BJC Azaa065

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

doi:10.

1093/bjc/azaa065 BRIT. J. CRIMINOL

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


CRIME PREVENTION, SWARM INTELLIGENCE AND
STIGMERGY: UNDERSTANDING THE MECHANISMS OF SOCIAL
MEDIA-FACILITATED COMMUNITY CRIME PREVENTION
Mark A. Wood and Chrissy Thompson*

Social media are now utilized extensively by Neighbourhood Watch-style initiatives; however, the
impact social media have on the practices and mechanisms of community crime prevention remains
under-theorized. Drawing on our observations of an Australian-based community crime preven-
tion group over two-and-a-half years, this article develops a grounded theory of the mechanisms
underpinning the group’s social media-facilitated practices of responding to local crime. We find
that social media-facilitated Neighbourhood Watch is shaped by two phenomena that have yet to
receive sustained attention in crime prevention research. These are swarm intelligence—a form of
self-organization wherein collectives process information to solve problems that members cannot
solve individually—and stigmergy: work that stimulates further work. In explaining how swarm
intelligence and stigmergy interact with several of the long-acknowledged mechanisms and issues
associated with Neighbourhood Watch, we emphasize the importance of examining how the media
context of community crime prevention groups shapes their practices, behaviour and (in)efficacy.
Key Words:  community crime prevention, neighbourhood watch, swarm intelligence,
social media, stigmergy, digital criminology

Introduction
Hailing from a municipality in Victoria, Australia, Springsclare Crime Prevention
is a community crime prevention group that has accumulated over 13,000 members
since being established in 2016. Unlike traditional Neighbourhood Watch schemes,
Springsclare Crime Prevention is run entirely through social media, using a semi-
public Facebook group to quickly spread information about crimes in the area, crime
prevention tips and other local concerns. As groups like Springsclare Crime Prevention
demonstrate, mobile and social media represent potentially disruptive technologies for
community crime prevention. Despite this, the potentially transformative impact of
mobile and social media on community crime prevention has received relatively limited
attention within criminology (Akkermans and Vollaard 2015; Lub 2017; 2018; Mols and
Pridmore 2019; Pridmore et  al. 2019). In particular, few studies have conceptualized
the mechanisms underlying social media-facilitated community crime prevention and
how they interface with well-established mechanisms, such as natural surveillance and
collective efficacy (Cozens 2008; Sampson 2012). Better understanding this interface
between old and new mechanisms provides much-needed knowledge on whether, when

*Mark A.  Wood, School of Humanities and Social Sciences, Deakin University, 221 Burwood Hwy, Burwood, Victoria 3125,
Australia; [email protected]; Chrissy Thompson, Independent Researcher.

Page 1 of 20
© The Author(s) 2020. Published by Oxford University Press on behalf of the Centre for Crime and Justice Studies (ISTD).
All rights reserved. For permissions, please e-mail: [email protected]
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


and how social media may be more productively integrated into community crime pre-
vention initiatives.
To better understand this interface, we draw on observations of Springsclare Crime
Prevention to develop a grounded theory of the mechanisms underpinning the group’s
crime prevention practices. Through doing so, we argue that social media-facilitated
Neighbourhood Watch initiatives can be better understood using two mechanisms
that have yet to be applied to understanding community crime prevention: swarm in-
telligence and stigmergy. These mechanisms help us understand the distinctly self-
organized nature of crime prevention behaviour that can be fostered through social
media-facilitated Neighbourhood Watch initiatives, as well as their use of collective in-
telligence to respond to local crime (see Estellés-Arolas 2020). They also help us under-
stand how social media can amplify some of the long-standing issues associated with
Neighbourhood Watch programs, including their potential to increase fear of crime
and exclusionary sentiments (Rosenbaum 1987; Laycock and Tilley 1995).
First used to describe the collective behaviour of social insects (Bonabeau and Meyer
2001), the notion of swarm intelligence represents a form of self-organization wherein
‘two or more individuals independently collect information that is processed through
social interaction and provides a solution to a cognitive problem that is not available to
single individuals’ (Beni and Wang 1993; Krause et al. 2010: 28). In this sense, swarm
intelligence is intentional: it is put to use in solving problems faced by a group. In the
animal world, one key problem addressed by swam intelligence is protecting groups
against threats, primarily through what Lima (1995) has termed the ‘many eyes effect’
or ‘many eyes theory’. Put simply, the many eyes effect posits that larger groups of ani-
mals will detect predators and other threats sooner through information diffusion,
with swarm intelligence facilitating escape by both the individual who initially detects
the threat and the other members of the swarm. Such swarm intelligence can be under-
pinned by stigmergy: a mechanism of indirect coordination where ‘work … stimulates
more work’ (Lewis and Marsh 2016: 1). Introduced by biologist Pierre-Paul Grassé
(1959) to explain termite behaviour, the concept of stigmergy has since been exam-
ined as a ‘universal organizing principle’ (Heylighen 2016a; 2016b) relevant to under-
standing issues as diverse as building design, Wikipedia editing and graffiti writing
(MacDowall 2014; see also Nieto-Gomez 2016; Caskey et al. 2018).
To demonstrate the stigmergic and swarm intelligence characteristics of community
crime prevention facilitated through social media, our article unfolds as follows. In the
following literature review section, we review research on the mechanisms underpin-
ning, and issues associated with, Neighbourhood Watch-style community crime pre-
vention groups. After explaining our critical realist grounded theory methodology,
the three findings sections of our article then examine, in turn, how the concepts of
swarm intelligence and stigmergy can be utilized to better understand the mechanisms
underpinning social media-facilitated crime prevention on groups such as Springsclare
Crime Prevention. Bringing the ‘many eyes’ theory of swarm intelligence into conversa-
tion with Jacobs’ (1961) notion of ‘eyes on the street’, we examine the impact social and
mobile media may have on community crime prevention initiatives, given their influ-
ence on the nature, reach and speed of information diffusion among members of com-
munity groups. While, on the one hand, social and mobile media enable many ‘eyes on
the street’ to meet on the screen, this can generate hyperconsciousness of crime that
is conducive to ‘suspicion creep’—a widening of the parameters of suspicion among
Page 2 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


members of a group. As we detail in the final findings section on stigmergy, this suspi-
cion creep is amplified by a reliance on first-order crime prevention measures: activities
that primarily aim to reduce crime. When self-organized through stigmergic systems,
direct crime prevention activities, such as neighbourhood patrols, have a tendency to-
ward positive feedback, for they lack feasible negative feedback mechanisms. This issue
is less likely to affect second-order crime prevention measures: activities that prevent
crime as an indirect by-product of pursuing other social aims. Such crime prevention
initiatives have greater scope to be stigmergic but not stigmatic—i.e. to enjoy the bene-
fits of stigmergic self-organization without increasing exclusionary and hypervigilant
sentiments.

Literature Review
Neighbourhood Watch-style community crime prevention groups have been the subject
of extensive research, with criminologists examining their aims, preventative efficacy,
mechanisms and pitfalls (see Rosenbaum 1987; Bennett 1990). In this research, scholars
have proposed numerous mechanisms to explain why Neighbourhood Watch programs
might lower crime rates. These include that Neighbourhood Watch programs may rep-
resent a visible deterrent (Laycock and Tilley 1995; Newburn 2007; see Jacobs 1961);
that they may limit opportunities for crime to occur through informing residents about
security (Cirel et al. 1977; Bennett 1990) and that they may indirectly affect crime rates
through providing the police with information about suspicious activities (Clarke and
Hough 1984; Garofalo and McLeod 1989). Furthermore, scholars have argued that,
Neighbourhood Watch groups may lower crime rates through enforcing standards of
desired behaviour and engaging in other direct interventions (Rosenbaum 1987; Lub
2017). Neighbourhood Watch groups may, in other words, encourage greater informal
social control among residents (Rosenbaum 1987; Sampson 2012), often through
increasing the collective efficacy of communities, i.e. ‘social cohesion among neigh-
bors combined with their willingness to intervene on behalf of the common good’
(Sampson et al. 1997: 918; Sampson 2012). Collective efficacy has been employed exten-
sively in the fields of crime prevention and social ecology, with empirical studies finding
strong empirical support for collective efficacy as a mechanism of informal social con-
trol (Pratt and Cullen 2005; Sampson 2012).
These theoretical foundations of Neighbourhood Watch have, however, been prob-
lematized by studies examining the limitations and potential pitfalls of such programs.
Much of this literature notes that Neighbourhood Watch programs can only be suc-
cessful when a very specific set of conditions are met (Palumbo et al. 1997). Thus, whilst
evidence suggests that Neighbourhood Watch programs can have a modest effect
in reducing crime (Garofalo and McLeod 1989; Bennett et al. 2008), such programs
are not viable for all neighbourhoods. Research has, e.g., consistently indicated that
Neighbourhood Watch programs are considerably more common in relatively affluent
and comparatively low-crime areas (Husain 1988; Hope 1995; Laycock and Tilley 1995;
Brunton-Smith and Bullock 2019) and face considerable challenges in disadvantaged
and high-crime neighbourhoods. For these reasons, Rosenbaum (1987: 127) also ques-
tions several of the core foundations of the Neighbourhood Watch ‘Implant Hypothesis’,
i.e. that ‘citizen participation, and eventually informal social control mechanisms,

Page 3 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


can be “implanted in neighbourhoods where they do not currently exist”’. Similarly,
important questions have also been raised about whether Neighbourhood Watch-
associated practices, such as Block Watches, are self-sustaining (see Rosenbaum 1987;
Hope 1995). As Skogan (1988: 49) notes, ‘concern about crime simply does not provide
a basis for sustained individual participation’, with many evaluations of community
crime prevention initiatives indicating that participants’ interest in the initiative waned
over time (Lindsay and McGillis 1986; Garofalo and McLeod 1989).
Finally, scholars have noted that, contrary to Neighbourhood Watch’s aims of
increasing social cohesion and lowering residents’ fear of crime, such schemes can
serve to amplify exclusionary sentiments (Laycock and Tilley 1995; Crawford 1997;
1999) and increase residents’ fear of crime (Rosenbaum 1987; Rosenbaum and Heath
1990). Research has indicated that fear of crime is lower in neighbourhoods where
residents feel that they have a greater degree of informal control over behaviour in
the area (Skogan and Maxfield 1981; Greenberg et al. 1982). However, several studies
have indicated that Neighbourhood Watch programs have the potential to increase
residents’ fear of crime through increasing their awareness of local crime (Rosenbaum
and Heath 1990). Indeed, as Rosenbaum (1987: 128) notes, mechanisms intended to
decrease fear of crime, such as residential meetings, may actually serve as forums that
‘heighten residents’ fears and prejudices’.
As noted earlier, few studies have examined community crime prevention groups on
Facebook, and only a number of studies have examined similar groups established on
WhatsApp: an instant messaging app now owned by Facebook (Dixon 2017; Lub 2018;
Pridmore et al. 2019). Much of this literature examines the implications social media
have had in making it easier to report crimes or ‘suspicious’ behaviour to members of
a community; as Lub (2017: 136) notes, ‘today, reporting to the police or fellow resi-
dents about suspicious behaviour or persons is just a phone swipe or a mouse click away,
lowering the threshold for reporting “suspicious” acts or persons’. This lowering of the
threshold for reporting crime and ‘suspicious’ behaviour can lead to both positive and
negative outcomes (Lub 2017). Positively, it may increase group members’ ability to
engage in situational control. Though efficacy-focussed studies into this issue remain
scant, Akkermans and Vollaard’s (2015) research indicates that WhatsApp neighbour-
hood crime prevention groups in the Dutch city of Tilburg contributed to a decline in
burglaries within some neighbourhoods. Negatively, it may calcify ingroup/outgroup
divisions and intensify the ethnic profiling and stigmatization of marginalized popu-
lations. Moreover, as Mols and Pridmore (2019: 280)  indicate, the use of WhatsApp
for the purpose of crime prevention can lead to the normalization of surveillance and
vigilance among members of the community—‘a neighbourhood where lateral surveil-
lance becomes the status quo’.

Methodology
Background: Springsclare Crime Prevention
Located in a suburb of Melbourne and home to over 100,000 people, the pseud-
onymous Springsclare is one of several local government areas in Victoria. Springsclare
is a relatively affluent municipality with a low crime rate and bears many of the demo-
graphic hallmarks of areas where community crime prevention groups are more likely
Page 4 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


to emerge (see Skogan 1988; Hope 1995: 44). Australian Bureau of Statistics census
data indicates that Springsclare has higher than Australian average rates of residents
who are married, own their own homes, have children in a nuclear family and have
a university level of educational attainment. Following an increase of burglaries in
the area and intense media coverage of crime rates within Victoria, the municipality
came to host one of the largest Facebook crime prevention groups within the state. As
noted above, Springsclare Crime Prevention was established and operates primarily
through social media. With the exception of regular in-person meetings between the
small number of page administrators who moderate posts and comments on the page,
group members do not meet regularly face-to-face. Rather, the overwhelming ma-
jority of Springsclare Crime Prevention-related activity is self-organized through the
Facebook page.
We purposively selected Springsclare Crime Prevention as our case study as it rep-
resented a relatively large semi-public Facebook group. Our choice of case study was
guided by considerations pertaining to the ethics of ‘online’ observation. The ethics of
‘online’ observation are hotly debated, in particular the practice of covertly observing
or ‘lurking’ on Internet groups and pages (Reilly and Trevisan 2016). As Willis (2019)
argues, in spaces where users would reasonably expect their posts to be viewed by a
large number of individuals, we can consider such spaces semi-public and view research
examining such spaces as comparable to observational research in a public space. For
these reasons, we, like other scholars concerned with digital communities (Willis 2019),
eschewed private groups and selected a semi-public case study group, i.e. a group that
allows anyone to join without a vetting process. Furthermore, we sought, and were
granted, ethical approval by our university’s research ethics committee to conduct our
study of Springsclare Crime Prevention.
To protect the identity of group members, we have, first, anonymized all posts we
quote directly, second, assigned the group itself a pseudonym and, third, made slight
alterations to the syntax of direct quotes from members to prevent the identification
of their authors through Internet searches. Following the long tradition of providing
pseudonyms for research locations (Crow and Maclean 2000), we have elected not to
identify the name of the municipality to protect the identity of its members and to avoid
subjecting the crime prevention group to undue external attention. Using a random
name generator, we allocated the municipality the pseudonym Springsclare and the
Facebook group the pseudonym Springsclare Crime Prevention.

Data collection and analysis


To investigate the mechanisms underpinning Springsclare Crime Prevention, we
adopted a critical realist grounded theory methodology (Oliver 2011) that drew on
naturalistic observations of the group between August 2016 and February 2019. In prac-
tice, our methodology combined a critical realist ontology and its attendant use of
retroduction (Bhaskar 1986) with the well-established grounded theory conventions of
coding, memo writing and theoretical sampling (Glaser and Strauss 1967).
In August 2016, we both joined Springsclare Crime Prevention. Upon joining, we
regularly began receiving updates from the group, which we collected and coded
using the procedure detailed below. To avoid our presence affecting the behaviour of
Page 5 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


group members (see Webb et al. 1966), we took an unobtrusive approach to our ‘digital
fieldwork’ (see Hine 2015) and did not interact with the members of Springsclare
Crime Prevention. However, rather than retroactively ‘scraping’ archived data from
Springsclare Crime Prevention, we analysed all material posted to the page as it was
posted. Collecting and analysing our data in this fashion enabled us as to experience
Springsclare Crime Prevention as members of the group would; during some weeks,
we received and analysed few comments and updates from group members, whilst, in
others, we analysed a considerable quantity of user-generated content. In this manner,
we observed and analysed Springsclare Crime Prevention daily between August 2016
and February 2019. Textual and visual media content posted to Springsclare Crime
Prevention—including the ‘like’ and ‘share’ metrics of posts and comments—were
downloaded and analysed employing a grounded theory procedure, in which codes
were assigned to data excepts through the ‘constant comparison’ of new to existing
findings (Glaser and Strauss 1967: 114). In line with this grounded theory method-
ology (Glaser and Strauss 1967), incidents were our unit of analysis. In our study, such
incidents were comprised of (1) an individual post in the group and (2) any com-
ments, shares and ‘reaction’ responses to the post by other group members or the
original posting member. In examining these incidents to understand the mechanisms
of Facebook-facilitated community crime prevention, we analysed three of their key
characteristics:
(1)  G
 roup members’ practices on Springsclare Crime Prevention: how, if at all, were
individuals using Springslare Crime Prevention to engage in crime prevention?
(2)  Group members’ interactions through Springsclare Crime Prevention: how did
group members coordinate crime prevention activities and interact with other
members?
(3) The architecture of Facebook and Springslare Crime Prevention: how did the
architecture of the Springslare Crime Prevention Facebook group shape members’
practices and interactions?
In line with a critical realist grounded theory approach, our analysis was primarily
retroductive, asking ‘what must be true for this to be the case?’ before abstracting po-
tential causal mechanisms underpinning Springsclare Crime Prevention and seeking
empirical evidence for these abstractions (Bhaskar 1986; Oliver 2011). As Oliver (2011:
379–80) details, critical realism’s emphasis on retroduction is highly compatible with
grounded theory’s increasing emphasis on abduction over induction (Charmaz 2006;
Reichertz 2007). Indeed, as Oliver (2011: 380) puts it, ‘retroduction is simply abduc-
tion with a specific question in mind’. In line with an abductive/retroductive ap-
proach, we recognised that coding is always informed by the researchers’ pre-existing
knowledge, but that concepts have to ‘earn their way’ into the analysis through being
evidenced in the data (Oliver 2011: 381). To increase the reliability of our findings,
we analysed data independently before discussing and resolving any discrepancies in
our analysis by consensus (Syed and Nelson 2015). We concluded our observations of
the group in February 2019 after deciding that we had reached theoretical saturation:
the point in a critical realist grounded theory project where the theory generated by
the study ‘has, for the time being, greater explanatory power than its rivals’ (Oliver
2011: 379).

Page 6 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


The swarm intelligence of social media-facilitated crime prevention
In this section, we detail, first, the applicability of swarm intelligence as a mechanism
for understanding collective behaviour on Facebook community crime prevention
pages and, second, several of the implications swarm intelligence holds for crime pre-
vention practice. As detailed earlier, when applied to human collectives, swarm intel-
ligence has been defined by Krause et al. (2010: 28) as the phenomenon whereby ‘two
or more individuals independently collect information that is processed through social
interaction and provides a solution to a cognitive problem that is not available to single
individuals’. Underlying this definition of swarm intelligence are several necessary con-
ditions that must be met for collective behaviour to be considered an example of the
phenomenon. These conditions are: 1)  collective intelligence; 2)  self-organised behaviour
and 3)  a medium that allows members to coordinate in real-time. Moreover, it is a neces-
sary condition that the collective intelligence emerges out of self-organized behaviour,
which is fostered by this medium for real-time coordination and collaboration. Each
of these conditions, we argue, are met by the behaviour we observed on Springsclare
Crime Prevention.
First, to be considered swarm intelligence, collective behaviour must be driven by the
collective intelligence of a group: i.e. intelligence that emerges out of the collaboration
of individuals in the collective but which is irreducible to the individual collective mem-
bers’ intelligence (Lévy 1997). Springsclare Crime Prevention meets this condition for
swarm intelligence as the collaborative and self-coordinated behaviour of members
hinges on intelligence that is irreducible to that of individual group members. Indeed,
Springsclare Crime Prevention’s potential as a crime prevention platform, we argue,
lies primarily in its fostering of collective intelligence, wherein intelligence between
actors enables novel forms of coordination, cooperation, deliberation and collabor-
ation in responding to local crime. Here, Springsclare Crime Prevention’s potential as
a crime prevention platform arises not simply through the aggregation of information
about crime but through facilitating real-time collective behaviour in responding to
this information.
Second, to be considered swarm intelligence, this collective intelligence must arise
out of behaviour that is self-organized—‘organization without an organizer’ (Garnier
et al. 2007: 4). As we detail when we turn to the mechanism of stigmergy on Springsclare
Crime Prevention, members’ behaviour in the group was overwhelmingly self-organized.
Members of Springsclare Crime Prevention do not issue commands to other members
when they report a crime or a suspicious occurrence. Nonetheless, by reporting such
behaviour, they may set off a chain of self-organized activity among other users, whose
own additional behaviour may lead to real-time interventions that would not have oc-
curred otherwise.
Third, for swarm intelligence to emerge between humans, a medium is required that
allows members of a collective to coordinate in real-time (Lévy 1997). Human swarm
intelligence is, therefore, an emergent property of three combined elements: 1) the
individuals who make up the swarm; 2) the technological infrastructure that allows
members to self-coordinate in real-time and 3) the information shared between actors.
Any change in one of these three elements will, therefore, shape the behaviour and
outcomes of the swarm. As we detail further below, this medium for real-time coord-
ination between members has the potential to fundamentally change the temporality

Page 7 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


of community-based crime prevention. On Springsclare Crime Prevention, it is the
speed at which information about a crime can be transmitted between geographic-
ally dispersed local actors that sets the group apart from traditional Neighbourhood
Watch groups, where such information is shared to members primarily at scheduled
meetings and via newsletters (Hope 1995; Wickes 2010). If a property crime, such as
the theft of a vehicle, is reported by one member of the group, a large number of in-
dividuals in the area can quickly be alerted to the offence and be on the lookout for
individuals matching the description of ‘suspects’. In September 2016, a user posted
the following:
I know most of you are asleep right now, but I just had a car pull up in the middle of the street next to
my work car. A guy got out and was looking through my canopy with a small torch/phone light. I just
so happened to be on my way to bed when I heard the car. I quickly flicked on the outside light, un-
locked the front door and made my way out (in just my jocks mind you) and the guy quickly jumped
back in and took off...If anyone has cameras in the area that can shed some light on the car’s details,
I’m sure the police will be grateful.

This post elicited over 100 reactions from members and 28 comments, including imme-
diate responses from members asking what time the attempted burglary occurred so
that neighbours on the street could check their own CCTV footage. Here, we see a clear
example of members combining and processing information about potential criminal
activity in their area in real-time. In the discussion between members on this post, in-
formation is exchanged:
If you end up identifying the vehicle let me know. Did it look like a station wagon by any chance?
About 6 months ago I was driving up a side street … around 4am. As I was driving through the street
I had to come to a complete stop because there was a SILVER HOLDEN STATION WAGON parked
in the middle of the road
I live on [street name] and the house next door got his work car broken into a few nights ago and stole
all his tools. It may be the same person

In the following sections, we examine, in turn, two of the key implications of this swarm
intelligence: the creation of a ‘many eyes’ effect in detecting and responding to crime;
and the fostering of ‘suspicion creep’—a widening of what group members perceived
to be suspicious behaviour.

Many eyes on the street


In studies of threat detection and avoidance among animal swarms, ethnologists have
demonstrated that the speed and accuracy of group decisions increase with group size
(Ward et  al. 2011). Ward et  al. (2011) put such increases down to the ‘self-organized
division of vigilance’ within groups, drawing on a variant of the ‘many eyes’ theory. In
explaining threat detection and avoidance in animal swarms, the ‘many eyes’ theory
considers two key factors: the size of a group and the speed at which information is
transmitted within the group. As Ward et al. (2011) explain, ‘the “many eyes” theory
predicts that predators will be detected sooner by larger groups, increasing the prob-
ability of escape for the individual that first detects them, as well as for other group
members as the information is transmitted across the group’.

Page 8 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


With the rise of social media-facilitated Neighbourhood Watch groups, the ‘many
eyes’ theory can be readily translated into the domain of crime prevention—perhaps,
unsurprising given the frequency with which the figure of citizen ‘eyes’ is referred to
in crime prevention literature. As Smith et al. (2002: 83) note, Neighbourhood Watch
is often described as extra ‘eyes and ears’ for the police. More substantially, the no-
tion of ‘eyes on the street’ figures centrally within Jacobs’ (1961) explanation of nat-
ural surveillance—a core technique employed in Neighbourhood Watch schemes.
Insofar as it has been taken up within Crime Prevention Through Environmental
Design literature, Jacob’s notion of promoting ‘eyes on the street’ has focussed on
developing physical spaces that better facilitate natural surveillance (Cozens 2008;
Cozens and Davies 2013). Jacob’s (1961) notion of ‘eyes on the street’, therefore,
frames ‘eyes’ as potential guardians of spaces that may, when a critical mass is
achieved, serve to discourage certain forms of offending within well-surveilled areas
(Cozens 2008).
Building on Jacob’s emphasis on ‘eyes on the street’, we suggest that there is an add-
itional need to consider how these ‘eyes’ acquire, combine and process the information
that they see (and hear). Moreover, we suggest that changes in information communi-
cations technology create a renewed need to consider the morphology of information
diffusion in crime prevention groups. Together, social and mobile media introduce
three key changes to the ‘media context’ (Hampton 2010: 1129) of community crime
prevention groups that can fundamentally disrupt the nature, reach and speed of in-
formation diffusion within them.
First, social media facilitates a pattern of many-to-many information diffusion that
facilitates swarm intelligence. Swarms, as Thacker (2004) explains, are characterized
by leaderless decentralized coordination and, for this reason, swarm intelligence is oc-
casionally said to generate a hive mind: forms of collective intelligence or conscious-
ness, where intelligence is located between actors (Rosenberg and Baltaxe 2016). By
changing the pattern of information sharing from a one-to-many format to a many-
to-many real-time format, such crime prevention pages may produce human swarms
that are collectively aware of changes in information regarding a crime and are able
to (self-)coordinate in responding to it. Second, social media can greatly increase the
potential reach of information provided by individual group members. Members of
crime prevention groups are able to self-broadcast information about crime and sus-
picious behaviour to a large quantity of individuals who, informed about an incident,
may furnish the group with additional information. In line with the ‘many eyes’ theory,
this increase in the size of the crime prevention group increases the probability that
one member of the group will detect an incident and, subsequently, relay information
regarding it to the remaining members of the group. Finally, social and mobile media
increase the speed of information diffusion within citizen crime prevention groups. In
line with the ‘many eyes’ theory, this increase in the speed of information transmission
enables users to respond to incidents ‘in real-time’, providing additional clarifying in-
formation that not only enables members of the ‘swarm’ to avoid unfolding risks but
also to obtain additional information that may assist criminal investigations. Bringing
Jacob’s notion of ‘eyes on the street’ into dialogue with the ethnological ‘many eyes’
theory, we might understand social media crime prevention groups as producing many
eyes on the street that meet on the screen.

Page 9 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Suspicion creep
As a result of this swarm intelligence, members of Facebook crime prevention groups
may  receive a more informed picture of crime within their area than they would by
consuming news media. However, this more accurate picture of crime can come at a
cost. As members became more conscious of local crime, receiving sometimes mul-
tiple reports of crimes within their area, many also became hypervigilant, distrustful
and fearful—an issue long associated with Neighbourhood Watch programs (see
Rosenbaum 1987; Rosenbaum and Heath 1990) but amplified by the real-time updates
and lowered threshold for reporting ‘“suspicious” acts or persons’ afforded by social
media. Early in 2018, one member of Springsclare Crime Prevention articulated this
sentiment explicitly, before ‘signing out of’ (leaving) the group:
The Age reported today that Victoria is rated one of the safest places yet Victorians are the most
scared. This is not a Crime Prevention page it’s a crime reporting page and it’s making me paranoid.
This type of page just encourages people to worry about what could happen but may not happen ...
not healthy. Signing out of living in fear.

This post, which was deleted within a day by administrators, highlights several short-
comings of Facebook-facilitated crime prevention groups: their propensity to increase
members’ fear of crime, their tendency to only report rather than actively prevent
crime and, subsequently, their failure to systematically engage in actual crime preven-
tion behaviours.
In addition to these shortcomings, swarm intelligence contributed to what we term
suspicion creep. In suspicion creep, the forms of behaviour perceived to be suspicious by
group members expands as individual members report witnessing new behaviours that
they perceived to be suspicious. These behaviours are then recoded by sections of the
group as suspicious, leading to  an increased repertoire of behaviours deemed suspi-
cious and, by corollary, reportable. Though some of the behaviours reported by mem-
bers are undoubtedly crime-related, many are not. Many of the posts we analysed, e.g.,
concerned groups of young people doing nothing more than occupying public space.
Others concerned individuals wearing hoodies at night, Pokémon GO players or individ-
uals taking pictures of flowers in an individual’s garden. Suspicion creep can, therefore,
reinforce an ‘us vs them’ mentality, an assertion of community and national borders as
in need of protection from outsiders—a key issue long identified with Neighbourhood
Watch (Laycock and Tilley 1995; Crawford 1997; 1999) but, in this case, amplified by
the many-to-many and real-time communication afforded by social media.

Social media-facilitated crime prevention as stigmergy


One of the key differences between groups such as Springsclare Crime Prevention and
traditional Neighbourhood Watch schemes lies in the organizational mechanisms they
employ. Whereas Neighbourhood Watch schemes tend to plan organized patrols and
other timetabled activities centrally, Springsclare Crime Prevention and similar social
media groups promote a form of community crime prevention that is overwhelmingly
self-organized. Preventative actions, in other words, emerge ‘from local interactions,
without any centralised control directing the activity’ (Heylighen 2016a: 14). As one
moderator wrote, in a post celebrating the group reaching 5000 members:
Page 10 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


This page was designed for the best interest of our beloved shire … but more importantly the com-
munity who make it tick. It’s fantastic to see the interaction, contribution, education and aware-
ness displayed by all.… keep up the great work everyone and let’s try and look after one another.
Springsclare needs this, no wait Springsclare deserves this symbiotic relationship between its people
and the shire for a healthy and prosperous future.

In Springsclare Crime Prevention, we see a shift from scheduled preventative activity to


stigmergic preventative activity: crime prevention practices that are enacted in response
to traces left in the medium and, in being enacted, themselves leave a trace that stimu-
lates further preventative action from others. Heylighen (2016a: 5)  defines stigmergy
as ‘an indirect, mediated mechanism of coordination between actions, in which the
trace of an action left on a medium stimulates the performance of a subsequent action’.
As Heylighen (2016a) notes, stigmergy describes a feedback loop in which marks pro-
duced in an action incite further action of the same kind, which, in turn, produces a
further action-inciting mark (see Figure 1). Importantly, stigmergy does not describe a
deterministic process. For the most part, marks that incite a particular action represent
neither a necessary nor sufficient condition for that action to occur (Heylighen 2016a).
To use the terms of Heylighen’s (2016a) model of stigmergy, social media introduces
a medium for action-inciting traces to be registered. Though there exist a variety of ex-
amples of human stigmergy that are not facilitated by digital technologies, the digital
traces that characterize many computational media can offer a fertile environment for
stigmergy. On Springsclare Crime Prevention, stigmergic crime prevention is a direct
product of the visible digital footprint left by members’ actions on the page, which
serves to stimulate further crime preventative actions by members. In one example
from August 2016, a moderator posted the following with a screenshot of Google Maps
showing their current location:
Just patrolled surrounding areas, nothing suspicious.
Feel free to patrol your surrounding area and upload a screenshot so everyone know when and where
you patrolled.

Another member responded in the comments that they had ‘also patrolled around’,
while another raised alarm over a car driving around the streets:
Not sure if one of us, but a beige Honda CR-V was driving in and out of streets inside Primrose drive
just after 6am this morning. Didn’t get plate as it was pretty dark.

Figure 1.  Heylighen’s (2016a) stigmergic feedback loop model.

Page 11 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Here, we can see how the digital footprint of the original post triggered further crime
prevention actions by other members. For this reason, social media-facilitated crime pre-
vention represents a form of marker-based stigmergy in that the marker (social media
posts) is produced indirectly by the action (engaging in lateral surveillance). In other
words, where engaging with ‘analog’ crime prevention media, such as Neighbourhood
Watch pamphlets (see Wickes 2010), does not generate a trace that stimulates fur-
ther trace-generating action, engaging with the digital equivalents of such material
does. This has several implications that have a bearing on members’ participation in
group practices—a longstanding issue faced by community crime prevention initiatives
(Rosenbaum 1987; Skogan 1988). Most notably, owing to their stigmergic character,
social media-facilitated crime prevention techniques do not require planning or antici-
pation from different members, nor do they require centralized control or supervision.
This lends the crime prevention practices of such groups a high degree of flexibility
as members’ cooperative actions do not need to be coordinated or task allocated, nor
do members even need to be mutually aware of one another’s actions (Zamfirescu and
Filip 2010; Heylighen 2016b).
However, on account of its propensity for positive feedback loops (Heylighen 2016b),
stigmergy is a risky tool for crime prevention practice. As Dipple (2015: 3)  explains,
feedback mechanisms are crucial to stigmergy:
Stigmergy creates equilibrium between positive and negative feedback within the system. Positive
feedback is provided by agents contributing to a signal and results in the signal strength of a trail
increasing as more agents contribute. This promotes rapid and successful task completion. In op-
position to this, the environment provides a negative feedback that will diminish the contributions
making up the signal as they become unused and irrelevant. Signal decay ensures old information
dissipates as it becomes redundant.

In stigmergy, goal attainment provides atrophic negative feedback that moderates


amplification generated by positive feedback loops. This introduces a key problem
for Springsclare Crime Prevention and similar groups. Namely, individual tasks
on Springsclare Crime Prevention may be completed by identifying individuals or
establishing that ‘suspicious’ behaviour represents a false alarm. The completion
of such individual tasks, however, does not necessarily bring into play negative feed-
back, for their very completion leaves a residuum of suspicion and vigilance directed
at the overarching goal of creating a ‘crime free’ neighbourhood. In August 2016, a
moderator acknowledged this positive feedback loop, noting that ‘some members are
afraid. Scared and anxious about incidences happening a little too frequently, that
they have told me they left the page’. As we witnessed through following Springsclare
Crime Prevention, such stigmergic feedback loops were readily apparent as a driving
force behind the amplification of surveillance, vigilance and fear on the platform.
Members of the group came to ‘see’ crime everywhere, as they received constant re-
minders that crime was occurring near them, and exhortations to engage in lateral
surveillance. As one member who joined the group soon after its establishment in
July 2016 stated:
Reading through these posts has now scared me half to death and I have locked my doors! Up to now
I have been very slack with locking etc. but I might now be a bit more careful. I don’t want to become
paranoid though.

Page 12 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Another user lamented how the popular augmented reality smartphone game Pokémon
GO had resulted in more people walking and driving around the neighbourhood:
How can we know if people/cars are suspicious or Pokemon GO hunters? Maybe we all need to down-
load ‘Pokemon GO’ and check for Pokemon in the street when we see something suss

This last example is particularly noteworthy in demonstrating how new mobile technolo-
gies—in this case, augmented reality smartphone applications—can disrupt expected
uses of public space. The temporary flood of people in public space eagerly hunting
Pokémon on their phone application, utilizing public space and coming together to
participate in a collective activity is ostensibly one of the key aims of Springsclare Crime
Prevention, i.e. to bring the community together. However, due to the positive feedback
loop produced on this page, the presence of Pokémon GO players sparked disquiet
and suspicion such that locals were encouraged to download the Pokémon GO app to
better distinguish ‘suspicious’ behaviour from the Pokémon hunting behaviour associ-
ated with the augmented reality game.
This propensity for amplifying vigilance indicates that stigmergy may be a pro-
ductive force for crime prevention when the preventative efficacy of an action is an
indirect by-product of the stigmergic behaviour rather than the direct aim of that be-
haviour. Here, we can productively distinguish between first-order and second-order
forms of crime prevention. First-order crime prevention represents practices and ini-
tiatives aimed at reducing crime. Examples of first-order crime prevention include
neighbourhood watch patrols and sharing crime prevention tips. Second-order crime
prevention, by contrast, represents practices and initiatives that prevent crime as a
byproduct of pursuing other aims. Here, we can include initiatives that foster com-
munity cohesion, which consequently increases informal social control and collective
efficacy.
In being narrowly conceived as a crime prevention page, Springsclare Crime
Prevention did little to foster such forms of second-order crime prevention; its remit
was practices  aimed directly at preventing crime. As demonstrated by the group, this
narrow remit can be circumvented through branding not as a crime prevention page
specifically but as a broader local community group concerned with a range of issues
affecting an area. Indeed, the utility of this approach has long been recognized by
researchers examining crime prevention. As Bursik and Grasmick (1999: 154)  note,
‘crime prevention programs have the greatest likelihood of success if they are inte-
grated into the activities of more general, multi-issue neighbourhood organizations’.
Such an approach is precisely what occurred with Springsclare Crime Prevention within
the group’s second year, when it rebranded as Springsclare Community Hub. Upon re-
branding, the group flourished as a platform where residents of Springsclare assisted
one another in a variety of tasks, from offering advice on service providers in the area
to locating lost animals. Such actions still benefit from the stigmergic, marker-based
mechanism of self-organization provided by Facebook groups; members’ trace-making
actions are stimulated by traces left by other members, which, in turn, stimulate further
trace-making actions. But unlike first-order crime prevention, they do not simultan-
eously amplify hypervigilance and fear of crime among group members. The second-
order crime prevention they allow for stokes a very different affective register among
members, for the positive feedback of members’ contributions is put in check by the
attainment of small-scale goals that do not overtly pertain to crime prevention. The
Page 13 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


running newsfeed is now more balanced with just as many (if not more) news stories
about good deeds, found pets, successful charity drives and good Samaritans handing
in lost belongings:
Tommy has come home, thank you for everyone’s help!! – Feb 2018
I would like to thank the lovely lady and her son who paid for my parent’s groceries at the super-
market at The Bentley Square. What a totally inspiring and amazing thing you did today. My Parents
(although embarrassed) were so chuffed with your act of kindness. As my Grandmother used to say
‘Your blood’s worth bottling’. Good things will come your way, thank you xxx – June 2018

Coupling traditional Neighbourhood Watch crime prevention techniques with technolo-


gies that facilitate human swarming and stigmergic self-organization can allow groups
to more effectively harness the four key functions of swarm organization: coordination,
cooperation, collaboration and deliberation (Garnier et al. 2007). Furthermore, through
generating marker-creating action that stimulates the performance of subsequent
marker-generating action, technologies that promote stigmergic self-organization have
the potential to address one of the long-standing issues identified in Neighbourhood
Watch groups: their struggle to sustain participation among members over time (Lindsay
and McGillis 1986; Rosenbaum 1987; Skogan 1988; Garofalo and McLeod 1989). As Hope
(1995: 49)  notes, ‘the longevity of neighborhood watch schemes may have most to do
with their not requiring much of participants in practice’. Stigmergic crime prevention
can serve to meet this crucial condition for longevity through shifting from programmed
interventions—such as scheduled block watches requiring ongoing commitment from
members—to self-organized interventions that respond to markers left by members.
Though human stigmergy is far from a deterministic process, stigmergic markers can
serve to ‘nudge’ members to engage in preventative practices (Thaler and Sunstein 2009).
These practices, in turn, may generate positive feedback loops of self-organized first- and
second-order preventative behaviour that responds to members’ concerns as they arise.
In doing so, groups built around swarm intelligence and stigmergic second-order crime
prevention have the potential to increase the collective efficacy of communities. Indeed,
as the findings of other studies indicate, networked digital technologies have the poten-
tial to enhance local community by increasing the social capital of neighbourhoods and
the community attachment of their members (Stern and Adams 2010). For these reasons,
Hampton (2010: 1129) concludes that ‘media play an integral role in collective efficacy’
and that current theories of neighbourhood effects need to be revised to better account
for the role of ‘media context’ (Hampton 2010: 1129). Harnessed in the service of second-
order crime prevention, then, stigmergy may have the potential to increase the collective
efficacy of communities. Harnessed in the service of first-order crime prevention initia-
tives, stigmergy has the potential to increase fear among residents and further stigmatize
those at risk of offending. The difficulty, therefore, lies in generating crime prevention
programs that are stigmergic but not stigmatic.

Conclusion
As our analysis demonstrates, the swarm intelligence of Facebook crime prevention
groups can generate hyperconsciousness of crime among members. Viewed purely
from an instrumental perspective, this hyperconsciousness may increase the efficacy of

Page 14 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


local crime prevention techniques that thrive on information sharing. However, taking
a broader view of what constitutes effective crime prevention, this hyperconsciousness
may amplify members’ fear of crime, foster punitive sentiments and lead to ‘suspicion
creep’—a widening of the parameters of suspicion, wherein increasingly innocuous
acts are viewed as suspect. This widening of the parameters of suspicion is, further-
more, a product of the stigmergic self-coordination that operates on community crime
prevention pages such as Springsclare Crime Prevention. In explaining the stigmergic
dimensions of such pages, we have advanced six arguments:
(1) Crime prevention on Facebook community crime prevention groups is stigmergic
as it leaves markers in a medium that, in turn, stimulates further marker-generating
activity.
(2)  Stigmergic organization has different implications for first-order community crime
prevention activities and second-order community crime prevention activities.
(3) When self-organized through stigmergic systems, first-order crime prevention ac-
tivities, such as neighbourhood patrols, have a tendency towards positive feedback.
This is because such forms of direct crime prevention often lack a feasible negative
feedback mechanism indicating that such activities have achieved their goals.
(4)  A key consequence of this tendency towards positive feedback is that stigmergic
first-order crime prevention can lead to (a) an amplification of members’ fear of
crime and vigilance and (b) ‘suspicion creep’.
(5)  This consequence does not similarly affect second-order community crime pre-
vention activities. When self-organized through stigmergic systems, second-order
crime prevention activities can have a tendency towards equilibrium, with the posi-
tive feedback of members’ contributions checked by the attainment of small-scale
goals.
(6) W hilst retaining the benefits of stigmergic self-coordination, second-order
crime prevention techniques do not carry the previously detailed negative con-
sequences stemming from the application of stigmergic self-coordination to
first-order crime prevention activities. Such crime prevention initiatives are more
likely to be stigmergic but not stigmatic and, consequently, have the potential to
reduce crime through facilitating certain forms of community cohesion and col-
lective efficacy.
In examining the stigmergic properties of social media-facilitated crime prevention,
it is worth briefly examining how the concept of stigmergy connects with and departs
from other key concepts that criminologists have drawn upon to understand human
interaction effects, such as interaction ritual chains (see Collins 2004; Rossner 2011),
emotional and social contagion (Fagan et al. 2007) and cascades (Braithwaite 2020).
First, stigmergy as a coordination mechanism can be distinguished from Collins’
(2004) influential concept of interaction ritual chains in that, whereas interaction rit-
uals operate through physically co-present face-to-face encounters (Collins 2004: 19;
Henry 2020), stigmergic cooperation operates without direct encounters; communica-
tion between members is asynchronous and is the product of traces one actor leaves in
a medium. However, like interaction rituals, stigmergic behaviour can produce ‘emo-
tional energy’ among actors—indeed, such emotional energy can be key to stigmergic
mechanisms ‘stimulating’ indirect cooperation between individuals.

Page 15 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


For this reason, certain forms of stigmergy also bear an ostensible similarity to emo-
tional and social contagion: the transmission of behaviours—or, in the case of emo-
tional contagion, emotions—through ‘contact, communication, and imitation’ (Fagan
et al. 2007). However, there are again key distinctions between these concepts, as well
as Braithwaite’s 2020: 138)  similar notion of cascades, which emphasizes that behav-
iour diffuses through emulation and that ‘All contagions are cascades and contagions
are the most important kinds of cascades’. The most notable of these is that, unlike
cascades and social and emotional contagion, stigmergy does not occur through a pro-
cess of one actor imitating another. Indeed, stigmergic coordination does not occur
through individuals witnessing the behaviour of marker-leaving individuals—rather
it occurs through engaging with the markers they leave behind. Furthermore, as
Drury et  al. (2019) highlight, one of the key issues associated with contagion-related
metaphors is that they can imply an automatic transmission of affect and behaviour
between parties. Contrary to contagion’s imitative automaticity, stigmergy does not ne-
cessarily require an individual to undertake the same kind of action involved in cre-
ating the marker. Thus, whilst emotional contagion can arise out of what Theraulaz
and Bonabeau (1999) terms quantitative stigmergy—wherein traces stimulate the same
kind of action but at different intensities and frequencies—it can never represent a
form of what they term qualitative stigmergy: where a trace stimulates a different kind
of action (Heylighen 2016b: 54). Here, it is important to emphasize that unlike cascades
and contagion, the concept of stigmergy is concerned with behaviour coordination ra-
ther than behaviour diffusion. Thus, whilst stigmergic crime prevention practices can
cascade through contagion (see Braithwaite 2020), stigmergy is not reducible to behav-
ioural contagion, nor is it reducible to Braithwaite’s related notion of crime prevention
as a cascade phenomenon.
In examining the stigmergic properties of Springsclare Crime Prevention, our re-
search highlights the need to further examine the mediatic conditions and media
context of community crime prevention: the role different media and communica-
tion technologies play in shaping community crime prevention by altering the speed,
morphology and reach of information diffusion and coordination between commu-
nity members. As philosophers of technology and perspectives such as actor-network
theory have long emphasized (see Brown 2006), technologies play a key role in shaping
human sociality and interactions—interactions that are central to community crime
prevention initiatives. As a small qualitative study into a specific spatially bounded com-
munity, our findings are not generalizable; however, our insights into the underlying
mechanisms of swarm intelligence and stigmergy are transferable to similar crime
prevention groups employing Facebook and similar social media platforms. In exam-
ining the potential impact of social media in shaping community crime prevention
initiatives, however, we want to emphasize that care needs to be taken in avoiding an
approach characterized by technological determinism, i.e. an approach that treats so-
cial media, such as Facebook, as having a uniform effect in shaping community crime
prevention initiatives. As social events are multiply determined (see Elder-Vass 2010),
the mechanisms we have examined here do not operate in isolation from a variety
of other mechanisms pertaining to the characteristics of spatially bounded commu-
nities. Such mechanisms will shape the impact that stigmergic mechanisms have on
events arising from social media-facilitated community crime prevention initiatives.
Future studies examining the mediatic dimensions of community crime prevention
Page 16 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


might, therefore, add measures of Internet-facilitated collective intelligence to the list
of variables they examine (see Malone 2018). Moreover, by measuring the reach and
speed of information-diffusion among crime prevention groups on social media, they
might test whether ‘many eyes on the street’ meeting on the screen does increase the
efficacy of such groups—and whether it leads to a form of ‘suspicion creep’ in doing so.
For as our research indicates, the mechanisms of swarm intelligence and stigmergy may
represent a double-edged sword for community crime prevention. Though powerful in
increasing the ability for community crime prevention groups to collectively respond
in real-time to crime, swarm intelligence and stigmergy can contribute to a hive mind
buzzing with fear, hypervigilance and exclusionary sentiments.

References
Akkermans, M. and Vollaard, B. (2015). Effect van het WhatsApp-project in Tilburg op het
Aantal Woninginbraken—Een Evaluatie. Tilburg University.
Beni, G. and Wang, J. (1993), ‘Swarm intelligence in cellular robotic systems’, in P. Dario,
G.  Sandini and P.  Aebischer, eds, Robots and Biological Systems: Towards a New Bionics?,
703–12. Springer.
Bennett, T. H. (1990), Evaluating Neighborhood Watch. Cambridge University Press.
Bennett, T., Holloway, K. and Farrington, D. (2008), ‘The Effectiveness of Neighborhood
Watch’, Campbell Systematic Reviews, 4: 1–46.
Bhaskar, R. (1986), Scientific Realism and Human Emancipation. The Thetford Press.
Bonabeau,  E. and Meyer,  C. (2001), ‘Swarm Intelligence: A  Whole New Way to Think
About Business’, Harvard Business Review, 79: 106–15.
Braithwaite,  J. (2020), ‘Crime as a Cascade Phenomenon’, International Journal of
Comparative and Applied Criminal Justice, 44: 137–69.
Brown, S. (2006), ‘The Criminology of Hybrids: Rethinking Crime and law in Technosocial
Networks’, Theoretical Criminology, 10: 223–44.
Brunton-Smith,  I. and Bullock,  K. (2019), ‘Patterns and Drivers of Co-Production in
Neighbourhood Watch in England and Wales: From Neo-Liberalism to New Localism’,
The British Journal of Criminology, 59: 85–106.
Bursik, R. J. and Grasmick, H. G. (1999), Neighborhoods & Crime. Lexington Books.
Caskey, T. R., Wasek, J. S. and Franz, A. Y. (2018), ‘Deter and Protect: Crime Modeling
With Multi-Agent Learning’, Complex & Intelligent Systems, 4: 155–69.
Charmaz,  K. (2006), Constructing Grounded Theory: A  Practical Guide Through Qualitative
Analysis. Sage Publications.
Cirel, P., Evans, P., McGillis, D. and Whitcomb, D. (1977), Community Crime Prevention
Program, Seattle: An Exemplary Project. Government Printing Office.
Clarke, R. and Hough, M. (1984), Crime and Police Effectiveness, Home Office Research Study
No. 79. Home Office.
Collins, R. (2004), Interaction Ritual Chains. Princeton University Press.
Cozens, P. (2008), ‘Crime Prevention Through Environmental Design’, in R. Wortley and
L. Mazerolle, eds, Environmental Criminology and Crime Analysis, 153–77. Willan.
Cozens, P. and Davies, T. (2013), ‘Investigating “Eyes on the Street,” Perceptions of Crime
and the use of Security Shutters—Insights from a Residential Suburb in Perth (WA)’,
Crime Prevention and Community Safety, 15: 175–91.

Page 17 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Crawford, A. (1997), The Local Governance of Crime: Appeals to Community and Partnerships.
Clarendon Press.
—— (1999), ‘Questioning Appeals to Community Within Crime Prevention and Control’,
European Journal on Criminal Policy and Research, 7: 509–30.
Crow, G. and Maclean, C. (2000), ‘Community’, in G. Payne, ed., Social Divisions, 223–41.
Macmillan.
Dipple,  A.  C. (2015), ‘Collaboration in Web N.0: Stigmergy and Virtual Pheromones’, doctoral
dissertation, Queensland University of Technology.
Dixon,  N. (2017), ‘Stranger-ness and Belonging in a Neighbourhood WhatsApp Group’,
Open Cultural Studies, 1: 493–503.
Drury, J., Ball, R., Neville, F., Reicher, S. and Stott, C. (2019), Re-Reading the 2011 Riots:
ESRC Beyond Contagion Interim Report. University of Sussex. Available online at https://
www.sussex.ac.uk/webteam/gateway/file.php?name=beyond-contagion-report-forthe-
guardian-january-2019.pdf&site=557 [accessed: 20 July 2020].
Elder-Vass, D. (2010), The Causal Power of Social Structures: Emergence, Structure and Agency.
Cambridge University Press.
Estellés-Arolas, E. (2020), ‘Using Crowdsourcing for a Safer Society: When the Crowd
Rules’, European Journal of Criminology, 1–20. doi: 10.1177/1477370820916439.
Fagan,  J., Wilkinson,  D.  L. and Davies,  G. (2007), ‘Social Contagion of Violence’, in
D. J. Flannery, A. T. Vazsonyi and I. D. Waldman, eds, The Cambridge Handbook of Violent
Behavior and Aggression. Cambridge University Press. p. 688–723.
Garnier, S., Gautrais, J. and Theraulaz, G. (2007), ‘The Biological Principles of Swarm
Intelligence’, Swarm Intelligence, 1: 3–31.
Garofalo, J. and McLeod, M. (1989), ‘The Structure and Operations of Neighborhood
Watch Programs in the United States’, Crime & Delinquency, 35: 326–44.
Glaser,  B. and Strauss,  A. (1967), ‘Grounded Theory: The Discovery of Grounded
Theory’, Sociology: the Journal of the British Sociological Association, 12: 27–49.
Grassé, P. P. (1959), ‘La Theorie de la Stigmergie: Essai d’interpretation du Comportement
des Termites Constructeurs’, Insectes Sociaux, 6: 41–81.
Greenberg S. W., Rohe W. M. and Williams J. R. (1982), Safe and Secure Neighbourhoods:
Physical Characteristics and Informal Territorial Control in High and Low Crime Neighbourhoods.
Department of Justice.
Hampton, K. N. (2010), ‘Internet Use and the Concentration of Disadvantage: Glocalization
and the Urban Underclass’, American Behavioral Scientist, 53: 1111–32.
Henry,  A. (2020), ‘Interaction Rituals and “Police” Encounters: New Challenges for
Interactionist Police Sociology’, Policing and Society. doi: 10.1080/10439463.2020.1791861.
Heylighen, F. (2016a), ‘Stigmergy as a Universal Coordination Mechanism I: Definition
and Components’, Cognitive Systems Research, 38: 4–13.
—— (2016b), ‘Stigmergy as a Universal Coordination Mechanism II: Varieties and
Evolution’, Cognitive Systems Research, 38: 50–9.
Hine, C. (2015), Ethnography for the Internet: Embedded, Embodied and Everyday. Bloomsbury
Publishing.
Hope, T. (1995), ‘Community Crime Prevention’, Crime and Justice, 19: 21–89.
Husain, S. (1988), Neighbourhood Watch in England and Wales: A Locational Analysis. Home
Office.
Jacobs, J. (1961), The Death and Life of Great American Cities. Jonathon Cope.

Page 18 of 20
CRIME PREVENTION, SWARM INTELLIGENCE AND STIGMERGY

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Krause,  J., Ruxton,  G.  D. and Krause,  S. (2010), ‘Swarm Intelligence in Animals and
Humans’, Trends in Ecology & Evolution, 25: 28–34.
Laycock, G. and Tilley, N. (1995), Policing and Neighbourhood Watch: Strategic Issues. Home
Office Police Research Group.
Lévy, P. (1997), Collective Intelligence: Mankind’s Emerging World in Cyberspace. Perseus Books.
Lewis,  T.  G. and Marsh,  L. (2016), ‘Human Stigmergy: Theoretical Developments and
New Applications’, Cognitive Systems Research, 38: 1–60.
Lima, S. L. (1995), ‘Back to the Basics of Anti-Predatory Vigilance: The Group-Size Effect’,
Animal Behaviour, 49: 11–20.
Lindsay,  B. and McGillis,  D. (1986), ‘Citywide Community Crime Prevention: An
Assessment of the Seattle Program’, in D.P. Rosenbaum, ed., Community Crime Prevention:
Does it Work?, 46–67. SAGE.
Lub,  V. (2017), Neighbourhood Watch in a Digital Age: Between Crime Control and Culture of
Control. Palgrave Macmillan.
—— (2018), ‘Neighbourhood Watch: Mechanisms and Moral Implications’, British Journal
of Criminology, 58, 906–24.
MacDowall, L. (2014), ‘Graffiti, Street Art and Theories of Stigmergy’, in J. Lossau and W.
Stevens, eds, The Uses of Art in Public Space, 45–60. Routledge.
Malone,  T.  W. (2018), Superminds: The Surprising Power of People and Computers Thinking
Together. Little, Brown and Company
Mols, A. and Pridmore, J. (2019), ‘When Citizens Are “Actually Doing Police Work”: The
Blurring of Boundaries in WhatsApp Neighbourhood Crime Prevention Groups in The
Netherlands’, Surveillance & Society, 17: 272–87.
Newburn, T. (2007), Criminology. Willian Publishing.
Nieto-Gomez,  R. (2016), ‘Stigmergy at the Edge: Adversarial Stigmergy in the War on
Drugs’, Cognitive Systems Research, 38: 31–40.
Oliver,  C. (2011), ‘Critical Realist Grounded Theory: A  New Approach for Social Work
Research’, British Journal of Social Work, 42: 371–87.
Palumbo, D., Ferguson, J. L. and Stein, J. (1997), ‘The Conditions Needed for Successful
Community Crime Prevention’, in S. P. La, ed., Crime Prevention at a Crossroads. Anderson.
pp. 79–98.
Pratt, T. C. and Cullen, F. T. (2005), ‘Assessing Macro-Level Predictors and Theories of
Crime: A Meta-Analysis’, Crime and Justice, 32: 373–450.
Pridmore,  J., Mols,  A., Wang,  Y. and Holleman,  F. (2019), ‘Keeping an Eye on the
Neighbours: Police, Citizens, and Communication Within Mobile Neighbourhood Crime
Prevention Groups’, The Police Journal, 92: 97–120.
Reichertz, J. (2007), ‘Abduction: The Logic of Discovery of Grounded Theory’, A. Bryant
and K. Charmaz, eds, The Sage Handbook of Grounded Theory, 214–28. SAGE.
Reilly, P. and Trevisan, F. (2016), ‘Researching Protest on Facebook: Developing an Ethical
Stance for the Study of Northern Irish Flag Protest Pages’, Information, Communication &
Society, 19: 419–35.
Rosenbaum, D. P. (1987), ‘The Theory and Research Behind Neighborhood Watch: Is it a
Sound Fear and Crime Reduction Strategy?’, Crime & Delinquency, 33: 103–34.
Rosenbaum,  D.  P. and Heath,  L. (1990), ‘The “Psycho-Logic” of Fear-Reduction and
Crime-Prevention Programs’, in Social Influence Processes and Prevention, 221–47. Springer.
Rosenberg, L. B. and Baltaxe, D. (2016), ‘Swarm Intelligence and Morality of the Hive
Mind’, paper presented at Collective Intelligence, 1–9. New York University, 1–3 June.
Page 19 of 20
WOOD AND THOMPSON

Downloaded from https://academic.oup.com/bjc/advance-article/doi/10.1093/bjc/azaa065/5935390 by Auckland University of Technology user on 27 October 2020


Rossner,  M. (2011), ‘Emotions and Interaction Ritual: A  Micro Analysis of Restorative
Justice’, British Journal of Criminology, 51: 95–119.
Sampson,  R.  J. (2012), Great American City: Chicago and the Enduring Neighborhood Effect.
University of Chicago Press.
Sampson,  R.  J., Raudenbush,  S.  W. and Earls,  F. (1997), ‘Neighborhoods and Violent
Crime: A Multilevel Study of Collective Efficacy’, Science, 277: 918–24.
Skogan, W. G. (1988), ‘Community Organizations and Crime’, Crime and Justice, 10: 39–78.
Skogan W. G. and Maxfield M. G. (1981), Coping With Crime. SAGE.
Smith, M. J., Clarke, R. V. and Pease, K. (2002), ‘Anticipatory Benefits in Crime Prevention’,
Crime Prevention Studies, 13: 71–88.
Stern, M. J. and Adams, A. E. (2010), ‘Do Rural Residents Really Use the Internet to Build
Social Capital? An Empirical Investigation’, American Behavioral Scientist, 53: 1389–422.
Syed, M. and Nelson, S. C. (2015), ‘Guidelines for Establishing Reliability When Coding
Narrative Data’, Emerging Adulthood, 3: 375–87.
Thacker, E. (2004), Networks, Swarms, Multitudes: Part Two. CTHEORY.
Thaler, R. H. and Sunstein, C. R. (2009), Nudge: Improving Decisions About Health, Wealth,
and Happiness. Penguin.
Theraulaz, G. and Bonabeau, E. (1999), ‘A Brief History of Stigmergy’, Artificial Life, 5:
97–116.
Ward, A. J., Herbert-Read, J. E., Sumpter, D. J. and Krause, J. (2011), ‘Fast and Accurate
Decisions Through Collective Vigilance in Fish Shoals’, Proceedings of the National Academy
of Sciences, 108: 2312–5.
Webb, E. J., Campbell, D. T., Schwartz, R. D. and Sechrest, L. (1966), Nonreactive Measures
in the Social Sciences. Rand McNally.
Wickes,  R.  L. (2010), ‘Generating Action and Responding to Local Issues: Collective
Efficacy in Context’, Australian & New Zealand Journal of Criminology, 43: 423–43.
Willis,  R. (2019), ‘Observations Online: Finding the Ethical Boundaries of Facebook
Research’, Research Ethics, 15: 1–17.
Zamfirescu, C. B. and Filip, F. G. (2010), ‘Swarming Models for Facilitating Collaborative
Decisions’, International Journal of Computers Communications & Control, 5: 125–37.

Page 20 of 20

You might also like