Is it possible to use mathematical tools to predict human behavior? This is the approach proposed by the scientific discipline ‘social physics’, a trend studied by BBVA’s New Digital Businesses that seeks to understand and predict human behavior through the analysis of large amounts of data, similar to what is done in the fields of physical and biological sciences.
BBVA’s New Digital Businesses (NDB) has carried out a series of investigations and analyses to understand what this scientific trend consists of, how it is applied and in which spheres it could have significant results. “Social physics is actually a very old concept, which has very interesting new applications today,” explains Raúl Muñoz, who is responsible for this research along with Gonzalo Santoyo, both members of the Research and Development team at NDB.
Social physics: what is it?
Social physics or socio-physics is a field of science that uses physics-inspired mathematical tools to understand the behavior of human groups.
Traditionally, physicists have developed methodologies that enable theoretical physics to operate with extremely complex phenomena, such as climate or astrophysical phenomena. However, many objects of study – both in physics and in the social sciences – are dynamic systems in which different elements interact with one another and are influenced by external factors. From this parallelism arises the idea that many social systems can be modeled and studied according to physical laws.
Social physics, for example, has been used to study how opinions are propagated through a social network or how bubbles or economic crises are generated. Also its utility has been explored in improving the sustainability of urban populations, in the study of market investment strategies, or for the planning and design of cities.
More than two centuries old
The concept of ‘social physics’ is not a new one. We could say that it is more than two centuries old. The first approaches to social physics were described by the French thinker Henri de Saint-Simon in 1803. However, it was his student and collaborator, Auguste Comte, considered the father of sociology, who coined the term.
Since then, social physics has aroused interest over time, as in the mid-twentieth century, or recently with a new wave of “socio-physics”.
The professors at the Massachusetts Institute of Technology (MIT), Alex “Sandy” Pentland and Dr. Yaniv Altshuler, were responsible for reviving interest in this discipline in recent years. At the human dynamics laboratory of MIT’s Media Lab, these scientists have spent years trying to quantitatively measure the consequences of different types of social interactions (such as interactions within organizations, or in cities). Consequently, they have been the precursors of social physics, understood as the analysis of social phenomena from ‘big data’, in the course of research aimed at overcoming the traditional limitations of ‘machine learning’.
Since then, several companies like Endor, Algowave, Thasos and Ginger, to mention a few, have developed the concept using technologies that offer mathematical solutions for studying diverse social phenomena. Some of them emerged as MIT spin-offs.
Why is it a trend again?
“The main innovation lies in the emergence of ‘big data‘. There is now a large amount of social data that can be easily captured – from new sources such as social networks, cell phones, credit cards… Ubiquitous digital data is available in all aspects of human life and enables quantitative and predictive answers to be found to many questions in the social sciences,” says Muñoz.
Similarly, in 2001, in his book ‘Social Physics: How Good Ideas Spread – The Lessons from a New Science‘, Pentland explains that the proliferation of mobile sensors capable of collecting data in the Internet era, as well as the arrival of new mathematical tools capable of analyzing interdependencies in networked systems, added to the new computing capacity of modern computers; are the ingredients that make this old branch of knowledge closer than ever to materializing.
From theory to practice
The discovery of social physics is based on the premise that all human behavioral data contain a set of common “laws of social behavior”; that is, mathematical relationships that arise when a large enough number of people operate in the same place.
But beyond this theory, what applications can social physics have? Some examples where social physics can be or has been used are:
- Detection of cyberterrorism. The ability to analyze semantically structured data flows (e.g., from social networks) allows social physics engines to process written texts in foreign languages, such as Arabic, that many conventional data analysis tools cannot easily process.
- Financial Health. The scientific article ‘Money Walks: Implicit Mobility Behavior and Financial Well-Being’ shows how by studying credit card transactions in a social physics model, it was possible to predict user overspending with 83% accuracy, financial problems with 77% accuracy, and late payment with 78% accuracy.
- Propagation of ‘viral’ opinions. In social networks such as Twitter or Facebook (as in condensed matter physics) online individuals can spread opinions by sharing them with each other through “word of mouth”; a phenomenon that could also be studied through social physics.
- Organization and business efficiency. Social physics can be used to reveal how work is actually done, to decide on the types of interventions to be made, and to measure the impact of change management initiatives on productivity and performance. The application of social physics in this field has been one of the applications that the NDB team has analyzed most deeply, finding that a greater number of social interactions in the work environment would improve work performance.
Many of these applications are the result of studies and research conducted by MIT. In order to understand the scope and practical utility of this scientific discipline, the NDB team has replicated and continued some of the research work published by this university, obtaining similar results and conclusions.
Opportunities and challenges
“Thanks to the case studies, we have seen that social physics could have useful applications within the financial sector,” Muñoz explains. For example, the use of engines that take into account data on user transactions (along with other demographics, in this case) could help to better understand people’s financial behavior and, “through this, help us create more useful financial health solutions.” In addition, understanding statistical patterns in human dynamics can be key to the creation of a more responsible society, “committed to the environment and sustainable in the long term,” he adds.
In short, social physics is a broad science that uses modern mathematical techniques to solve problems. It has been shown that it provides information on complex collective behavior and could improve accuracy in certain cases. “However, it also poses some challenges, especially because of the amount of data needed for its use,” Muñoz explains. In this respect, at NDB they are also interested in the possible combination, in the future, of this technology with new privacy-enhancing techniques (PETs), which the unit is also researching.