© Copyright JASSS

JASSS logo ------

Social Physics: How Good Ideas Spread--the Lessons from a New Science

Pentland, Alex
Penguin: London, 2014
ISBN 9781594205651 (hb)

Order this book

Reviewed by Helder Coelho
Departamento de Informatica, Universidade de Lisboa

Cover of book Around 2014, two books were released that are most relevant for social simulation and complexity science; one by Joshua Epstein (2013) “Agent_Zero: Toward Neurocognitive Foundations of Generative Social Science” and this one by Alex Pentland on human behavior and social networks capable of collecting and refining decision strategies. The first one has already been reviewed by Bruce Edmonds in a previous issue of JASSS ( http://jasss.soc.surrey.ac.uk/17/4/reviews/2.html, and I had the opportunity to read it before I started with this review.

First of all, I picked up Pentland´s book after three intense years working with my PhD student Carlos Lemos on Civil Street Protests and Violence in NetLogo (see our papers at ResearchGate). Last year, I read some parts of his book, but I made no real efforts in taking a critical view on the whole content. I also visited YouTube to get the video talks of Pentland, and I became even more curious about its real impact and value. A graduate student of my Faculty was also responsible to call my attention to Pentland’s book when discussing the implementation of a desire mechanism (mimetic, character, conflict) (Antunes et al., 2014). But, the influence of the book was nil. Pentland never started any experimentation.

Social physics is a label made by Auguste Comte 165 years ago to designate a new understanding of human behavior and society (system of relationships). But now we are living in a different era, the so-called digital one, i.e. modern societies represent a combination of humans and technologies, and we don´t know well how to deal with it. “Markets” are everywhere and they seem more relevant than the will of politicians and citizens. Migration seems the black plague in a region with 2000 years of population movements, wars, battles and diseases. Today something is different from yesterday: cities have millions instead of thousands of human beings and healthcare is stronger.

Is Social Physics a science of intuition (direct insight) or of rational insight? Are the equations a tool to acquire new knowledge without inference or the use of reasons, or are they means to arrive fast to some conclusion: to see ahead? Or, is it a new theory of information propagation in communities?

In our world, real crowds can appear and disappear, but our understanding must include not only economic and political aspects: human beings are no longer independent decision makers or perfect rational individuals. Simon (1957) thought of a bounded rationality to dismiss a sort of imperialism, but he forgot the dynamic social consequences of “millions of people learning from each other (social learning) and influencing each other´s opinions” (Pentland). Ideas flow in all directions, are shaped, and turn out to be the driving force behind the radical changes (political revolutions, cultural moves, religious malaises, economic crisis) as well as innovations (Internet, computer design, business). There is not only fair competition behind any market, but also other forms such as greed, cheating, corruption, and even Adam Smith´s “invisible hand”. The absence of established rules/regulations provoke all sorts of evil deviations.

When we study human behavior we must do it step by step, taking into account a computational theory of behavior based upon a mathematical predictive science of society, and also the mechanisms of social interactions and of societal changes:

Flow of ideas and information -> Changes in behavior -> Understanding society

Reactions (financial decision making, social influence, political attitudes, purchasing behavior, health) are the consequences which are to be analyzed, yet taking that there is an engine that drives Social Physics: Big Data.

Social Physics can be defined as quantitative social science that mathematically describes connections between the flow of ideas and information among individuals (investors, traders, brokers, and also through social networks) and the people´s behavior (norms, moves, messages, social learning, productivity, creativity).

We can observe two main mechanisms in human societies and animal populations, the mimetic effect or social learning (copying successful individuals) (see Antunes et al., 2014) and the contagious communication (diseases, emotions such as fear, anger, or envy, sympathy, rivalry; this is discussed in René Girard’s (1965) book: Deceit, Desire and the Novel). These mechanisms are discussed extensively in Epstein´s last book. His aim is to find a different architecture for agents (Agent_Zero), built with three main components (rational, affective, social) instead of a reactive or cognitive (BDI: Believe, Desire, Intention) form (see also the architecture of PECS: Physical conditions, Emotional state, Cognitive capabilities, Social status; with four blocks, including also a physical component beyond the rational, affective and social ones).

Those involved in problem solving know that it is advantageous to see a situation from different perspectives (e.g. a photographer takes different angles to support the opinion expressed by the words of a journalist) in order to generate better solutions. And, by flowing ideas, science can help people to bypass “impossible” difficulties. However, without energetic people the flow is stopped, and without tolerant guys no evolution occurs. The same happens in a dictatorship with the flow of beliefs. More oppression and violence decreases the legitimacy of power and increases anger. Explanations are not sufficient and justifications make the point to struggle for a change.

When discoveries are performed the (wise) crowd believes that search is a trick, but immediately they agree that exploration is necessary, and trial-and-error is one of the ways out of it. With the Portuguese sea discoveries (the route to India) intelligence behind the path-finding task: which spot in the Brazilian coast was responsible for changing the wind, i.e. the turning point from where navigators could appoint to Cape of Good Hope (South Africa): “to bet against the common sense” in order to explore the entire Indian Ocean, and “to follow the contrarian consensus”. “People act like idea-processing machines, combining individual thinking and social learning” (all quotes from Pentland). Often, sophisticated mathematics is behind, where equations are able to quantify the process of social learning and exploration (e.g. Epstein´s formula is a good example apart of the social physics and the mathematics of social influence).

Which are the building blocks of collective intelligence? Looking to successful enterprises we discover energetic and creative feelings that are the output of good management and communication of ideas: the flow is easy and simple (jobs had used often the mantra “focus and simplicity”).

Habits (health), preferences (politics), tendencies, values, ideas and beliefs are influenced by social interactions: there is a collective intelligence. Those who live with people with the same opinions tend to become more radical. The behavior flows and becomes bigger with size of exposition. Contagion works well either with opinions or viruses. “Human behavior is determined as much by social context as by rational thinking or individual desires. Continual exploratory behavior of humans is a quick learning process that is guided by apparent popularity among peers, meanwhile adoption of habits and preferences is a slow process that requires repeated exposure and perceptual validation within a community of peers” (Pentland).

After a small introduction about the title, Pentland explains how everything moves from ideas to actions and how Big Data is used to understand the evolution of human societies (see also the potential for a huge impact on scientific discoveries, medicine, engineering, business models, urban planning and policy). The book is divided into four parts. Part I covers the discipline of Social Physics along three directions: exploration (finding good ideas and making better decisions, social learning, tuning networks), flow of ideas (building blocks of collective intelligence with habits, preferences and curiosity, common sense), and engagement (social pressure, subjugation and conflict, rules of engagement). Part II deals with “idea machines” by going through collective intelligence, the patterns of interaction, the shaping of organizations (engagement, exploration, diversity, social intelligence) and organizational change (via social networks to create a new kind of institutions). Finally, Part III invests in data-rich cities (sensing cities, city science) and Part IV in data-rich societies (privacy, living with data).

The following two pieces of Pentland´s book show the movement from ideas to actions:

“The rate of idea flow is intrinsically a function of the ease of access and interaction between residents living in the same city.”

And, the chain of transformations:

“… fine-grain characteristics of human social interaction: the distribution of social ties, the flow of ideas along these ties and the means by which those ideas are converted into new behaviors and new social norms by engagement within peer group.”

The role of social ties is immense when we are looking attentively to communities and take the social fabric into account created by the exchange of goods, ideas, gifts and favors (corruption). Also, the solutions for the goods of a community were deeply influenced by the forms of capitalism and socialism. At the micro level, human behavior was not only guided by economic exchanges, but also by other types of exchanges (e.g., information, ideas, and norms), and engagement enhances trust. Over time, we were driven by cascades, bursts of engagement (repeated cooperative interactions) and a lot of social pressure.

Engagement requires interaction (measure of the number of interactions in function of social pressure) and cooperation (to support the increase of social capital and to maintain team work) to build trust. Groups have collective and social intelligence which is dependent of cohesion, motivation, satisfaction and equality of conversational turn taking). And, the physics of social interaction aided by measures of the patterns of idea flow and how people interact (socio metric data).

How do individuals in general decide about group actions? In the animal world the answer is by signaling mechanisms for dominance, interest, agreement in order to find resources, make decisions and coordinate action. Yet, humans do the same. There is a science for building great teams supported by our knowledge of individual behaviour, and directed to improve the flow of ideas and the creative output. We know also which are the important factors for predicting productivity, namely the amount of interaction and the level of engagement (idea flow).

The question is always how concepts and ideas are translated into mathematical models and governed by equations (instead of models) to be sufficiently precise and sound. When we read Pentland’s book and compare it with Epstein´s we observe some substantive differences and we become more impressed by the articulation of ideas and the approach of “generative explanations” in Epstein´s book (see also the four appendixes with code and other resources at the Web site of Princeton University Press http://press.princeton.edu/). As a matter of fact, justifications are essential to convince a reader to start soon an implementation, for example by adopting NetLogo as a tool.

One critical point of Pentland’s book is that one cannot go behind the mathematical relationships and the descriptions attached (a lot of explanations are missing and replication is almost impossible). The level of generality of the equations for diverse phenomena can also intrigue readers of Social Physics and force them to make a step backward and thinking of the essentials presented. In fact, after reading the book it is difficult to go ahead without fear. The opposite impression occurs after reading Epstein. Often, engineering depends mostly on the will to go ahead, and this is a good reason to evaluate the book. We can love and feel enthusiastic with a content when reading it, but thereafter there is some need to going forward with that knowledge and apply it, make the necessary transformations, and propose something very different. The leap in Pentland´s book is tremendous, yet the beauty is more associated to the generative social science of Epstein, able to "grow" the phenomena of interest in an artificial society of interacting agents.

Finally, books are always directed to specific audiences and these two have definitely different target groups. Social Physics aims at business, marketing, city planning and health care professionals, an audience of university graduates and professionals. Agent-Zero is more demanding, because it is the third one of a trilogy started in the 1990s with Growing Artificial Societies. It refers to post-graduate university students and requires knowledge in mathematics, computer science, and cognitive science; moreover, it is valuable for engineers eager to build nice software installations.

* References

ANTUNES, L., Nunes, D. and Coelho, H. (2014), The Geometry of Desire. Proceedings of AAMAS 2014, Blue Sky Ideas Track, Paris, May 5-9, 2014.

EPSTEIN, J., (2013), Agent_Zero: Toward Neurocognitive Foundations of Generative Social Science. Princeton: Princeton University Press.

GIRARD, R. (1965), Deceit, Desire, and the Novel: Self and the Other in Literary Structure. Baltimore: John Hopkins University Press.

SIMON, H. (1957), Models of Man. Social and Rational. Hoboken, New Jersey: John Wiley.


ButtonReturn to Contents of this issue