6
KEY DEFINING CONCEPTS
Collective intelligence,
democracy, and governance
Stephen Boucher, Carina Antonia Hallin, Lex Paulson,
Nino Javakhishvili-Larsen, and David Leal Garcia
This handbook aims to synthesize key ideas from the emerging field of collective intelligence and place them more clearly in the context of public problem-solving. Groups, under
certain conditions, can solve problems and accomplish complex tasks in a way that surpasses
the abilities of any individual within them. How can we understand this phenomenon, and
what are the conditions necessary to bring it about?
The notion of collective intelligence is polysemic, informed by disciplines as diverse
as conflict mediation and computer science. This poses challenges as well as opportunities: Though our notions of cognition, collaboration, and citizen engagement are enriched
by studies from a diversity of fields, there is also a risk that these fields talk past one another
or that crucial links go unexplored.
Our intent in this handbook of collective intelligence for democracy and governance is
to add clarity and coherence to the field of collective intelligence as it matures into a scientific discipline in its own right. The increased salience of the field in recent decades may be
correlated with some of the key trends of our time, notably the revolution in digital technology that has enabled new modes of large-scale collaboration (McNutt et al., 2016), the
growing complexity and interdependence of public problems (Noveck, 2015), and a crisis of
trust in public institutions that has inspired a “wave” of deliberative experiments worldwide
(OECD, 2020).
The present moment has been described as a transition from the fourth industrial
revolution, characterized by the increasing automation and interconnection of economic
functions (Brynjolfsson and McAfee, 2014), to an upcoming fifth industrial revolution
centered on the need to make organizations and societies more intelligent (Noble, Mende,
Grewal et al., 2022). Authors have referred to this new technological paradigm as the “deep
minds era,” in which deep knowledge is harnessed, and collective consciousness created,
from both humans and machines (cf. Friedman, 2019). See further Part 7 of the Handbook
on Collective Intelligence, Technology and Collective Consciousness.
This fifth industrial revolution encompasses the notion of harmonious human–machine
collaborations with a specific focus on the well-being of multiple stakeholders (i.e., society,
companies, employees, and customers). For some, this “deep minds era” holds the promise
of a revolution in how we learn and think through the leveraging of human–machine
108
DOI: 10.4324/9781003215929-7
Collective intelligence, democracy, and governance
collaborations – a period of collective thinking and learning, harnessing unconscious and
tacit knowledge of humans facilitated by AI and deep learning, yielding better solutions to
society’s complex problems.
In policymaking, a shift is possible from the old paradigm of centralized decision-making
and a narrow understanding of expertise, to more decentralized forms of knowledge creation where humans and machines think together in symbiosis. A key question for policymakers
could become, “How can the combination of human and machine intelligence help us
go further than we’ve ever gone in serving the public”? (Friedman, 2019) In this context, we offer an overview of definitions and typologies of the field of collective intelligence as it
stands today. In doing so, we hope to clarify and explain some of the most important concepts at
the foundation of this handbook, noting some of the emerging trends in politics and governance
that could pave the way for a more inclusive and smarter paradigm.
This overview is structured into four sections. The first section presents definitions of
collective intelligence offered by some of the field’s pioneers. The second section takes a
deeper look into the conceptual frameworks around human intelligence and how these have
been applied to groups in different contexts, noting especially the key factors that have been
identified as reinforcing or hindering group performance. The third section examines key
concepts in democratic innovation that will be touchstones for many of the cases in this
handbook. The final section looks at the frontier of human and machine intelligence, a
burgeoning area of interest for researchers and practitioners alike.
We do not aim here to be encyclopedic in our presentation of these definitions, but rather
to provide a structured overview of notions that may be of greatest interest to the reader.
Seminal definitions of collective intelligence
Pre-1970
The notion of collective intelligence has important antecedents prior to the popularization of the term itself in the 1980s. Many of these concepts are explored in greater detail in
chapter 1 of this handbook.
Theoretical arguments for collective intelligence may originate with the ancient Greek
philosopher Protagoras, who according to Plato argued that politikē technē, or the expertise
to govern a city, was found not in exceptional individuals but in the community as a whole
(Kenney, 2008; Landemore, 2013). In the third book of his Politics, Aristotle makes more
detailed arguments for the governing capacity of large groups, who “though not excellent
as individuals, nevertheless can, when they have come together, be better than the few best
people, not individually but collectively, just as feasts to which many contribute are better
than feasts provided at one person’s expense” (3.11). Waldron (1995) more recently revived
interest in what he characterized as Aristotle’s “doctrine from the wisdom of the multitude.”
An important non-Western antecedent is found in the doctrines of the Chinese philosopher
Confucius, whose ethical ideal of ren has been translated as “co-humanity” (Puett and
Gross-Loh, 2016).
In the common era, political and philosophical touchstones for CI include Machiavelli’s
arguments for the superior wisdom of the masses when compared with single rulers
(Mansfield, 2001); Ibn Khaldun’s notion of ‘asabiyah or “group spirit” as the key driver of
political change (Irwin, 2018); Spinoza’s argument that aggregated judgments are more likely
to produce rational outcomes the larger the body of citizens (Tractatus Theologico-Politicus
16.16; see Landemore, 2013); J. S. Mill’s arguments on the superior power of representative
109
Boucher, Hallin, Paulson et al.
government to create a framework for diverse ideas to compete, thereby advancing the
knowledge available to all (Mill, 1865; Gray, 2013); and Dewey’s (1927) argument that on
any public issue, the cognitive capacities of the concerned population are essential to identifying and defining the problem before technical expertise can be applied.
Two antecedents to the field of CI deserve special mention. In the 18th century, the mathematician and statesman Nicolas de Condorcet was the first to offer a mathematical justification of the arguments for aggregative intelligence prefigured by Aristotle, Machiavelli, and
Spinoza. His “Jury Theorem” (1785) proposes that under certain conditions, the majority
opinion of a group whose individuals have only slightly better-than-average accuracy
approaches a probability of 1 as the size of the group increases. Condorcet has the distinction
of being a democratic theorist who attempted to put his ideas into practice, as co-author of
the Girondin proposal for the French constitution in 1792, which emphasized the role of
participatory local assemblies, a free press, and mass public education (Miller, 2018). The
second is Francis Galton, an English polymath whose 1907 article on a crowd’s accurate
prediction of the weight of an ox is perhaps the most cited early reference in the field (see the
discussion in chapter 1 of this handbook).1
It should be noted that despite their foundational importance to the concepts in this handbook, the term “collective intelligence” is not found in any abovementioned text. Mulgan
(2018) notes that the phrase “appears to have first been used in the nineteenth century by a
doctor, Robert Graves, referring to the advancing state of medical knowledge, and separately by a political philosopher, John Pumroy, to refer to popular sovereignty.” If true, it is
noteworthy that these two early references correspond to two principal forms of collective
intelligence discussed herein: cumulative collective intelligence (the body of shared knowledge that accumulates within a given society), and the synchronous collective intelligence
of a group engaged in a common task at a given time.
1970 to 2010
It was in the information age that “collective intelligence” as such became an object of study.
Describing the goal of a digital conferencing system, Hiltz and Turoff (1978) defined collective intelligence as “a collective decision capability [that is] at least as good as or better
than any single member of the group.” Douglas Engelbart, whose team at Stanford Research
Institute invented the mouse, hyperlink, video conferencing, and document co-drafting
(among much else), coined the term “collective IQ” to refer to the potential of computers
to enable large groups to solve increasingly complex tasks (Engelbart, 1995).
Kierstead (1974) may have been the first to use the term “collective intelligence” in the context of democratic theory. In his doctoral thesis, he used the term to characterize Dewey’s
theories on civic participation as a form of experiential education for citizens (often called
“learning by doing”). Greater and higher-quality civic participation, in the Deweyan view, is
not merely a form of consultation, but is rather a mechanism for society to develop its intelligence at the largest scale. Though theorists such as Habermas (1985), Mansbridge (1983),
and Cohen (1986) would develop a range of arguments for more deliberative and inclusive forms of democracy from the 1970s, until Landemore and Elster (2012) these theorists
largely did not refer to “collective intelligence” as such, nor engage closely with the field of
human– computer interaction (for discussion, see Spada, chapter 5). Nevertheless, arguments
developed in favor of participatory and deliberative democracy in this period – including the
benefits to society from diversifying its store of ideas and combining them in new ways –
remain important touchstones for the field of CI as it relates to democracy and governance.
110
Collective intelligence, democracy, and governance
Lévy’s book Collective Intelligence: Mankind’s Emerging World in Cyberspace (1997) theorized
collective intelligence principally in the context of the internet and its possibilities. His
definition of collective intelligence as “an intelligence distributed everywhere and constantly valorized,” whose goal is “the recognition and mutual enrichment of people,” carried
Engelbart’s optimism to near-utopian heights.
The text that introduced the idea of collective intelligence to a global audience was
Surowiecki’s The Wisdom of Crowds: Why the Many Are Smarter than the Few and How Collective
Wisdom Shapes Businesses, Economies, Societies and Nations (2004). He centers his book on “a
simple but powerful truth…: under the right circumstances, groups are remarkably intelligent, and are often smarter than the smartest people in them.” Opening his book with the
Galton study, and using illustrations from business, prediction markets, and deliberative
democracy, Surowiecki proposes a concise list of key drivers of group intelligence: independence of judgment, decentralized information, diversity of opinion, a fair and accurate
aggregation mechanism, and shared trust. While digital platforms feature in Surowiecki’s
account, he does not qualify collective intelligence, like Lévy, as a novel phenomenon of the
internet age. Collective intelligence, for Surowiecki, is deeply ingrained in human societies
that various forms of organization can hinder or unlock.
Sunstein (2006) in Infotopia: How Many Minds Produce Knowledge explores the human
potential to pool information and to use it for public good. He gives special attention to
the emergence of wikis, open-source software, and prediction markets, highlighting the
astonishing productivity and accuracy of these tools while also pointing to the risks of group
polarization and “echo chambers” that often emerge online. Howe (2006) coined the term
“crowdsourcing” in his article for Wired entitled “The rise of crowdsourcing”. According
to Howe, crowdsourcing is an act whereby an organization or institution takes a function
which was once performed by employees and outsources them to a broader network, generally in the form of an open call.
2010 to present
Landemore and Elster’s co-edited volume Collective Wisdom 2 (2012) was a watershed
moment that brought democratic theorists, cognitive scientists, and technologists into a
common conversation on the dynamics of group intelligence. Enriched by the pioneering
work of Hong and Page (2004) on cognitive diversity, the authors argue that CI should be
understood not merely as the intensification or amplification of individual intelligence, but
as “an emergent phenomenon that cannot be traced simply to individual minds, but rather to
the interaction between those minds and between them and their constructed environment.”
Landemore’s Democratic Reason: Politics, Collective Intelligence, and the Rule of the Many (2013)
develops these arguments in the context of democratic innovation and gives a genealogy
linking Protagoras, Aristotle, Condorcet, Dewey, and others to contemporary developments
in the field.
In their Handbook of Collective Intelligence (2015), Malone and Bernstein survey the research
frontiers of the emerging field of CI, drawing principally from the disciplines of computer
science, biology, economics, and psychology. They present a stripped-down definition of collective intelligence: “groups of individuals acting collectively in ways that seem intelligent.”
Though noting that this phenomenon has existed “for a very long time,” they posit that with
the rise of digital technologies, a “new kind of collective intelligence has emerged.”
In her book Smart Citizens, Smarter State: The Technologies of Expertise and the Future of
Governing, Noveck (2015) argues for a transformation in governance “that takes seriously the
111
Boucher, Hallin, Paulson et al.
capacity, intelligence, and expertise of all people and forges institutions that know how to
marshal and use that capital.” Drawing on her own expertise in the Obama Administration
creating open innovation challenges, she proposes “crowdlaw” – organized citizen input at
all stages of policymaking – as a way both to address society’s most complex problems and to
ease the “pandemic of distrust” in government.
The field of cultural evolution has added an important perspective on group intelligence: how knowledge can accumulate over many generations. In The Secret of Our
Success: How Culture Is Driving Human Evolution, Domesticating Our Species and Making Us
Smarter (2015), Henrich observes that
the striking technologies that characterize our species, from the kayaks and compound bows used by hunter-gatherers to the antibiotics and airplanes of the modern
world, emerge not from singular geniuses but from the flow and recombination of
ideas, practices, lucky errors and chance insights among interconnected minds and
across generations.
Just as the “collective brain” of a society can augment its intelligence through greater size,
interconnection of group members, and sociality (the motivation of any member to share
what she knows), Henrich argues, so too can it lose its intelligence if those conditions go
unmet.
Mulgan’s Big Mind: How Collective Intelligence Can Change Our World (2017) synthesizes
work from several domains to propose new conceptual frameworks for the field. He defines
collective intelligence as “the capacity of groups to make good decisions – to choose what
to do, and who to do it with – through a combination of human and machine capabilities.” The type of tasks taken on by that group – observing, prediction, generating ideas,
and so on – are called in this handbook its “core functions” (see Mulgan, chapter 4), and
“functional capabilities” in the earlier text. Those functions are supported in turn by collective intelligence “infrastructures” (tools, objects, common rules, and institutions) and
organizational models that can favor or hinder group integration and learning. Mulgan
concludes that solving 21st-century problems will require a new generation with skills in
“intelligence design,” creating “assemblies” of the above elements adapted to the needs of
a given problem.
Andersen and Hallin’s (2017) book on “Global strategic responsiveness: Exploiting
frontline information in the adaptive multinational enterprise” presents tools to help design
responsive and intelligent organizations. The book provides models to inform strategic
decisions through the crowdsourcing of frontline information. The subsequent “A network of brains” (Andersen, Hallin and Fredens, 2018) presents ideas on how organizations
can draw on distributed knowledge to make better decisions and more quickly adapt to a
changing environment.
In his Superminds (2018), Malone – founder of the MIT Center for Collective
Intelligence – suggests that human and artificial intelligence (AI) will be combined in
increasingly complex ways, yielding exponentially more powerful forms of cognition.
As part of Britain’s public innovation agency, the NESTA Centre for Collective
Intelligence Design (co-founded by Mulgan) supports the use of collective intelligence on
complex social problems. They currently define collective intelligence as something “created
when people work together, often with the help of technology, to mobilize a wider range
of information, ideas and insights to address a social challenge” (Peach et al., 2019). The
range of disciplines seeking to understand the mechanisms of collective intelligence has thus
112
Collective intelligence, democracy, and governance
become vastly richer, with important insights from computer science, information science,
political science, management, mathematics, social psychology and neuroscience.3
Foundational concepts of collective intelligence
The definition of human intelligence is continuously evolving. Definitions have ranged
from a mental quality that consists of abilities to learn from experience, to the ability to adapt
to new situations, understand and handle abstract concepts, and to the use of knowledge to
create new knowledge and to be able to manipulate one’s environment (Britannica, 2022).
Broadly speaking, human intelligence can be understood as a toolkit of intellectual capabilities marked by complex problem-solving behaviors and high levels of self-awareness (Tirri
and Nokelainen, 2011). Correlations have been posited between higher levels of intelligence
and better life outcomes (Brown and Wai, 2021).
What is individual intelligence (and how do we measure it)?
Over the past century, psychologists have made progress in defining and systematically measuring general intelligence. Spearman (1904) found that intelligence can be measured using
different kinds of IQ tests adapting a variety of test tasks in experiments. The “g factor,”
also referred to as the general intelligence, general mental ability, or general intelligence
factor, is a variable that summarizes positive correlations among different cognitive tasks,
reflecting the fact that an individual’s performance on one type of cognitive task tends to
correlate with performance on others. Spearman labeled this correlative trend a “positive
manifold” (Spearman, 1904).
Over the last century, scholars have emphasized different aspects of the intelligence construct. For example, in a 1921 symposium, the American psychologist Terman stressed the
ability to think abstractly, while his colleague Torndike emphasized learning and
the ability to give good responses to questions. Later in the century, a variety of
other tests were introduced to probe intelligence from different angles, starting notably
with the Torrance tests of creative thinking in the 1950s (Runco et al., 2010) to the more
recent Cognitive Reflection Test (CRT), which seeks to measure people’s ability to override
instinctive responses in order to arrive at the correct answer (Frederick, 2005).
Psychologists have generally agreed that human adaptation to the environment is the
key to understanding both what intelligence is and what it does (e.g. Greenspan, 1979).
Such adaptation may occur in a variety of settings: a citizen learns to access an online public
consultation and respond to questions of relevance to her community; a physician treating
a patient with unfamiliar symptoms learns about the underlying disease; or a member of a
creative crowd reworks an online painting to convey a more coherent impression. For the
most part, adaptation involves changing one’s own behaviors and strategies to cope more
effectively with the environment, but it can also mean changing the environment or finding
an entirely new one.
For the purposes of this handbook, we follow Mulgan (2018) in conceiving of intelligence “not just as a matter of extraordinary memory or processing speeds,” but rather as “our
ability to use our brains to know which path to take, who to trust, and what to do or not
do.” As Mulgan notes, the Latin roots of the word – inter, “between,” and legere, “to choose
or gather”– suggest the essence of the idea of intelligence as a “gathering together” of
information from our environment and experience, combining diverse cognitive capacities in a manner that enables us to make choices favorable to our survival.
113
Boucher, Hallin, Paulson et al.
Types of individual knowledge
Individual intelligence can be aided by explicit (codifiable) and tacit (unconscious)
knowledge (see the introduction to Part 3 for a review).
Explicit knowledge is defined as conscious knowledge that is independent of the individual and contains identifiable facts, information, procedures, and routines, which can be
codified, processed, stored, and easily exchanged. Explicit knowledge (facts, procedures, and
routines) can be articulated, codified, stored, and processed (Nonaka, 1991).
Tacit knowledge is defined as subconscious knowledge that is within the individual
and contains habits, experience, skills, emotions, intuition, sensing, and intuition, which
is difficult to codify and make explicit and communicate to others (e.g., how to drive a
car). The term “tacit knowledge” was introduced in 1958 by Michael E. Polanyi. He later
characterized tacit knowledge with the maxim “we can know more than we can tell,” referring to the difficulty in systematically expressing much of what we know, such as how to ride
a bicycle (Polanyi, 1966). Such tacit and unconscious processes generate intuitive thoughts
that are “the end product of an implicit learning experience” (Reber, 1989). In other words,
intuitive processing and judgments are the product of one’s tacit knowledge stock, which is
stored subconsciously, and intuition and tacit knowledge are thus interconnected in creating
judgments (Shirley and Langan-Fox, 1996).
Embodied knowledge is defined as the constant flow of senses, actions, and experiences
that encodes how the bodies should act without representation in a given situation. All
together our bodies acquire the knowledge that resides within the body or is gained through
the body (Nagatomo, 1992), when senses, emotions, experiences, and action flow within
an individual, grasping and embodying in the individual cognition, knowing how to act
without thinking, for example, using the motor skills necessary for the given circumstances,
such as moving fingers while typing, or pedaling while riding the bike (Tanaka, 2011). As
Tanaka (2011) describes, embodied knowledge is visible during the performance and has
no “verbal explanation,” as our bodies act, and act competently, prior to our conscious
awareness.
Types of organizational knowledge
Distributed knowledge is knowledge dispersed within a group, organization, network,
or community. Distributed knowledge may take the form of tacit, explicit, or embodied
knowledge shared among and across the members. According to Ågotnes and Wang (2017),
this is the segmented knowledge dispersed across a group. When combined, these segments
comprise the full shared knowledge of the group, just as the pieces of a puzzle together
create a full picture. Critically, individual segments may have only partial knowledge of the
others, limiting the potential for knowledge to be combined in ways that benefit the whole.
Distributed knowledge is typically “situated,” meaning that it is dependent on the contextual
understanding, culture, language, time, location, and other variables that make up our individual knowledge, subject to our own mental models (Haraway, 1988). Collective intelligence, therefore, may emerge from distributed knowledge when agents aggregate what they
know to solve a common challenge. Knowledge can also be distributed not only among
many minds, but different cognitive artifacts – such as language, books, tools, and maps –
capture information and help tackle complex tasks.
Collective intelligence emerges not only when distributed knowledge is brought together
to solve a problem, but also when a symbiotic relationship is created between distributed
114
Collective intelligence, democracy, and governance
and centralized knowledge. Andersen and Hallin (2017) argue that sustainable organizational performance is achieved when decision-makers coordinate processes anchored in
emerging opportunities from the frontline (e.g., citizens) with forward-looking activities
by decision-makers (e.g., policymakers). The fast and emergent processes performed by
actors at the frontline observe and respond to environmental stimuli that form distributed
knowledge and the slow processes initiated by decision-makers interpret events and reasons
about updated actions. Current insights from the fast response processes can be aggregated
systematically from the distributed knowledge and fed into the slow process of reasoning
for policymaking. When the fast and slow processes interact, they form a dynamic system
that adapts organizational activities to changing conditions and the problems that emerge
from them.
Collective intelligence
For the purposes of this book, we define collective intelligence as the capacity of groups
to outperform individuals in problem-solving, innovation, prediction, creativity,
and other cognitive tasks. This capacity, in turn, can be derived from the various kinds of
knowledge listed above: explicit, tacit, centralized, distributed, or embodied.
A recent empirical study of a collective intelligence “c factor” shows that a group’s
general ability to perform a wide range of tasks expands the areas of human intelligence
research applying similar methods and concepts to groups (Woolley et al., 2010). The
“c factor” they identify adopts a similar definition and methodology as the psychometric
approach of general individual intelligence “g factor.” These studies have shown that a
“c factor” can explain between-group differences in performance as well as other potential
causes such as task structure or group composition (Woolley, Aggarwal, and Malone, 2015).
The original 2010 study demonstrates that the smartest groups, controlling for individual
intelligence, were distinguished by three factors: by the average social sensitivity of group
members, the equality in distribution of conversational turn-taking, and the proportion of
females in the group (Woolley et al., 2010).
The wisdom of crowds, while often used synonymously with collective intelligence,
most typically describes a specific instance of synchronous CI (further explained in the next
section) in which the predictions or estimations of a crowd outperform its average (or, alternatively, its best) member. Such was the case in the famous Galton experiment with which
Surowiecki (2005) opens his work. Condorcet’s Jury Theorem (1785) is often cited as the
first mathematical demonstration of crowd wisdom; the three conditions he gave for the
effect are described in the section above. A more recent contribution to the understanding of
this effect is Hong and Page’s (2004) Diversity theorem, sometimes called “Diversity Beats
Ability.” It proposes that for any group, the squared error of a collective prediction equals
the squared error of the average member minus the diversity of predictions. The theorem
predicts the counterintuitive outcome that a crowd can be made collectively wiser by adding
members who are individually less accurate than the average as long as their contributions
increase the overall diversity of ideas in the pool. This perhaps unexpected benefit of cognitive diversity has since been demonstrated empirically (see Servan-Schreiber chapter 32 in
this handbook).
Do other species demonstrate collective intelligence? Opinions are divided. Some have
asserted that intelligence is a uniquely human quality, in that only our species can manipulate symbols and recombine ideas in a manner that is deliberate and self-aware (Pinker,
2010). Others point to a range of collective behaviors in terms of swarm intelligence
115
Boucher, Hallin, Paulson et al.
whose highly complex outcomes far surpass the ability of individual members. Such
behaviors include the capacity of honeybee colonies to select among alternative hive sites
(Seeley and Buhrman, 1999); army ants to construct bridges (Reid et al., 2015); and birds
to synchronize migration choices over great distances (Sliwa, 2021). Regardless of whether
these animal and swarm behaviors rise to the level of intelligence, they are impressive
demonstrations of the powerful capacities that emerge from the efforts of seemingly powerless individuals. Humans are indisputably unique in their ability to create cumulative culture, that is, for each generation to make deliberate improvements in the tools and processes
they inherit (Henrich, 2015). This special capacity, which Henrich links to physiological
and social changes that followed the domestication of fire by early hominids, has turned
culture into a “second track” of our evolution alongside the genetic selection that produced
the behaviors described above.
The term “collective wisdom” has been used somewhat loosely in the literature to date.
It can suggest a depth and durability of intelligence that goes beyond the capacities of a
single group on a single problem (to which the phrase “wisdom of crowds” is commonly
applied). In their co-edited volume on collective wisdom, Landemore and Elster explain
that they privilege the term “wisdom” over “intelligence” because the former is “a more
encompassing notion than the apparently more technical concept of intelligence.” Wisdom
also “evokes a larger temporal horizon,” suggesting “the intelligence of a collective extending
not just through space (including many people) but through time as well (including many
generations)...” Finally, Landemore and Elster suggest that the term wisdom goes beyond
matters of formal correctness or verifiable fact to include moral or political questions on
which there may not be a universally satisfactory answer.
Taking a holistic approach to highly complex moral problems, based on a wealth
of diverse experience, are elements that have characterized wisdom in many human
cultures (Grossman, 2017; see the discussion in Baum case, chapter 27 in this handbook).
Alternatively, Sternberg (2000) suggests that wisdom can be understood as a subset of
intelligence related to solving practical problems, as opposed to theoretical ones. For the
purposes of this handbook, then, we understand intelligence to refer to the ability to solve a
given problem or accomplish a given task, drawing on the accumulated culture and diverse
abilities mentioned above. In complementary fashion, wisdom can be understood as the
sedimentation of learnings from many instances of intelligence over time, its successes and
failures, our own and those of our community and ancestors. For our purposes, collective
intelligence thus describes the capacity to solve a public problem currently facing us, and
collective wisdom the treasury of past intelligence, especially of a moral or ethical kind,
that we may draw upon.
How collective intelligence works
A potentially important distinction arises regarding the different time scales on which collective intelligence can operate.
Tomasello (2019) proposes that the knowledge-sharing functions of human culture can
be seen along two dimensions: a “coordinative dimension” of culture, including the
range of cooperative structures and conventions we use to collaborate synchronously (i.e.,
focusing on the same problem or task at the same time); and what he calls a “transmittive
dimension” of culture, in which members of a group pass along skills and knowledge to
its junior members, potentially creating a “ratchet effect” where each generation improves
on the technologies and institutions of all previous ones. This “ratchet effect” is a main
116
Collective intelligence, democracy, and governance
feature of Muthukrishna and Henrich’s (2016) arguments concerning the “collective brain”
of a society that each member draws from and to which each may contribute (see also Vale
et al., 2012).
Within this “coordinative dimension” an additional nuance can be applied: Members of a
group can collaborate on the same problem or task asynchronously, that is, make a sequence
of contributions that are combined toward a single main goal. Mapping the human genome,
for example, was a highly complex task requiring the intelligence of a large community
to operate, some performing actions synchronously (such as a laboratory team conducting
a single study), and others asynchronously (several teams adding pieces of information
regarding a certain gene). And each of these teams benefit from the transmission of a huge
store of accumulated scientific knowledge from previous generations, transmitted via peerreviewed journals, embodied in laboratory equipment, and so on.
We can see, therefore, how collective intelligence operates in three modes:
1. Synchronous collective intelligence, where members of a group interact to solve a
problem or accomplish a task, or where members make individual contributions at the
same time4 and those contributions are aggregated (e.g., a prediction contest).
2. Asynchronous collective intelligence, where members of a group make a sequence of
contributions for a given problem or task (e.g., 700 attendees making guesses about the
weight of an ox over the course of a day).
3. Cumulative collective intelligence, where members of a group create and transmit
knowledge across time (e.g., how to build a compound bow, treat an illness, or manage
a project spreadsheet).
In the context of democracy and governance, as in all forms of human collaboration,
these three modes of collective intelligence work in combination. In preparing their
recommendations, for instance, the 99 members of the Irish Constitutional Assembly of
2012–2014 thought together synchronously (during meetings in which members proposed
and recombined their ideas); they thought together asynchronously (the input for their work
were eight agenda items established by the Oireachtas Éireann, Ireland’s bicameral parliament), and benefitted in their work from the accumulated expertise of civil servants,
advocates, and academics. Among the results was a national referendum legalizing same-sex
marriage (a synchronous moment of decision), which many hailed as a breakthrough after
decades of political stalemate (Elkink et al., 2017).
As described above, synchronous collective intelligence can be produced whether or not
the members of a group know one another or interact in any way. Thus, another potentially
useful distinction arises.
On the one hand are aggregative methods of collective intelligence, in which individuals contribute independently to a common task and the sum total of these contributions
produces a level of accuracy or performance that surpasses the average (or even the best) individual. Such was the case in the Galton experiment, in which 787 attendees of the county
fair produced a collective estimate of the ox’s weight far more accurate than the average individual guess, without knowing or interacting with one another. Since the scale and diversity
of the crowd are two principal factors in producing this “wisdom of crowds,” digital technologies like prediction markets often draw upon aggregative methods to produce intelligent
outcomes.5
On the other hand are deliberative (or synergistic) methods of collective intelligence
whose effectiveness depends on the quality of interactions among the members of a group.
117
Boucher, Hallin, Paulson et al.
The Irish Constitutional Assembly of 2012–2014 presents an exemplary case: Of the 99
members, 66 were selected randomly from the general population, balanced by age, gender,
and region, and 33 were members of parliament selected across various parties. As Farrell and
Suiter (2019) explain, citizens’ assemblies have been able to produce groundbreaking policy
outcomes because citizens are welcomed into a framework within which they can share
knowledge, explore differing perspectives, and bring tacit knowledge to the surface in a collaborative and non-judgmental way. On questions of deep moral complexity, from marriage
to reproductive rights, it is the high quality of interactions among assembly members that
enables new insights to emerge, relationships of trust to form, and proposals for public good
to be defined. We may contrast this outcome with another national exercise on a similarly
divisive issue, that of Britain’s referendum on membership in the European Union in 2016.
In this superficially “democratic” exercise, a lack of informed deliberation among citizens of
differing views produced an outcome that has not only deepened the nation’s divides but has
produced at least as many policy challenges as it has resolved (Clarke et al., 2017).
Drivers of collective intelligence
The past two decades have seen a remarkable series of advances in our understanding of
how group intelligence is produced. Several conditions have been identified that can make a
group smarter than the sum of its parts, or conversely, that can hinder collective intelligence
from taking shape.
Based on our analysis of the different contributing fields, we propose that the factors
driving collective intelligence can be usefully divided into three categories, each of which
answer one of the following questions: (1) “Who should be in the group?,” (2) “How should
the group be organized?,” and (3) “How many should be in the group?”
Figure 6.1 Principal drivers of collective intelligence.
Source: Authors’ synthesis.
118
Collective intelligence, democracy, and governance
(1) The “who”
(a) Members should be at least minimally competent for the task at hand.
Condorcet (1785) theorizes that the judgment of the average member of his “wise
jury”6 will need to be at least slightly better than random. Similarly, Hong and Page
(2004) affirm that collective performance will grow either as the crowd becomes
more diverse or as the competence of its individual members improves. Following
Dewey (1927), Noveck (2015) argues that “competence” in public affairs is often
misconstrued as requiring special credentials or technical expertise, whereas even
citizens who are not themselves “expert shoemakers,” to borrow Dewey’s metaphor,
can nevertheless be good judges of “whether the shoe fits and how it fits.”
(b) Members should have access to decentralized sources of information. Hayek
(1945) argued that markets are superior to centralized planners in producing desired
social outcomes because they simply and rapidly synthesize a vast quantity of
decentralized information held by local actors in a single form: the market price. As
Surowiecki (2005) observes, high-performing organizations have tended to be good
aggregators of the distributed knowledge and local information of their members
(see also Argyris, 1995). Conversely, if members of a diverse and competent group
are all using the same flawed information source, collective intelligence will suffer.
(c) The group should be cognitively diverse. Implicit in Aristotle’s metaphor of the
“feast to which many contribute” (Politics, 3.11) is the argument that certain types of
tasks are better accomplished by a group that combines diverse preferences and talents.
If the previous category considers the variety of information sources available to a
group, cognitive diversity refers to the variety of ways a group can interpret and use
that information. Page (2017) proposes that each member of a group brings their “cognitive repertoire” to group tasks, a unique package of mental models, representations,
and heuristics that can be applied to a given type of problem. A group becomes potentially more intelligent in proportion to the variety of the cognitive tools at its disposal.
Importantly, this kind of diversity is linked to, but distinct from, other indicators of
diversity such as age, gender, and ethnicity. Page argues that a group that is diverse
according to these measures may indeed be cognitively diverse, given the range of life
experiences and cognitive skills that can arise from those differences. So too, however,
can a group that “looks” diverse actually think in similar ways – for example, if they
share a similar educational or professional background.
(d) Members should be motivated to contribute what they know. Even the most
diverse, competent, and well-informed group will not be intelligent if individual
members do not share what they know. Motivation for a given activity can be either
intrinsic, that is, meaningful based on one’s values or identity, or extrinsic, derived
from external rewards or consequences (Ryan and Deci, 2000). A feeling of shared
identity and responsibility can be a powerful motivator (Van Bavel and Packer, 2021).
Alternatively, aggregative methods like prediction contests often rely on monetary or
prestige-based incentives to motivate participants (see Servan-Schreiber case, chapter 27
in this handbook). Gamification methods are increasingly used by governments to
stimulate citizen engagement based on the enjoyment linked to solving puzzles,
amassing points, and/or engaging in friendly competition (Sgueo, 2018).
(2) The “how”
(a) Members must be able to communicate efficiently and at a reasonable cost.
As Raymond (1999) posits in his study of open-source communities, the rise of
peer-to-peer platforms dramatically lowered the costs of collaboration, making it
119
Boucher, Hallin, Paulson et al.
possible for entire operating systems to be created by large groups of volunteers.
Conversely, processes that are inaccessible to certain populations or require too
much time from each member are unlikely to gain a critical mass (Noveck 2015).
An effective CI method makes it easy to access the common task; critical factors may
include choosing the right language, time, and physical spaces, as well as planning
for different levels of digital literacy and connectivity.
(b) Members should be able to form judgments independently and share them
without fear. One of Condorcet’s (1785) three conditions for his “wise jury” was
that the judgment of each member should be free from influence by the others. In
aggregative methods such as the Galton example, or in prediction markets today,
the existence of a prize can incentivize participants to bring their best thinking to
bear, regardless of if their ideas diverge from others. Nevertheless, as Kahneman
(2011) and others have shown, anchoring effects, information cascades and a range
of unconscious biases can diminish our independence of judgment, putting collective intelligence at risk. In other contexts where the impulse to conform may
be even stronger, a climate of psychological safety has been shown to be critical in
bringing good ideas to the fore, especially when these challenge the conventional
view (Edmonson, 1999).
(c) There should be a fair and effective method to aggregate contributions. Such
aggregation methods can be low-tech, such as the basket and pieces of paper in the
Galton case (Surowiecki, 2004), or high-tech, such as the software for evaluating
team contributions to the Netflix prize (Page, 2017), or the Pol.is platform used to
map policy preferences in Taiwan (see Taiwan case, chapter 7 in this handbook).
Deliberative methods of CI typically require substantial facilitation to assure the
quality of interaction within the group, such as in Deliberative Polls (Fishkin et al.,
chapter 13 in this handbook) or citizens’ assemblies (Lironi, chapter 8 in this handbook; Farrell and Suiter 2019). Regardless of the degree of facilitation or technology
required, participants’ contributions should be reflected accurately, and similarly
situated participants should be given equal opportunity to contribute.
(d) Collective thinking is often enhanced by specially designed spaces and artifacts.
From neighborhood maps to whiteboards to circular amphitheaters, well-designed
spaces can enhance group interaction and well-chosen objects can usefully focus group
attention (see Bergman, chapter 35, in this handbook). Paul (2021) gathers a range of
studies of “embodied cognition” and “situated cognition,” that is, how thinking
may be conditioned and potentially enhanced by the tools and spaces available to the
thinker. Many of these recent studies have drawn on Clark and Chalmers’ (1998)
arguments for the “extended mind”, that is, an understanding of our cognitive
processes not strictly limited to the neurochemical activities of our brains; according
to this view, human thinking can be understood as incorporating our bodies, tools,
spaces, and other minds. This is an emerging and contested area of cognitive science
(see Colombo et al., 2019), with much to offer to theorists and practitioners of CI.
(3) The “how many”
(a) Larger groups can be more intelligent…but not always. In classical
demonstrations of crowd wisdom, if all other variables are held constant, a larger
group will outperform a smaller one. Servan-Schreiber (2018) notes the statistical
outcome of higher group intelligence but with diminishing returns: for example,
each new member of a Condorcetian jury improves collective performance, but in
smaller and smaller increments. For ideation or problem-solving tasks, on the other
120
Collective intelligence, democracy, and governance
hand, scaling up could make it more likely that a specific problem is spotted (the
“many eyes” effect, see Raymond 1999), or that a “needle-in-a-haystack” solution
is identified. This capacity of larger groups (n=20 to n=200) was demonstrated by
Mercier and Claidière (2022) regarding the quality of answers to factual problems,
where members were able to discuss with one another. In many cases, digital technologies have radically lowered the costs of adding participants, especially for aggregative CI methods; this has enabled prediction markets such as Hypermind7 or
citizen-science platforms like Zooniverse8 to aggregate large numbers of estimations
or contributions at very low cost. Large-scale participation can itself be an important
goal of a collective intelligence process, such as when the target population has been
marginalized or excluded from such participation in the past (see Du Bois, 1920; Gret
and Sintomer, 2005; Tawa-Lama, chapter 20 in this handbook); or when the goal is to
raise awareness of a larger set of issues, as may be the case for citizen science initiatives
(see Kragh, chapter 38 in this handbook). Deliberative methods, unlike aggregative
ones, have proven notoriously difficult to scale (see Hartz-Karp and Sullivan, 2014).
Maintaining a high-quality deliberative experience has proven challenging in online
settings, since these settings do not easily capture the full range of gesture and tone
that are critical to human communication (Tannen, 2005). Three of the cases in this
handbook show promising approaches to maintaining a high-quality deliberative
experience using scalable online tools (see Fishkin, chapter 13; Spada, chapter 5;
Liu, chapter 7). Cognitive biases such as groupthink, information cascades, and
conformity bias may also be harder to mitigate at larger scales, which can lead to
disastrous results (Sunstein, 2007). Strategies to preserve independence of judgment
can include facilitation techniques such as asking participants in a workshop to write
down their opinions before group deliberation begins. Edmonson (2019) proposes a
series of strategies to favor psychological safety in group settings where the predominance of a “majority view” risks stifling constructive dissent.
Finally, some processes, such as visioning processes in a small community (Baum, chapter 27
in this handbook) or citizens’ assemblies, do not require large-scale participation for group
intelligence to emerge. In such cases, resources are more appropriately invested in maximizing interaction quality among a smaller number of participants. Where appropriate,
therefore, scaling up participation can be an important lever for augmenting collective intelligence; nevertheless, any scaling strategy should account for these potential risks and costs.
Core functions of collective intelligence
One way of understanding the capacities that underlie human intelligence is to answer the
question: What did these capacities enable us to do? How did their development allow certain individuals and groups to outcompete their peers and transmit these capacities, genetically and culturally, to their descendants?
Feldman Barrett (2021) argues that from an evolutionary perspective, “brains are not
made for thinking.” Rather, the first animal brains evolved to coordinate allostasis, that is,
managing the balance of energy needed to keep a body alive and functional. As the perceptive and motor abilities of early animals developed hundreds of millions of years ago,
these coordinative tasks expanded to include analysis and decision-making – “Is that a
threat?” “Is this food?”9 – and to execute the proper motor response. More complex forms of
thinking such as memory and prediction – “Is this likely to be food? Was it good to eat last
121
Boucher, Hallin, Paulson et al.
time?” – were selected for because these modes of thinking helped reduce the depth of analysis needed in everyday situations, conserving energy and thus better managing allostasis.
Tomasello (2019) picks up the evolutionary story to explain how and why humans
developed unique capacities for cognition in a highly competitive environment. Based on a
wealth of experimental data comparing responses of chimpanzees and human children on
a range of collaborative tasks, he proposes that it was through the development of shared
intentionality – the ability to conceptualize a common task and play complementary roles
in achieving it – that our unique form of collective intelligence emerged. Unlike other
social mammals, who live together but collaborate only when direct benefits are possible for
each individual, humans developed “ultrasociality,” including the ability to sacrifice individual interests for the sake of group success and transmit bodies of cultural knowledge to
the next generation (Haidt, 2012). Along the same lines, the works of Haidt (2012), Boehm
(2009), and Henrich (2015) trace how the core functions of intelligence of our early animal
ancestors – observing the environment, predicting the probable outcomes of different courses
of action, making and executing decisions, and learning from them – became functions that
humans took on collectively, causing a re-arrangement of human societies that made us look
very different from our primate cousins.
In this sense, our complex language, social institutions, abstract concepts, and moral
values are all adaptations that have allowed this evolutionary “bet on collaboration” to pay
off. Whether the rise of Homo sapiens has been a good thing for our planet is a separate matter;
nevertheless, our spectacular success from an evolutionary perspective cannot be denied, and
collective intelligence lies at its core.
Building off these advances in the cognitive and evolutionary sciences, Mulgan (2018)
proposes the following “core functions” of collective intelligence10:
1. Observing or gathering data on a given problem or need.
2. Modeling and predicting the likelihood of risks and the probability of different
scenarios coming to pass.
3. Generating options for action, based on past experience or creating new approaches
that have not yet been tried.
4. Filtering better potential courses of action from worse ones, deliberating on the relative virtues and risks of the better options, and deciding which option to adopt.
5. Implementing the decision by mobilizing the necessary resources and taking action.
6. Learning from the results of the actions taken by making sense of the resulting data,
remembering the lessons learned, and synthesizing past lessons into wisdom that can
be transmitted to others.
To put this in the context of previous sections, if the drivers of collective intelligence are the
answers to the question, “What makes a group smart?”, these core functions can be understood as answering the question, “What task will that group accomplish?” The answers to the
latter question will, in turn, guide the answers to the former one: Who should be in the group;
how should it be organized; and how many individuals will be needed for the task at hand.
Impediments to collective intelligence
Not all crowds are wise. The seminal studies of Mackay (1841), Le Bon (1905), Janis (1982),
and Sunstein (2006) have shed light on some of the factors that can hinder group intelligence
or, at worst, transform otherwise intelligent people into an unthinking mob.
122
Collective intelligence, democracy, and governance
In this and the next sections we will analyze impediments to CI that are intrinsic to
group behavior and groups’ information processing. We should keep in mind that there are
many other factors beyond the scope of this chapter that might hinder the implementation
of a CI process. Examples include the role of political competition, corruption, and capture
often discussed in the field of community-driven development (Mansuri and Rao, 2012).
Other examples can be found in studies of public policy, economics, and political science
analyzing the difficulty of implementing reforms in public administrations (Cairney, Akkila,
and Wood, 2019; Acemoglou et al., 2008; Rodrik, 2018; Mayka, 2019). Lastly, within the
field of organizational management lies an entire subfield on managing change, a body of
research that gives important insights for those interested in reforming or reinventing government (see, for instance, Stouten, Rousseau, and De Cremer, 2018). Additional details on
the barriers to effective implementation of collective intelligence processes can be found
in the chapter on measuring impacts (see chapter 5) and in chapter 3 in relation to political
obstacles to CI.
While genetic and cultural evolution has bestowed us with powerful cognitive tools,
so too has it resulted in mental “shortcuts” which conserve precious energy but can also
produce bad reasoning. Such cognitive biases and heuristics include the use of stereotypes, confirmation bias, and motivated reasoning, among many others (Kahneman, 2011).
Enhancing collective intelligence often requires becoming aware of, and taking steps, to
mitigate the effects of such biases on collective thinking.
Several case studies in this handbook illustrate the difficulties of bringing together all the
principles above to foster collective intelligence.
Groupthink and polarization might become manifest in the cognitive mode of action.
This is because humans are generally homophiles, which means that people prefer to link
up with others who are similar to them (i.e., in terms of race, nationality, ideology, gender,
age, etc.). This human trait can manifest itself most visibly when many people are involved.
Groupthink and polarization are especially common in large groups. “The term [groupthink] refers to a deterioration in mental efficiency, reality testing and moral judgements
as a result of group pressures” ( Janis, 1971). Groupthink is the result “when the members’
strivings for unanimity override their motivation to realistically appraise alternative courses
of action” ( Janis, 2008). Groupthink is especially prevalent when group cohesiveness is high,
that is, when members share a “we-feeling” of solidarity and desire to maintain relationships
and status within the group. In the course of collaborative work, therefore, it is key to create
conditions where individuals of large groups maintain independence of judgment and are
protected from these unhelpful forms of peer influence.
Sunstein (2006) presents, among others, the key notions of echo chambers (in which
beliefs are reinforced or amplified inside an insulated circle of participants), information
cascades (when people form judgments based solely on information provided by others
without verifying or questioning it) and groupthink. As a remedy to such “jumping off a
cliff in unison,” Sunstein (2005) insists on the importance of nurturing a culture of healthy
dissent.
Gammelgaard and Hallin (2018) show that barriers to effective crowdsourcing processes
can include the following:
Ambiguity: Lack of common language and shared understanding leading to sparse common
ground.
Multiplicity: A diverse set of practices and approaches producing a lack of coordination and
consistency.
123
Boucher, Hallin, Paulson et al.
Complexity: Poor insight into one’s own organization tied to a lack of innovation management and governance processes.
Inconsistency: Lack of combined efforts and priorities leading to imprecise objectives and
poor implementation.
Uncertainty: When crowdsourcing is associated with conditions of risk.
Problem complexity
A core argument for the importance of collective intelligence in governance is that increasingly
complex public problems require diverse sources of information and expertise to be combined
and mobilized. In fact, the world is said to become increasingly volatile, uncertain, complex,
and ambiguous (VUCA) (Mack et al., 2016). As the Covid-19 pandemic illustrates, globally
disruptive events unfold ever faster, the outcomes of different types of events are unclear, the
interdependence between actors is greater, and, even when we have information, it is difficult
to interpret. A conventional riposte is that it is the very complexity of problems like public
health or fiscal policy that disqualifies the average citizen from making a useful contribution.
Rather, public decisions of greatest import should be left to technocrats or elected officials of
outstanding abilities (presuming such to exist). Others identify precisely in this complexity
the reasons why more inclusive forms of democracy may yield better decisions. Babbitt (2012)
exemplifies this view, arguing that “properly understood, democracy can address many of the
problems arising from governing a world increasing in complexity.”
Different ways of characterizing the complexity of public problems include:
Dynamic complexity, meaning a systematic distance or delay between cause and
effect in space or time. The dynamic complexities of climate change are a clear
example, as there is a distance and delay between the decisions of climate solutions
and the effect. According to Scharmer and Kaeufer (2010), dynamic complexity
often goes along with social complexity.
Social complexity is the result of diverse interests, preferences, and mental models
among the stakeholders concerned by a given problem. The greater the social complexity, the more important is the multi-stakeholder approach to real problem-solving.
The cooperation of all the relevant stakeholders is vital to include the interests of all.
Emerging complexity “is characterized by disruptive change”, according to
Scharmer and Kaeufer (2010). It is characterized by the fact that it does not occur
until a certain unpredicted event happens, and conditions for solving the problem
are unknown, often the problem changes its form and makes it difficult to define
both the problem statement and key stakeholders. “The greater the emerging complexity, the less we can rely on past experiences.” Furthermore, as we unfold new
solutions, the problem and the diagnosis keep evolving, therefore, such problems
require more cognitive and coordinative CI actions.
Wicked problems, as coined by Rittel and Webber (1974), are those for which each
attempt to create a solution changes the understanding of the problem. As is the case
with many social and political issues, wicked problems may have several competing
definitions; their sub-causes may be deeply interconnected, meaning that any intervention will ramify through the system in unexpected ways; their manifestations
may change rapidly and continuously; and stakeholders may disagree on what a
124
Collective intelligence, democracy, and governance
good solution looks like. Because of these characteristics, wicked problems like
poverty or climate change are unlikely to ever be solved “once and for all.” Rather
than applying a linear model of problem-solving, therefore, wicked problems
demand methods of continuous learning and experimentation (see Noveck 2021).
Organizational learning requires groups to be able to reflect not only on the new
data arising from the field, but also on their own capacities to model and interpret
this data, “learning how to learn.” This approach of “learning loops” has been
developed and refined to help organizations and communities address problems of
this complexity.
(Argyris, 1995; Edmondson, 1998; Mulgan, 2018)
Surowiecki (2005) proposes that collective intelligence is especially effective in resolving
three types of problems:
Cognitive - These are problems that concern matters of local knowledge and
insights, where there is a right and a wrong answer, for instance how will infectious
diseases evolve (as in Servan-Schreiber and Larmanou, Handbook). Large crowds
can be surprisingly smarter at producing the right information than smaller groups
of individuals.
Coordinative - This is when a group of people is better at coordinating behaviors
than individuals. For example, what matters is not where people work in a hybrid
work environment, but who performs the tasks and how the tasks are coordinated
by the team in order to finish tasks (Riedl, Kim, Gupta, Malone, Woolley, 2021).
Cooperative / collaborative - This is when a group wishes to work together
to find and assess solutions. Whereas coordination can happen organically - like
pedestrians naturally self-regulating on the sidewalk - and the resolution of cognitive challenges will be served by actors sharing their knowledge independently,
collaboration requires the active interaction of the group members.
Policymakers may be facing all types of problems, therefore requiring a diversity of CI skills
and approaches throughout the policy cycle. However, the cooperative / collaborative type
tends to be particularly frequent, imposing on policymakers to pay particular attention to the
conditions for quality deliberation set out by Fishkin.
Design principles and frameworks
In 2010, Malone, Laubacher, and Dellarocos published a paper entitled “The Collective
Intelligence Genome” in the MIT Sloan Management Review. The article gathered nearly
250 examples of web-enabled collective intelligence, such as Threadless, Wikipedia, and
Innocentive. After examining the examples in depth, the authors presented the following
framework for designers of collectively intelligent system:
1. What is being done by the crowd? That is a question about the purpose of the task.
The many organizational tasks encountered in collective intelligence systems can be
boiled into two basic genes. One gene is aiming at creating something new, such as a
piece of software or a T-shirt design. The other gene is related to how the actors evaluate
and select alternatives, such as deciding on whether a new module should be included
125
Boucher, Hallin, Paulson et al.
in the next release of Linux open-source software, which T-shirt to manufacture, or
deciding which Wikipedia article to delete.
2. Who is doing it? This question relates to who undertakes the task on the platform.
Here are two sub-questions to be addressed: Is it the crowd who decides or is it a
hierarchy? If it is the crowd, the task is undertaken by a member of a large group who
has chosen to do so. However, if the task is performed by the hierarchy, it is typically identical to a traditional hierarchical organization with few decision-makers (e.g.,
policymakers) on the top of the hierarchy and answered when they assign a crowd to
perform a certain task for policymaking.
3. Why are they doing it? This question concerns what motivates the crowd to take
part in the activity? And what incentives are at work for the crowd? Examples of
high-level motivations are money and the promise of financial gains by fulfilling
the task. It could also be the feeling of love by being motivated from their intrinsic
enjoyment of an activity, love for the community and because they contribute to
something social or to a good cause larger than themselves. A third motivator is glory
in the sense that one becomes recognized for one’s performance within a crowd or
community.
4. How is the task being done? In traditional organizations, the how question is typically responded to by describing the organizational structures and processes. The
question concerns insight into four components to be evaluated related to the how genes
for crowds:
(a) Collection occurs when the crowd performs tasks independently from each other,
such as creating a video or suggesting an idea on a crowdsourcing platform happens
independently among the crowd members. It could also be in the form of an idea
contest among the crowd members.
(b) Collaboration takes place when members of a crowd work together to create
something, where the individuals are dependent on each other to reach an outcome,
such as when participants create a new Wikipedia page together and build on each
other’s contributions.
(c) Group decision occurs when inputs from members of the crowd are assembled to
make a decision that concerns the group as a whole. It could be a deliberation of
a political issue by a crowd on a platform. Other variations of group decisions are
voting, consensus, averaging, and prediction markets, examples of which will be presented
throughout this handbook.
Mulgan (2018) also suggests the following five key principles for those seeking to orchestrate
collective intelligence:
1. Autonomous commons – Create shared resources open to all group members that
capture past learnings and allow for information, ideas, and arguments to flow freely and
be organized for later use.
2. Balanced use of core functions – Identify which core function or functions are most
critical to the task at hand and choose tools and processes most likely to support those
functions.
3. Focus and right granularity – Adopt tools and facilitation methods that help the
group focus on the most important elements of the task, avoid potential distractions and
ask no more from group members than what is needed.
126
Collective intelligence, democracy, and governance
4. Reflexivity and learning – Keep in mind that knowledge production is a dynamic
flow (“knowledge needs knowledge about knowledge”) and that refining a thought process may require learning and sense-making in iterative loops.
5. Integrate for action – Combine tools, methods, participants, and commons in the
manner most suited for the task at hand, being ready to abandon past methods and adopt
new ones as needed for the circumstances at hand.
These frameworks are not exhaustive, as their proponents admit. Rather, they are intended
to serve as potential sources of inspiration for designers of collectively intelligent systems, to
be augmented by new and emerging frameworks such as those represented in this handbook.
Innovations in democracy and governance
In a growing number of political systems, there is an emerging awareness of the benefits of
large-scale participation in decision-making processes, the allocation of public resources, and
the resolution of disputes. These activities can be organized such that they are performed by
an individual, a designated subset of the community, or the whole community. In developing
governance innovations, different democratic principles can also be applied to any scale of
community and decision-making process for organizations, according to Malone (2004).
He distinguishes between the following forms of power distribution in the decision-making
process:
Democracy: A system of government, organization, or any type of community,
where all eligible members of the respective society contribute, typically through
elected representatives with the majoritarian principle.
Loose hierarchies: Delegation of the decision-making power to the subordinates
with the bottom-up process, where the managers and leaders approve the final
decisions.
Free market principle: Internal trade of the ideas and decisions in the large-scale
involvement.
Today, democracy is often equated in people’s minds with the practice of elected officials
making public decisions, executed by unelected bureaucracies, in which the mass of citizens
has limited power beyond voting periodically – “uploading one bit of data every four years,”
as Taiwan’s Digital Minister Audrey Tang puts it – and voicing their opinions, which have
no binding force in the public debate, with the excuse that large-scale participation justifies
a limited form of representation and delegation of power. CI designs can be applied to many
forms of power distribution and use the frameworks laid out by political theorist Robert
Dahl (2020) to strengthen democracy’s five essential principles:
Effective participation: In an ideal democracy, before a policy is adopted by the
association of citizens, all its members must have equal and effective opportunities
for making their views known to the other members as to what the policy should be.
Equality in voting: When the moment arrives at which the decision about policy
will finally be made, every member must have an equal and effective opportunity
to vote, and all votes must be counted as equal.
127
Boucher, Hallin, Paulson et al.
Gaining enlightened understanding: Within reasonable limits as to time, each
member must have equal and effective opportunities for learning about the relevant
alternative policies and their likely consequences.
Exercising final control over the agenda: The members must have the exclusive opportunity to decide how, and if they choose, what matters are to be placed
on the agenda.
Including all adults (a criterion largely unapplied in the case of women before
the 20th century).
Many stories in this handbook demonstrate efforts to address each of these pillars of democracy, focusing primarily on effective participation and gaining enlightened understanding;
examples range from the jan sunwai consultations and the Mobile Vaani phone voice
response device in India to the crowdsourcing platform in the Danish town of Slagelse.
Other dimensions in fact invite further developments. Fishkin (2018) notes in particular how
efforts to involve citizens in policymaking have so far been limited when it comes to binding
impact on the policy agenda.
With today’s historic model of representative democracy being challenged, this handbook
explores a variety of democratic innovations. Although they are attracting significant academic interest, there is considerable debate as to what can be qualified as a democratic innovation. Elstub and Escobar (2017) define democratic innovations as “processes or institutions
developed to reimagine and deepen the role of citizens in governance processes by increasing
opportunities for participation, deliberation and influence.” We will understand them as new
approaches to decision-making being explored around the world and designed to increase,
diversify, and deepen opportunities for citizen participation in governance, policy, and
public administration processes.
In understanding how different democratic innovations and CI approaches can help
strengthen democracy, it is useful to consider how they might play out at different phases of
the policy cycle. Mulgan (2018) usefully distinguishes between the following six stages of
the democratic process:
• Problem framing, or, in Mulgan’s words, “framing questions and determining what is
worthy of scarce attention and through what lens it is to be seen.” Gitte Kragh, chapter 38
in this handbook, tells us how participatory citizen science can help understand the magnitude and speed of climate change, which ultimately helps society decide what is the issue
at hand and how much of a problem it is. The Hypermind platform similarly provides
an example of collective prediction applied to disease outbreaks (see Servan-Schreiber
chapter 32 in this handbook).
• Agenda-setting: “Identifying and nominating issues that might be amenable to action.”
In this handbook, the story of the International Panel on Climate Change illustrates
how we try, collectively, to define what is the policy agenda on climate change based
on current scientific knowledge, while the Fab Global City Initiative shows how we can
purposefully elicit collective stories of our shared challenges to create the momentum for
collective action to tackle them.
• Ideation: “Generating options to consider.” The example of the Danish Alternative
party, of the German Agora Energiewende stakeholder platform, of an early 20th century
public challenge in Australia, of the Danish Sager der Samler collaborative, or of the UN’s
128
Collective intelligence, democracy, and governance
UNLEASH innovation lab are all examples in this book of assemblies designed to foster
new solutions, on climate change and other topics.
• Deliberation: “Scrutinizing options.” Many CI formats presented here focus precisely
on enabling careful examination of policy options from a variety of angles, from school
participatory budgeting to online deliberative polls and civic tech platforms.
• Decision: “Deciding what to do” is complex and paramount to policymakers’ job. Some
will focus on creating the right context for open-minded dialogue, as in the examples here
of the Open European Dialogue or of Helmut Kohl’s “gastrosophy,” while the Scottish
administration sought to enable “collectively brave” decisions thanks to the Theory U
approach.
• Evaluation: “Scrutinizing what’s been done and judging whether it’s working” may be
less common, but reflecting on performance, accumulating and sharing knowledge of the
experience gained is crucial to developing better solutions together, as exemplified here,
for instance, by the case of the “zero long-term unemployment” experiment in France.
While common sense invites us to consider how to foster collective intelligence at each step
of the policy cycle, complexity invites us to be recursive in our approach. In the complex,
often confusing world we live in, the OODAL loop – which stands for Observe, Orient,
Decide, Act & Learn (Boyd 2018) – proposes a five-step approach to help decision-makers
filter available information, put it in context, and quickly make the most appropriate decision while also understanding that changes can be made as more data becomes available.
Applying this recursive examination of CI devices at each step of the policy cycle should
become central to public administration’s mandate.11
These activities are organized via different participatory practices, such as:
Representative practices include those where decisions are made, resources
allocated, or disputes resolved by an individual or body selected by the community
for this purpose.
Administrative practices include those where decisions are made, resources
allocated, or disputes resolved by an individual or team that is employed by the state.
Participatory practices include those where the entire community (as construed
by the rules of that community) are invited to contribute to making a decision,
allocating a resource, or resolving a dispute. These can include actions taken by the
community as a whole, or by a subset that chooses to participate. (Examples: town
halls, participatory budgeting, the jan sunwai…).
Deliberative practices include those where a subset of the community (or the
entire community, in smaller-scale societies) gathers to consider an issue, weigh
arguments, and make decisions or recommendations. In contemporary usage, as
well as in ancient societies such as Athens, such groups can be chosen by sortition to
create a resemblance with the community as a whole. (Examples: citizens’ assemblies, consensus conferences, citizens’ juries, deliberative polling).
Participatory democracy thus gives citizens an opportunity to take a direct role in political decisions and policies that affect their lives, rather than through elected representatives.
Deliberative democracy focuses instead on developing the conditions by which the
considered judgment of a “mini-public” (a representative sample of the population) can
129
Boucher, Hallin, Paulson et al.
be brought to bear on a public issue (Landemore, 2013). As such, deliberative democrats
insist on the quality of the deliberation leading up to those considered judgments, evaluating the pros and cons of different alternative views, and are often less concerned with
maximizing the total number of people taking part in the process. Direct democracy,
in the famous “ladder of citizen participation” sketched out by Sherry Arnstein (1969),
can be understood as any mechanism which grants the electorate power to decide on policies without the intermediation of elected representatives. Referenda and plebiscites are
often criticized, however, for the manipulation of facts and superficial debates which often
characterize these exercises, and the fact that they are suspiciously popular among antidemocratic movements and regimes (Mounk, 2018). Some scholars evaluate democratic
innovations not through the lens of whether a given process is deliberative or participatory
but rather if it increases the system’s overall deliberativeness (Mansbridge and Parkinson,
eds., 2012). This could be the case, for instance, with a protest initiated by a trade union
elite. Overall, adapting the adage that with greater power comes greater responsibility,
we would say that with greater delegation of power comes greater responsibility for ethical and neutrally informed deliberation. Landemore (2017) argues that it is important to
create the proper conditions for effective deliberation, including creating an alignment
around common goals and constraints, and facilitation techniques that depersonalize ideas,
allowing better arguments to naturally come to the fore. Fishkin (2018) identifies five key
conditions for quality deliberation:
Quality information: The extent to which participants are given access to reasonably
accurate information relevant to the issue at hand.
Substantive balance: The extent to which arguments offered by one side or perspective are
answered by considerations offered by those who hold other views.
Diversity: The extent to which all major positions taken on the issue in the public sphere
are represented by participants in the discussion.
Conscientiousness: The extent to which participants sincerely weigh the merits of the
arguments.
Equal consideration: The extent to which arguments offered by all participants are
considered on the merits regardless of which participants offer them.
The examples provided in this handbook suggest that some policymakers are exploring
new ways of including external voices – citizens, experts, professionals… – in their decisionmaking and governance, predictions and innovation. Both participatory and deliberative
processes are visible in the handbook, from participatory budgeting in schools to rich dialogue formats at community, national and even transnational level (see Ringler, in this handbook, chapter 31). As pointed out by Mulgan (2018), “If we see democracy not only as an
expression of popular views but also as a collective thinking process, then (...) the quality of
deliberation matters as much as the quantity of people involved.” As this new paradigm takes
shape, our understanding of democracy could increasingly be oriented toward methods and
processes that harness collective intelligence, supported where appropriate by machine intelligence, to strengthen accurate and innovative policymaking.
Elstub and Escobar (2017), drawing upon a variety of authors, invite us to examine such
democratic innovations with the following questions.
130
newgenrtpdf
Low
High
Election
Purposive selection / sampling
Sortition
Self-selection
MODE OF PARTICIPATION
How do the participants
communicate with each other and
how intense is the participation?
Observation
Listening
Voting
Asking questions, making
comments, engaging in
deliberation
INVOLVEMENT
How are participants involved and
how intense is their contribution?
No decision
Aggregation of preferences
Bargaining / negotiation
Deliberation
AUTHORITY & POWER
How much influence do the
participants have over what public
authorities do?
Personal
benefits: Citizens
have no influence
but may gain
personal benefits
Advise and
consult: Participants
provide advice and
consultation for public
authorities who retain
decision-making power
Co-governance: Citizens do not
formally influence decisions
but may informally influence
them and public opinion can
be transformed through the
process
Direct authority and
co-decision: Citizens
join public officials to
make decisions
Collective intelligence, democracy, and governance
131
SELECTION
How are the participants selected and
how inclusive is the process?
Boucher, Hallin, Paulson et al.
Table 6.1 Collective intelligence methods for democracy and governance
Core Functions
Methods and Tools
Observing/
gathering data
•
•
•
•
•
•
•
•
•
•
•
Modeling and
predicting
Generating
options /
creativity
Filtering/
deliberating/
deciding
Implementing/
pooling labor
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Crowd-mapping
Crowd deliberation and participation on platforms
Citizens’ assemblies or juries
Open data
Citizen science
Participatory Action Research
Data mining
Prediction markets
e-petitions
Crowd surveys (off-/online)
Prediction and forecasting methods: prediction markets,
contests, swarm intelligence, future studies, scenario building
(e.g., Futurecasting, transformative scenario planning,
delphi method, future workshop, causal layered analysis, etc.)
Collaborative mapping
Predictive modeling, predictive analytics
AI functions such as NLP
Lean and prescriptive analytics, behavioral insights, ethnography
Open calls for creative solutions
Open government
Participatory budgeting
Citizens’ assemblies or juries
Collaborative platforms
Hackathons
Design thinking (e.g., living labs, etc.)
Public challenges and other challenge mechanisms
Ideation techniques (e.g., design sprints, hackathons, creative
problem-solving, etc.)
Policy labs
Think tanks
Citizen juries
Mini-public deliberations (Citizens’ assemblies or juries, deliberative
polls, planning cells, etc.)
Focus groups
Opinion surveys
Data collaboratives
Decision matrix, multi-criteria decision analysis
Transformative scenario planning
Door-to-door canvassing
Citizen hearings
Participatory Budgeting
Nudging and monetary incentives
Mutual aid networks
Pol- /Legal-/Sup-tech
Regulatory sandboxes and innovation hubs
Randomized control trials
132
Collective intelligence, democracy, and governance
Table 6.1 Cont.
Core Functions
Methods and Tools
Learning/sensemaking/
memory/wisdom
•
•
•
•
•
•
•
•
Wikis
Communities of practice
Collective learning and sense-making
Community science
National archives
Research, training
Serious games
User committees
The democratic innovations encountered in this handbook usually demonstrate hybrid
combinations of these different options. Generally speaking, democratic innovation is eminently, as evidenced in all the stories, characterized by hybridity and experimentation: We
see mini-publics selected by lot combined with self-selected participants online and referenda in the case of Iceland’s constitutional crowdsourcing initiative. We have examples of
collaborative governance combined with crowdsourcing, deliberation, and digital participation, for instance, in the municipality of Slagelse.
The following table similarly categorizes some of the most commonly encountered collective intelligence methods according to the schema of “core functions” of group thinking
as described by Mulgan (2018). Of course such categories are overly simplified, as some
methods can contribute to more than one function, which is why you will find some methods
listed in more than one line. A number of devices, such as participatory budgeting or citizens
assemblies can have several functions depending on how they are implemented.
This evolving toolbox is being constantly enriched due to new abilities provided by
machines working with humans.
Humans and machines thinking together
Hybrid intelligence
Hybrid intelligence (HI) is the combination of human and machine intelligence,
expanding human intellect instead of replacing it. HI takes human intelligence and
intentionality into account when making meaningful decisions and performing appropriate
actions, together with ethical, legal, and societal values.12
Artificial intelligence can facilitate the surfacing of collective tacit knowledge for policymaking in numerous ways (cf. Peeters et al., 2021). Examples of such hybrid forms of
intelligence include Amazon Mechanical Turk (AMT) for workflow, MIT’s Superminds,
and other crowdsourcing platforms such as CitizenLab that enhance citizen participation in
democratic processes through natural language processing (NLP) features or AI algorithms.
Peeters et al. (2021) suggest the following ways in which artificial intelligence can function
in a symbiotic relationship with human intelligence and thereby advance decision-making:
In a dyadic way, a policymaker can ask a decision support system what the right
decision would be under a given context and for a given problem.
133
Boucher, Hallin, Paulson et al.
At the team level, an example could be a swarm of public drones that perform
sensing surveillance of floods and then transmit data to a platform on the ground
used by rescue teams to help move people from potential flood areas.
At the organizational multi-team level, such as when policymakers carry
out new policy reforms but tap into different groups’ judgments using CI
methods, such as swarm intelligence when each participant connects to the
server and is allocated a controllable graphical magnet that allows the user to
freely apply force vectors on the puck in real-time (Rosenberg, 2015). It could
also be citizens’ deliberation on participatory platforms, supported by natural
language processing (NLP) to sort and categorize the open-text contributions of
interested citizens.13
At the societal and cultural level, such as when multiple systems collaborate
and interact with one another supported with AI. Collective intuitions of policy
questions can be transformed into hypotheses that are testable using AI. For
example, collective intuitions and predictions by citizens can point in the direction
of a potentially remarkable discovery for humanity.
A rapidly expanding variety of technologies and algorithms can help to harness the vast
amount of information contained in collective intelligence and to make sense of it to transform it into knowledge that carries meaning and which is actionable. For instance, readers
may come across the following concepts in this and other handbooks:
• Artificial intelligence and machine learning (an AI technology that allows computers
to learn beyond their initial programming), in particular due to neural networks (groups
of algorithms inspired by the human brain), with specific developments such as convolutional neural networks, a type of artificial neural networks in which the pattern of
connections between neurons is inspired by the visual cortex of animals.
• Predictive analytics, which is a branch of advanced analytics that makes predictions
about future outcomes using historical and big data combined with statistical modeling,
data mining techniques and machine learning. This discipline also includes methods
such as lagged causal relationships from time series, which is a way to infer causal
relationships between two time series.
• Heuristic search, which is a type of search performed by AI that looks to find a good
solution, not necessarily a perfect one.
• Logistic regression, decision trees, and logic trees are all statistical models that help
study the relationship between various groups of variables that are particularly common
for machine learning.
• In data science, Bayesian networks are a type of probabilistic graphical model that uses
Bayesian inference for probability computations, and backpropagation is a method to
train neural networks.
• Data mining uses sophisticated mathematical algorithms to segment data and assess the
likelihood of future events. Data mining is also known as knowledge discovery in data.
• Sentiment analysis (or opinion mining) is an NLP technique used to determine
whether data is positive, negative, or neutral.
• Natural language processing, which we encounter in two stories in this handbook,
refers to the ability of a computer program to understand, create patterns, and sort human
language as it is spoken and written.
134
Collective intelligence, democracy, and governance
These technologies may be supported by human–computer interaction principles and
methods that act as aggregators of collective intelligence:
Civic and gov-tech platforms: Citizen feedback, proposals, discussion forums,
complex deliberation, voting. “Civic technology” is a nascent force in the relationship between governments and communities. Elements of the civic technology
ecosystem include “open data, related information and communications technology (ICT) innovations and the organizational boundary-spanning practices of
civic technology” (McNutt et al., 2016).
Crowdsourcing platforms: Crowdsourcing is a “blend of bottom-up open,
creative process with top-down organizational goals” (Brabham, 2013). It is also
described as “a process of accumulating the ideas, thoughts or information from
many independent participants, with the aim to find the best solution for a given
challenge” (Guazzini et al., 2015). Crowdsourcing is also defined as the act of
submitting a task or problem to the public for completion or solution (e.g., Howe,
2006), such as knowledge commons in the form of wikis, Quora, user communities, StackOverflow, etc. Such platforms can be enhanced by gamification
features. Crowdsourcing is a model for integration of the “creative energies of
online communities into day-to-day operations” in order to “leverage the collective intelligence” of the online communities (Brabham, 2013).
In recent years, crowdsourcing has been noted as the preferred search tool for harnessing
collective intelligence (Certomá, Dyer, and Pocatilu, 2017) and has been seen as a point of
cost-effective value creation through collecting relevant knowledge for improving policy
and citizens engagement in policymaking (Aitamurto and Chen, 2017).
Collective consciousness
Emile Durkheim was the first to introduce the theory of collective consciousness in his 1893
book The Division of Labor in Society. Later, he would also rely on the concept in other books,
including Rules of the Sociological Method, Suicide, and The Elementary Forms of Religious Life.
In these books, he explains that the phenomenon is “the totality of beliefs and sentiments
common to the average members of a society.” Durkheim based his theory on observations
that in traditional or primitive societies, religious symbols, discourse, beliefs, and rituals
fostered the collective consciousness. In such cases, where social groups were quite homogeneous, the collective consciousness resulted in what Durkheim termed a “mechanical solidarity” – in effect an automatic binding together of people into a collective through their
shared values, beliefs, and practices.
Collective consciousness has its roots in shared awareness. The word “conscious”
derives from the Latin cum (“with”) and scire (“to know”). In the original sense, two
people who know something together are said to be conscious of it “to one another.”
Combining the two terms “collective” and “consciousness” means shared ideas by individuals relating as interdependent elements within groups or larger structures (Smith and
Thomasson, 2005).
In the Durkheimian perspective, collective consciousness can be created from intentional
human interactions and learning. When such consciousness is absent, members of a community fail to see the context for their actions; as a result, behaviors become automatic and more
135
Boucher, Hallin, Paulson et al.
difficult to adapt to new conditions. Conversely, communities become collectively conscious
through processes of self-reflectivity and meta-cognition (see Paulson in this handbook, ch.
1), and can use these learnings to alter negative patterns that stand in the way of collective
goals (Pór, 2017).
According to Hallin, collective consciousness can be identified and congregated from the
following levels and dependent on the common course of collective intelligence for each
level, such as the individual level, in dyad relationships, in groups, in organizations, in the
neighborhood, in the community, for the city, for the country/shire, for the state/province,
network, and global brain (e.g., the whole internet). (See the introduction to Part 7 on collective intelligence, technology and collective consciousness)
As the cases of this handbook show, democratic innovation is as its core a redistribution
of power outward, where participation, inclusiveness, and diversity are each core drivers of
collective wisdom. Moreover, technological progress combining collective intelligence and
AI provides opportunities to create learning processes that allow a community to become
more self-aware, and therefore potentially more resilient and fulfilled.
This chapter on foundational definitions of collective intelligence has sought to provided
a selection of the key definitions and concepts in the emerging field of CI as they relate to
democracy, governance, and public problem-solving. Testing these concepts in real-world
conditions, exploring the impacts achieved and lessons learned, is the aim of the case studies
which follow.
Notes
1 It should be noted that from the perspective of democracy and governance, the body of Galton’s
career as a whole is abhorrent: the coiner of the term “eugenics,” Galton was an unabashed racist
who argued for the weeding out of “undesirable traits” from the English population.
2 The conceptual distinctions between “collective knowledge,” “collective intelligence,” and “collective wisdom” are interesting and still somewhat unsettled. We take up this point in Section
2.4 infra.
3 NESTA blog, “Using machine learning to map the field of collective intelligence research”,
Nov. 22, 2018. www.nesta.org.uk/blog/mapping-col lect ive-intel ligence-resea rch/, accessed July
3, 2022.
4 The phrase “at the same time” is, of course, dependent on circumstances and context.
5 Again, depending on one’s understanding of “at the same time,” prediction markets could either be
considered a synchronous or asynchronous mode; regardless, the drivers and overall outcome are
the same.
6 Our phrase, not Condorcet’s.
7 See www.hyperm ind.com/.
8 See www.galaxyzoo.org/.
9 Of course the “thoughts” of these early animals were not expressed in the form of human language;
these are illustrations only.
10 As noted above, Mulgan in this handbook adopts the term “core function” in place of his earlier
term “functional capabilities” (Mulgan, 2018).
11 It is interesting to note here how public administrations around the world are creating spaces
to evaluate, learn, and improve on new governance approaches, from the UK What Works
centers, to the EU’s Competence Centre on Participatory and Deliberative Democracy, or
the French “Direction interministérielle de la transformation publique,” among many other
examples.
12 For an overview on this matter, see the introductory chapter to Part 3 on the collective intelligence
society and the use of collective tacit knowledge and artificial intelligence for policymaking.
13 Examples of this are the stories included in this handbook on the Danish municipality of Slagelse
as well as on deliberative polling online.
136
Collective intelligence, democracy, and governance
References
Acemoglu, D., Johnson, S., Querubin, P. and Robinson, J.A., 2008. When does policy reform work? The
case of central bank independence (No. w14033). National Bureau of Economic Research.
Ågotnes, T., and Wáng, Y.N. (2017). Resolving distributed knowledge. Artificial Intelligence, 252, 1–21.
Doi: http://dx.doi.org/10.1016/j.art int.2017.07.002
Aitamurto, T. and Chen, K. (2017). The value of crowdsourcing in public policymaking: epistemic,
democratic and economic value. The Theory and Practice of Legislation, 5(1), 55–72.
Andersen, T.J. and Hallin, C.A. (2017). The adaptive organization and fast-slow systems. Oxford
Research Encyclopedias: Business and Management. Oxford University Press.
Andersen, Torben J., Hallin, C.A. and Fredens, K. (2018). Et netværk af hjerner. Copenhagen: Gyldendal.
Argyris, C. (1995). Action science and organizational learning. Journal of Managerial Psychology,
10(6), 20–26.
Arnstein, S. (1969). A ladder of citizen participation. Journal of the American Planning Association, 35(4),
July 1969, 216–224.
Atlee, T., Benkler, Y., Homer-Dixon, T., Levy, P. Malone, T.W., et al. (2008). Collective intelligence: creating a prosperous world at peace. Earth Intelligence Network. 1st edition (March 19, 2008),
Oakton, Virginia.
Babbitt, P. (2012). Complexity and democracy: pathologies of modern governance. Prepared for
delivery at the 2012 Annual Meeting of the American Political Science Association, Aug. 30–Sept.
2, 2012.
Barrett, L.F. (2021). Seven and a half lessons about the brain. Picador.
Bennett R. (1998). The importance of tacit knowledge in strategic deliberations and decisions.
Management Decision, 9(3), 589–597.
Bertomá, C., Dyer, M., and Pocatilu, L. (2017). Citizen empowerment and innovation in the data-rich city,
London: Springer.
Boehm, C. (2009). Hierarchy in the forest: The evolution of egalitarian behavior. Harvard University Press.
Bound, K., and Mulgan, G. (2019). A compendium of innovation methods. NESTA.
Boyd, J. (2018). A discourse on winning and losing. Maxwell Air Force Base, AL: Air University Press.
Brabham, D.C. (2013). Crowdsourcing. MIT Press.
Brown, M.I., and Wai, J. (2021). Can you ever be too smart for your own good? Comparing linear
and nonlinear effects of cognitive ability on life outcomes. Perspectives on Psychological Science, 16(6),
1337–1359. doi:10.1177/1745691620964122. PMID 33682520. S2CID 232142799
Brynjolfsson, E., and McAfee, A. (2014). The second machine age: work, progress, and prosperity in a time of
brilliant technologies. First edition. New York: W.W. Norton & Company.
Cairney, P., Heikkila, T., and Wood, M. (2019). Making policy in a complex world. Cambridge
University Press.
Certomà, C., Corsini, F., and Frey, M. (2020). Hyperconnected, receptive and do-it-yourself city. An
investigation into the European imaginary of crowdsourcing for urban governance. Technology in
Society, 61.
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19.
Clarke, H.D., Goodwin, M., and Whiteley, P. (2017). Brexit: Why Britain voted to leave the European
Union. Cambridge: Cambridge University Press.
Cohen, J. (1986). An epistemic conception of democracy. Ethics, 97(1), 26–38.
Colombo, M., Irvine, E., and Stapleton, M. (Eds.). (2019). Andy Clark and his critics. Oxford
University Press.
Dahl, R. (2020). On Democracy. Yale University Press.
Dewey, J. (1927). The public and its problems. Athens, OH: Swallow Press.
Du Bois, W.E.B. (1920). On the ruling of men. From Darkwater: voices from within the veil. Harcourt Brace.
Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science
Quarterly, 44(2), 350–383.
Edmondson, A. (2019). The role of psychological safety. Leader to Leader, 2019: 13–19. https://doi.
org/10.1002/ltl.20419
Edmondson, A., and Moingeon, B. (1998). From organizational learning to the learning organization.
Management Learning, 29(1), 5–20.
Elkink, J.A., Farrell, D.M., Reidy, T., and Suiter, J. (2017). Understanding the 2015 marriage referendum in Ireland: context, campaign, and conservative Ireland. Irish Political Studies, 32(3), 361–381.
137
Boucher, Hallin, Paulson et al.
Elstub, S., and Escobar, O. (2017). A typology of democratic innovations. Paper for the Political
Studies Association’s Annual Conference, 10th–12th April 2017, Glasgow.
Encyclopaedia Brittanica, online, “Human Intelligence”, https://www.britann ica.com/science/
human-intel ligence-psychology, accessed January 10, 2022.
Engelbart, D. (1962). Augmenting the human intellect: a conceptual framework. Summary report
AFOSR-3223, Stanford Research Institute, Menlo Park, CA.
Engelbart, D. (1995). Toward augmenting the human intellect and boosting our collective IQ.
Communications of the ACM, 38(8), 31–33.
Farrell, D., and Suiter, J. (2019). Reimagining democracy: lessons in deliberative democracy from the Irish front
line. Cornell University Press.
Fishkin, J.S. (2018). Democracy when the people are thinking: Revitalizing our politics through public deliberation. Oxford University Press.
Foss, K., and Foss, N.J. (2002). Authority in the context of distributed knowledge. DRUID - Danish Research
Unit for Industrial Dynamics.
Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19(4),
25–42. doi:10.1257/089533005775196732
Friedman, T.L. (2019). Warning! Everything is going deep: the age of surveillance capitalism. The
New York Times. January 2019. Retrieved from: www.nyti mes.com/2019/01/29/opin ion/art ific
ial-intel ligence-surveilla nce.html
Galton, F. (1907).Vox populi. Nature, 75, 450–451.
Gammelgaard, F.B., and Hallin, C.A. (2018). Changing the innovation game-crowdsourcing in
incumbent firms. In ACM Collective Intelligence Conference.
Gray, J. (2013). Mill on liberty: a defence. Routledge.
Greenspan, S.I. (1979). Intelligence and adaptation: an integration of psychoanalytic and Piagetian
developmental psychology. Psychological Issues, 12(3–4, Mono 47/48), p. 408.
Gret, M., and Sintomer, Y. (2005). The Porto Alegre experiment: learning lessons for better democracy.
Zed Books.
Grossmann I. (2017). Wisdom in context. Perspectives on Psychological Science, 12(2), 233–257.
doi:10.1177/1745691616672066
Guazzini, A., Vilone, D., Donati, C., Nardi, A., and Levnajić, Z. (2015). Modeling crowdsourcing as
collective problem solving. Scientific Reports, 5(1), 1–11.
Habermas, J. (1985). The theory of communicative action, volume 1: reason and the rationalization of society.
Boston, MA: Beacon Press.
Haidt, J. (2012). The righteous mind: why good people are divided by politics and religion. Vintage.
Haraway, Donna. (1988). Situated knowledges: the science question in feminism and the privilege of
partial perspective. Feminist Studies, 14(3), 575–599.
Hartz-Karp, J., and Sullivan, B. (2014). The unfulfilled promise of online deliberation. Journal of Public
Deliberation, 10(1), 1–5.
Hayek, F.A. (1945). The use of knowledge in society. The American Economic Review, 35(4), 519–530.
Henrich, J. (2015). The secret of our success. Princeton University Press.
Hiltz, S.R. and Turoff, M. (1978). The network nation: human communication via computer. Addison-Wesley.
Hong, L., and Page, S. (2004). Groups of diverse problem solvers can outperform groups of highability problem solvers. Proceedings of the National Academy of Sciences, 101 (46), 16385–16389.
Howe, J. (2006). The rise of crowdsourcing. Wired Magazine, 14(6), 1–4.
Irwin, R. (2018). Ibn Khaldun. Princeton University Press.
Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and fiascos. Boston: Houghton Mifflin.
Janis, Irving L. (1971). Groupthink. Psychology Today, 5(6), 43–46.
Janis, Irving L. (2008). Groupthink. IEEE Engineering Management Review, 36(1).
Kahneman, D. (2011). Thinking, fast and slow. Macmillan.
Kenney, E.J., Denyer, N., Easterling, P.E., Hardie, P., and Hunter, R. (Eds.). (2008). Plato: Protagoras.
Cambridge University Press.
Kierstead, F. (1974). Education for a transitional democracy: a comparison of Jean Jacques Rousseau’s concept of
general will to John Dewey’s concept of collective intelligence. PhD thesis, University of Oklahoma.
Landemore, H. (2013). Democratic reason: Politics, collective intelligence and the rule of the many. New Jersey,
Princeton: Princeton University Press.
Landemore, H. (2017). Beyond the fact of disagreement? The epistemic turn in deliberative democracy. Social Epistemology, 31(3), 277–295, DOI:10.1080/02691728.2017.1317868
138
Collective intelligence, democracy, and governance
Landemore, H., and Elster, J. (2012). Collective wisdom: Principles and mechanisms. Cambridge
University Press.
Leal, D. (2022). Speech acts and collective intelligence methods: impact on leadership and innovation in
organizations and systems. PhD thesis, Universitat de Barcelona.
Le Bon, G., (1905) Psychologie des foules. Paris : Édition Félix Alcan
Mack, O., Khare, A., Krämer, A., and Burgartz, T. (2016) Managing in a VUCA World, Springer.
Malone, T.W. (2004). The future of work: How the new order of business will shape your organization, your
management style and your life. Harvard Business School Press.
Malone, T.W., and Bernstein, M.S. (Eds.). (2015). Handbook of collective intelligence. MIT Press.
Malone, T.W., Laubacher, R., and Dellarocas, C. (2010). The collective intelligence genome. MIT
Sloan Management Review, 51(3).
Mansbridge, J.J. (1983). Beyond adversary democracy. University of Chicago Press.
Mansfield, H.C. (2001). Machiavelli’s new modes and orders: A study of the discourses on Livy. University of
Chicago Press.
Mansuri, G., and Rao, V. (2012). Localizing development: Does participation work? Development
Economics Research Department Policy Research Project. Washington, DC: World Bank.
Mayka, L. (2019). Building participatory institutions in Latin America: Reform coalitions and institutional
change. Cambridge University Press.
McNutt, J.G., Justice, J.B., Melitski, J.M., Ahn, M.J., Siddiqui, S.R., Carter, D.T., and Kline, A.D.
(2016). The diffusion of civic technology and open government in the United States. Information
Polity, 21(2), 153–170.
Mercier, H., and Claidière, N. (2022). Does discussion make crowds any wiser? Cognition, 222, 104912.
Metcalf, L., Askay, D.A., and Rosenberg, L.B. (2019). Keeping humans in the loop: Pooling knowledge
through artificial swarm intelligence to improve business decision making. California Management
Review, 61(4), 84–109.
Mill, J.S. (1865). Considerations on representative government by John Stuart Mill. Longman, Green,
Longman, Roberts, and Green.
Miller, J. (2018). Can democracy work?: A short history of a radical idea, from ancient Athens to our world.
Simon & Schuster.
Millotay, N., and Sgueo, G. (2020). Collective intelligence at EU level - Social and democratic
dimensions. European Parliamentary Research Service.
Morel, C., and Oury, J.M. (2011). Absurd decisions and highly reliable processes.
Mounk, Y. (2018). The people vs. democracy. Harvard University Press.
Mulgan, G. (2018). Big mind: How collective intelligence can change our world. Princeton University Press.
Muthukrishna, M., Henrich, J. (2016). Innovation in the collective brain. Philosophical Transactions of
the Royal Society B: Biological Sciences, 371. 10.1098/rstb.2015.0192
Nagatomo, S. (1992). An eastern concept of the body: Yuasa’s body scheme. In M. Sheets-Johnstone
(Ed.), Giving the body its due (pp. xi, 233). Albany: State University of New York Press.
Noble, S., Mendez, M., Grewal, D., and Parasuraman, A. (2022). The fifth industrial revolution: How
harmonious human-machine collaborations are triggering a retail and service r(evolution). Journal
of Retailing, 98(2), 199–208.
Nonaka, I. (1991). The knowledge creating company. Harvard Business Review, 72(6), 104–114.
Nonaka, I., and Takeuchi, H. (1995). The knowledge-creating company: How Japanese companies create the
dynamics of innovation. New York, NY: Oxford University Press.
Noveck, B.S. (2015). Smart citizens, smarter state: The technologies of expertise and the future of governing.
Harvard University Press.
Noveck, B.S. (2018). Crowdlaw: Collective intelligence and lawmaking. Analyse & Kritik, 40(2),
359–380.
Noveck, B.S. (2021). Solving public problems: A practical guide to fix our government and change our world.
Yale University Press.
OECD (2020). Innovative citizen participation and new democratic institutions. Catching the Deliberative
Wave, Paris.
Page, S. (2017). The diversity bonus. Princeton University Press.
Page, Scott (2007). The difference: How the power of diversity creates better groups, firms, schools and societies.
Princeton, NJ: Princeton University Press.
Parkinson, J., and Mansbridge, J. (Eds.). (2012). Deliberative systems: Deliberative democracy at the large
scale. Cambridge University Press.
139
Boucher, Hallin, Paulson et al.
Paul, A.M. (2021). The extended mind: The power of thinking outside the brain. Eamon Dolan Books.
Peach, K., Berditchevskaia, A., and Bass, T. (2019). The collective intelligence design playbook. Nesta
Centre for Collective Intelligence Design. Available at www.nesta.org.uk/tool kit/col lect ive-intel
ligence-design-playbook/
Peeters, M.M.M., van Diggelen, J., van Den Bosch, K., and Bronkhorst, A. (2021). Hybrid collective
intelligence in a human–AI society. AI & Society, 36(1), 217–238.
Pinker, S. (2010). The cognitive niche: Coevolution of intelligence, sociality, and language. Proceedings
of the National Academy of Sciences, 107(supplement_2), 8993–8999.
Polanyi, M. (1958). Personal knowledge: Towards a post-critical philosophy. Chicago, IL: University of
Chicago Press.
Polanyi, M. (1966). The tacit dimension. Chicago, IL: University of Chicago Press.
Pollok, P., Lüttgens, D., and Piller, F.T. (2019). How firms develop capabilities for crowdsourcing to
increase open innovation performance: The interplay between organizational roles and knowledge
processes. Journal of Product Innovation Management, 36(4), 412–441.
Pór, G. (2017). On the verge of collective awakening. Spanda Journal, VII, 1/2017.
Puett, M., and Gross-Loh, C. (2016). The path: What Chinese philosophers can teach us about the good life.
Simon & Schuster.
Raymond, E. (1999). The cathedral and the bazaar. Knowledge, Technology & Policy, 12(3), 23–49.
Reber, A.S. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology: General,
118, 219–235.
Reid, C.R., Lutz, M.J., Powell, S., Kao, A.B., Couzin, I.D., and Garnier, S. (2015). Army ants dynamically adjust living bridges in response to a cost–benefit trade-off. Proceedings of the National Academy
of Sciences, 112(49), 15113–15118.
Riedl, C., Kim, J.Y., Gupta, P., Malone, T.W., and Woolley, A.W. (2021). Quantifying collective
intelligence in human groups. Proceedings of the National Academy of Sciences, 118 (21). https://doi.
org/10.1073/pnas.200573711
Rittel, H.W., and Webber, M.M. (1974). Wicked problems. Man-made futures, 26(1), 272–280.
Rittel, H.W., and Webber, M.M. (1974). Wicked problems. Man-made Futures, 26(1), 272–280.
Rodrik, D., 2018. Understanding economic policy reform. In Modern Political Economy and Latin
America (pp. 59–70). Routledge.
Rosenberg, L. (2015). Human swarms, a real-time paradigm for collective Intelligence. Collective Intelligence
Conference, 2015.
Runco, M.A., Millar, G., Acar, S., and Cramond, B. (2010). Torrance tests of creative thinking as
predictors of personal and public achievement: A fifty year follow-up. Creativity Research Journal, 22
(4). DOI: 10.1080/10400419.2010.523393
Ryan, R.M., and Deci, E.L. (2000). Intrinsic and extrinsic motivations: Classic definitions and new
directions. Contemporary Educational Psychology, 25, 54–67.
Ryan, M., Gambrell, D., and Noveck, B.S. (2020). Using collective intelligence to solve public problems.
Nesta: London, UK. www.nesta.org.uk/report/using-col lect ive-intel ligence-solve-publicproblems/
Scharmer, C. and Kaeufer, K. (2010). In front of the blank canvas: sensing emerging futures. Journal of
Business Strategy, 31(4), 21–29. https://doi.org/10.1108/02756661011055159
Seeley, T.D., and Buhrman, S.C. (1999). Group decision making in swarms of honey bees. Behavioral
Ecology and Sociobiology, 45(1), 19–31.
Servan-Schreiber, É. (2018). Supercollectif. La nouvelle puissance de l’intelligence collective. Fayard.
Shirley, D.A., and Langan-Fox, J. (1996). Intuition: A review of the literature. Psychological Reports,
79, 563–584.
Sgueo, G. (2018). Games, power and democracies. Bocconi University Press.
Sliwa, J. (2021). Toward collective animal neuroscience. Science, 374(6566), 397–398.
Smith, D.W. and Thomasson, A.L. (2005). Phenomenology and philosophy of mind. New York: Oxford
University Press.
Spearman, C. (1904). General intelligence: Objectively determined and measured. American Journal of
Psychology, 15, 201–293.
Sternberg, R. J., & O’Hara, L. A. (2000). Intelligence and creativity. In R. J. Sternberg (Ed.),
Handbook of intelligence (pp. 611–630). Cambridge University Press. https://doi.org/10.1017/
CBO9780511807947.028
140
Collective intelligence, democracy, and governance
Stouten, J., Rousseau, D.M., and De Cremer, D., 2018. Successful organizational change: Integrating
the management practice and scholarly literatures. Academy of Management Annals, 12(2), 752–788.
Sunstein, C.R. (2005). Why societies need dissent. Harvard University Press.
Sunstein, C.R. (2006). Infotopia: How many minds produce knowledge. London: Oxford University Press.
Sunstein, C. R. (2007). Republic.com 2.0. Princeton University Press
Surowiecki, J. (2004). The Wisdom of Crowds: Why the Many Are Smarter Than the Few. London.
Abacus. 295.
Surowiecki, J. (2005). The wisdom of crowds. Anchor.
Tanaka, S. (2011). The notion of embodied knowledge. In P. Stenner, J. Cromby, J. Motzkau, J. Yen, &
Y. Haosheng (Eds.), Theoretical psychology: Global transformations and challenges (pp. 149–157).Concord,
Canada: Captus Press.
Tannen, D. (2005). Conversational style: Analyzing talk among friends. Oxford University Press.
Tirri, K., and Nokelainen, P. (2011). Measuring multiple intelligences and moral sensitivities in education. Moral Development and Citizenship Education. Springer.
Tomasello, M. (2019). Becoming human. Harvard University Press.
Torras-Gómez, E., Guo, M., and Ramis, M. (2019). Sociological theory from dialogic democracy.
International and Multidisciplinary Journal of Social Sciences, 8(3), 216–234. https://doi.org/10.17583/
rimcis.2019.4919
Vale, G.L., Flynn, E.G., and Kendal, R.L. (2012). Cumulative culture and future thinking: Is mental
time travel a prerequisite to cumulative cultural evolution? Learning and Motivation, 43(4), 220–230.
https://doi.org/10.1016/j.lmot.2012.05.010
Van Bavel, J.J., and Packer, D.J. (2021). The power of us: Harnessing our shared identities to improve performance, increase cooperation, and promote social harmony. Little, Brown Spark.
Waldron, J. (1995). The wisdom of the multitude: Some reflections on book 3, chapter 11 of Aristotle’s
politics. Political Theory, 23(4), 563–584.
Woolley, A., Aggarwal, I., and Malone, T.W. (2015). Collective intelligence and group performance.
Current Directions in Psychological Science, 24(6), 420–424. doi:10.1177/0963721415599543
Woolley, A., Chabris, C.F., Pentland, A., Hashmi, N., and Malone, T.W. (2010). Evidence for a collective intelligence factor in the performance of human groups. Science, 330 (6004), 686–688.
141