured clarity, consistency, and allowed for reliable calculation of Krippendorff's Alpha.

The findings demonstrate that YouTube comment sections function not merely as participatory extensions of journalism but as arenas where emotional discourse and symbolic contestation shape civic engagement. Rather than passive consumption, comments reveal active involvement through sarcasm, ad hominem attacks, and symbolic support for ordinary citizens, extending the digital public sphere beyond deliberation into affective politics.

Emotional discourse and symbolic contestation evident in the comments operate as mechanisms of civic accountability. Sarcasm, humor, and ad hominem attacks not only express frustration but also transform ordinary citizens into symbolic figures who embody collective grievances. These affective responses highlight how digital publics hold policymakers accountable, demonstrating that emotional intensity itself can function as a form of political participation and governance oversight.

This contribution is particularly significant because YouTube has rarely been examined ethnographically, and its comment culture illustrates how digital publics articulate frustration and accountability in distinctive ways. The platform's participatory affordances highlight cultural practices of engagement that differ from other social media, underscoring its importance as a site for symbolic activism and affective politics.

Based on the findings and preceding discussion, several recommendations can be proposed to improve the quality of public communication and the management of digital spaces, particularly in the context of citizen participation through social media. One essential step is the enhancement of digital literacy, which should not only focus on technical competencies but also emphasize ethical communication, empathy, and critical thinking. A comprehensive approach to digital literacy will empower citizens to engage constructively in public discourse, avoid destructive expressions, and foster healthier dialogue. In addition, mass media and digital platforms must implement more advanced comment moderation systems, ideally powered by artificial intelligence, to systematically filter out comments that are sexual, physically abusive, or contain hate speech. Such moderation should be supported by clear editorial policies to ensure that comment sections remain productive spaces for discussion rather than arenas of verbal violence.

Within this framework, it is equally important to empower citizens as public communication actors. Figures like Effendy, who emerged as a symbol of courage in voicing public grievances, demonstrate that ordinary individuals can play a meaningful role in representing collective concerns. Therefore, governments and media institutions should establish more formal and structured channels for public participation, such as policy forums, online consultations, and citizen feedback platforms. To mitigate the negative effects of anonymity, which often encourages destructive commentary, social media and news platforms could develop verified discussion features that require identity confirmation. This would promote accountability and ethical engagement in digital communication.

Furthermore, continued research into the dynamics of emotion and digital identity is essential. Interdisciplinary studies combining political communication, social psychology, and digital anthropology can offer deeper insights into how collective emotions are formed and expressed online, and how these affect perceptions of policy and governmental legitimacy. Lastly, governments should consider integrating sentiment analysis from social media as a real-time tool for evaluating public policy. Spontaneous and widespread citizen commentary on digital platforms can serve as an early indicator of policy effectiveness, public reception, and societal impact. By incorporating these insights into decision-making processes, policymakers can become more responsive to public needs and reinforce accountability in governance. Overall, these recommendations aim to build a healthier, more inclusive, and democratic digital communication ecosystem, where citizen participation is not only valued but systematically accommodated in the development of public policy.

These findings resonate with Habermasian notions of the public sphere while simultaneously complicating them through emotion and symbolism. By foregrounding affective discourse and symbolic contestation, the study extends theoretical understandings of digital publics, demonstrating that civic engagement in online spaces is shaped not only by deliberation but also by emotional intensity and symbolic representation. Taken together, these insights conclude the discussion and set the stage for the study's final implications.

This study demonstrates that YouTube comment sections on the LPG subsidy regulatory adjustment in Indonesia operate as arenas of affective politics, where emotional discourse and symbolic contestation shape civic engagement. By applying ECA, the research highlights how sarcasm, humor, ad hominem attacks, and symbolic support for ordinary citizens function as mechanisms of civic accountability. These findings extend theoretical understandings of the digital public sphere, showing that online participation is driven not only by deliberation but also by emotional intensity and symbolic representation.

Beyond theoretical contributions, the study offers practical implications for improving digital communication ecosystems. Enhancing digital literacy, strengthening moderation systems, and creating structured channels for citizen participation are essential steps toward fostering constructive engagement. Moreover, integrating sentiment analysis into policy evaluation can help governments respond more effectively to public needs and reinforce accountability in governance.

Overall, the research underscores the importance of recognizing emotion and symbolism as central dimensions of digital civic participation. By situating YouTube comments within broader debates on digital journalism, digital public communication, and governance, the study contributes to building a healthier, more inclusive, and democratic digital environment where citizen voices are acknowledged as integral to policy discourse.