Big data-driven computational politics decodes human society. Photo: IC

With the rapid development of the Internet of Things, cloud computing, social networks, and information acquisition technologies, data volumes have exploded, becoming a vital asset for society. The big data era offers new opportunities to study social dynamics, attracting significant attention from computer scientists and sociologists regarding its societal, economic, and research potential. Institutions such as the Santa Fe Institute in the United States, Google Research, HP Social Computing Laboratory, and Harvard University have adopted complexity science to describe the complex phenomena within social systems and propose theories such as complex adaptive systems. Using computers as research tools, they are pioneering new methods of computational social science and deeply integrating social computing into social research.

Interdisciplinary advancement

As research objectives in political science evolve and its methodologies transform, political scholars have moved beyond traditional quantitative research methods relying on small sample sizes. By analyzing massive datasets, they are able to uncover increasingly complex social phenomena, giving rise to the field of computational politics. As an interdisciplinary field bridging computing science and political science, computational politics combines cutting-edge data analysis techniques with social science methodologies, leveraging multi-source and heterogeneous big data resources, increasingly powerful computational capabilities, and continually optimized algorithms to advance political science research. This has led to the development of methods such as political computing, social simulation modeling, and internet-based experimentation.

Currently, national governance is undergoing profound structural and operational changes, with data capabilities emerging as a key factor in modern countries for addressing internal issues such as information asymmetry, information fragmentation, and communication barriers between society and the government. The mechanisms for data acquisition and processing are important components in the development of national data capabilities and are crucial for driving the digital transformation of national governance. Through digital means, countries can more efficiently collect, analyze, and utilize information, thereby improving governance effectiveness. For example, the application of digital technologies allows governments to monitor social dynamics in real-time and to respond quickly to public needs, which enhances the scientific and targeted nature of decision-making. Additionally, open information policies foster greater government transparency and increase public trust and support.

In summary, as a key element of national governance, information can facilitate vertical communication to address information asymmetry, control, and accountability issues within the state, while enabling horizontal communication to address information flow between the state and society. The enhancement of digital capabilities is of great significance for improving national governance effectiveness. The study of computational politics, which engages in participatory observation and scholarly responses to the dual digital transformation of governance structures and operations in the digital age, upholds and advances the computational tradition within political science, and represents an important subfield of the broader domain of computational social science.

Era of data

From the perspective of research methodology, large-scale, high-dimensional, and multi-source data sets have greatly enhanced the external validity of political science research samples, enabling a broader reflection of real-world situations. These data sets include not only traditional structured data, such as election results or economic indicators, but also unstructured data from social media, such as text, images, videos, and complex social network information. This diverse data allows researchers to monitor changes in political behavior in real-time, enhancing both the immediacy and relevance of their findings. By leveraging cutting-edge data analysis techniques such as natural language processing, image recognition, and social network analysis, researchers can process and mine these complex and diverse data types. This not only expands the research paradigm of political science, but also provides new perspectives on traditional topics. For example, in election forecasting and public opinion analysis, analyzing emotional shifts on social media can more accurately predict voter behavior. In international relations research, sentiment analysis of diplomatic statements offers insights into the dynamics of interactions between countries. Parametric modeling methods, combined with machine learning algorithms, enable researchers to adjust variables within complex systems and explore the interactions between subjects and environments under different conditions. Given that human society is a complex system, individuals, organizations, and nations can spontaneously generate high-level structures, patterns, or behaviors that are not directly determined by the attributes of the constituent units but emerge from countless micro-interactions. Therefore, computational political science can not only reveal individual-level behavioral characteristics but also understand the behavior patterns of different groups within the system, as well as the social phenomena that emerge from these countless micro-interactions. This provides powerful tools for understanding the mechanisms behind the emergence of complex human social systems.

The debate over research paradigms is a crucial stage for emerging disciplines. The term “computational politics” may indeed lead to the misconception that “methodological innovation alone can foster a new discipline.” However, it marks the emergence of a field where interdisciplinary approaches drive the accumulation of knowledge. Sociologist Duncan J. Watts argues that solving complex social problems often requires the complementary application of multiple research paths, but researchers often lack the expertise to apply multiple methods simultaneously. It is evident that, during the collection stage, “what you see is what you get.” Big data offers researchers more opportunities, driven by fundamental shifts from the analog to the digital world. During the analysis stage, however, big data weakens the role of theoretical hypotheses in political science research, expands the model selection space, and uncovers broader possibilities for political research.

Challenges & opportunities

From the current perspective, computational political science faces several methodological and technical challenges. First, data quality remains a crucial issue in political computing research. The authenticity, representativeness, and generalizability of big data directly affect the reliability of the results. Although big data offers advantages such as massive volume, persistence, and non-responsiveness, problems like incompleteness, difficult access, non-representativeness, drift, algorithm interference, dirty data, and sensitivity can undermine its reliability. Data from diverse sources—including government databases, opinion polls, news outlets, and social media—often vary in quality, which can reduce model accuracy, a problem further exacerbated by biases such as social bias, sample selection bias, and algorithmic bias. Second, models may be influenced by theoretical and data selection biases, leading to limitations and distortions. Human behavior and decision-making are shaped by complex factors, including cognition, emotion, and social context, which makes it difficult to model them in a straightforward way. Models based on historical data and assumptions have limited predictive power and are often constrained by quantitative data, making them inadequate for explaining qualitative political issues. Finally, normalcy bias presents a significant obstacle. People tend to assume that the future will mirror the past, overlooking new circumstances. Political computing models based on historical data tend to reflect historical norms, making it difficult to anticipate sudden and unexpected events. Cognitive biases also make individuals more likely to accept information that is consistent with their experiences, while ignoring contradictory information, which further skews predictions toward historical norms and hinders the identification of new possibilities.

To address the difficulties facing the development of computational politics and ensure its long-term growth, several steps must be taken. First, it is essential to enhance the discipline’s intrinsic motivation, clearly defining the role of computational technologies within political science research and their unique contributions. Specifically, at the level of talent cultivation, efforts should focus on curriculum design, faculty development, textbook innovation, professional training, and admission requirements. Computational politics aims to cultivate high-quality interdisciplinary composite talent with expertise in both political science theory and the ability to handle and analyze big data, as well as to predict and explain political behavior using computational models. This calls for universities to establish interdisciplinary programs that combine elements of engineering with management, and science with the arts. Curriculum design should balance practical training courses with traditional political theory courses, guiding students to transition from traditional research-based skills to innovative and applied competencies suited to the big data era. Faculty development is also crucial, with a need to attract and nurture outstanding scholars and teachers with backgrounds in computational social science while encouraging existing faculty to actively update their knowledge systems to meet evolving teaching needs. The academic community should publish accessible, popular science materials to lower the entry threshold of the discipline, helping non-experts understand how computational models can be applied to explain and predict political phenomena, thereby stimulating broader interest and support for the field. Schools equipped with the necessary resources should play a leading role by setting up interdisciplinary research centers or laboratories and conducting joint projects to lead the new trend of interdisciplinary integration. In addition, combining qualitative research methods with computational politics is an important future direction. While traditional qualitative research methods excel at deeply exploring complex causal relationships in the context of case studies, computational methods are adept at handling large datasets and discovering macro-level patterns. The organic combination of these approaches not only supplements the shortcomings of a single methodology, but also provides a more comprehensive and profound perspective for solving complex social and political problems. For instance, qualitative interviews can provide in-depth insights into stakeholders’ needs, while computational models can simulate various policy scenarios, helping decision-makers evaluate the likely impacts of strategies.

In the wave of globalization, the emergence of computational politics marks a new stage in the study of social sciences. The development of this discipline not only requires political scientists and computer scientists to cross traditional disciplinary boundaries, break the institutional constraints of the past, and truly collaborate, but also requires close coordination within the social science research field. Theories and methodologies from economics, sociology, etc. can be exchanged to inject new vitality into computational politics, jointly pushing forward the deepening and expansion of scientific research. Faced with future challenges, computational politics needs to continuously absorb the latest sci-tech achievements, such as artificial intelligence and blockchains, to enhance its research capabilities. At the same time, it should pay attention to ethical considerations to ensure that technology applications conform to the principles of social fairness and justice. Through such efforts, computational politics can reveal more subtle patterns of political behavior and provide solid data support for government decision-making, promoting the modernization of social governance. Only by building a solid foundation of interdisciplinary communication can the strengths of each discipline be fully leveraged, integrating diverse perspectives and advanced technological means to achieve a more comprehensive and in-depth understanding of complex social phenomena.

Ling Zheng is an associate professor from the Zhou Enlai School of Government at Nankai University (NKU). Mu Jiaxiao is an assistant research fellow from the Center for Digital Government and Data Governance at NKU.