What do the magnetism of materials, the outcome of a referendum, and the dynamics of a geopolitical conflict have in common? At first glance, nothing. Yet, according to Serge Galam – French theoretical physicist, senior researcher emeritus at the Centre National de la Recherche Scientifique (CNRS) and at the Centre de Recherches Politiques de Sciences Po (CEVIPOF), and founder of sociophysics – they are all governed by underlying mechanisms of interaction among their elementary constituents, that can be described using the same mathematical principles. Just as statistical physics explains how microscopic interactions among particles give rise to macroscopic phenomena, it can also shed light on how individual opinions evolve into collective decisions. Widely regarded as one of the pioneers of the mathematical modelling of social phenomena, Galam founded sociophysics, a new field of research, in the 1980s by developing a physics-inspired approach to social systems and building new mathematical frameworks to describe how collective phenomena emerge from individual interactions, in particular to explain opinion dynamics, consensus formation, elections, polarization, and conflict. His groundbreaking models have uncovered counterintuitive hidden mechanisms shaping collective behaviour and have attracted international attention for their ability to predict collective outcomes for specific ranges of model parameters and to explain the unexpected occurrence of major political events, including Brexit and the 2016 U.S. presidential election. In this exclusive interview, Galam retraces the intellectual journey that led him to challenge the traditional boundaries between the physical and social sciences. From his influential theory of democratic opinion reversals driven by committed minorities and unconscious biases to the application of sociophysics to contemporary geopolitical crises, he reflects on both the promise and the limitations of mathematical models for understanding human behaviour. Along the way, he discusses the impact of artificial intelligence, the spread of misinformation, the future of democracy, and the ethical challenges raised by the possibility of predicting collective behaviour through mathematical models. The relevance and power of these models are also reflected in Italy’s scientific and cultural debate, as illustrated by Serge Galam’s recent participation in the Dyses 2026 conference in Naples. Known for their historically dense community networks, strong social ties, and complex relational dynamics, Naples and Southern Italy provide a privileged lens through which to explore how consensus forms and collective transformations unfold across Southern territorial contexts.
Professor Galam, you are widely regarded as the father of sociophysics, a field of research that you began developing at a time when few imagined that physics could be applied to social phenomena. What inspired you to pursue this path, and how did the physics and sociology communities react to your ideas at the time?
The idea came from a simple observation. Statistical physics shows how simple local interactions between elementary constituents can generate complex collective phenomena. I wondered whether a similar way of thinking could help us understand how individual opinions evolve into collective decisions. Of course, people are not atoms, but when we study large groups, collective patterns are found to emerge from repeated interactions among small groups of people. At the time, the reaction was largely negative. Most physicists considered social phenomena outside the scope of physics, while many social scientists feared that such an approach would reduce human behaviour to mathematical equations. I found myself rejected by both communities. Yet I was convinced that mathematical modelling could complement, rather than replace, the social sciences by revealing hidden collective mechanisms. It took many years before sociophysics gained recognition as a legitimate field of research within physics. Today, it is an established area of research, with its own international community and scientific contributions. However, it is still confined mostly to physicists, computer scientists and lately mathematicians. At Sciences Po, my current challenge is to build bridges between sociophysics and political science. The situation is encouraging, as some political scientists are becoming open to mathematical modelling that complement traditional analyses. Yet, much remains to be done.
Sociophysics may seem counterintuitive to those unfamiliar with the discipline. In simple terms, how can the laws of physics effectively describe complex phenomena such as political decision-making, opinion formation, and collective behaviour?
First, it is important to stress that I don’t apply the laws of physics to human interactions. I try to figure out what are the laws of human interactions along the way physics did it for atoms. In fact, sociophysics does not attempt to predict what a particular individual will do. Instead, it studies how interactions among many individuals can generate collective patterns. This is exactly what statistical physics does with particles, where macroscopic phenomena emerge from the interactions among microscopic constituents. However, above collective patterns are not obtained by statistics but from the individual opinion changes driven by local interactions. People influence one another through discussions, social pressure, shared beliefs, and group identities. By modelling these interactions with simple rules, we can identify mechanisms that explain consensus, polarisation, sudden public opinion shifts, or persistent divisions. The objective is not to capture every psychological detail but to uncover the fundamental mechanisms driving collective dynamics. In many cases, relatively simple models are sufficient to reproduce complex social outcomes, which often appear counterintuitive.
You gained international attention for predicting surprising political outcomes, such as Brexit and the 2016 U.S. presidential election, through your theory of opinion dynamics and minority spreading. How can a minority turn into a majority democratically?
One of my most influential findings is that democratic opinion dynamics are not shaped by the aggregation of initial individual choices, but are reshaped by invisible mechanisms emerging from voluntary and uncoerced social interactions. Most people assume that the initial majority always wins. My work shows that this is not necessarily true. When a small minority consists of inflexible individuals who never change their opinion, while the majority remains open to discussion, repeated local interactions can progressively shift the balance in favour of the minority. An equally important ingredient, which also operates through repeated local interactions, is what I called the “tie-breaking rule”. Whenever a discussion ends in a deadlock, the group does not remain neutral. The decision is unconsciously resolved according to the dominant shared prejudice or cultural bias already present in the related social population. This hidden mechanism systematically favours one opinion over another without people being aware of it. Either committed agents, unconscious tie-breaking mechanisms, or the combination of both, may eventually produce a complete reversal of the initial majority preference. This does not happen in every situation, since some critical thresholds must be reached, but when the conditions are met the outcome can appear highly surprising, as illustrated by political events such as Brexit or the 2016 U.S. presidential election.
When constructing a mathematical model of social reality, what is the greatest methodological challenge? How do you translate seemingly intangible concepts such as trust, loyalty, or propensity for conflict, into measurable variables?
The greatest challenge is deciding what to leave out and how to account for others. A model is not a copy of reality; it is a simplified representation designed to isolate the mechanism responsible for a given phenomenon. If we try to include every aspect of human behaviour, the model quickly becomes impossible to analyse and loses its explanatory power. Concepts such as trust or loyalty are not measured directly. Instead, they are represented through model parameters or variables that influence how individuals interact, whether they are willing to change their opinion, or how strongly they resist social pressure. The purpose is not to quantify emotions themselves, but to capture their effects on collective dynamics. A useful model is one that identifies the essential mechanism while remaining simple enough to generate testable predictions. For instance, while stubbornness, inflexibility, or commitment have different origins, in sociophysics, what matters is not why a person has such a characteristic but the fact that such a person keeps their opinion and does not change it. And then to find out what is the associated impact on the dynamics of opinion during a public debate.
Your work has also examined the conflict between Russia and Ukraine. According to your model, what are the key factors that determine the stability or fragmentation of international alliances? Are there indicators capable of anticipating their evolution?
My recent work suggests that geopolitical stability is not primarily a matter of the intentions or preferences of individual states. It is a structural property of the network formed by their historical bilateral relations. Some networks are intrinsically unstable because they contain what I call structural frustration, where patterns of alliances and rivalries cannot all be simultaneously satisfied regardless of the decisions made by individual states. Given this situation, I showed that supranational institutions can overcome this instability without eliminating historical antagonisms. Instead, they superimpose a new layer of interactions based on sustained institutional interactions, including political, economic, and security cooperation. When these institutional ties become sufficiently strong, they transform the overall structure of the network into a stable configuration, even among countries with a long history of conflict. This framework provides a structural explanation for Europe’s transition from recurrent wars before 1945 to lasting cooperation afterwards, and also helps explain why several Eastern European countries stabilised after joining NATO and the European Union, while the Russia-Ukraine relationship remained structurally unstable in the absence of an equivalent institutional framework. From this perspective, the most informative indicators are not isolated political events but changes in the architecture of interstate interactions themselves.
Today, Artificial Intelligence and social media algorithms amplify the spread of information, but also of disinformation. How do these new factors affect the models of sociophysics and the dynamics of opinion polarisation?
Artificial intelligence and digital platforms do not fundamentally change the mechanisms of sociophysics, but they significantly modify the environment in which social interactions occur. They increase the speed, scale and frequency of opinion exchanges while creating highly connected networks that differ from traditional face-to-face interactions. Algorithms tend to expose individuals to information consistent with their existing beliefs, reinforcing polarisation and making consensus more difficult. At the same time, misinformation can spread much faster than in the past, especially when it activates strong emotions, which in turn activates hidden prejudices. These developments require refining our models to account for the structure of digital networks and algorithmic amplification. The underlying statistical principles remain the same, but the dynamics become faster and often more extreme.
Looking at practical applications, to what extent could the predictive models of sociophysics become operational tools for governments, international institutions, and diplomats in preventing or managing geopolitical crises?
Sociophysics should not be seen as a tool for predicting the future with certainty. Its real value lies in identifying possible collective outcomes under different conditions and in uncovering the mechanisms that may lead to instability, escalation, or consensus. Used responsibly, these models could help decision-makers evaluate the potential consequences of different strategies and identify critical thresholds beyond which conflicts become much harder to contain. They may also reveal situations where small changes in the structure of interactions can trigger major collective shifts. Sociophysics does not eliminate uncertainty, but it can improve our understanding of the conditions under which different outcomes become more or less likely. That is where its practical value lies.
If collective behaviour can be described and, at least in part, predicted through statistical models, what are the implications for the concept of free will? Are our individual decisions truly as autonomous as we tend to believe?
The term “statistical models” is misleading. My work is not based on statistics, Big Data, or data fitting. Instead, I develop theoretical mathematical models that identify the mechanisms through which collective behaviour emerges from individual interactions. I often summarize my approach in a simple formula: “No statistics. No data. But universal mathematical modelling”. Regarding free will, it is fundamental to emphasize that there is no contradiction between individual freedom and the predictability of collective behaviour. Sociophysics does not claim that specific individual choices are predetermined. Each person remains free to make their own decisions, often for reasons that cannot be captured by a mathematical model. However, the collective outcomes resulting from the aggregation of individual choices can become predictable under given conditions, in particular through the collective patterns that emerge from social interactions. Furthermore, this collective predictability does not require millions of individuals; in many situations, populations of a few hundred individuals can already be sufficient for robust collective regularities to appear. This principle is similar to that of statistical physics: we cannot predict the behaviour of a single particle, yet we can accurately describe the behaviour of a gas. Likewise, individual freedom remains intact, while collective behaviour follows mechanisms that can often be identified and analysed mathematically. Paradoxically, discovering why and how individual choices are self-manipulated by hidden psychological biases, may open the way to a liberated free will.
Every predictive technology inevitably raises ethical questions. Have you ever considered whether the models of sociophysics could be used not only to understand mass behaviour but also to manipulate it, for instance, by authoritarian governments or major digital platforms?
Any scientific knowledge can be used for beneficial or harmful purposes. Sociophysics is no exception. Models that improve our understanding of collective behaviour could indeed be exploited to influence or manipulate public opinion. However, I believe their greatest value lies elsewhere. By revealing the mechanisms through which opinions spread, biases emerge, and collective decisions evolve, sociophysics helps us recognize the forms of influence and manipulation that are already operating today. Many opinion dynamics are perceived as the spontaneous expression of democratic choice, whereas they may also be influenced by structural mechanisms, social biases, or asymmetric influences that remain largely invisible. Understanding these mechanisms is therefore not only a scientific objective but also a way to strengthen democratic awareness and make public debate more transparent. Indeed, sociophysics could help restore and strengthen the genuinely democratic dimension of collective choices.
Beyond geopolitics and elections, what phenomena of everyday life can be explained through sociophysics? From financial crises to fashion trends, from viral content on social media to cultural shifts, are there common dynamics that regulate these processes?
Many apparently unrelated phenomena may share common underlying mechanisms. Financial bubbles, the spread of fashions, viral content on social media, technological adoption, and even certain cultural changes can all result from repeated interactions between individuals who influence one another. The specific subject may differ, but the collective dynamics can be remarkably similar. Small initial differences can be amplified through imitation, social influence, or network effects, eventually producing large-scale collective outcomes. Sociophysics seeks to identify these common mechanisms rather than explain each phenomenon separately. This is precisely why approaches inspired by statistical physics can provide insights into such diverse domains of human activities.
Looking ahead to the next ten years, what do you believe is the greatest blind spot in our understanding of human collective behaviour? And what contribution can sociophysics offer to bridge this gap?
The greatest blind spot is the persistent confusion between individual and collective behaviour. Knowing how individuals think does not tell us how societies behave. Collective phenomena emerge from interactions and obey mechanisms that cannot be reduced to individual psychology, although individual traits must be incorporated into the modelling. This fundamental idea is still far from being fully accepted. This is precisely where sociophysics can make a decisive contribution. Rather than replacing sociology or psychology, it complements them by identifying the universal mechanisms underlying polarization, consensus, minority influence, and sudden social shifts. In a world increasingly shaped by social media and artificial intelligence, understanding these collective dynamics is becoming not only relevant but necessary. Ultimately, this knowledge may help our societies avoid falling into forms of collective madness that appear to result from democratic choices, while in reality they are often driven by hidden prejudices, unconscious biases, or the disproportionate influence of tiny but highly determined minorities.
Photo Credits: EMEDIA
© All Rights Reserved
Lorena Coppola, Fondatrice e Presidente della Fondazione Léonide Massine, giornalista, esperta di comunicazione, international project management e direzione artistica, ha al suo attivo una lunga esperienza nel settore editoria. Traduttrice, scrittrice, docente, autrice e curatrice di diversi testi, si occupa da anni del settore stampa, pubbliche relazioni e content management. Danzatrice e coreografa, esperta di studi coreutici, ha collaborato con diversi teatri e compagnie internazionali, tra cui il Vancouver City Dance Theatre e il Théâtre de l’Opéra Metz-Metropole. Dal 2010 è Vicedirettore del Giornale della Danza. Attualmente fa parte della redazione partenopea di Politica Meridionalista.