Understanding Social Systems Using Computational Approaches

Robert Hacker
12 min readApr 20, 2023

--

Credit: MPIDR

“What was traditionally considered separate objects of study―on one hand, the freedom of human action, with all its burdens of uncertainty and unpredictability, and, on the other hand, nature, with its inner order―had created a gap between the social and the natural sciences for centuries. Today, Complexity and Chaos Theory shows this gap to be highly artificial, redeeming the social sciences from being in a position of scientific minority, or, as Kiel and Elliott would say, in a position of scientific stepchild compared to the so-called hard Sciences).”[1]

For about the last fifteen years, first at MIT and then at Florida International University (FIU), I have been involved in technology, commercializing faculty research and social entrepreneurship. During that time science has nearly completed the transition from a model of reality based on energy and matter to John Wheeler’s model of information and computation.[2] More recent advances in artificial intelligence (AI), such as multimodal AI, have only accelerated this transition by providing new means to discover scientific knowledge. Computational physics or synthetic chemistry and biology might be the proof and the future of commercialization in the sciences.

Coincident with AI at commercial strength, another scientific framework has gained increasing recognition — complexity science. Complexity science has many definitions partly because of the intriguing concept of emergence. Emergence endeavors to explain the many examples in natural and manmade systems where the whole is not derived simply by adding the components. For example, a tree is more than the additive outcome of a seed’s components.[3] Complex systems are perhaps more easily explained in terms of three concepts — networks, agent-based modeling (ABM) and scale — according to Dr. David Krakauer, the President of Santa Fe Institute.[4]

If we examine these three concepts, I am particularly intrigued by the ability to better understand manmade systems. Effectively, I see complexity moving the approach to manmade systems away from behavioral frameworks and more toward computational approaches. Both the polymaths John von Neumann and Herbert Simon (Nobel Laureate) criticized social science for a lack of computational methods (although progress has been made since both men made their remarks). I believe that the lack of progress in social science is equally explainable by the focus on government policy as the solution to many problems, e.g., healthcare, racial discrimination, gender inequality, disability, etc. The government’s record in successfully addressing social problems is meagre at best. Perhaps we can move away from the “social determinants” and use more computational tools which have less subjective interpretations (and bias) to understand social problems. A recent report by the National Academies Press, “Toward a 21st Century National Data Infrastructure: Enhancing Survey Programs by Using Multiple Data Sources (2023)[5]”, supports this move toward more computational approaches:

“Much of the statistical information currently produced by federal statistical agencies — information about economic, social, and physical well-being that is essential for the functioning of modern society — comes from sample surveys. In recent years, there has been a proliferation of data from other sources, including data collected by government agencies while administering programs, satellite and sensor data, private-sector data such as electronic health records and credit card transaction data, and massive amounts of data available on the internet. How can these data sources be used to enhance the information currently collected on surveys, and to provide new frontiers for producing information and statistics to benefit American society?” The international consulting firm McKinsey takes the point a step further in a recent article wherein it urges that the new means of data analysis actually shape the policy and implementation of social projects.[6]

This article presents some examples of how these computational tools may bring new insight to social problems and more specifically that: 1) Complex systems can explain human systems computationally. 2) All human systems are networks, similar to computer networks, for the exchange of information. 3) The most productive networks are open networks that scale superlinearly, which permits productive exchange and innovation. 4) ABMs capture these networks and model their scaling features through stochastic simulations using algorithms (as opposed to equations).

Networks

It is well-known that both energy and information move over networks. Energy and information are the foundation of reality according to physicists.[7] Therefore, understanding networks better should help us to understand reality and by extension all human or social systems. We know that there are two types of networks — open and closed.

· Open networks are self-organizing, non-hierarchical structures with superlinear returns to scale (explained below). This structure permits the easy ingress and egress of people and information, which spawns innovation from the exchange of new ideas. Open networks arise in part because they foster the natural explore-exploit behavior wired into all natural systems by evolution.

· “Closed networks are systems where a set of entities or nodes are interconnected in a closed loop, such that the flow of information within the network is self-contained and does not [significantly] involve external entities or inputs. Examples of closed networks include computer networks, electrical circuits, and transportation systems, among others. One of the key characteristics of closed networks is that the amount of flow or traffic within the network is limited by the network’s capacity, which is determined by the number and capacity of the nodes and links within the network. As a result, the amount of flow or traffic within the network is not proportional to the number of nodes in the network but is instead sublinear.”[8] (explained below)

Scaling

The noted physicist Geoffrey West explains scaling well:

“Even though the conceptual and mathematical structure of the growth equation is the same for organisms, social insect communities, and cities, the consequences are quite different: sublinear scaling and economies of scale that dominate biology lead to stable bounded growth and the slowing down of the pace of life, whereas superlinear scaling and increasing returns to scale that dominate socioeconomic activity lead to unbounded growth and to an accelerating pace of life.”[9]

Closed networks are sublinear in terms of growth and eventually die. A single human is a closed network and sublinearity explains why we each die. Our body is trying to maximize benefits from a given physical base until we exhaust this fixed base. Companies built on what HBS taught us — increased economies of scale and efficiency — also die as they mirror the behavior of living things.

Open networks are superlinear — increasing returns to scale. A simple equation demonstrates an example of superlinearity.

Y=X^B

B>1 (1.1, 1.2, 1.25, …)

Superlinear networks have increasing returns to scale brought about by an increasing number of nodes, new information and the stimulus to explore for innovation. Cities are the best example of open networks, and much to my surprise so are universities. In both cases, the ability to innovate from new information and people extends the life of these two types of systems in theory indefinitely. The first universities were founded in medieval times and the first cities date back almost 10,000 years. Open source software development is another example of an open network with superlinear scaling.

However, some human systems have the unfortunate characteristic of being closed systems. In these systems the information that is most likely to be believed is the information from another person in the network. [10][11][12][13][14] Examples of closed systems might be dictatorships, prisons and ghettos determined by race, ethnicity or the culture of a minority. It is in these closed systems that many social problems fester and resist new approaches almost by their nature or design.

Agent-based Models

An Agent-based Model (ABM) is a “computational simulation in which the individual components (“agents”) of a system are represented and interact explicitly. An agent-based model is typically iterated over time steps, with aspects of the agents updated at each time step. Agent-based models can be contrasted with models in which the behavior of the system is based on equations and individuals are not represented explicitly.”[15] ABMs can model the behavior of any living thing, but they are perhaps the best type of model to show the purposeful behavior characteristic of humans. Thomas Conway’s famous “Game of Life” model is now considered to be the earliest example of an ABM. ABMs are stochastic by nature.[16]

“Agent-based modeling grew out of early work on Cellular Automata, pioneered by John Von Neumann in the 1950s. Cellular Automata (CA) is a discrete model composed of individual “cell-like” entities that perform a range of functions according to a predetermined set of coded instructions.”[17] ABMs recognize that human systems are very similar to ecologies. “New strategies, new things are coming and going and striving to survive and do well in a situation they mutually create. We can describe this algorithmically, but not easily by equations, not just because the situation is complicated to track but because new behaviors and categories of behavior are not easily captured by equations.”[18] Applications of ABM include modeling transportation systems in urban settings, the spread of contagious disease and simulating an ecosystem.

To summarize to this point. Complex systems can explain human systems computationally. All human systems are networks, similar to computer networks, for the exchange of information. The most productive networks are open networks that scale superlinearly, which permits productive exchange and innovation. ABMs capture these networks and model their scaling features through stochastic simulations using algorithms (as opposed to equations).

An Example of a Closed Network — China

Rather than highlight again the benefits of open networks, I prefer to illustrate the drawbacks of closed network communities such as the dictatorship in China. China is particularly interesting because they had a period of phenomenal economic growth which ended when the country reverted to the worst characteristics of a closed network. I intentionally chose China as the example of closed network, but I could have used some of the African American communities in the U.S, or the indigenous populations in Latin America to illustrate my points.

China has been a Communist dictatorship since the end of WWII. Deng Xiaoping introduced dramatic reforms while preserving the Communist form of government on December 18, 1978.

“The reforms carried out by Deng and his allies gradually led China away from a planned economy and Maoist ideologies, opened it up to foreign investments and technology, and introduced its vast labor force to the global market, thus turning China into one of the world’s fastest-growing economies.”[19] — NBC News

In 1978 China had a GDP of $149.5 billion and a per capita income of $156.40. By 2020 GDP had reached $14.7 trillion and per capita income was $10,409.[20] This dramatic success can be described simply as the “opening”, wherein China took on many of the characteristics of an open network. Foreign investment was encouraged, foreign partnerships flourished, foreign visitors increased dramatically and information for technological and economic benefit moved without excessive government interference. Effectively, China innovated its way out of poverty to become the second largest economy in the world. (I often wonder if they took a lesson from Singapore.)

Xi Jinping, the current President of China, started significantly changing policies beginning in 2020–2021. He increased regulation of private sector companies especially in social media and information, favored state-owned companies for new opportunities, encouraged aggressive mistreatment of Moslem minorities in western China and generally stifled the economy to the point where some commentators worry that China GDP growth might now stall or even decline. Xi’s “closing of the network” resulted in the expected negative consequences when one moves away from the open network model and has resulted in reduced foreign investment. Direct foreign investment since 2016 when Xi assumed power has been markedly lower than before his Presidency, declining from an average $3.3 billion (2006–2015) to $1.6 billion (2016–2021).[21] Many factors may be cited to explain this dramatic drop off, but stifling private sector innovation and the flow of information through regulation create a more closed network that is not attractive to foreign corporations.

While the peculiarities of the Chinese dictatorship might make it hard to generalize about all closed networks, we can draw some meaningful conclusions about closed networks from the China case.

1. Closed networks create a natural asymmetry of information. The lack of information reduces innovation and opportunities for the members of the closed network.

2. New information, such as new health treatments, political views or social media, are evaluated by the collective historical wisdom of the network rather than in a more open-minded, progressive way.

3. The inability to change views and methods reduces the ability of a modern, complex society to improve and challenges the economic and political viability of the network.

4. The natural tendency for humans to explore and then exploit programmed in by evolution, which fuels experiment and innovation, is reduced by the lack of stimulation and the uncertainty caused by the lack of information and changes in policy.

Franck Billé and Caroline Humphrey make the same points talking about another modern dictatorship — Russia.

“As professor of law and politics Sergei Radchenko writes, in order to truly become a Eurasian power, Russia should focus instead on opening borders, on removing barriers to the flow of goods, capital, and people, and on encouraging cross-border communities that would “tie Russia to this region linguistically and culturally in ways that would support the notion of an Asian identity.”[22]

The Path Going Forward

Closed networks are the cause of many social problems because they naturally lead to an asymmetry of information. I believe that the solution to any social problem must address the asymmetry of information, in large part through a four-part framework of 1) self-esteem, 2) education, 3) inclusion and 4) information. I developed this framework in much detail in an earlier article on Medium[23], “A Primer for the 21st Century: The 5 Systemic Changes”.

I think in this article I have shown how open networks are built on inclusion and information. I think we can all agree that information is better processed with education. Also, the greater one’s self-esteem, the more likely one is to expand their network.[24] From the Medium article referenced above, one realizes that my four-part framework is also the basis for what Nobel economist FA Hayek called individual empowerment. Essentially, individual empowerment was strongly preferred by Hayek as the alternative to government and its policy-based approach.

With advances in computational methods, increasing datasets and increasing awareness of ABM, I think that more rigorous approaches to social problems can be substantiated and used to accelerate the solutions to the pressing social issues.

“The most modern definition of work is an interactive exchange in which the participants benefit from the interaction. Interestingly, cooperation is also described as an interactive exchange in which the participants benefit from the interaction.

What if performance is incorrectly attributed to win-lose competition and is, in effect, more a result of diversity, self-organizing communication and non-competitive processes of creative cooperation?”[25] — Esko Kilpi

Notes:

This article by Dr. Erlijn van Genuchten, “8 Ways Artificial Intelligence Can Help Recognize Poverty”[26], is an excellent example of how modern computational methods can be applied to a social science problem.

An article from Santa Fe Institute (SFI) illustrates the application of new computational methods in music. “So it is hard to imagine a more alluring topic for an SFI working group than “Complexity and the Structure of Music: Universal Features and Evolutionary Perspectives Across Cultures.”[27] Co-sponsored by SFI and the Institute for Advanced Studies of Aix-Marseille University, France (IMéRA), this forum brought together network and complexity scientists, musicologists, music theorists, composers, performers, and neuroscientists to trade licks about the intersections of music and complexity from as many angles as possible.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

[1] https://www.scirp.org/journal/paperinformation.aspx?paperid=69101

[2] https://philpapers.org/archive/WHEIPQ.pdf

[3] Cesar Hidalgo, Why Information Grows, (Basic Books, 2015), 196

[4] Santa Fe Institute is one of the leading institutions investigating complexity. Max Planck Institute for Physics of Complex Systems and the CABDyN Complexity Center at Oxford University are other leading complexity research organizations.

[5] https://nap.nationalacademies.org/catalog/26804/toward-a-21st-century-national-data-infrastructure-enhancing-survey-programs-by-using-multiple-data-sources

[6] https://www.mckinsey.com/industries/public-and-social-sector/our-insights/reimagining-public-health-programs-to-deliver-equitable-impact?cid=app

[7] https://philpapers.org/archive/WHEIPQ.pdf

[8] ChatGPT, perhaps derived from TechTarget.com

[9] Geoffrey West, …, The Universal Laws of Growth, Innovation, Sustainability, and the Pace of Life in Organisms, Cities, Economies, and Companies

[10] “Strong and Weak Ties” by Mark Granovetter published in The American Journal of Sociology May 1973

[11] “The Social Ecology of Disadvantaged Neighborhoods” by Robert J. Sampson and Jeffrey D. Morenoff, published in the journal Science in 2002.

[12] “Social networks and health: A systematic review of sociocentric network studies in low- and middle-income countries” by Margaret J. E. White and colleagues, published in the journal Social Science & Medicine in 2019.

[13] “Segregation, social ties, and neighborhood effects” by Christopher Winship and Robert Sampson, published in the journal The Urban Sociology Reader in 2008.

[14] https://iopscience.iop.org/article/10.1088/1742-5468/2013/02/P02043

[15] https://simudyne.com/resources/conways-game-of-life-and-the-birth-of-agent-based-modeling/

[16] https://www.publichealth.columbia.edu/research/population-health-methods/agent-based-modeling#:~:text=Agent%2Dbased%20models%20are%20computer,epidemiology)%20are%20assigned%20certain%20attributes.

[17] https://simudyne.com/resources/conways-game-of-life-and-the-birth-of-agent-based-modeling/

[18] W. Brian Arthur, Eric D. Beinhocker, and Allison Stanger, Complexity Economics

[19] https://www.nbcnews.com/news/world/china-xi-jinping-third-term-rcna53346

[20] https://www.macrotrends.net/countries/CHN/china/gdp-gross-domestic-product'>China GDP 1960–2023</a>

[21] https://www.macrotrends.net/countries/CHN/china/foreign-direct-investment'>China Foreign Direct Investment 1979–2023</a>

[22] Franck Billé and Caroline Humphrey, On the Edge

[23] https://roberthhacker.medium.com/a-primer-for-the-21st-century-the-5-systemic-changes-7e710e3c2321

[24] https://scholarworks.lib.csusb.edu/cgi/viewcontent.cgi?article=4506&context=etd-project

[25] https://medium.com/@EskoKilpi/new-economic-spaces-b5fd19f6a668

[26] https://medium.com/@ErlijnG/8-ways-artificial-intelligence-can-help-recognize-poverty-1278f6357d68

[27] https://www.santafe.edu/news-center/news/integrated-mess-music-lovers-science

--

--

Robert Hacker
Robert Hacker

Written by Robert Hacker

Director StartUP FIU-commercializing research. Entrepreneurship Professor FIU, Ex IAP Instructor MIT. Ex CFO One Laptop per Child. Built billion dollar company