See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/349636618

The Scope of Computational Social Science

Chapter · February 2021

CITATIONS READS
5 1,537

1 author:

Claudio Cioffi
George Mason University
147 PUBLICATIONS 3,868 CITATIONS

SEE PROFILE

All content following this page was uploaded by Claudio Cioffi on 26 February 2021.

The user has requested enhancement of the downloaded file.

 The Scope of Computational Social Science
Claudio Cioffi-Revilla
February 21, 2021

Abstract
Computational social science (CSS) is a field that integrates the
study of humans and groups at all scales through the formal method-
ology of computational and mathematical models. Its purpose is to
advance scientific understanding of social phenomena using computa-
tion, both as a conceptual and theoretical paradigm, based on informa-
tion processing, and as a methodological tool. Recently, a restrictive
version of CSS has been proposed, driven by “big data” from social
media, and progress in algorithms from computer science, eschewing
theory, mathematical models, and computational simulations—which
are other core areas of CSS. This chapter presents a comprehensive
and balanced scope of CSS that is theoretically and methodologically
guided by theory, enriched by analytical models, and enabled by com-
puter simulations; all three drawing on empirical data, be it big or
small. After comparing the restrictive approach to CSS to two in-
sightful metaphors, this chapter presents CSS as (a) an empirical sci-
ence seeking deep and testable knowledge, (b) an integrative social
science of humans and groups, and (c) a computational science that
uses computer and mathematical models to advance our understand-
ing of human and social dynamics. The scope of CSS is defined by
five areas that constitute its core and borderlands that straddle the
boundaries of CSS and other fields, as illustrated by contemporary
theory and research on complex crises.1
1
Chapter prepared for: Handbook of Computational Social Science, Volume 1: Theory,
Case Studies and Ethics, edited by Uwe Engel, Anabel Quan-Haase, Sunny Xun-Liu, and
Lars Lyberg. Oxon, UK: Routledge, forthcoming. Comments are welcomed by the author
at ccioffi1571 AT gmail DOT com

1
 It was six men of Indostan
To learning much inclined,
Who went to see the Elephant,
(Though all of them were blind)
That each by observation
Might satisfy his mind.

John Godfrey Saxe (transl.), The Blind Men and the Elephant

1 Introduction
Understanding, defining, and reflecting on the scope of computational social
science (CSS) could not be more timely, given the rapid and highly dynamic
expansion of this fledging field in recent years and current complex crises in
our epoch. This chapter provides context for and a description of the scope
of CSS, based on its history and evolving investigations from the core to the
frontiers of the field.
The scope of CSS is diverse, because the subject matter under investi-
gation is vast and its methods are numerous. For example, big data is ex-
ceptionally valuable for enriching the core and expanding the frontiers of the
field. It can be said that big social data, or big data about social phenomena,
is already beginning to play a transformative role as a driver of growth in
social knowledge. However, CSS is an interdisciplinary field of social science
that integrates individual social science disciplines as well as neighboring dis-
ciplines. The purpose of CSS is to advance scientific understanding of society
and social dynamics using the computational paradigm of complex adaptive
systems—as pioneered by Herbert A. Simon and his contemporaries—and to
make advanced use of computation in all its functions.
Recently, a more restrictive version of CSS has been proposed, driven
mostly by the novelty of “big data” from social media and other sources
(e.g., geospatial remote sensing, the Internet of Things (IoT), multimedia
and multi-sensor data), leveraging progress in algorithms from computer sci-
ence, while largely ignoring rigorous theory, models, or computational social
simulations—all three of which have provided powerful and lasting contribu-
tions.2 This chapter argues for a comprehensive and balanced scope for CSS,
2
Controversies over the proper scope of CSS would be entirely unnecessary were it

2
one that is paradigmatically guided by theory (“there is nothing so practical
as a good theory,” Lewin 1945: 128), enriched by analytical models, and en-
abled by computer simulations; all three drawing on data, be it big or small
(Kitchin and Lauriault 2014). The role of data in CSS is fundamental, as
it is in astronomy, biology, linguistics, or any other field of empirical and
computational science.
The chapter continues as follows: the next section describes two metaphors
that shed light on how to define the contemporary scope of CSS. Section
3 highlights the scientific nature of CSS and what that entails. Section 4
presents the main content, by providing a comprehensive and balanced scope
for CSS. The last section summarizes the main conclusions.

2 Two Metaphors: On Elephants and Italian

Pasta
Describing the scope of contemporary CSS in light of a plurality of con-
tending descriptions or specifications is challenging and reminiscent of two
insightful metaphors. Both metaphors, by analogy, illustrate how parochial
or unnecessarily narrow perspectives erroneously claim to represent what in
reality is a much bigger and richer reality.
The first and more familiar of the two metaphors is the proverbial parable
known as “The Blind Men and the Elephant,” cited at the beginning of this
chapter.3 Each blind man examines the elephant in its various constituent
components, including tusks, feet, ears, back, tail, and other parts. They
then compare notes and conclude that they are in disagreement, based on
their individual perceptions. But the elephant’s sighted owner has a realistic
not for the fact that several recent publications by a number of accomplished scholars
have presented a description of CSS focused entirely on the new big data (a sort of “Big
Data CSS”; e.g., Alvarez 2016; Lazer et al. 2009, 2020). By well-established scientific
and epistemological standards, such a view is incomplete, short-sighted, biased, and less
productive as a viable description of CSS, because it excludes other exciting constituent
components of the emerging field, such as theories, mathematical models, simulations, and
new analytical methods.
3
See www.wordfocus.com/word-act-blindmen.html for John G. Saxe’s English transla-
tion of the parable of The Blind Men and the Elephant. In the same website it is observed
that “the story is also used to teach tolerance for other cultures. We only ‘see’ the cul-
ture in which we are immersed.” Wise advice for computational social scientists, of all
persuasions.

3
understanding of the whole animal, because his comprehensive perspective
is based on direct observation.
The second metaphor is more scholarly and (regrettably) not as well
known. Here, a group of scientists from a variety of disparate disciplines
take the place of the men in the elephant’s metaphor. In this case, the
scientists congregate for an Italian pasta meal consisting of pappardelle alla
bolognese (noodles with delicious meat and tomatoes sauce from Bologna)
accompanied by some hearty Amarone red wine.4 The physicist sees how
heat applied to salted water makes the latter boil sooner when cooking the
pasta, and how torque exerted by his fork gathers each clump of pasta while
wondering whether surface tension may attract more sauce. The chemist un-
derstands how chewing the pasta begins to break down the grain flour glutens,
assisted by the sauce’s own acidity (pH ≈ 4.2). The management scientist, a
former systems engineer, thinks about the complex supply chains that make
possible the delicious meal, providing pasta, San Marzano tomatoes, olive
oil, ground meats, and all other necessary ingredients for a proper bolognese
sauce, not to mention the cheese and other tasty ingredients. The sociologist
enjoys her pasta while observing the table manners of her colleagues, two of
which assist their fork with a soup spoon while another cuts his pasta in-
stead of twisting it unbroken, indicative of their likely socioeconomic status
(SES). The chef knows all these (as well as other) perspectives, so he has
the most comprehensive and, therefore, most realistic understanding of the
whole situation.
Metaphorically speaking, CSS is like all elephants and all pasta dishes. It
is a fertile field consisting of several areas of investigation that offer immense
and exciting domains and tools for research, such as algorithmic information
extraction from big and small data sets, network models, social complexity,
and various well-established and novel computer simulation approaches, as
described in section 4 below. Each of these areas within CSS—grounded on
computational foundations—can be pursued individually or in combination,
depending on the goals of investigation. For one area to claim a monopoly
on the whole field is like one of the actors in the two metaphors claiming
that their’s is the whole field, ignoring other equally valid perspectives. CSS
is immensely exciting in proportion to the variety and power of its founda-
4
My summary version is adapted from McCain and Segal (1988). I am grateful to
Nancy McGlen, who many years ago at the University of Buffalo brought this amusing
and deeply insightful little book to my attention.

4
tional concepts, algorithms, data, theories, models, and simulations, espe-
cially when leveraged in combination and through creative synergies.

3 The Nature of CSS

We know what astronomy, organic chemistry, and archaeology are about,
which helps understand their respective scientific scope. As we approach the
scope of CSS, we need to clearly ask and answer the question: What is CSS
about? There are three classes of epistemological reasons why CSS requires—
and in fact demands inclusion of—theories, models, simulations, and other
scientific constructs, besides big data and data mining algorithms. These
reasons can be summarized in terms of three simple arguments stated in the
form of claims, or even testable (i.e., refutable and verifiable) hypotheses:

1. CSS is an empirical science.

2. CSS is a social science.

3. CSS is a computational science.

Each of these claims on the nature of CSS has been documented in a

library of volumes (cf., e.g, Bankes et al. 2002; Bennato 2015; Castelfranchi
2001; Cioffi 2009, 2010, 2014; Conte et al. 2011, 2012; Epstein 2006; Gilbert
2010; Hedström and Bearman 2009; Kuznar 2006; Manzo 2014; Squazzoni
2008; Torrens 2010; Trobia 2001), as well as in thousands of CSS papers
in professional journals and conference proceedings.5 What follows in this
section is a summary of “what CSS is about” intended for reference.
5
A small representative sample of CSS journals includes Computational and Mathemat-
ical Organization Theory (CMOT), founded by Herbert A. Simon, arguably the leading
CSS theoretician of the founder’s generation), the Journal of Artificial Societies and So-
cial Systems (JASSS), Social Science Computer Review (SSCR), and others in areas of
applied computing, network science (Social Networks), complexity science (Advances in
Complex Systems), applied physics (Physical Reviews E ), and simulation modeling (In-
ternational Journal of Modelling and Simulation). The Journal of Computational Social
Science (JCSS Springer) and IEEE’s Transactions on Computational Social Systems are
among the newest periodical journals specifically focused on CSS.

5
3.1 CSS is an empirical science: Testable knowledge
on humans and groups
CSS is part of science, as opposed to the fine arts or humanities, although
computational investigations in these two are growing. As an empirical sci-
ence (as opposed to pure mathematics), this basic fact indicates that the
epistemology of CSS has to do with elements such as systematic, testable,
and reproducible descriptions, explanations, and predictions (or forecasts)—
although the third class is not strictly necessary for CSS to qualify as a
science.6
Description, a foundational element of CSS qua empirical science, requires
observation and measurement, the results of which are often formalized in
terms of mathematically stated laws that describe how phenomena occur.
Duverger’s Law of Party Systems, Zipf’s Rank-Size Law, Pareto’s Law on
Income Inequality, and Richardson’s Law of War Magnitudes, among many
others, are examples (Arrow 1956; Kemeny & Snell 1962; Simon 1957). Al-
though not all scientific laws are stated in mathematical form, those that are
provide many advantages, not least of which is the power to generate addi-
tional logically valid and often empirically testable deductions for expanding
frontiers of knowledge.
Scientific laws that are stated mathematically are commonly referred to
as models (Arrow 1956; Kemeny et al. 1962; Lave and March 1993; Maki
and Thompson 1973; Olinick 2014), although the term “model” also applies
to other objects, not all of them necessarily mathematical—analogous to
the term “mechanism” in analytical social science and CSS. For example, a
flowchart, a Gantt chart, and any Unified Modeling Language (UML) dia-
gram (class, sequence, state, use case, and others) are also models (Bruschi
1971). Models provide ways of describing phenomena, sometimes crossing
the boundary between description and explanation.
Explanation in science means providing a causal mechanism that accounts
for observed phenomena, such as events and other occurrences, features, and
laws. A causal mechanism can be deterministic or probabilistic (Eels 1991;
Good 1980; Salmon 1980; Suppes 1984), depending on its causality, where
the “or” refers to the Boolean logic inclusive OR, also known as “and/or” in
common language. Laws require theoretical explanation, because by them-
6
The argument that CSS is part of science remains valid regardless of whether predic-
tion or forecasting are viewed as required features of science.

6
selves they only describe; they do not explain why a given pattern holds as a
law. A scientific explanation is normally in the form of a process composed
of one or more causal mechanisms (e.g., Lave and March 1993: ch. 1, on
how and why the theory of plate tectonics (due to Alfred Wegener and oth-
ers) is a valid scientific explanation for earthquakes, the intensity of which is
described [but not explained] by Richter’s well-known power law of seismic
magnitude).
Simulations are used in every field of science where there is need to ana-
lyze high dimensionality models or theories, because traditional closed form
solutions are unavailable or undesirable, for whatever reasons. High dimen-
sionality can be a sufficient condition for using simulations, which enable the
deduction of conclusions by running a simulation, assuming computational
tractability.
In all science, therefore, laws describe and theories explain—paraphrasing
S. Toulmin (1960). Laws and theories are constitutive or defining features of
science, not optional activities or exotic embellishments created for esthetic
purposes (although elegance is a desirable attribute of social theory; Cioffi
2017: ch. 8; specifically applying Lave and March 1975 to CSS). Simula-
tions are necessary only when scientific analysis requires many variables or
parameters that are relevant to the phenomenon under investigation, as are
all complex systems and many complicated ones. This is the first reason why
theories, models, and simulations are part of and necessary for CSS.
The essential role of social theory—i.e., the use of theoretical concepts,
causal mechanisms, and scientifically viable explanations—in understanding
and making sense of big data should be obvious. Consider the use of data
mining and dimensionality reduction algorithms for extracting information
from big data such as large-scale networks (e.g., Gleich and Mahoney 2016).
A common formalism used in algorithmic information extraction is to com-
pute the Laplacian matrix Ln×n of a large, weighted network graph G(V, E),
where n = |V |, defined as the difference between the diagonal or degree D
and the adjacency A matrices of G, or L := D − A. The common inter-
pretation of L when applied in data mining is as a network of springs with
different tensions, which is impossible to understand without a proper theo-
retical grasp of such dynamics and how spring-linked networks behave. (By
contrast, the mathematical interpretation of L is as a graph-based gradi-
ent equivalent to a field gradient in continuous space; which is not a theory.)
Similarly, results from most other complex data mining algorithms cannot be
understood without reference to constructs that constitute theoretical mod-

7
els, such as metric spaces, mathematical functions (and functionals), forces,
preferences, energy, entropy, and dynamics, among others.

3.2 CSS is a social science: The “Big Five” and all

others
Like all sciences, social science is articulated through disciplines. The uni-
verse of social science includes many large and well-established disciplines.
These are the “big five,” which are frequently defined as anthropology, eco-
nomics, political science, social psychology, and sociology—however, other
classifications are also common.7
The universe of social sciences also includes rich interdisciplinary fields,
such as human geography, management science, linguistics, communication
science, social science history, among others.
CSS is a social science, analogous to the claim that computational as-
tronomy, computational biology, and computational linguistics are part of
astronomy, biology, and linguistics, respectively. Any computational science
is, by definition, part of some science.
Each social science discipline includes a set of social theories, models,
and simulations, more in some disciplines and areas and less in others, albeit
some more than others.8 Social theories from “The Big Five” and other social
sciences are used in CSS to inform the causal mechanisms and processes under
investigation. Conversely, CSS contributes to social science by formulating
new theories that explain social phenomena.

3.3 CSS is a computational science: The value propo-

sition of computing
Social simulations are like computational theories; they are used when models
and theories contain too many variables for mathematical or analytical ap-
7
For example, at the US National Science Foundation, the Directorate of Social, Be-
havioral and Economic Sciences comprises two divisions, each containing a myriad of
disciplines.
8
If anything, it could be argued that a fundamental problem in social science is the
proliferation of unviable theories lacking in testability or falsifiability, as well as the survival
of demonstrably inferior, flawed, or false theories. CSS can play a useful role in this area
by demonstrating or exposing logical or empirical errors in putative social “theories”—i.e.,
debunking faulty theories

8
proaches leading to closed form solutions. This situation is common in many
social contexts, given the intrinsic complexity of even (seemingly) simple
human phenomena (e.g., aggregating votes, small group behavior, decision-
making under risk), as discussed below (section 4.1) in greater detail.
There is no single instrument that is used in science for producing new
knowledge. Conversely, no single tool has a monopoly on the production of
new science. Every science—natural, social, engineering, pure or applied—
relies on sets or ensembles of specialized instrumental methodologies and
methods to investigate its subject matter and achieve progress in the advance-
ment of knowledge. In the natural sciences, initially philosophical methods
of inquiry and, subsequently, after many centuries, mathematical, statisti-
cal, and computational methods provided such instruments (Casti 1992); in
addition to actual physical instruments. In the social sciences, roughly the
same instrumental sequence—philosophical, historiographic, statistical, and
mathematical tools—has occurred, albeit with somewhat different priorities
(statistics has prevailed over mathematics and historiographic methods are
not used in natural science, except in the field of history of science, which is
more akin to history).
Social science became computational after centuries of having used his-
toriographic, statistical, and mathematical methods—somewhat similarly to
the natural and engineering sciences—soon after the invention of automatic
computing; i.e., since the early 1960s (e.g., Guetzkow et al. 1963; Simon 1969
[first edition of Simon 1996]; Deutsch 1963; Messick 1963), about a decade
after John von Neumann’s (1966) pioneering Theory of Self-Reproducing Au-
tomata. The fact that the term “computational” was not used from the
very first day when computers were used to conduct social research does not
detract from the fundamental and historically verifiable claim.
As has been observed elsewhere, “Field Theory (Lewin 1952), Structuralist-
Functionalist Theory (Radcliffe-Brown 1952), Conflict Theory (Richardson
1952a, 1952b), the Theory of Groups (Simon 1952), Political Systems Theory
(Easton 1953), as well as Behavioral Decision Theory (Allais 1953), among
others, required new formalisms that could treat conceptual and theoretical
complexity of human and social dynamics, beyond what could be accom-
plished through systems of mathematical equations solved in closed form”
(Cioffi 2017: 36). Each of these social theories and related models, as well
as numerous others, eventually underwent computational implementation
and simulation became a new methodology (Benson 1961; Borko 1962), be-
fore object-orientation was introduced into social research through R, Java,

9
Python, and other programming languages.
Earlier it was stated that a computational approach—including but not
limited to simulation modeling methods—is required when encountering prob-
lems of high dimensionality, as is the case with many social phenomena.
However, the computational approach in social science has a dual nature,
encompassing a theory as well as a tool. This is because CSS is both a field
of science enabled by computing, as well as a field of social science informed
by a computational paradigm.
The methodological aspect of CSS is well-known obviously due to the
fact that, by definition, computing requires data processing hardware and
software. Less appreciated or understood, but far deeper from a scientific
perspective, is the computational paradigm of CSS introduced by Herbert A.
Simon (1996 [1969]) and others (Augier and March 2005; Batty 2006; Gilbert
and Troitzsch 2005; Holland 1975, 1995; Miller and Page 2007). This aspect
of CSS relies on the interdisciplinary theory of complex adaptive systems
and the role played by information processing at all scales of human and
societal behavior (Cioffi 2017: 331–334; Simon 1996). This computational
paradigm also includes a specific Theory of Artificial Systems that explains
how and why social complexity originates and evolves. Recent developments
of the same computational and information processing paradigm include the
Canonical Theory of Social Complexity (Cioffi 2005, 2014: 214–220, 2017:
335–341), which aims to explain the genesis and evolution of coupled and
complex systems-of-systems that include human, artificial, and natural enti-
ties.
Again, big data also plays a major role at this computational level, any
time that large amounts of information are necessary for numerous reasons
that may range from creating new theories and simulations to testing hy-
potheses and models of social systems and processes.

4 The Scope of CSS

The scope of any field can be described or specified in two ways: intensively
or extensively, similar to the way in which the definition for a given concept
is constructed (Collier and Gerring 2009; Goertz 2006; Hempel 1952). Here
we will approach the scope of CSS in both ways, because each perspective
elicits complementary information and jointly increases overall clarity. The
last part of this section illustrates the scope of CSS by considering what is

10
required to investigate and understand the subject of complex crises, which
are replete with challenging human and social dynamics.
For heuristic purposes, we shall describe the scope of CSS as consisting of
a core, comprised of five areas, surrounded by boundaries and borderlands.

4.1 Core
The core of contemporary CSS consists of the following five main areas of
investigation: computational foundations, algorithmic information extrac-
tion, networks, social complexity, and social simulations. Each of these in-
terrelated areas contains additional components—in terms of key concepts,
principles, and applications—such that the entire core comprises a system-
of-systems (Ackoff 1971; Cioffi 2017, with bibliographies for each area of the
CSS core). The five areas are conceived in roughly ordinal sequence, as in
an ordered set, because it is often the case that each area depends on or can
be enriched by one or more of the previous areas (as in a Likert scale); and,
conversely, each area can be investigated without recourse to subsequent ar-
eas (e.g., computational research on social networks, or on social complexity,
without reference to simulations).

Area 1: Computational foundations in CSS consist of ideas com-

mon to all computing systems and programs, and their basic characteristics
and relevance for humans and social systems. These include elementary
notions familiar to all computer scientists (computers, operations, memory,
programs, languages, abstraction, data structures, algorithms, and heuristic
tools such as UML [Unified Modeling Language]; Ambler 2005) and Systems
Modeling Language (SysML) (Fridenthal et al. 2015), especially how these
apply to humans and social systems (Cioffi 2017: ch. 2). In other words, CSS
foundations include all the building blocks of the field which define the com-
putational paradigm (not upper constructs). This is therefore different from
other formal or even mathematical approaches to doing social science. This
first, foundational area within the CSS core is conceptual, methodological,
procedural, and theoretical, consistent with the information processing and
complex adaptive systems approach of Herbert A. Simon and founders of the
field. Concepts, principles, and methods from computational foundations are
used in all other areas of CSS.9
9
The historical origins by founders of each area of CSS are summarized by chronological

11
 Area 2: Automated (or algorithmic) information extraction is
the second area within the core of CSS (Cioffi 2017: ch. 3). This field con-
cerns the use of algorithms, data structures, and associated methods (e.g.,
visualization, sonification) to extract information from social data, be it in
the form of large or small data sets. This is often seen as today’s “home of
big data,” with all its diversity (textual, digital, numerical, audio, graphi-
cal, geospatial, sensor data, as well as other forms).10 Research in this area
requires an understanding of principles of linguistics and content analysis,
semantic dimensions of meaning, and data mining in a technical sense. In
general practice, data mining takes on the framework of a process whereby
there is an initial corpus (or corpora) of input data pertaining to some human
or social phenomena (events, transactions, interactions, and other directional
data; but can also be attributes, such as geospatial features) and the auto-
mated information extraction proceeds through various stages which include
but are not limited to data preparation, algorithmic selection and prepara-
tion, the actual running of algorithms, data collection and, finally, analysis
and dissemination. These procedures often involve sequential spirals, similar
to a learning process, so they are not always as linear as they may sound.
Multidimensional cognitive spaces are used in this area of CSS (e.g., so-called
“sentiment analysis,” among others), as well as behavioral theories—e.g., Af-
fect Control Theory (Heise 1987), including Osgood’s foundations and Hoey’s
current developments (e.g., Hoey et al. 2013)—that explore and extend such
spaces. Results from automated information extraction are sometimes used
in other areas of CSS, such as in social network investigations, in research on
social complexity, or in simulations.

Area 3: Social networks are the third area of the CSS core (Cioffi
2017: ch. 4). Fundamental concepts and principles in this area include basic
network ideas as they apply to human and social science, as well as met-
rics, mathematical models, and associated theories. Although social network
analysis dates to the 1950s (i.e., prior to the availability of computers), an
interesting and telling reason why this area of CSS did not take off until
the advent of computational methods in social science is because it often
tables at the beginning of each chapter in Cioffi (2017).
10
In practice, however, social simulation (Area 5) generates just as much volume of data
from running models, so social media alone does not have a monopoly on “big data,” as
mentioned below.

12
requires the manipulation of large, non-sparse matrices (for specifying ad-
jacency, distance, or other networks of social relations). For example, in
the case of everyday social transactions, such large matrices have dimension
defined by N × N actors, requiring large and computationally expensive pro-
cesses, which—for all intents and purposes—are impractical or sometimes
simply not feasible to analyze other than computationally. Not until the
advent of computers, especially current fast computers within the past 10–
20 years, did social scientists gain access to such tools. This core area has
numerous applications in domains that range from individual-level human
cognition and belief systems to global networks in the world system. Any
given investigation in the area of social networks may be self-contained, or
it may relate to one or more of the other areas: social data mining, social
complexity, or simulation.

Area 4: Social complexity is the fourth area in the core of CSS (Cioffi
2017: 5–7; Simon 1965). This area is so extensive that it is often divided into
three subareas of learning and investigation: (1) origins and measurement of
social complexity (lines of evidence, indicators, organizational structures);
(2) laws of social complexity (both empirical and mathematical patterns);
and (3) theories of social complexity (explaining and understanding the ori-
gin and long-range development of social complexity). From a temporal
perspective, the computational science of social complexity covers the social
history of the past 12,000 years, since the earliest phase transitions from
kin-based, simple societies of hunter-gatherers to the first socially complex
human communities (simple chiefdoms), which first occurred during the late
Upper Paleolithic (ca. 9000–7000 BC) in Anatolia and adjacent regions of
west Asia. This long-range formative phenomenon (origin of social complex-
ity) occurred independently as the first archaic state-level polities formed in
separate parts of the world (west Asia, followed by east Asia, south Amer-
ica, and Mesoamerica) and throughout the millennia it has evolved to the
current aggregate and highly interdependent system of the world commu-
nity of countries (nation-states) and organizations that exhibit diverse forms
of social complexity. Measuring, describing, and explaining social complex-
ity is similar to cosmology, only focused on social as opposed to physical
(and much later biological) systems (Bellwood 2018; Christian 2004; Mithen
2004). Computational theories of social complexity are informed by both
archaeological and environmental data, as well as by formal theories based
on mathematical models.

13
 Area 5: Computational social simulations constitute the fifth area
of the CSS core. Beginning with pioneering investigations in the 1950s, this
area contains numerous concepts, principles, methodologies, and types of
simulation models. This area is divided into (1) simulation methodology
(motivation, design, verification, validation, and analysis of simulation mod-
els [the so-called MDIVVA methodology of simulation development], which is
independent of the specific type or simulation); (2) variable-oriented models
(system dynamics models, queueing models, microsimulations); (3) object-
oriented models (cellular automata, agent-based models); (4) learning and
evolutionary models; and (5) hybrid models that combine modules of differ-
ent kinds (e.g., system dynamics combined with agent-based modules, or any
of these augmented by learning or evolutionary algorithms). Social simula-
tion models are created in native code or using a modeling environment such
as Stella, Vensim, or Wolfram SystemModeler (for system dynamics models)
and MASON (Luke et al. 2019), Netlogo (Laver 2020; Wilensky & Rand
2015), or Repast (for agent-based models), among others.11

Not surprisingly, the above five areas of the CSS core also constitute its
modern academic curriculum. The whole core is often taught in a single
course (often called Introduction to Computational Social Science), whereas
each area is subsequently amplified by ad hoc courses. After mastering the
core, students are further trained in additional topics, such as geospatial
science, data visualization, evolutionary computation, epidemiology, or other
related topics discussed in the next section.

4.2 Boundaries and Borderlands

Beyond the core of CSS lie its boundaries and borderlands, where CSS meets
other fields and disciplines. While avoiding distraction by nominal cate-
gories and classifications, it is useful to clarify the boundaries of CSS. To
begin, consider the CSS core as the principal scientific territory of the field
(its “turf,” similar to mechanics being within the turf of physics or genet-
ics within biology), with “provinces” constituted by the five core areas just
described. Based on the nature and scope of CSS, at the boundaries we see
other disciplines and fields from neighboring natural sciences and engineer-
11
A useful resource in this area of CSS is the SIMSOC listserv.

14
ing sciences—since social complexity is always lodged within some natural
environment or ecosystem. Artifacts (“artificial systems,” in the sense of H.
Simon [1996]) provide the interface between society and nature. In addition,
the borderlands of CSS contain complexity science, applied mathematics,
and operations research, systems engineering, artificial intelligence, among
other fields.
Borderlands, which straddle the boundaries of CSS, are fascinating be-
cause they witness the interactive flow of people, information, energy, and
other resources—tangible and intangible, natural and engineered. The bor-
derlands of CSS are populated by researchers and ideas from CSS (“CSS
natives,” so to speak) and by researchers and ideas from neighboring dis-
ciplines (“collaborative foreigners” or partners). Examples of borderlands
include complexity science applications in CSS, any number of areas from
applied mathematics used in CSS (e.g., queuing theory, dynamical systems,
stochastic processes), and optimization problems encountered in CSS, among
many others.
Scientific interactions among core areas and borderlands generate CSS
investigations and new frontiers with increasingly universal scope, as follows:
Disciplinary CSS The most narrowly focused work refers to single-discipline-
based CSS, such as is the case for computational archaeology, computa-
tional economics, computational geography, computational linguistics,
computational political science, computational social choice, or com-
putational sociology. Each of these draws on concepts and principles
within the confines of a specific social science discipline or specializa-
tion.

Multidisciplinary CSS A higher level of universality refers to CSS that

is constituted by two or more social science disciplines, such as po-
litical sociology, social psychology, human geography, communication
science; or by one or more social science plus one or more natural or
engineering discipline, such as socio-natural systems, human factors,
computational conflict science, or cognitive neuroscience. The distinc-
tive feature of these is that these disciplines are aggregated without
much intent to integrate in a unified sense; i.e., the unity of science is
not the primary intent or result in multidisciplinary research.

Interdisciplinary CSS This is the version of CSS which seeks the highest
degree of scientific integration across disciplines and fields, based on

15
 unifying concepts and principles. Complexity science, network science,
systems theory, and information science, as well as applied mathemat-
ics, are often essential in supporting interdisciplinary CSS.

4.3 Complex crises

The scope of CSS just presented is well-illustrated by considering the nature
and scientific research demands of complex crises. While the core, bound-
aries, and borderlands of CSS describe its spatial scope on the scientific land-
scape, complex crises occur also in time, such that they represent a uniquely
challenging and currently active subject of CSS investigation. Not all crises
are complex (Hermann 1990; McClelland 1961; Snyder and Diesing 1977).
However, all crises are space-time defined by a large and partially interrelated
set of interesting features. Every crisis has the following phenomenology:

Initiating event(s) One or more specific historic event always initiates a

crisis, although it may take a while to recognize such events. Initiat-
ing events are frequently sudden or unexpected (e.g., the 9/11 terrorist
attacks, the 1962 discovery Soviet missiles directed to Cuba, the 2019
COVID-19 outbreak in Wuhan, China [Chang et al. 2020; Enserink &
Kupferschmidt 2020; Jager 2020], and similar “triggers”)12 , but they
may also occur gradually (e.g., human migrations, climate change, sub-
stance abuse epidemics, depletion of nonrenewable resources, increasing
illiteracy in science, decreasing technology skillsets).

Branching process The lifecycle of a crisis involves a branching process,

where the root of the tree is given by the initiating event(s), branches
are represented by the various sequences of events that follow the ini-
tiation, and the leaf events define an outcome space that is highly
contingent or path-dependent. The probability of each outcome is de-
termined by the probability of all prior events, according to the tree’s
structure—by probabilistic causality.
12
Another term used in the crisis literature is “precipitating event.” In the case of
international relations crises, the initiating event is generally a compound event consisting
of other events (Cioffi 1998: 160–163; Snyder and Diesing 1977: 15). Such compound
events frequently encapsulate an underlying complex structure that can be challenging to
understand and model.

16
Phase transitions The branching process of a crisis includes transition
from pre-crisis normality, to the crisis itself (in a narrow sense), to the
crisis aftermath. This strictly ordered sequence of phase transitions is
another universal characteristic in the phenomenology of crises. The
timing of ordered phase transitions is often unpredictable, although the
necessary or sufficient conditions associated with each transition may
be known or predictable.

Uncertainty increase Uncertainty increases in a crisis, because the variety

and number of possible outcomes in the outcome space—including some
hazardous subspaces—is high relative to normal, non-crisis conditions.
Increases in uncertainty (entropy) are known to naturally cause stress
in humans, whereas a decrease in variety or number of outcomes can
decrease stress.

Possible extreme outcomes While some outcomes of a crisis may be ben-

eficial (e.g., win-win resolution outcomes), others frequently generate
calamities with a range of negative payoffs. The (ordinal) level of
calamity ranges from emergency (relatively minor consequences well
within management’s resources), to disaster (significant but manage-
able consequences with available resources), to catastrophe (unman-
ageable consequences beyond all available resources), to extinction (so-
cietal collapse).13

Time pressure Stress experienced by crisis participants and stakeholders

has the effect of reducing the perceived time available to make deci-
sions. Scientifically valid information can reduce such stress by adding
verifiable information, realism, and comparative perspective to the sit-
uation.

High stakes Low stakes are inconsequential, by definition, so it takes high

stakes to produce a crisis. High stakes can be generated by skewed
distributions of events as well as other sources of extreme values or
realizations, such as cultural values, features of complexity (discussed
below) and other sources. High stakes may not be the same for all ac-
tors, so relative comparisons can clarify a crisis and sometimes facilitate
its resolution.
13
This is a simplified version of the ordinal scale of calamity used in the area of disaster
research and emergency management.

17
Game structure and evolution Every crisis is formally characterized by
some game structure in terms of actors, strategies, outcomes, pay-
offs, and situational information such as communication, perceptions
of length of game, and other characteristics. Often the crisis game
structure is some combination of elementary 2 × 2 game in the sense
of Rapoport. However, the game structure of a crisis is typically non-
stationary, since it evolves through several forms from the onset phase
to termination, and is not always known in real-time (because it de-
pends on assessments and estimates of actors’ strategies, preferences,
and other defining features of the crisis game).

While the above are universal features of all crises, the subset of complex
crises have the following additional characteristics:

Many actors The simplest crises occur between two actors. Complex crises
involve three or more actors or components (i.e., high cardinality). An
analogy in natural science would be the so-called n-body problem in
astronomy, where the future trajectory of three or more bodies in a
gravitational field is impossible to obtain in closed form, but can be
rigorously analyzed via a computational model.

High dimensionality Numerous variables and parameters (both continu-

ous and discrete) are necessary to model and understand a complex cri-
sis, making it more difficult and frequently impossible to derive closed
form solutions. Formally, actors and entities in a complex crisis encap-
sulate numerous attributes (and operations/methods); more so than in
a simple crisis.

High heterogeneity The diversity of entities in the ontology of a complex

crisis is higher than in a simple crisis. Classes of heterogeneous actors,
multiplex network relations (each relation, in turn, having multiple
attributes and dynamics), time scales (from short to long, depending
on strategies considered, spanning several orders of magnitude), and
other heterogeneous entity categories are examples.

Large outcome space While simpler crises may have a relatively small
outcome space, a complex crisis always has a larger outcome space, base
on the above features. Outcome space size grows exponentially with
cardinalities, so in a complex crisis the size of the outcome space can

18
 easily be orders of magnitude greater. Due to these challenging count-
ing problems, regions of such outcome spaces are often unknown or
even unknowable, especially in real (i.e., actionable) time. Metastabil-
ity—the possibility of or potential for transitioning to other states—is
a related feature of a complex crisis.

Networked interdependence Connectivity among actors and other enti-

ties in a complex crisis is manifested through interdependence.14 While
game theory provides some assistance in understanding this aspect,
there are myriad other aspects of networked interdependence that re-
main analytically intractable in closed form, which calls for computa-
tional simulation methods.

Nonlinearity Finally, numerous relations and, therefore, many effects are

not linear, but rather exponential, hyperbolic, periodic, stochastic, or of
other forms. Nonlinearity can be a source of surprise and uncertainty,
adding to overall stress. Hierarchies, feedback loops, bifurcations, tip-
ping points, and phase transitions are manifestations of nonlinearity.

While the theoretical analysis of crises for cases that have minimal, re-
duced form—such as in 2-actor 2 × 2 games and, more recently, mean field
games—real world complex crises are more analytically tractable via increas-
ingly powerful approaches provided by CSS. Climate change, epidemics rang-
ing from local to global pandemics, increasing challenging urban and critical
infrastructure systems, and the challenges of spacefaring civilization, among
others, provide examples of contemporary complex crises that are being inves-
tigated in collaboration with other fields—at the CSS core, the borderlands,
and beyond.

5 Conclusions
Computational social science is an empirical science, a social science, and a
field similar to other computational sciences where computing plays a dual
and defining role, both as a theoretical paradigm and a methodological in-
strument. The former regards the fundamental role of information-processing
for describing, explaining, and understanding social phenomena, in this case
14
The interdependence of states is known as “entanglement” in quantum science.

19
through human cognition and decision-making at the micro-level that gen-
erates societal phenomena at the macro-level. The latter regards the role of
the computer as an enabling instrument allowing us to reach and investigate
frontiers of knowledge far beyond what is possible through historical, statis-
tical, or even mathematical methods. Both aspects define the character of
CSS.
Today, theories, models, and data (both “small” and “big”) comprise the
complex landscape and rich ecology of CSS. The only difference in terms
of big data is that their sheer size and rate of growth (first and second
derivatives, respectively) present major and exciting challenges to the other
components. Theories and models are challenged and enabled by big data,
and vice versa. Computational social simulations use and also create new
demands for big data.
There is good reason to think that “the science of the twenty-first cen-
tury will be computational” (Denning 2007). Such a big CSS perspective in-
cludes big data, algorithms, theories, models, and simulations, among other
scientific constructs and instruments. Denning’s insightful prediction (and
conjecture) is arguably becoming increasingly true of social science, as for
most other domains of science. In the future, social science will prosper if
it adopts the computational approach with more robust understanding of
its historical origins and rich diversity and, even more so, if CSS leverages
more formal approaches from applied mathematics, just as during the twen-
tieth century the social sciences learned how to apply statistics. This is also
necessary with the addition of virtual worlds and robots (Parisi 2014) to
the domain of social science investigations (Bainbridge 2007). The synergy
of current statistical approaches, long-range historical and comparative per-
spectives, enhanced variety of mathematical approaches (beyond differential
equations and game theory), and the full spectrum and scientific power of
computational approaches cannot but generate an explosive abundance of
new discoveries and deeper understanding of social phenomena, both simple
and complex.

So oft in theologic wars,

The disputants, I ween,
Rail on in utter ignorance
Of what each other mean,
And prate about an Elephant
Not one of them has seen!

20
 J. G. Saxe (transl.), The Blind Men and the Elephant

Computational social scientists should work together to describe, explain,

and scientifically understand the whole elephant of social science.

Acknowledgment
Dedicated to the late Rosaria Conte[1954–2016], interdisciplinary computa-
tional social scientist per eccellenza. The earliest and shorter version of this
chapter was presented at the panel on “Computational Social Sciences: A
Bricolage of Approaches,” 8th International ACM Web Science Conference
(WebSci’16), Hanover, Germany, May 22–25, 2016, chaired by Paolo Parigi.
An expanded version was presented at the 2016 Lipari International Summer
School in Complex Systems and Computational Social Science, Lipari, Italy,
directed by Alfredo Ferro. The author would like to thank panel chair Paolo
Parigi for the invitation to write and present this paper, and Robert Axtell,
Joshua Epstein, Nigel Gilbert, Dirk Helbing, David Masad, J. Daniel Rogers,
Larry Kuznar, Vahab Mirrokni, Chris Rouly, Flaminio Squazzoni, and Klaus
G. Troitzsch for comments and discussions. The final version of this chapter
was presented and developed for the Digital Traces Workshop, University of
Bremen, Germany, November 8–10, 2018, at the kind invitation of Professor
Uwe Engel. This work was originally funded by NSF CDI Type II grant no.
IIS-1125171 and by the Center for Social Complexity at George Mason Uni-
versity. The author is solely responsible for the views and opinions expressed
in this paper.

References
Ackoff, Russell L. (1971). Towards a system of systems concepts. Manage-
ment Science 17(11): 661–671.

Allais, Maurice. (1953). Le comportement de l’homme rationnel devant le

risque, critique des postulats et axiomes de l’ecole Americaine [Rational hu-
man behavior under risk: critique of postulates and axioms of the American
school]. Econometrica 21: 503–546.

21
Alvarez, R. Michael (2016). Computational Social Science: Discovery and
Prediction. Cambridge University Press.

Ambler, S. W. (2005). The Elements of UML 2.0 Style. Cambridge and New
York, Cambridge University Press.

Arrow, K. J. (1956). Mathematical models in the social sciences. General

Systems: Yearbook of the Society for General Systems Research 1: 29–24.

Augier, Mie, & March, James G. (Eds.). (2004). Models of a Man: Essays
in Memory of Herbert A. Simon. Cambridge, MA: The MIT Press.

Bainbridge, W. S. (2007). The scientific research potential of virtual worlds.

Science 317(5837): 472–476.

Batty, Michael. (2006). Cities and Complexity: Understanding Cities with

Cellular Automata, Agent-Based Models, and Fractals. Cambridge, MA: The
MIT Press.

Bankes, Steven C., Lempert, Robert J., & Popper, Steven W. (2002). Mak-
ing Computational Social Science Effective: Epistemology, Methodology, and
Technology. Social Science Computer Review, 20(4), 377–388.

Bellwood, P. (2018). Agricultural Origins. International Encyclopedia of

Anthropology, 1–9.

Bennato, Davide. (2015). Il computer come macroscopio [The Computer as

a Macroscope]. Milan, Italy: Franco Angeli.

Benson, Oliver. (1961). A Simple Diplomatic Game. In James N. Rosenau

(Ed.), International Politics and Foreign Policy (pp. 504–511). New York:
Free Press.

Borko, Harold (Ed.) (1962). Computer Applications in the Behavioral Sci-

ences. Englewood Cliffs, NJ: Prentice-Hall.

Bruschi, A. (1971). La Teoria dei Modelli nelle Scienze Sociali [The Theory

22
of Models in the Social Sciences]. Bologna, Italy, Il Mulino.

Cangelosi, A. and D. Parisi, eds. (2012). Simulating the evolution of lan-

guage. Cham, Switzerland, Springer Science & Business Media.

Castelfranchi, Cristiano. (2001). The theory of social functions: challenges

for computational social science and multi-agent learning. Journal of Cogni-
tive Systems Research, 2, 5–38.

Casti, John (1992). Reality Rules, Picturing the World in Mathematics, Vol-
umes 1 (The Fundamentals) and 2 (The Frontiers). New York: Wiley.

Chang, Sheryl L., Nathan Harding, Cameron Zavhreson, Oliver M. Cliff,

Mikhail Propopenko (2020). Modelling transmission and control of the COVID-
19 pandemic in Australia. arXiv, March 24, 2020. Preprint.

Cioffi-Revilla, Claudio. (2005). A Canonical Theory of Origins and Devel-

opment of Social Complexity. Journal of Mathematical Sociology, 29(April-
June), 133–153.

Cioffi-Revilla, Claudio. (2009). On Social Complexity: A Manifesto for

Computational Social Science. Presidential Address. Annual Conference of
the North American Association of Computational Social and Organizational
Sciences, Phoenix, Arizona.

Cioffi-Revilla, Claudio. (2010). Computational Social Science. Wiley Inter-

disciplinary Reviews (WIREs): Computational Statistics, 2(3), 259–271.

Cioffi-Revilla, Claudio (2017). Introduction to Computational Social Science:

Principles and Applications. 2nd edition. Cham, Switzerland: Springer.

Collier, D. and J. Gerring (2009). Concepts and Method in Social Science.

New York and London, Routledge.

Conte, Rosaria, Gilbert, Nigel, Bonelli, Giulia, & Helbing, Dirk. (2011).
FuturICT and Social Sciences: Big Data, Big Thinking. Zeitschrift fur Sozi-
ologie, 40(5), 412–413.

23
Conte, R., Gilbert, G. Nigel, Bonelli, G., Cioffi-Revilla, C., Deffaunt, G.,
Kertesz, J., . . . Helbig, D. (2012). Manifesto of Computational Social
Science. European Physical Journal Special Topics, 214, 325–346.

Christian, D. (2004). Maps of Time: An Introduction to Big History. Los

Angeles, University of California Press.

Deutsch, Karl W. (1963). The Nerves of Government. New York: Free Press.

Easton, David. (1953). The Political System: An Inquiry into the Study of
Political Science. New York: Alfred A. Knopf, Inc.

Eells, E. (1991). Probabilistic causality (Vol. 1). Cambridge University

Press.

Enserink, M. & K. Kupferschmidt (2020). With COVID-19, modeling takes

on life and death importance. Science 367(6485): 1414–1415.

Epstein, Joshua. (2006). Generative Social Science: Studies in Agent-Based

Computational Modeling. Princeton, NJ: Princeton University Press.

Freidenthal, S., Alan Moore, andRick Steiner (2015). A Practical Guide to

SysML: The Systems Modeling Language. Amsterdam, Netherlands, Else-
vier.

Gilbert, Nigel (Ed.) (2010). Computational Social Science. Los Angeles:

Sage.

Gilbert, N. & K. Troitzsch (2005). Simulation for the Social Scientist. Buck-
ingham and Philadelphia, Open University Press.

Goertz, G. (2006). Social Science Concepts: A User’s Guide, Princeton Uni-

versity Press.

Good, I. J. (1980). Some comments on probabilistic causality. Pacific Philo-

sophical Quarterly 61: 301-304.

Gleich, David F., & Mahoney, Michael W. (2016). Mining Large Graphs.

24
In Peter Bühlmann, Petros Drineas, Michael Kane, & Mark van der Laan
(Eds.), Handbook of Big Data. Boca Raton, London, and New York: CRC
Press. Pp. 192–215.

Guetzkow, Harold, Alger, Chadwick F., Brody, Richard A., Noel, Robert C.,
& Snyder, Richard C. (Eds.). (1963). Simulation in International Relations:
Developments for Research and Teaching. Englewood Cliffs, NJ: Prentice-
Hall.

Heise, David R. (1987). Affect Control Theory: Concepts and Model. Jour-
nal of Mathematical Sociology 13(1–2): 1–33.

Hedström, Peter, & Bearman, Peter (Eds.). (2009). The Oxford Handbook
of Analytical Sociology. Oxford University Press.

Hempel, C. G. (1952). Fundamentals of Concept Formation in Empirical

Science. Chicago, University of Chicago Press.

Hermann, C. F. (1990). International Crisis as a Situational Variable. Clas-

sics of International Relations. J. A. Vasquez. New Jersey, Prentice-Hall,
Inc.

Hoey, J., et al. (2013). Bayesian affect control theory. 2013 Humaine Asso-
ciation Conference on Affective Computing and Intelligent Interaction, IEEE.

Holland, John H. (1975). Adaptation in Natural and Artificial Systems. Ann

Arbor, MI: University of Michigan Press.

Holland, John H. (1995). Hidden Order: How Adaptation Builds Complexity.

Reading, MA: Addison-Wesley.

Jager, W. (2020). Predictions on the spread of COVID-19 bear many uncer-

tainties. Elephant in the Lab. https://doi.org/10.5281/zenodo.3724048.

Kemeny, J. G. & J. L. Snell (1962). Mathematical Models in the Social

Sciences. Cambridge, Mass., The MIT Press.
Kitchin, Rob, & Lauriault, Tracey P. (2014). Small data in the era of big
data. GeoJournal, 80(4), 463–475.

25
Kuznar, Lawrence A. (2006). High-Fidelity Computational Social Science in
Anthropology. Social Science Computer Review, 24(1), 15–29.

Lave, Charles A., & March, James G. (1993 [1975]). An Introduction to

Models in the Social Sciences. New York: Harper and Row.

Laver, M. (2020). Agent-Based Models of Social Life: Vol. 1 (Fundamen-

tals), Vol. 2 (Polarization and Ethnocentrism). Cambridge University Press.

Lazer, David, Pentland, Alex, Adamic, Lada, Aral, Sinan, Barabasi, Albert-
Laszlo, Brewer, Devon, . . . Alstyne, Marshall Van. (2009). Computational
Social Science. Science, 323(5915), 721–723.

Lewin, Kurt. (1952). Field Theory in Social Science: Selected Theoretical

Papers. Chicago and London: University of Chicago Press.

Luke, S., et al. (2019). The MASON Simulation Toolkit: Past, Present,
and Future. Proceedings of the MABS (Multi-Agent-Based Simulation XIX)
Workshop, AAMAS, Stockholm, Sweden, 2018. Heidelberg, Germany, Springer.

Mithen, S. (2004). After The Ice: A Global Human History, 20,000–5000

BC. Cambridge, MA, Harvard University Press.

Maki, D. P. and M. Thompson (1973). Mathematical Models and Applica-

tions: With Emphasis on the Social, Life, and Management Sciences. En-
glewood Cliffs, New Jersey, Prentice-Hall, Inc.

Manzo, Gianluca (Ed.) (2014). Analytical Sociology: Actions and Networks.

New York: Wiley.

McCain, Garvin, & Segal, Erwin M. (1988). The Game of Science (5th edi-
tion ed.). Pacific Grove, CA: Brooks/Cole Publishing Co.

McClelland, C. A. (1961). The acute international crisis. World Politics

14(1): 182-204.

Messick, Samuel J. (1963). Computer Simulation of Personality: Frontier of

26
Psychological Theory. Philadelphia: American Psychological Association.

Miller, John H., & Page, Scott E. (2007). Complex Adaptive Systems: An
Introduction to Computational Models of Social Life. Princeton and Oxford:
Princeton University Press.

Mithen, S. (2004). After The Ice: A Global Human History, 20,000–5000

BC. Cambridge, MA, Harvard University Press.
Olinick, M. (2014 [1978]). An Introduction to Mathematical Models in the
Social and Life Sciences. Reading, Mass., Addison-Wesley.

Parisi, D. (2014). Future robots: Towards a robotic science of human beings.

Amsterdam and Philadelphia, John Benjamins Publishing Company.

Radcliffe-Brown, A. R. (1952). Structure and Function in Primitive Society.

Glencoe, Illinois: Free Press.

Richardson, Lewis Fry. (1952a). Is it possible to prove any general state-

ments about historical fact? British Journal of Sociology, 3(1), 77–84.

Richardson, Lewis Fry. (1952b). Contiguity and deadly quarrels: The lo-
cal pacifying influence. Journal of the Royal Statistical Society, Series A,
115(Part II), 219–231.

Salmon, W. C. (1980). Probabilistic Causality. Pacific Philosophical Quar-

terly 61: 50–74.

Simon, Herbert A. (1952). A formal theory of interaction in social groups.

American Sociological Review, 17(2), 202–212.

Simon, Herbert. (1957). Models of Man. New York: John Wiley and Sons.

Simon, Herbert A. (1965). The architecture of complexity. Proceedings of

the American Philosophical Society, 106, 467–482.

Simon, Herbert A. (1996 [1969]). The Sciences of the Artificial 3rd edition.
Cambridge, MA: MIT Press.

27
Snyder, G. H. and P. Diesing (1977). Conflict Among Nations: Bargaining,
Decisionmaking, and System Structure in International Crises. Princeton,
N.J., Princeton University Press.

Squazzoni, Flaminio. (2008). A (computational) social science perspective

on societal transitions. Computational and Mathematical Organization The-
ory, 14(4), 266–282.

Suppes, P. (1984). Probabilistic Metaphysics. New York and Oxford, Basil

Blackwell.

Toulmin, S. (1960). The Philosophy of Science: An Introduction. New York,

Harper and Row.

Torrens, P. M. (2010). Geography and computational social science. Geo-

Journal 75(2): 133–148.

Trobia, Alberto. (2001). La Sociologia Come Scienza Rigorosa: Modelli sim-

ulativi, intelligenza collettiva, forme del mutamento [Sociology as a Rigorous
Science: Simulation Models, Collective Inteligence, and Patterns of Change].
Milan, Italy: Franco Angeli.

von Neumann, John. (1966). Theory of Self-Reproducing Automata. Edited

and completed by Arthur W. Burks. Urbana, Illinois: University of Illinois
Press.

Wilensky, U. & W. Rand (2015). An Introduction to Agent-Based Modeling:

Modeling Natural, Social, and Engineered Complex Systems with NetLogo.
Cambridge, MA, MIT Press.

28
Index
9/11 terrorist attacks, 16 Bearman, Peter, 5
belief systems, 13
academic curriculum of CSS, 14, 16 Bellwood, Peter, 13
Ackoff, Russell L., 11 Bennato, Davide, 5
algorithms, 2, 5, 12 Benson, Oliver, 9
dimensionality reduction example, big data, 2, 4, 5, 10, 12
7 social theory for explanation, 7
in computational foundations of biology, 3, 14
CSS, 11 computational biology, 8
Laplacian matrix, 7 Boolean logic, 6
Allais, Maurice, 9 Borko, Harold, 9
Alvarez, Michael, 3 boundaries and borderlands of CSS,
Ambler, Scott W., 11 14
analytical social science, 6 branching process, 16
Anatolia, Turkey, 13 path dependency, 16
applied mathematics, 15, 16 Bruschi, Alessandro, 6
and future of CSS, 20
dynamical systems, 15, 20 calamities, 17
game theory, 18, 20 catastrophes, 17
graph theory (see also network sci- disasters, 17
ence), 19 emergencies, 17
optimization, 15 extinctions, 17
queuing theory, 15 calamitous outcomes, 17
stochastic processes, 15 Canonical Theory of Social Complex-
archaeology, 5 ity, 10
computational archaeology, 15 cardinality, 18
Arrow, Kenneth J., 6 and exponential growth, 18
artificial intelligence, 15 high cardinality in complex crises,
astronomy, 3, 5, 18 18
computational astronomy, 8, 18 Castelfranchi, Cristiano, 5
Augier, Mie, 10 Casti, John, 9
catastrophes, 17
Bainbridge, William S., 20 Chang, Sheryl L., 16
Bankes, Steven C., 5 chemistry, 4, 5
Batty, Michael, 10 chiefdoms, 13

29
Christian, David, 13 computational social choice, 15
Cioffi-Revilla, Claudio, 5, 7, 9–13, 16 computational social psychology, 15
climate change, 16, 19 computational socio-natural systems,
closed form, 7, 9 15
cognition, 13, 20 computational sociology, 15
Collier, David, 10 computational tractability, 7
communication science computer science, 2
computational communication sci- concepts, 5, 7
ence, 15 Conte, Rosaria, 5, 21
complex adaptive systems, 2, 10, 11 content analysis, 12
complex crises, 2, 11, 16–19 core areas of CSS, 4
complex systems, 7, 15, 16 computational foundations, 4, 11
bifurcations, 19 social complexity, 4, 9, 10, 13, 15
emergence, 20 chiefdoms, 13
metastability, 19 kin-based societies, 13
non-stationarity, 18 laws, 13
path dependency, 16 measurement, 13
phase transitions, 13 origins, 13
skewed distributions, 17 simple societies, 13
computational archaeology, 15 states, 13
computational cognitive neuroscience, theories, 13
15 social data mining, 4, 7, 12–13,
computational communication science, 20
15 Laplacian matrix, 7
computational conflict science, 15 learning process, 12
computational economics, 15 stages of analysis, 12
computational geography, 15 social networks, 4, 7, 12
computational human factors, 15 adjacency matrix, 7, 13
computational human geography, 15 degree matrix, 7
computational linguistics, 15 diagonal matrix, 7
computational paradigm, 2, 10, 11, distance matrix, 13
19 social simulation, 4, 8, 9, 12–14,
Simon’s Theory of Artificial Sys- 20
tems, 10 agent-based models, 14
computational political science, 15 learning and evolutionary mod-
computational political sociology, 15 els, 14
computational science, 3, 5, 8, 9, 20 object-oriented models, 14
case of CSS, 8–10 SIMSOC listserv, 14

30
 variable-oriented models, 14 graphical, 12
cosmology, 13 interactions, 12
coupled human-natural-engineered sys- numerical, 12
tems, 10 role of, 3
COVID-19 pandemic, 16 sensor data, 12
crises textual, 12
branching process, 16 transactions, 12, 13
complex crises, 16, 18–19 visualization, 14
high heterogeneity, 18 data sonification, 12
high cardinality, 18 data structures, 11, 12
high dimensionality, 18 data visualization, 12, 14
high heterogeneity, 18 decision-making, 17
large outcome space, 18–19 during crises, 17
networked interdependence, 19 under risk, 9
nonlinearity, 19 under uncertainty, 17
phenomenology, 18–19 Denning’s Conjecture, 20
decision-making conditions, 17 Denning, Peter, 20
extreme outcomes, 17 description, 6, 7
game structure, 18 Deutsch, Karl W., 9
high stakes, 17 Diesing, Paul, 16
initiating event(s), 16 dimensionality, 18
outcome space, 16, 17 high dimensionality, 18
phase transitions, 17 disasters, 17
phenomenology, 16–18 disaster research (see also calami-
time pressure, 17 ties), 17
uncertainty increase, 17 disciplinary CSS, 15
critical infrastructure systems, 19 duality of computation in CSS, 10, 19
Cuban Missiles Crisis of 1962, 16 dynamical systems, 15, 20
dynamics, 8
data, 5, 12
archaeological, 13 earthquakes, 7
attribute, 12 Richter’s Power Law of Seismic
audio, 12 Magnitudes, 7
both “small” and “big”, 20 Wegener’s Theory of Plate Tec-
digital, 12 tonics, 7
environmental, 13 east Asia, 13
events, 12 Easton, David, 9
geospatial, 12 economics

31
 computational economics, 15 geography
ecosystem, 15 computational geography, 15
Eells, Ellery, 6 George Mason University
emergence, 20 Center for Social Complexity, 21
emergencies, 17 geospatial data, 2
emergency management, 17 geospatial science, 14
empirical science, 5 Gerring, John, 10
energy, 8, 15 Gilbert, Nigel, 5, 10, 21
Engel, Uwe, 21 Gleich, David F., 7
engineering, 9, 15 global networks, 13
Enserik, Martin, 16 Goertz, Garry, 10
entropy, 8, 17 Good, I.J., 6
epidemics, 19 Guetzkow, Harold, 9
epidemiology, 14
epidemics, 19 Hedström, Peter, 5
pandemics, 19 Heise, David R., 12
epistemology, 3, 5, 6 Helbing, Dirk, 21
Epstein, Joshua, 5, 21 Hempel, Karl, 10
evolutionary computation, 14 Hermann, Charles F., 16
extreme outcomes, 17 heterogeneity, 18
extreme values, 17 high dimensionality, 7, 10, 18
high heterogeneity, 18
Ferro, Alfredo, 21 history, 9
flowchart, 6 history of science, 9
forces, social, 8 Hoey, Jesse, 12
forecasting, 6 Holland, John H., 10
formal methods, 9, 13 human geography
mathematics, 9 computational human geography,
formal models, 6, 9, 11, 13 15
Fridenthal, Sanford, 11 human migrations, 16
functions, mathematical, 8 hunter-gatherer societies, 13

game theory, 18, 20 Indostan, 2

and 2-actor 2 × 2 games, 19 information, 15
and complex crises, 18 information processing, human and so-
mean field games, 19 cietal, 10, 11, 19
Gantt chart, 6 information science, 16
genetics, 14 initiating event(s), 16

32
interdependence, 19 Lewin, Kurt, 3, 9
interdisciplinary CSS, 15–16 Likert scale, 11
unity of science, 16 linguistics, 3, 12
interdisciplinary nature of CSS, 2 computational linguistics, 8, 15
Internet of Things (IoT), 2 Lipari International Summer School
invention of automatic computing, 9 in Complex Systems and Com-
putational Social Science, Li-
Jager, W., 16 pari, Italy, 21
Java programming language, 9 Luke, Sean, 14
journals, 5
Advances in Complex Systems, 5 Mahoney, Michael W., 7
Computational and Mathematical Maki, Daniel P., 6
Organization Theory (CMOT ), Manzo, Gianluca, 5
5 March, James G., 6, 7, 10
IEEE Transactions on Computa- Masad, David, 21
tional Social Systems, 5 McCain, Garvin, 4
International Journal of Modelling McClelland, Charles A., 16
and Simulation, 5 McGlen, Nancy E., 4
Journal of Artificial Societies and measurement, 6
Social Simulation (JASSS ), 5 mechanism, 6–8
Journal of Computational Social deterministic, 6
Science, 5 probabilistic, 6, 16
Social Networks, 5 mechanisms, 7
Social Science Computer Review Mesoamerica, 13
(SSCR), 5 Messick, Samuel J., 9
metaphors, 3–5
Kemeny, John G., 6 The Blind Men and the Elephant,
kin-based societies, 13 3–4, 21
Kitchin, Rob, 3 The Italian Pasta Meal, 4
Kupferschmidt, Kai, 16 metastability, 19
Kuznar, Larry, 21 methodology, 9
Kuznar, Lawrence A., 5 metric spaces, 8
large outcome space, 19 Miller, John, 10
Lauriault, Tracey P., 3 Mirrokni, Vahab, 21
Lave, Charles A., 6, 7 Mithen, Steven, 13
Laver, Michael, 14 multi-sensor data, 2
Lazer, David, 3 multidisciplinary CSS, 15
Lempert, Robert J., 5

33
 computational communication sci- outcome space, 16–19
ence, 15 calamitous outcomes, 17
computational human geography, exponential growth with cardinal-
15 ities, 18
computational political sociology, large outcome space, 19
15 order of magnitude, 19
computational social psychology, win-win outcomes, 17
15
multimedia data, 2 Page, Scott, 10
multiplex, 18 Paleolithic period, 13
pandemics, 16, 19
natural science, 9, 14 Parigi, Paolo, 21
nature and definition of CSS, 5–10 Parisi, Domenico, 20
network science, 5, 16 path dependency, 16
large-scale networks, 7 perception, 3
multiplex, 18 phase transitions, 13, 17
non-stationarity, 18 phenomenology, 19
nonlinearity, 19 of crises, 16
and bifurcations, 19 philosophy, 9
and feedback loops, 19 physical instruments, 9
exponential, 19 physics, 4, 14
hierarchies, 19 political science
hyperbolic, 19 computational political science, 15
periodic, 19 political sociology
source of stress, 19 computational political sociology,
source of surprise, 19 15
source of uncertainty, 19 Popper, Steven W., 5
stochastic, 19 prediction, 6
tipping points, 19 preferences, 8
probabilistic causality, 16
observation, 6 programming languages
Olinick, Michael, 6 Java, 9
ontology, 18 object-oriented programming (OOP),
crisis ontology, 18 10
operations research, 15 Python, 10
optimization, 15 R, 9
origin of CSS, 9, 11 pure mathematics, 6
origin of term “computational”, 9 Python programming language, 10

34
queuing theory, 15 sentiment analysis, 12
Simon’s Theory of Artificial Systems,
R programming language, 9
10
Radcliffe-Brown, A. R., 9
Simon, Herbert A., 2, 5, 6, 9–11, 13,
Rand, William, 14
15
Rapoport, Anatol, 18
SIMSOC listserv, 14
remote sensing data, 2
skewed distributions, 17
reproducibility, 6
small groups, 9
resources
Snell, J. Laurie, 6
disaster response resources, 17
Snyder, Glenn H., 16
resource depletion, 16
social choice
Richardson, Lewis F., 9
computational social choice, 15
robots, 20
social knowledge, 2, 9
Rogers, J. Daniels, 21
social media, 2, 12
Rouly, Chris, 21
social psychology
Salmon, Wesley, 6 computational social psychology,
Saxe, John Godfrey, 2, 3 15
science illiteracy, 16 social science disciplines, 2, 8, 9
scientific explanation, 6–8, 21 anthropology, 8
Wegener’s Theory of Plate Tec- economics, 8
tonics and earthquakes, 7 interdisciplinary fields, 8
scientific laws, 6, 7 communication, 8
Duverger’s Law of Party Systems, human geography, 8
6 linguistics, 8
Laws of Social Complexity, 13 management science, 4, 8
Pareto’s Law on Income Inequal- social science history, 8
ity, 6 political science, 8
Richardson’s Law of War Magni- social psychology, 8
tudes, 6 sociology, 4, 8
Richter’s Power Law of Seismic social science models, 2, 5, 8
Magnitudes, 7 mathematical models, 3, 6, 11
theoretical explanation of, 6 social simulation toolkits
Zipf’s Rank-Size Law, 6 MASON Multi-Agent Simulator of
scope of CSS, 2–21 Networks or Neighborhoods,
Segal, Erwin M., 4 14
semantic dimensions of meaning, 12 Netlogo, 14
multidimensional semantic spaces, Repast, 14
12 Stella, 14

35
 Vensim, 14 11
Wolfram System Modeler, 14 systems of systems, 11
social simulations, computational, 2, systems theory, 16
3, 5, 7, 8, 10, 14 systems-of-systems (SOS), 10
social theories, 20
social theory, 2, 3, 5, 7, 8 technological ignorance, 16
Affect Control Theory, 12 terrorism, attacks, 16
Behavioral Decision Theory, 9 testable deductions, 6, 8
Conflict Theory, 9 The Blind Men and the Elephant, see
debunking unscientific theories, 8 also metaphors, 2
elegance, 7 Thompson, Gerald L., 6
Functionalist Theory, 9 Thompson, Maynard, 6
Lewin’s Field Theory, 9 time pressure, 17
Political Systems Theory, 9 tolerance, 3
proliferation of unsupported, faulty,Torrens, Paul, 5
or false legacy theories, 8 Toulmin, Stephen E., 7
Simon’s Theory of Artificial Sys- Trobia, Alberto, 5
tems, 10, 15 Troitzsch, Klaus, 10
theories of social complexity, 13 Troitzsch, Klaus G., 21
Theory of Groups, 9 uncertainty, 17
societal extinction, 17 increase as cause of stress, 17
socioeconomic status (SES), 4 uncertainty increase, 17
sociology Unified Modeling Language (UML),
computational sociology, 15 6, 11
south America, 13 class diagram, 6
space program, challenges of, 19 sequence diagram, 6
space-time, 16 state diagram, 6
spacefaring civilization, 19 use case diagram, 6
Squazzoni, Flaminio, 5, 21 unity of science, 15, 16
statistics, 9, 20 interdisciplinary CSS, 16
stochastic processes, 15 University of Bremen, Germany, 21
stress, 17 US National Science Foundation, 8,
substance abuse, 16 21
Suppes, Patrick, 6 Directorate of Social, Behavioral
supply chains, 4 and Economic Sciences (SBE),
synergies, 5 8
systems engineering, 4, 15
Systems Modeling Language (SysML), virtual worlds, 20

36
 von Neumann, John, 9
voting behavior, 9

Wegener, Alfred, 7
west Asia, 13
Wilenski, Uri, 14
win-win outcomes, 17
world system, 13
Wuhan, China, 16

37

View publication stats