Public discourses and government interventions behind China’s ambitious carbon neutrality goal

The study investigates public discourse surrounding China’s 2060 carbon neutrality goal and evaluates how government interventions shape public opinion online. Using Weibo as the public sphere, the research asks: which frames dominate public conversations about carbon neutrality (scientific, moral, economic, co-benefit, energy security, political, global), how attitudes (support, oppose, neutral) distribute and evolve, and to what extent government climate policies and political events influence discourse and consensus. Given the top-down nature of target setting in China but the importance of public support for costly climate policies, the work aims to map discourse dynamics, identify key influencers, and assess the relative effectiveness of concrete policy actions versus political signaling in mobilizing public engagement.

The paper situates itself within scholarship on public opinion and climate communication, highlighting prior evidence that public discontent can influence Chinese environmental policymaking and that framing is central to public engagement on climate issues. It notes that social media platforms like Weibo function as a public sphere for climate discussion in China and that climate skepticism online is typically marginal and often driven by individuals rather than institutions, consistent with earlier studies. The work also connects to research on polarization around climate change on social media and on how government policies and investment can affect public support for climate policy-making.

Data: The authors collected 996,675 Weibo posts using curated climate-related keywords and, after filtering unrelated content, analyzed 981,348 active posts from 235,467 accounts in China between March 2021 and December 2021 (covering the period after China’s September 2020 carbon neutrality announcement). Keywords were refined through an iterative process including expert review (e.g., climate change, climate crisis, global warming, double carbon, carbon emissions, climate action, carbon neutrality, carbon peaking, climate hoax, low-carbon development, energy transition, climate weapon, climate finance, climate summit). User types (individual vs. institutional) followed Weibo’s verification mechanism.

Coding and models: Posts were coded into seven frames (scientific, moral, economic, co-benefit, energy security, political, global) and three attitudes (support, oppose/object, neutral) using a three-stage active learning procedure combining human annotation and fine-tuning of a Chinese-MacBERT-Large model (a pre-trained Chinese RoBERTa). In each iteration, 300 posts were sampled, predicted by the current model (frames and attitudes with probabilities), human-coded with adjudication between two experts, and used to fine-tune the model. Iterations continued until performance stabilized. Final performance: frames model accuracy 84.89% with Cohen’s kappa 0.804; attitudes model accuracy 82.06% with kappa 0.613. Both accuracy and kappa were reported due to their differing assumptions about rater agreement.

Consensus metric: Internal frame consensus and internal attitude consensus were measured via normalized information entropy, where higher entropy indicated greater consensus (alignment) within a community’s distribution over frames or attitudes. Monthly dynamics of consensus were tracked across all users and by subgroups (institutions vs. individuals; supporters vs. neutral vs. opposers).

Event analysis: The study linked discourse changes to three categories of government actions: (1) major climate policies (e.g., halting financing for overseas coal power plants; the “1 + N” policy framework), (2) domestic political events (e.g., high-level meetings and speeches), and (3) international political events. Effects were evaluated using changes in posting volumes, attitudes, and consensus around event dates.

Data/code availability: Processed data are available at http://doi.org/10.6084/m9.figshare.24460692; code available on request.

• Participation and creators: Since China’s 2060 neutrality goal announcement, carbon neutrality accounts and content quadrupled, though only 0.006% of all Weibo users actively engaged. Individuals comprised 90% of creators, producing 75.20% of posts; institutions were fewer but more influential (higher average reposts). Climate posts by individuals increased markedly, highlighting the public’s growing role.
• Attitudes: Over 70% of posts clearly support the neutrality goal; only about 3% oppose. Supportive posts gain, on average, five times more reposts than opposing posts. Institutions’ attitudes are stable; individuals’ attitudes are more volatile over time.
• Frames: Among seven frames, economic, moral, and global dominate (>80% of posts). Security- and co-benefit-related frames are least common. Overall internal frame consensus is low; institutional actors frame issues more evenly and stably. Individuals emphasize economic and moral frames and less politics; a spike in the economic frame in March 2021 quickly faded.
• Opposers vs. supporters vs. neutral: 59.7% of opposing posts rely on global and moral frames, including arguments about Western hypocrisy and constraints on China’s development, and skepticism tied to local weather experiences. Supporters distribute frames more evenly. After March 2021, neutral users converge strongly on the economic frame and show the highest internal consensus over time.
• Declining consensus among opposers: Opposers exhibit a significantly declining internal frame consensus, indicating growing internal division.
• Influencers: The top 300 accounts (about 1.3%) created 3.25% of posts but generated 92.26% of reposts, underscoring disproportionate influence. Among individual influencers (often internet celebrities), focus converged on the economic frame; moral storytelling was marginal except for a brief October 2021 uptick. Institutional influencers include highly impactful international organizations and NGOs that leverage celebrity ambassadors (e.g., WWF with Zhu Yilong; UNEP with Wang Junkai). Prominent SOEs (e.g., State Grid, Jinneng) actively supported the neutrality target. Media were influential but less so than NGOs/IOs; government agencies posted heavily but earned fewer reposts, likely due to neutral/repetitive content and messaging style.
• Government agencies’ framing and influence: Ministries framed content according to mandates: MFA primarily global; NDRC political and economic; MEE political and global; MIIT favored transition-related framing; MOST emphasized scientific; SASAC leaned on energy security. MFA was the most influential government actor by reposts, outshining MEE and NDRC.
• Policy vs. politics impacts: Major, credible climate policies (stopping overseas coal financing; “1 + N” framework) drove multi-day increases in posts and energized both supporters and opposers. Symbolic policy inclusions (e.g., Five-Year Plan mentions) drew little attention. Domestic political events without concrete measures did not boost posts or had only short-lived spikes. International political events produced short-term double-sided impacts. Inconsistent domestic political messaging (e.g., emphasizing both neutrality and coal in Dec 2021) increased confusion, mobilized opposers, and reduced consensus.

Findings show broad online support for China’s carbon neutrality goal among engaged Weibo users, with opposition minimal and increasingly fragmented. Public discourse is dominated by economic, moral, and global frames, with institutions offering steadier framing than individuals. Neutrals are especially responsive to economic narratives, suggesting a path for mobilization, though this frame’s salience can quickly wane with evolving economic conditions. Government communication and action affect discourse differently: concrete, credible policies substantially shape public engagement and simultaneously activate supporters and opposers, potentially reducing consensus in the short term, while political signaling without measures has limited mobilizing power. The influencer ecosystem—particularly NGOs/IOs and certain SOEs, along with celebrity-driven amplification—plays an outsized role in shaping visibility and framing. The MFA’s dominance among government influencers underscores the importance of international framing in China’s climate discourse. Overall, the results address the research questions by mapping the framing landscape, identifying key actors, and demonstrating that substantive policy actions are more effective than political rhetoric in influencing public views and engagement on carbon neutrality.

The study provides a large-scale, data-driven analysis of China’s online discourse on carbon neutrality, revealing high support levels, dominance of economic/moral/global frames, and the pivotal role of top influencers. It shows that credible climate policies, more than political rhetoric, mobilize public engagement and shape attitudes, while opposition remains small and increasingly divided. Policy implications include prioritizing clear, credible measures and deploying economic framing to engage neutral audiences, while anticipating and mitigating backlash from opposers when new policies are introduced. Future research could expand across platforms, probe micro-level mechanisms among different actor types, incorporate exogenous factors (e.g., extreme weather, technology), and assess feedback from public discourse into policymaking processes.

• Platform scope: Data are from Weibo only, introducing selection bias and potentially missing populations active on other platforms; cross-platform comparisons are needed.
• Micro-level mechanisms: Lack of data limited exploration of detailed micro explanations across actor types.
• Omitted factors: The focus on government influence overlooks other drivers (e.g., extreme weather events, technological progress) that may affect public views concurrently.
• Policymaking feedback: The study does not examine how public views might reshape China’s climate policymaking process.
• Sample characteristics: The engaged community is a small subset of users and skews younger, more educated, and from developed regions; cultural discouragement of public objection may under-represent opposition.