Safety Science: Marie Nilsen, Trond Kongsvik, Petter Grytten Almklov

Safety Science 148 (2022) 105644
Contents lists available at ScienceDirect
Safety Science
journal homepage: www.elsevier.com/locate/safety
Splintered structures and workers without a workplace: How should safety

science address the fragmentation of organizations?
Marie Nilsen a, *, Trond Kongsvik a, Petter Grytten Almklov b
a
Department of Industrial Economics and Technology Management, Norwegian University of Science and Technology (NTNU), Norway
b
Department of Sociology and Political Science, Norwegian University of Science and Technology (NTNU), Norway
A R T I C L E I N F O A B S T R A C T
Keywords: Organizational fragmentation presents a challenge to prominent safety perspectives hinged upon the traditional
Organization concept of the organization. The continuing disjunction in the workplace has reached new heights in the recent
Contemporary work phenomenon of platform-mediated work (PMW), where workers engage in on-demand labor mediated by
Safety science
platforms. In this paper, the explanatory power of some influential organizational perspectives in safety science is
Platform economy
Goal conflicts
explored in relation to PMW.
Safety culture Neoteric business models combined with platform technology introduce alterations in accountability, goal
conflicts, and social relations. These changes necessitate adjustments in our perspectives to address the safety
challenges of a fissured, contemporary work-life.
This qualitative study based on interviews with 37 delivery platform workers and managers in the Nordic
region and observations of two online courier communities reveals features that diverge from traditional work
settings. Sociotechnical systems thinking is applied in examining goal conflicts arising from work contexts where
the traditional employer-employee relationships are becoming transformed into two-sided marketplaces for
clients and platform workers selling labor. The long tradition for addressing culture in safety science is then
considered in analyzing its applicability to PMW. Finally, we reflect on how the safety research community can
address the fragmentation of the organization.
1. Introduction inadequacies in training and safety management, psychosocial risks,

lack of bargaining power, and low social affinity (Anyfantis & Boustras,
In many sectors, the nature and organization of work are rapidly 2020).
changing. In particular, recent decades have witnessed a phenomenon The fragmentation of the organization has taken a new turn in the
denoted by Weil (2014) as the fissured workplace. Organizational frag emerging platform economy, where digital platforms owned by large
mentation involves an increasing tendency for companies to focus on companies are matching and mediating work by software applications
core activities and outsource specialized work to subcontractors, third (apps) to individuals selling labor (Fleming et al., 2019). An estimated
parties, and independent contractors. This development has gone hand 2% of the European labor force has platform-mediated work (PMW) as a
in hand with pervasive digitalization and new coordinative technolo primary source of income (Pesole et al., 2018), and the number of
gies. For safety science as a research field, understanding risk and safety workers selling labor through digital platforms is rising (Eurofound,
depends on the ability of theory, methods, and models to be applicable 2018). PMW represents a wide spectrum of work ranging from high-
for new ways of organizing work. skilled, professional services (e.g., programming and graphics design)
Growing evidence on the negative health and safety effects of non- to highly ‘taskified’ activities (e.g., answering online surveys, delivery,
standard work arrangements links these arrangements to higher acci and cleaning) (see Huws et al. (2017) and Schmidt (2017) for an over
dent and injury rates, hazard exposures, adverse consequences to view). A common feature is that platform companies profit from the
physical and mental health, and income unpredictability (Quinlan, transactions mediated through their digital platforms between the
2015). Some factors contributing to the increase in accidents include worker and the customer. The workers are generally self-employed, and
* Corresponding author at: Health, Safety, and Environment, Department of Industrial Economics and Technology Management, 12th floor, Sentralbygg I,
Gløshaugen, NTNU, 7491 Trondheim, Norway.
E-mail address: marie.nilsen@ntnu.no (M. Nilsen).
https://doi.org/10.1016/j.ssci.2021.105644
Received 4 May 2021; Received in revised form 9 December 2021; Accepted 21 December 2021
Available online 30 December 2021
0925-7535/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M. Nilsen et al. Safety Science 148 (2022) 105644
the platform owner has no responsibility for their working environment dilemmas (see Eisenhardt, 1989; Almklov et al., 2014) as well as goal
- or at least they rarely formally accept such responsibility. alignment and efficiency/safety trade-offs (see e.g., Almklov & Anton
The problem formulation guiding this paper is: How well-suited are sen 2010; Kongsvik et al., 2012).
influential organizational safety theories for explaining safety chal Platform-mediated work is a step further in this development since it
lenges in platform-mediated work, and what theoretical adaptions could is a technologically mediated form of outsourcing to single individuals
be fruitful for this type of work organization? Specifically, we will focus contributing to the company’s core production. For PMW, the organi
on STS and safety culture strands of research. We will argue that PMW zation as a formal structure and social arena surrounding the worker is
partly repeals the organization as the context of work and that theo no longer present. Essentially and functionally, the workers are still
retical development and adaptation to new work arrangements are connected to an organization, but only through the App’s coordination
needed. By way of a qualitative study involving interviews of 37 plat and control mechanisms. Distinguishing between a functional/struc
form workers and company managers and observations of their online tural understanding of an organization as a formal system and a socio
communities, our paper will discuss organizational safety perspectives logical notion of the organization can contribute to understanding PMW
in relation to platform work. and its consequences. Whereas the formal functions of the organization,
We have chosen to explore the explanatory power of two important the coordination, and payment of work are inscribed in the algorithms,
theoretical perspectives in safety research when faced with PMW: 1) the organization as a social institution and an arena for social interaction
sociotechnical systems thinking (STS) and 2) safety culture. These with colleagues and managers is more or less absent. This also means
strands of research were selected because of their prolonged and wide that the organizational ‘toolbox’ for achieving safety through organi
application in safety science and their inclusion of work context in their zational efforts is different.
propositions, which is highly relevant when discussing PMW. STS has
greatly impacted safety research (Leveson, 2017; Sheridan, 2017; 1.2. Platform-mediated work
Waterson et al., 2017) and highlighted the importance of the organi
zation’s environment, including technology development, regulations, Food delivery companies with neither food nor delivery employees
and laws (Carayon et al., 2015; Pasmore et al., 2019; Winby & Mohr exemplify an era of digital connectivity where global enterprises can be
man, 2018). System safety from a multi-level perspective (Rasmussen, built with nothing but code (Goodwin, 2015). Equipped with Big Data,
1997) has been adopted to underscore the social and organizational powerful algorithms, and cloud computing abilities, platform companies
factors in complex processes in system design (Leveson, 2004). In like aiming for market domination can easily surpass the transformative
manner, the concept of safety culture has endured as a distinct field of power exhibited by factory owners during the industrial revolution
research with a high and steady stream of publications since the 1980s, (Kenney & Zysman, 2016). Digital platforms are refashioning work into
embracing different industries and sectors (Le Coze, 2019). Despite the gigs, tasks, and favors (De Stefano, 2016); disrupting businesses by
lack of consensus in its definition, safety culture and cultural approaches lowering entry barriers and changing value creation and capture; and
continue to be intertwined with organizational structures in the litera capitalizing on regulatory grey areas by subscribing to the mantra ‘Don’t
ture (Antonsen, 2009b; Schulman, 2020). Safety culturés link to the ask permission; ask forgivenesś (Kenney & Zysman, 2016, p. 67).
notion of the organization poses the question of what happens when the Contingent work arrangements facilitated by platforms can
organization is subject to extreme fragmentation, as we will argue is the contribute to isolation, job insecurity and income unpredictability for
case of PMW. workers, and lack of workplace and social protection traditionally
In the following, we will present some developmental trends that covered by standard employment relations (Cherry & Aloisi, 2016; Tran
challenge the traditional concept of the organization and describe some & Sokas, 2017). An overrepresentation of young individuals furthers the
hallmarks of platform-mediated work. Then we will turn to the impli risk of accidents associated with contingent work in PMW (Bajwa et al.,
cations of PMW, particularly the idea of conflicting goals in STS and 2018; Eurofound, 2018; Huws et al., 2017). Moreover, platform control
limitations in safety culture applications in this new context of work. features such as piece-rate compensation, performance ratings, internal
Finally, we provide some reflections on how the safety research com ranking systems, and gamification elements encourage long work hours
munity can address the fragmentation of the organization. and intensify work performance (Griesbach et al., 2019). Research on
drivers and cyclists revealed workers experiencing impairment due to
1.1. The fragmenting organization fatigue and pressure to cut corners, further highlighting the safety im
plications of PMW (Christie & Ward, 2019).
The essence of an organization, in one sense, is that it is a system of Two important characteristics necessitate a further investigation of
division of labor and coordination (see e.g., Mintzberg, 1983). Organi PMẂ s impact on our understanding of safety in contemporary work.
zational maps outline formal systems consisting of functional entities First, platforms structure themselves into ‘lean platforms’ that ‘hyper-
and mechanisms for coordination and control as a rational machine outsource’ everything from workers to maintenance, training, and
designed to achieve a purpose. Real-life organizations, however, are also capital (Selznick, 2014). The business model excludes workers from
social arenas where workers interact with organizational cultures, employee rights and basic social protections like unemployment benefits
shared histories, and values. They are also arenas for power struggles, in many countries (De Stefano, 2016) and social exchange, learning, and
sub-groups, and the emergence of communities of practice (Lave & solidarity that a shared workplace offers (Graham et al., 2017). There
Wenger, 1991) among workers with common interests and work fore, the standard binary relationship between the employer and
situations. employee is transformed into a triad between the customer, the plat
There are several strands of research within safety science that can form, and the worker – creating ambiguity in how it fits into existing
broadly be seen as organizational approaches to safety in the sense that regulations (Eurofound, 2018).
they address organizational qualities as sources of accidents or as Second, platforms contribute to the ‘unbundling of work’ where
qualities that strengthen safety. In the quest to improve safety, an or work activities are disaggregated into tasks and distributed to workers
ganizatioń s management toolkit often consists of an arsenal of in who perform their work in isolation (Bajwa et al., 2018). PMW may
terventions related to safety culture, compliance, protective measures, exclude workers from the protection of health and safety standards;
or working conditions at the sharp end. further, platforms exercise a great degree of control through ‘Click-
With the increasing reliance on subcontracting and outsourcing and through’ agreements that protect the enterprise from potential liabilities
a movement from monolithic organizations with in-house activities to and justify ́ algorithmic management’ strategies (Lee et al., 2015).
more network-based forms of organizing, safety research has addressed The externalization of risks, the individualized organizing of work,
concerns regarding coordination challenges and principal-agent and the highly controlled labor raise questions on how safety is
2
prioritized and valued. Furthermore, the combined attributes of PMW balance various goals while at the same time ensuring safe operations. In
challenge safety science perspectives hinged upon the organization. The a dynamic working environment, the modelś key assumption is to make
following section will introduce the two theoretical strands for explo boundaries visible and provide ample space for local adjustments while
ration and discuss their appropriateness to PMW. remaining within the safety limits (Moorkamp et al., 2014; Rasmussen,
1997). Introducing counter gradients such as training, safety regula
tions, and safety measures implemented in the workplace provides an
1.3. Goal conflicts at the sharp end error margin that serves as a buffer for workers who perform close to the
‘fringes of the usual, accepted practice’ (Rasmussen, 1997, p. 379).
Sociotechnical systems thinking facilitates a multi-level perspective The risk management strategies inherent in the models rely heavily
(left in Fig. 1) to elucidate complex processes and interdependencies that on organizational activities such as safety campaigns or several methods
influence work at the sharp end where humans increasingly interact from the safety management toolbox to prevent injuries and accidents.
with technology (Carayon et al., 2015; Leveson, 2004). The higher the This includes using data from reporting systems that provide insight into
level, the greater degrees of freedom (design, decisions, and time hori past accidents.
zons) and uncertainties in foreseeing local contingencies at the lower How does this model apply to PMW? We propose that in PMW,
levels (Rasmussen, 1997). Worker latitude is constrained by the local additional individual boundaries emerge and need to be considered as
work situation resulting from decisions at the upper levels. In PMW, an these may be incongruent to organizational goals. Since the independent
independent contractor generally exists outside the platform organiza contractor status in PMW places the worker outside the organizatioń s
tion and its risk management system. Hence, the simple hierarchical safety management system, the error margin created using the organi
structure represented in the figure becomes a more elaborate structure zational safety toolbox may be absent. Thus, other goals may dominate
in PMW, consisting of individual workers linked to the platform orga as the worker navigates the local work situation. Furthermore, the
nization through a digital platform while individually responsible for platform interests may be at odds with worker goals when market
the risk management of their legal entity/sole proprietorship. domination is part of the business strategy – an oversupply of service
STS thinking emphasizes the importance of organizations’ external providers benefits both the platform and the customers while workers
environments in examining work-life. Environmental pressures such as risk lower profitability (Degryse, 2016; Muller, 2019).
market conditions and changes in technology can impact risk manage Since independent contractors are responsible for providing their
ment strategies (Rasmussen, 1997). Recent developments include equipment, maintenance, and other responsibilities as their own com
diverse stakeholder interests, temporary organizational memberships pany, the quest for platform domination and other features of PMW may
(Carayon et al., 2015; Pasmore et al., 2019), and customers as part of the significantly impact safe work performance at the sharp end. We pro
work system (Winby & Mohrman, 2018). We further this development pose a migration model that includes additional conflicts for indepen
by underscoring the significance of independent workers loosely tied to dent workers to understand safety in the context of splintered work
the organization in the platform ecosystem. Consistent with Cross and structures. Further, we call for a broader view of sociotechnical systems
Swarts’ (2021) view, we challenge assumptions in safety management to include independent workers and other stakeholders outside the or
and research to include those existing around the organization’s ganization but still part of a wider ecosystem.
periphery.
In discussing the trend towards fragmentation and conflicting goals
at the sharp end in PMW, we take Rasmussen’s (1997) migration model 1.4. Safety culture
(right in Fig. 1) as a starting point in analyzing tensions between safety
and other organizational goals. The model exhibits how individuals and The interest in cultural issues in safety science can be ascribed to the
organizations strive to achieve safety while working towards other search for new approaches for safety improvements in working life.
organizational goals such as economic viability and acceptable work Safety culturés appeal seems to lie on the assumption that a good safety
load (Hollnagel, 2017; Hu et al., 2020). It illustrates the adaptive nav culture values and prioritizes safety relative to other goals (Dahl &
igation occurring within the ‘space of possibilities.’ Rasmussen (1997) Kongsvik, 2018; Neal et al., 2000) and is reflected in positive safety
refers to local work adaptations individuals or groups make in order to performance and results. This assumption explains the broad range of
Fig. 1. Sociotechnical systems thinking (left) illustrates the interaction between the various levels and the environment. The migration model (right) demonstrates
the space of possibilities and gradients toward the boundary of acceptable performance (adapted from Rasmussen, 1997).
3
research related to how safety culture can be assessed and described in Thus, we will discuss how safety culture might apply as a concept in
organizations, the relation between culture and safety performance and PMW, where the employee-employer relationship is diminishing, and
results, and how safety culture can be improved. Done almost entirely in the level of interaction is reduced.
working life and industrial contexts and framed within a traditional
understanding of the organization, definitions of safety culture/climate 2. Method
(Guldenmund, 2000) and questionnaires for safety climate use the terms
‘employees,’ ‘managers,’ and ‘leaders.’ 2.1. Study design
In many instances, the analysis of safety culture has been informed
by a three-level conceptualization of the concept (Bisbey et al., 2021; Since research on the emerging phenomenon of platform-mediated
Guldenmund, 2000; Schein, 1985). 1.) Norms and artifacts are tangible work is still limited in safety science, an explorative research design
and observable reflections of what is prioritized in a work community, was adopted in this study. We conducted semi-structured interviews
2.) Values are less tangible but might be expressed and be reachable for with couriers from two food delivery platforms in the Nordic region
research. Safety climate research might be considered on this level, as it (hereafter DP1 and DP2) using a qualitative method. DP1 is a digital
sets out to reveal attitudes and perceptions in a work community, usu platform composed of two types of couriers – employees and freelancers
ally by means of questionnaire surveys. 3.) Basic assumptions relate to (self-employed and through a third-party provider). DP2 is solely
subconscious and tacit ideas but still guide actions and intentions. These composed of courier partners (freelancers). The interviews occurred
are less available for research, but participant observation over long between February and September 2020. Due to the COVID-19 pandemic,
periods is one approach applied. some interviews were conducted online through Zoom or Teams. This
A distinction is made between a functionalist and interpretive view shift, however, allowed for a more familiar setting since scheduling was
on culture (Glendon and Stanton, 2000). Functionalist perspectives see adapted to the time and place (at home) that was most convenient for
culture as something that can be managed or even manipulated the participants. A flexible scheduling approach was taken to avoid
top-down by management and a means to an end that can serve certain potential negative consequences for work performance and interruption
strategic goals. Interpretive perspectives view culture as an emergent, of company services. The individual interviews were supplemented with
complex, bottom-up phenomenon involving the development of observations of two worker-only online communities. Observation
commonly owned identities and beliefs (Fig. 2). The safety culture captures informal interactions and adds depth to the study.
models that have been developed have mostly been within the func
tionalist perspective, exemplified in Hudson’s (2007) cultural ladder 2.2. Data collection
and Reason’s (1997) model on engineering a safety culture.
‘Being your own boss’ is a recruiting slogan used by platform owners. The empirical data (see Fig. 3 for an overview) consists of 37 semi-
It illustrates that management in the traditional sense does not exist for structured interviews and observations from two online communities.
platform workers. Instead, platform and app algorithms perform much The interview participants consisted of DP1 managers and employees,
of the work management. We thus propose that in the context of DP1 freelancers through a third party, DP2 managers and freelancers,
platform-mediated work where workers generally perform tasks and freelancers working on both platforms.
distributed, coordinated, and controlled by the digital technology (App) The initial contact with potential interviewees applied various ap
in isolation, the applicability of functionalist/management approaches proaches such as contact information from platform websites, social
to safety culture is significantly reduced. media, face-to-face inquiry near restaurant clusters, snowballing (Bier
The interaction among workers is also a foundation for safety culture nacki & Waldorf, 1981), and availing of platform services. The recruit
and how it is conceptualized. A definition by Kongsvik et al. (2018) is an ment and subsequent interviews followed the guidelines of the
illustrative example. Based on Bang (2011), safety culture is defined as Norwegian Center for Research Data (NSD). Upon explicit consent, the
‘the shared safety-related values, norms and perceptions of reality that interviews were audio-recorded and converted into transcripts. Since
develop in an organization when its members interact with each other the interviews aimed to collect data for a set of publications covering
and the surroundings’ (Kongsvik et al., 2018, 222, our translation). various facets of PMW, the interview guide included open-ended ques
Hence, sharing something in common can only be accomplished by some tions that ensured coverage of several issues: (1) individual background
form of interaction (Boudreau & Newman, 1993). When organizational (e.g., when they started with PMW, other income-generating activities,
members interact over a period of time, patterns of relationships might education, immigrant/non-immigrant background, union membership,
develop and form social structures. Interactions are also prerequisites for contract type, motivation for joining the platform), (2) platform com
developing common worldviews, norms, and values. Simply put, and pany (e.g., recruitment, organization, and company characteristics), (3)
based on safety science literature, safety culture emerges in people courier work experience (e.g., work process, incentives, technology), (4)
interacting in a traditional workplace. working conditions (e.g., safety, responsibilities, communication, social
PMW has been characterized as hyper-outsourced and individual support) and (5) regulations (e.g., rights and social protection).
ized. To a large extent, workers are not employed but contracted as sole Data collection focused on how goals in PMW are aligned and
proprietors. We propose that individualization reduces the opportunities whether there is incongruency between the platform and the worker.
for interaction between the workers, and as a consequence, the possi How are the workers organized, and how is safety ensured? Are there
bility for cultural development will also be reduced. ways to monitor performance, and are incentives and sanctions present?
Fig. 2. Functionalist and interpretive perspectives on organizational culture (Glendon & Stanton, 2000).
4
Fig. 3. Data collection - interviews and observations.
How do technology and organization influence their work performance? and organization of work in platform-mediated delivery work. The use
In exploring how safety culture relates to PMW, the patterns of social of technology and potential hazards and constraints faced by the
interaction and information exchange were explored through interviews workers received special focus.
which probed into issues such as platform communication with workers, The data coding involved two coding cycles (Saldaña, 2016). The
face-to-face and digital opportunities for interaction between workers, iterative process strongly links the empirical data with our interpreta
and collective problem-solving. tion. In the first cycle, in vivo coding was performed, resulting in 1,945
The semi-structured interviews enabled the interviewees to pursue codes. The codes were then categorically themed into 52 categories that
an idea of interest or elaborate on issues emerging in the interview. The describe patterns observed, such as general topics and ideas that emerge
interviews lasted from 45 to 75 min. Practical considerations, sufficient in the data (Saldaña, 2016). The categories included features of the
data to cover the topics, and saturation determined the participants’ platform technology (e.g., difficulties in shift grabbing), distinct features
adequate number (Tjora, 2019). of PMW (e.g., performance-based staffing), and worker experiences (e.
An online ethnographic method, netnography or digital anthropology g., difficulties of being self-employed, freelancing as a choice). The
(Horst & Miller, 2012; Kozinets, 2010), was used to study the couriers’ second cycle of the coding involved further categorization according to
social interaction and online behavior. Digital traces from online social similarities in their themes. The nine concepts (standardized services,
interaction (e.g., photos, illustrations, text, emoticons, and links to performance pressures, ratings and ranking systems, scheduled shifts
videos and news articles) provide invaluable insight into platform- and changing conditions, individual investments, risk factors of delivery
mediated work and social relations that are otherwise difficult to work, constricted communication, creating communities, and workers
observe among geographically dispersed couriers. Netnography pro without a workplace) were further coded into three main themes: rele
vides access to vast data; however, collecting information from non- gated responsibility, algorithmic authority, and solitary service.
verbal communication may be subject to misinterpretation, and partic In analyzing the goals, a specific focus was directed at how PMW
ipants may not represent the entire group (Kausel & Hackett, 2015). provides and constrains financial opportunities, impacts workload and
Data collection for the observations occurred following community safety. For instance, we looked into how performance and scheduling
discourse and consent. The online observations of two DP1 commu systems relate to workers’ ability to determine their schedules and in
nities, one in Slack and another in WhatsApp, occurred from June to come. While in determining social aspects of PMW, the analysis focused
August 2020. DP2 couriers have an inactive WhatsApp group, so the on their perceptions of management communications, how the platform
possibility of observing the community was not pursued. Observation ensures their safety, whether or not they feel pressured to deliver quickly
data included thick descriptions of member interaction and online and how the online community addresses technical issues and other
content screenshots (with explicit consent). The researchers’ interaction concerns.
with the community was limited to keep the interactions as natural as The results from the data analysis were compared to the results from
possible. The community members were provided information the observations. Consolidating results from the interviews, available
regarding the study and contact information. documents, and online observations improve validity through method
Couriers were encouraged to participate in the interview with a gift ological triangulation (Yin, 2010).
card. Potential sources of bias include selecting participants based on
accessibility and communication using a second language. The small 3. Results
amount of the gift card may also have discouraged some full-time low-
wage workers from participating. The three main themes represent particular aspects of PMW that
Platform companies are known for keeping information about their deviate from conventional work organizations and processes that may
technology and algorithms guarded closely (Degryse, 2016; Muller, have implications for the selected safety science perspectives. First, an
2019). While DP1 managers agreed to be interviewed on several occa overview of the two delivery platforms is presented, followed by the
sions through a videoconference, DP2 data from the managers were three categories from our data analysis.
limited to text format received through email.
3.1. Overview of food delivery
2.3. Data analysis
The two food delivery platforms (DP1 and DP2) offer a low threshold
The data analysis used the computer-assisted qualitative data anal for entry into the labor market. Many couriers are young (i.e., 20–35
ysis software NVivo to identify patterns relevant to studying the nature years old) male students with little or no previous work experience.
5
Managers from both platforms confirmed that the turnover rate is high, Because of the CBA, we receive compensation for a certain amount per
with DP2 emphasizing that the term turnover is unsuitable since the order, and it is, to be honest, a joke. It may cover brake pads for that
word is associated with employees. Our data indicates two main types of month. If you are lucky, it covers brake pads. (DP1-W13)
couriers. The first group consists of individuals who perform courier
Although many use equipment they owned before joining the plat
work as an additional income source and usually work a few hours a
form, some have bought a car after getting the job at DP2. Moreover, it is
week. The second group consists of individuals having difficulties
reasonable to assume that freelance car drivers incur a higher cost since
finding work due to lack of language proficiency, low educational
they also pay for fuel, parking during pickup and delivery, toll, vehicle
background, or personal issues. These tend to work more than the
insurance, and potential parking and speeding tickets.
former.
DP1 applies a hybrid model consisting of part-time employees (a
3.2.2. Risk factors of delivery work
minimum of 10 h) with open-ended contracts and freelancers who
DP1 and DP2 managers identified several risk factors in courier
establish themselves as independent contractors or employees of a third-
work: (1) theft of the vehicle, primarily bikes, (2) traffic-related injuries,
party company. According to DP1 management, they have employee
(3) damage to one’s own or others’ vehicle or property, and (4) courier
safety representatives and regular meetings with shop stewards (union
safety in meeting customers and restaurant staff. During the COVID-19
representatives). DP1 management mentioned that they have a report
pandemic, virus infection also emerged as a highly relevant risk.
ing system for employees, and although they encourage reporting,
While many cyclists acknowledged a few scratches and near misses,
underreporting is a problem. DP1 management initially planned to have
online community posts revealed more serious injuries like broken ribs.
only employees, but competition and flexibility pressures necessitated
An interviewee explained:
the recruitment of freelancers.
Those of us who worked long for [DP1], I hardly know anyone who has
Other players operate at significantly lower costs because, among other
not broken one thing or another. I myself have almost been run over by a
things, they do not have the same health and safety requirements and
truck. My bike was destroyed, and I missed death by the skin of my teeth.
employer responsibilities. […] Subsequently, we operate at a greater cost
There are many stories … So far, no one has lost their life or has incurred
due to the operational split. (DP1_Management1)
lasting injuries, but I think it is just a matter of time. (DP1-W17)
Company communication primarily occurs through emails and the
Workers described delivery work as physically demanding, espe
Delivery App’s chat function with courier support (dispatch). The cou
cially for bicycle couriers. Despite awareness of personal injury and
riers established an unofficial Slack workspace dedicated to couriers
accident risks, many young workers indicate a higher preoccupation
(freelancers and employees). Also, they have informal groups in What
with income. The interviews revealed that many immigrants with PMW
sApp for couriers in their respective cities.
as their main income engage in courier work for 40–60 h a week. Some
DP2 couriers are independent contractors, and as a result of the
added having an additional job on the side, such as cleaning. Several
service agreement between DP2 and the courier partner, the Employ
interviewees registered seeing other couriers without helmets or badly
ment Contracts Act and other employment legislation (e.g., Annual
maintained bicycles. For instance, a young freelancer (DP1-FL4) said,
Holidays Act) do not apply. Both platforms have recently contracted
‘Health and safety is not my cup of tea,’ while also admitting to a ‘crazy’
professional occupational health services, but these services are limited
biking style.
to their in-house employees, not the freelance couriers. DP2 pays for
The interviews indicate underreporting, which may be due to their
accident insurance for the freelance couriers, while DP1 does not cover
belief in personal responsibility and a general acceptance of courier
this for their freelancers. Managers from both platforms emphasize that
work risks. Piece rate payment may also discourage reporting, as an
freelance couriers are otherwise responsible for all other insurances (e.
employee indicates, ‘If you stop and report, you earn less money.’ A few
g., statutory pension, social security insurance, liability insurance). Both
DP2 drivers have experienced traffic incidents that did not result in
companies do not include freelance couriers in their safety management
personal injury but resulted in car damage. Due to concerns about its
system, and management underscores freelancers’ responsibility for the
impact on contract renewal, some do not report such incidents.
risks associated with independent contractor activities.
Although both platforms have a reporting system, they underscore that
independent couriers are responsible for their safety. Individual re
3.2. Relegated responsibility - transferring risks to the individual sponsibility is reflected by the lack of requirements from both platforms
for independent contractors to wear helmets during delivery.
Relegated responsibility refers to the transfer of responsibility tradi
tionally held by the organization to ensure safe operations to the worker. 3.3. Algorithmic authority – Algorithms and performance pressures
This section illuminates the consignment of responsibility to the indi
vidual. In PMW, financial gains for the platform do not necessarily The theme algorithmic authority points to the performance pressures
coincide with profitability for the individual. The safety management introduced by the control mechanisms embedded in the technology that
toolbox is also restricted to the core organization in PMW. Those oper is easily reconfigured to add or remove features that affect the local
ating at the outskirts of the organization are left to their own devices in conditions in PMW. The pressure to work efficiently in an ever-changing
determining financial gains from PMW, workload, and safety. context constrained by standardized instructions can push the individual
closer to the safety boundary.
3.2.1. Individual investments
Regardless of the working agreement, employees and independent 3.3.1. Information imbalance and standardized services
contractors alike must individually provide their mode of transportation Our empirical data also revealed information asymmetry, worker
(bike, moped, or car) and communication equipment (smartphone and surveillance, and performance evaluation systems that scholars have
mobile subscription). Avid cyclists were positive about using their bikes, identified with algorithmic management in PMW. The delivery process
stating that theirs are presumably better than the platform would pro is illustrated in Fig. 4.
vide. However, many bike couriers find maintenance costly, especially There is consensus on the App’s centrality in organizing the work
new couriers with little experience performing the maintenance them process, with workers often instructed by dispatchers to ‘follow the
selves. Although the collective bargaining agreement (CBA) of DP1 App.’ The work processes embedded within the technology result in
employees includes additional payment for equipment maintenance, not highly standardized work, with workers describing the job as ‘easy’ in
everyone is convinced it is enough to cover actual costs. cognitive load. Also, many couriers appreciated the freedom of not
6
Fig. 4. Work process in food delivery.
having a supervisor. In this job, they perceive themselves as their ‘own of performance-based staffing (PBS), which tied shift availability to
boss’. courier performance. Interestingly, there was little resistance from the
While customers can see the courier movement on a map in the workers when PBS was introduced, which was explained by the man
Customer App, couriers receive piecemeal information regarding their agement as follows: explained the lack of backlash as follows:
task. DP1 freelancers receive more information than employed couriers
It is a just system that automatically rewards good employees and makes
and can decline requests. Employees are obligated to perform the de
leadership aware of improvement potentials. […] It has given all cyclists a
livery they receive. The decline function has been a cause of resentment
more predictable shift that is better than the previous system. […] It re
for some, saying it leads to inefficiencies when the job is potentially
duces the need to have several employees in an office and reduces our
rerouted to a courier located farther from the pickup location. To
costs. (DP1-Management1)
discourage requests from being declined, updates on the App dished out
15-minute sanctions (inability to receive requests) for consecutive re However, one of the employees of DP1 divulged that one colleague
jections, which also impacted their performance ranking. suffering from anxiety attacks was close to having a ‘nervous break
DP2 has a customer rating system and system for measuring per down.’ In Slack, the shop stewards backed management decisions
formance, although more detailed information regarding these systems arguing that PBS would reduce tardiness and no-shows among the em
and their use was not disclosed. DP1 only has an internal ranking system ployees and decrease freelancers from declining too many times. A shop
based on completed orders per working hours, late login, etc. Some steward assured the employees that they would examine if the changes
employees noted that the evaluation system inaccurately reflected per negatively affected shift availabilities and earnings.
formance as delays can be due to many factors such as slow restaurants Like DP1, DP2 releases available hours two weeks in advance. Cou
or uphill biking difficulties. Although DP1 employees officially cannot riers that manage to grab published shifts receive compensation if no
decline deliveries, some who receive extremely long-distance deliveries orders are received during their pre-scheduled shifts. In addition, DP2
contact dispatch to reroute the delivery to a car. However, the delivery couriers can also log on anytime, similar to Uber drivers. However, they
may end up with another cycling employee due to car unavailability. do not receive any compensation for zero deliveries during these hours.
Some do not bother asking dispatch, worried it would further delay food Like the DP1 couriers, many indicate shift schedules as a source of stress.
delivery. Others performed the long delivery out of respect for other Later interviews with DP2 freelancers revealed that compensated
colleagues. Still, another employee with an IT background said that shifts were greatly reduced to a few hours early morning and very late
accepting the delivery is troublesome as this will teach the platform’s shifts with low demand. Freelancers from both platforms also indicated
artificial intelligence (AI) that it is all right to distribute long-distance that incentives for reaching a certain number of deliveries per week
deliveries to cyclists. were removed. When asked about the changing conditions on shift
The couriers do not have insight into the distribution of orders release and incentives, the consequences of income unpredictability
among available couriers. While some assume it is the courier closest to vary among individual couriers. For DP1 employees, a quiet shift means
the restaurant, others believe proximity is only one factor among many: less additional payment per order but a guaranteed hourly wage. Free
lancers who relied heavily on PMW as primary income expressed
I feel like they are a bit discriminating … we were four or five [DP1]
dissatisfaction with the changes but understood the fluctuating de
workers [freelancers], and there was [an employee] sitting with us in the
mands. One DP2 driver explained that the reduced number of
same circle, and it was a really good sunny day, and we were waiting for
compensated shifts was perhaps due to courier abuse by pre-scheduling
like one and a half hours without any delivery. So it is like one and a half
hours, staying at home, and letting the App reroute requests while still
hours completely wasted, and during that time, I noticed that [the
receiving compensation.
employee] got three deliveries. (DP1-FL1)
3.3.3. Performance pressure

3.3.2. Scheduled shifts and changing conditions The App has a countdown timer showing the estimated delivery
Advanced scheduling allows flexibility for workers and gives the time. The majority of the interviewees explained that they view delivery
company control over courier availability. Shifts or working hours up for time as a guide. DP1 managers emphasized that although speed is a
grabs are released two weeks in advance for both platforms. DP1 em factor in performing efficiently, they consciously do not include speed in
ployees receive their designated contractual hours the same day and their KPIs or communication with workers.
‘grab’ extra shifts. Additional shifts are occasionally listed on ‘Shift Speed seems to be a higher risk factor for freelancers paid per de
swap’ or when demand requires more workers on short notice. Many livery. Among employees, the pressure to deliver fast was more evident
indicated that advanced shift scheduling could be stressful as shifts may among newcomers than seasoned couriers who have learned to pace
disappear quickly. themselves. The piece-rate system appears to have a higher impact on
Later online observations revealed DP1 management’s introduction speed than the estimated time displayed on the App.
7
Yes, one feels that, but I see it as more of a guide saying that you should be about company differences, such as incentives and support (e.g.,
able to do this [within the estimated time], in a way. I do not feel financial support and personal protective equipment) to couriers during
pressure in that sense. The only thing is that I don’t get paid unless I the pandemic. The DP1 WhatsApp group is another site for expressing
deliver. So, if it is very busy and it is Friday night, and there are many frustrations, sharing information, and asking for help (e.g., to borrow a
orders, it pays for me to cycle fast to deliver an extra order within the spare bike). These two external platforms were used for unofficially
hour. (DP1-FL7) swapping shifts, which was especially helpful when the App swapping
function was down. It also helped gather phone screenshots to document
Many have experienced late deliveries to customers without any
distance changes and App issues, such as the paid break becoming an
direct penalties. In many cases, food preparation is a significant
unpaid break. The Slack community also had channels for social events
bottleneck in the delivery process. Although not directly sanctioned,
and informally reporting accidents to safety representatives.
their statistics reflect decreased performance through fewer completed
Members also post information about places they need to be extra
orders per number of working hours.
mindful of, like newly asphalted roads and bike theft areas. Also, they
A DP1 employee also underscored the conflicting objectives in their
use Slack to gather bulk orders and propose issues that shop stewards
work, saying that the prioritization of safety is communicated during
can raise during the next meeting with management. Unlike DP1 cou
recruitment, in push messages regarding extreme weather conditions,
riers, the DP2 WhatsApp city group is inactive. A courier explained the
and in meetings with shop stewards and safety representatives. How
inactivity:
ever, receiving emails containing their recent performance reminds
them of the need for efficient delivery and indirect pressure to increase We had one. We wrote on our WhatsApp group that if we all at the same
speed. time quit on DP2 and we went to the office, and we asked them for more
pay, then maybe they would help. Because if one or two people do that, it
3.4. Solitary service – Social relations in food delivery will not work. If everyone closed their App and went to the office, it could
help. But many were from [country] and [country], and they were
The third and final theme, solitary service, features the isolated scared to lose their job … many disagreed with the situation, and we
worker in PMW. The individualized work structure acts as a stumbling deleted it. (DP2-FL6)
block to forming social relations, worker protection achieved through Despite having a social platform, collective action among the DP2
collective voice, and potential benefits offered by safety culture. workers may have been hindered by their lack of knowledge in the
Nordic working life and fear due to lack of job alternatives. The social
3.4.1. Workers without a workplace relations among the DP2 couriers were primarily due to their back
DP1 and DP2 do not have a place for workers to rest in most of the ground as immigrants from the same countries. Those who do not share
cities they operate. Still, the couriers highlighted the importance of these may feel individualistic tendencies in their work.
having a common place to gather and partake in social exchange. An
employee voiced this need by saying:
4. Discussion
We don’t have anything in terms of physical space … On days like today,
it is not a problem, but if you have a public holiday in winter, the shopping As identified through our analysis, PMW has features that deviate
centers are shut, and then there is nowhere to shelter. Obviously, the from the traditional concept of the organization in three main ways. In
company always wants to get their starting levels at the right levels, at PMW, equipment, maintenance, and safety responsibilities are mainly
least two orders an hour, and then there is no need for shelter in their eyes. relegated by the platforms to the individual. Also, control, reward, and
(DP1-W10) communication systems are highly embedded in the platform technol
ogy. Workers function remotely, managed by the algorithms to perform
For the DP2 drivers interviewed, the car is the workplace. DP2 standardized services. Finally, the service is primarily organized in an
drivers in one city mentioned that they were mostly adult immigrants individualistic fashion, with limited opportunities for socialization.
from developing countries. Their inadequate education and language Although PMW is an extreme case of legal, functional, and social frag
proficiency make landing a permanent job difficult. mentation, workplace fissuring is observable in many sectors. Hence,
organizational fragmentation implies that some influential perspectives
3.4.2. Constricted communication in safety science might lose some of their explanatory power. We discuss
During a shift, the couriers’ main communication line is through a some examples in the following.
live chat with a dispatcher, while face-to-face encounters with fellow
couriers are highly sporadic. For the DP1 hybrid model, shop stewards
and employee safety representatives communicate information from 4.1. PMW and the sociotechnical model
their meetings through Slack and occasionally through email. DP1 em
ployees and freelancers alike found it strange not to have a phone 4.1.1. PMW and STS thinking
number to call headquarters. From a traditional hierarchical structure with distinct levels, some
Many couriers criticized poor management communication. Two present-day and future organizations may constitute flatter structures,
employees from different cities referred to the lack of information as more in line with dynamic networks weakly linked to smaller units (e.g.,
‘Mushroom management – keep them in the dark and feed them bull independent contractors) through digital technology. Our empirical
shit.’ This lack of information was apparent upon learning that PBS was study reveals that in PMW, safety management is consigned to the in
already implemented in another city. The increased delivery distance dividual contractor. Accordingly, workers outside the organization’s
and the PBS implementation ignited an online feud between shop legal boundaries may not be protected by established safety margins and
stewards and dissatisfied employees who believed that the company workplace regulations (Garben, 2017). In delivery work, this is reflected
intended to pressure employees to choose freelancing due to its by multiple freelancer allusions to safety as an individual responsibility,
advantages. coinciding with references to self-determination or being one’s ‘own
boss.’
3.4.3. Creating communities The platform delivery case also underlines the expedited move to
The DP1 Slack community members include employees and free detailed standardization where operations are reduced to discrete tasks
lancers who work for various reasons. Some members are full-time and labor into atomized products to be sold (Almklov & Antonsen,
freelancers of both delivery platforms. Through Slack, couriers learn 2019). Not only does platform technology have the ability to constrain
8
behavior, but it also has become a business strategy in delimiting the income-producing activities. Together with low piece rates that
organization’s responsibilities (Pujadas & Curto-Millet, 2019). encourage long hours of unregulated hard physical labor, independent
The organization’s perpetual struggle to balance safety with pro contractors may find themselves closer to their workload limits. More
duction goals is well-established within safety literature (Hollnagel, over, efficiency pressures from evaluation systems linked to acquiring
2017; Hu et al., 2020). We proposed that in platform-mediated work, future shifts and income move the worker closer to the functional per
where the platform strategy is to reap the benefits of network effects, formance boundary. In the absence of counter gradients applicable to
new conflicting goals emerge (Fig. 5). Economic failure is a common independent workers, couriers face an increasingly limited space of
boundary shared by the employer and the employee since organizational possibilities to safely conduct their activities (Nilsen et al., 2020).
viability is a goal that benefits both parties. However, our empirical Gaining an overview of workload and safety performance is arduous
results demonstrate two new boundaries emerging in PMW: the plat for platform management and safety regulators because of potential
form’s viability and the individual’s economy. The means of production multi-apping, moonlighting, and bifurcated reporting and SMS systems,
and maintenance are relegated to the worker. In addition, low piece excluding independent contractors. Moreover, platform control features
rates and temporary contracts may create greater financial pressures on and piece-rate payment contribute to underreporting even among em
independent contractors than employees receiving an hourly wage and ployees. Safety management systems restricted to employees and mea
open-ended contracts. sures based on accident reporting systems may thus prove to be
Moreover, as the platform expands its market, the increasing number perfunctory actions that merely satisfy regulatory requirements. To
of available workers may exceed market demands for the service, better understand risks in specific types of platform work, independent
leading to fewer tasks and a lower income. Platforms backed by venture contractors need to be included in the management of safety and greater
capitalist funding enable them to lose money as they conquer the market transparency and sharing of data on working hours across platforms are
(Cusumano, 2015; Muller, 2019). Hence, the goal of financial viability in critical.
the platform economy has metamorphosed into a game of market Sociotechnical systems thinking enables a multi-level work and work
domination, where a platform’s win is an independent contractor’s po context analysis by considering the organization’s environment. As seen
tential loss, at least in the shorter run. Our proposal on new conflicting in our study, participation in PMW may be transient, and individuals
goals, therefore, hold. may also be members of more than one organization at a given time.
Although some platforms like DP1 are open to relations similar to Hence, there is a need to acknowledge transitory organizational mem
traditional employer-employee relations, expansion pressures and tight berships (Pasmore et al., 2019). Moreover, the triadic relations and
competition result in a divided management system where freelancers relegated responsibility to individuals in PMW call for broadening our
fall outside organizational responsibilities. Hence, a need for higher- lenses so that our concept of work systems includes individuals at the
order control, such as regulatory or state intervention, may level the fringes of organizations (Cross & Swart, 2021; Winby & Mohrman,
playing field. Kenney and Zysman (2015) point out that social protection 2018). These actors are likely to be in greater need of social and work
based on employer contributions may no longer suffice; hence, the place protection but fall outside safety management paradigms delimi
gaping hole in the social safety net may be mended by a more extensive ted by traditional boundaries of the workplace and employment
welfare system that ensures basic rights and benefits for all, regardless of relations.
organizational affinity.
While the migration model has been applied to accident prevention, 4.2. PMW and safety culture
the model fails to highlight the long-term consequences on health and
safety from prolonged exposure to stressful conditions caused by per Safety culture has been addressed almost exclusively within a
formance within an extremely restricted space of possibilities. The lack traditional work context in safety science. This is not surprising, as ‘the
of job security and income unpredictability, in the long run, can be parent concept’ is organizational culture (Antonsen, 2009a), a concept
deleterious for worker health and well-being (Quinlan, 2015). The in that has been applied mainly to work organizations. Thus, when some
terviews illustrate that PMW also introduces a different concept of central aspects of the organization dissolve, safety culture might lose
workload. The migration model addresses workload as work performed relevance and explanatory power.
for the organization. Thus, it becomes inadequate in considering the We proposed that in PMW, the applicability of management ap
compounded workload through multi-apping or engaging in other proaches to safety culture is significantly reduced, as the work is coor
dinated and controlled by the technology. The empirical findings
correspond to our proposition. Platform relegation of responsibility to
the individual is one aspect of the organization’s dissolution. To some
extent, PMW, from the workers’ perspective, means being your own boss
and being responsible for your own safety. Within the functionalist
perspective on culture, safety culture development is related to man
agement activities, such as providing resources and tools for safety,
solving goal conflicts, and involving employees in safety issues. When
the responsibility for safety is relegated in platform companies, pre
scriptive models like Hudson’s (2007) cultural ladder and Reason’s
(1997) engineering of a safety culture become of limited relevance.
Management in its traditional form and the employer-employee rela
tionship no longer apply.
We also proposed that the opportunities for interaction between
platform workers are reduced, and consequently that the possibility for
cultural development is also reduced. A necessary condition for safety
culture development is interaction and communication between in
dividuals, so that shared values, norms, and perceptions of reality can
develop. The platform workers report that they have limited opportu
nities to interact. In line with our proposition, couriers have no common
physical workspace, and their main line of communication when prob
Fig. 5. Rasumssen’s migration model adapted to platform-mediated work. lems arise is the dispatcher.
9
Although the company receives customer feedback, it may not be work will be more common. Platform-mediated work is an emerging
conveyed to the courier. Several informants also complain about poor work arrangement that breaks with the logic of dyadic employer-
management communication and ‘being kept in the dark.’ App de employee relations and confronts established views on the monolithic
scriptions from the workers indicate features of algorithmic manage concept of organizations at the heart of safety science research.
ment with unclear logic behind performance calculations. Such a system Applying sociotechnical systems thinking, we underscored the limi
rewards blind compliance to the App and provides a breeding ground for tations of the organization’s safety management system, which does not
competitive and individualistic behavior. include independent workers. Using the migration model, we provided
A general finding from the interviews is that there is limited insight into two-sided markets where expanding markets create an
communication between couriers, and to a large extent, one-way oversupply of highly dispersed workers governed by algorithmic con
communication between couriers and the company. The functional trols and face financial risks traditionally belonging to the firm and a
models on safety culture tend to emphasize communication. Hudson’s more compounded concept of workload. Safety management systems
(2007) model highlights communication as key to developing trust and and safety research need to include independent contractors with loose
‘climbing up the cultural ladder.’ The importance of communication and ties to the organizations, especially those who perform PMW due to a
feedback is also highlighted by Reason (1997). lack of alternatives. These individuals may be less capable of balancing
Behind PMW lies a business model designed to reduce interaction safety with other goals and need additional safety nets in the absence of
between individual workers and the company. The responsibility that an employer. Although STS enables multi-level analysis and considers
rests on the employer in traditional ways of organizing work is relegated the work system’s environment, we propose a further broadening of our
to the individual worker. Such a consignment makes functional models lenses to capture the granularities existing within the ecosystem.
on safety culture less relevant for analyzing PMW. As organizational boundaries dissolve, responsibility for risks
There are still some symbols that could foster a perception of com incurred in their operations may also dissolve. PMW is part of the
munity among the couriers. They usually wear an easily recognizable growing trend in the outsourcing of activities as well as safety man
uniform. Also, the App gives some common guidelines and formats for agement to independent contractors. Airline companies, health services,
communication. Such factors could provide a basis for group identity, and other industries are a few examples of the splintering of organiza
even if the high turnover among the couriers makes it reasonable to tions. Thus, other actors such as policymakers, labor institutions, and
assume that it is fragile and unstable. However, we observed some sense non-governmental organizations may need to address issues that emerge
of community in informal online forums such as WhatsApp and Slack, within the work ecosystem explicitly. If platform companies are un
where experiences on work and the companies were exchanged. willing to take on this responsibility, how will it be distributed? In most
Interpretive perspectives of safety culture might thus be better suited cases, leaving it to individual platform workers exposed to conflicting
for studying PMW. According to Smircich (1983), such perspectives are pressures is not a viable solution. Both nationally and internationally,
grounded in anthropological approaches to culture, in essence viewing unions, NGOs, and public agencies need to find ways to address HSE
cultural development as a bottom-up process. Culture is used as a root issues and the social protection of these groups.
metaphor for the organization, i.e., something that an organization is Accelerated by technology, the organization’s continued fragmen
(rather than has in the functional view). For instance, the symbolic tation confronts us with the relevance of functional views on safety
perspective sees cultures as systems of shared symbols and meanings. culture and communities of practice within its boundaries. Our study
Analyzing culture involves interpreting such symbols and their under indicates that functionalist safety culture models and approaches may
lying themes. The cognitive perspective involves seeing cultures as have limited applicability in work settings where infrastructures
distinct systems of shared cognition, knowledge, and beliefs. The cul constrain social relations and social ties and interactions are transitory,
tural analysis then explores the members’ world views and common but interpretive approaches to culture are interesting to pursue.
logic and rules valid in a community. Still, technology also offers avenues for studying safety in PMW. In
Thus, culture as a concept is interesting as an analytical concept, but the absence of a workplace, observing social interactions and commu
more in a descriptive rather than normative way. What characterizes nities of practice may require specific forms of ethnographic studies,
platform worker communities as cultures? Are there any bottom-up such as netnography (Kozinets, 2010). Safety science can potentially
constructed common beliefs and world views developing in informal contribute to investigating the potential for technoregulation (Lettieri
arenas that have implications for safety? To answer such questions will et al., 2019; Pagallo & Durante, 2016) and universal social protection
require ethnographic research strategies and anthropological ap (Dijkhoff, 2019), as well as coming up with ways to enrich the safety
proaches to culture. Our interviews and observations point toward some regulatorý s toolbox through digital technology. Though legally chal
elements of a community of practice among platform workers, where lenging given the platform providers’ secretive business models, one
they, primarily online, share experiences among their peers. This can be such possibility is to draw on the extensive amounts of data collected by
seen as a kernel that could improve their work situation. If we look at the apps regulating the PMW and possibly utilize the information
ethnographic literature, there are some similarities to be seen with Orr’s infrastructure to implement better worker protection. Safety science
(1996) copy machine maintenance workers, exchanging experience researchers can also investigate risk governance beyond national bor
through radios, and Palsson and Helgason (1998) fishing boat captains. ders and provide foundations for developing holistic policies to protect
Both are examples of communities of practitioners not working physi workers in a fissuring workplace.
cally together but still sharing experiences. The online communities we Developing safety theories and addressing issues at the nexus be
observed were not as developed as these examples but could indicate tween humans, technology, organizations, and society will benefit from
that there might be a way forward, e.g., for unions or authorities to broader sociotechnical systems thinking that includes independent
foster a more cultural or practice-oriented community among the contractors operating outside the organizational boundaries. By
workers. focusing on humans, regardless of membership and spatial distribution,
we can truly address contemporary work issues and move towards the
5. Conclusion goal of safety in‘all fields of human activity’(Boustras et al., 2020).
While safety science can contribute to alleviating some of the downsides
This paper has demonstrated that influential organizational per or symptoms of the platform economy for the workers in the most pre
spectives in safety science have the traditional organization as their carious situations, the main tasks lie with authorities and regulators to
context and the resulting limitations necessitate some adaptations to ensure more general sustainability of the work situation and lives of the
emerging work arrangements such as PMW. As technology further en platform workers.
ables the performance of functions, network-based ways of organizing
10
CRediT authorship contribution statement Garben, S., 2017. Protecting Workers in the Online Platform Economy: An overview of
regulatory and policy developments in the EU. Publications Office of the European
Union, Luxembourg.
Marie Nilsen: Conceptualization, Methodology, Formal analysis, Glendon, A.I., Stanton, N.A., 2000. Perspectives on safety culture. Safety Science 34 (1),
Investigation, Resources, Writing – original draft, Writing – review & 193–214. https://doi.org/10.1016/S0925-7535(00)00013-8.
editing, Visualization, Project administration. Trond Kongsvik: Super Goodwin, T., 2015, March 4, 2015. The Battle Is For The Customer Interface. Retrieved
from https://techcrunch.com/2015/03/03/in-the-age-of-disintermediation-the-battl
vision, Conceptualization, Methodology, Investigation, Validation, e-is-all-for-the-customer-interface/?guccounter=2.
Writing – original draft, Writing – review & editing. Petter Grytten Graham, M., Hjorth, I., Lehdonvirta, V., 2017. Digital labour and development: impacts
Almklov: Conceptualization, Investigation, Validation, Writing – orig of global digital labour platforms and the gig economy on worker livelihoods.
Transfer 23 (2), 135–162. https://doi.org/10.1177/1024258916687250.
inal draft, Writing – review & editing. Griesbach, K., Reich, A., Elliott-Negri, L., Milkman, R., 2019. Algorithmic control in
platform food delivery work. Socius 5, 1–15.
Declaration of Competing Interest Guldenmund, F.W., 2000. The nature of safety culture: a review of theory and research.
Saf. Sci. 34 (1-3), 215–257.
Hollnagel, E., 2017. The ETTO Principle: Efficiency-Throoughness Trade-Off: Why
The authors declare that they have no known competing financial Things That Go Right Sometimes Go Wrong. CRC Press.
interests or personal relationships that could have appeared to influence Horst, H., Miller, D., 2012. Normativity and materiality: A view from digital
anthropology. Media Int. Australia Incorporating Cult. Policy 145 (1), 103–111.
the work reported in this paper. https://doi.org/10.1177/1329878X1214500112.
Hu, X., Casey, T., Griffin, M., 2020. You can have your cake and eat it too: Embracing
References paradox of safety as source of progress in safety science. Saf. Sci. 130, 104824.
https://doi.org/10.1016/j.ssci.2020.104824.
Hudson, P., 2007. Implementing a safety culture in a major multi-national. Saf. Sci. 45
Almklov, P.G., Antonsen, S., 2010. The commoditization of societal safety.
(6), 697–722. https://doi.org/10.1016/j.ssci.2007.04.005.
J. Contingencies Crisis Manage. 18 (3), 132–144. https://doi.org/10.1111/j.1468-
Huws, U., Spencer, N., Syrdal, D.S., Holts, K., 2017. Work in the European gig economy:
5973.2010.00610.x.
research results from the UK, Sweden, Germany, Austria, the Netherlands,
Almklov, P. G., & Antonsen, S. (2019). Standardisation and Digitalisation: Changes in
Switzerland and Italy. Retrieved from https://uhra.herts.ac.uk/bitstream/handle/22
Work as Imagined and What This Means for Safety Science. Safety Sci. Res.: Evolut.,
99/19922/Huws_U._Spencer_N.H._Syrdal_D.S._Holt_K._2017_.pdf.
Challenges New Directions 1.
Kausel, C.L., Hackett, P.M., 2015. Netnography. In: Hackett, P.M. (Ed.), Qualitative
Almklov, P.G., Rosness, R., Størkersen, K., 2014. When safety science meets the
Research Methods in Consumer Psychology: Ethnography and Culture, Vol. 2, pp.
practitioners: Does safety science contribute to marginalization of practical
252.
knowledge? Saf. Sci. 67, 25–36. https://doi.org/10.1016/j.ssci.2013.08.025.
Kenney, M., Zysman, J., 2015. Choosing a future in the platform economy: the
Antonsen, S., 2009a. Safety Culture : Theory, Method and Improvement. Ashgate,
implications and consequences of digital platforms. In: Paper presented at the
Farnham.
Kauffman Foundation New Entrepreneurial Growth Conference.
Antonsen, S., 2009b. Safety culture and the issue of power. Saf. Sci. 47 (2), 183–191.
Kenney, M., Zysman, J., 2016. The rise of the platform economy. Issues Sci. Technol. 32
https://doi.org/10.1016/j.ssci.2008.02.004.
(3), 61–69.
Anyfantis, I.D., Boustras, G., 2020. The effects of part-time employment and employment
Kongsvik, T., Albrechtsen, E., Antonsen, S., Herrera, I., Hovden, J., Schiefloe, P.M., 2018.
in rotating periods on occupational accidents. The case of Greece. Saf. Sci. 121, 1–4.
Sikkerhet i arbeidslivet. Fagbokforl, Bergen.
Kongsvik, T., Fenstad, J., Wendelborg, C., 2012. Between a rock and a hard place:
Bajwa, U., Gastaldo, D., Di Ruggiero, E., Knorr, L., 2018. The health of workers in the
Accident and near-miss reporting on offshore service vessels. Saf. Sci. 50 (9),
global gig economy. Globalization Health 14 (1), 124.
1839–1846. https://doi.org/10.1016/j.ssci.2012.02.003.
Bang, H., 2011. Organisasjonskultur. Universitetsforlaget, Oslo.
Kozinets, R.V., 2010. Netnography : Doing Ethnographic Research Online. Sage, Los
Biernacki, P., Waldorf, D., 1981. Snowball sampling: Problems and techniques of chain
Angeles.
referral sampling. Sociol. Methods Res. 10 (2), 141–163.
Lave, J., Wenger, E., 1991. Situated Learning: Legitimate Peripheral Participation.
Bisbey, T.M., Kilcullen, M.P., Thomas, E.J., Ottosen, M.J., Tsao, K., Salas, E., 2021. Safety
Cambridge University Press.
culture: an integration of existing models and a framework for understanding its
Le Coze, J.-C., 2019. How safety culture can make us think. Saf. Sci. 118, 221–229.
development. Hum Factors 63 (1), 88–110. https://doi.org/10.1177/
0018720819868878.
Lee, M.K., Kusbit, D., Metsky, E., Dabbish, L., 2015. Working with machines: The impact
Boudreau, F.A., Newman, M., 1993. Understanding Social Life: An Introduction to
of algorithmic and data-driven management on human workers. In: Paper presented
Sociology. West Publishing Company, Minneapolis.
at the Proceedings of the 33rd Annual ACM Conference on Human Factors in
Boustras, G., Grote, G., Reniers, G., Maiti, J., Cozzani, V., Jain, A., Newnam, S.,
Computing Systems.
Aneziri, O., Le-Coze, J.C., Arne, P.G.K., 2020. Safety science new scope. Saf. Sci. 121,
Lettieri, N., Guarino, A., Malandrino, D., Zaccagnino, R., 2019. Platform economy and
651. https://doi.org/10.1016/j.ssci.2019.08.011.
techno-regulation-Experimenting with reputation and nudge. Future Internet 11 (7),
Carayon, P., Hancock, P., Leveson, N., Noy, I., Sznelwar, L., van Hootegem, G., 2015.
163. https://doi.org/10.3390/fi11070163.
Advancing a sociotechnical systems approach to workplace safety - developing the
Leveson, N., 2004. A new accident model for engineering safer systems. Saf. Sci. 42 (4),
conceptual framework. Ergonomics 58 (4), 548–564. https://doi.org/10.1080/
237–270. https://doi.org/10.1016/S0925-7535(03)00047-X.
00140139.2015.1015623.
Leveson, N., 2017. Rasmussen’s legacy: A paradigm change in engineering for safety.
Cherry, M.A., Aloisi, A., 2016. Dependent contractors in the gig economy: a comparative
Appl. Ergon. 59 (Pt B), 581–591. https://doi.org/10.1016/j.apergo.2016.01.015.
approach. Am. UL Rev. 66, 635.
Mintzberg, H., 1983. Structure in Fives: Designing Effective Organizations. Prentice-Hall,
Christie, N., Ward, H., 2019. The health and safety risks for people who drive for work in
Englewood Cliffs, N.J.
the gig economy. J. Trans. Health 13, 115–127. https://doi.org/10.1016/j.
Moorkamp, M., Kramer, E.-H., van Gulijk, C., Ale, B., 2014. Safety management theory
jth.2019.02.007.
and the expeditionary organization: A critical theoretical reflection. Saf. Sci. 69,
Cross, D., Swart, J., 2021. The (ir)relevance of human resource management in
71–81. https://doi.org/10.1016/j.ssci.2014.05.014.
independent work: Challenging assumptions. Human Resour. Manage. J. https://doi.
Muller, Z., 2019. Algorithmic Harms to Workers in the Platform Economy: The Case of
org/10.1111/1748-8583.12389.
Uber. Colum. JL & Soc. Probs. 53, 167.
Cusumano, M.A., 2015. Technology strategy and management how traditional firms
Neal, A., Griffin, M.A., Hart, P.M., 2000. The impact of organizational climate on safety
must compete in the sharing economy. Commun. ACM 58 (1), 32–34. https://doi.
climate and individual behavior. Saf. Sci. 34 (1), 99–109. https://doi.org/10.1016/
org/10.1145/2688487.
S0925-7535(00)00008-4.
Dahl, Ø., Kongsvik, T., 2018. Safety climate and mindful safety practices in the oil and
Nilsen, M., Kongsvik, T., Antonsen, S., 2020. Working Conditions and Safety in the Gig
gas industry. J. Saf. Res. 64, 29–36. https://doi.org/10.1016/j.jsr.2017.12.009.
Economy - A Media Coverage Analysis. In: Paper presented at the e-proceedings of
De Stefano, V.M., 2016. The Rise of the ‘Just-in-Time Workforce’: On-Demand Work,
the 30th European Safety and Reliability Conference and 15th Probabilistic Safety
Crowd Work and Labour Protection in the ‘Gig-Economy’. Comparative Labor Law
Assessment and Management Conference (ESREL2020 PSAM15), Venice, Italy.
Policy J. 37 (3), 471–504.
Orr, J.E., 1996. Talking about machines: an ethnography of a modern job.
Degryse, C., 2016. Digitalisation of the Economy and Its Impact on Labour Markets. ETUI
Pagallo, U., Durante, M., 2016. The Pros and Cons of Legal Automation and its
Research Paper-Working Paper.
Governance. Eur. J. Risk Regul. 7 (2), 323–334. https://doi.org/10.1017/
Dijkhoff, A.A., 2019. Universal Social Protection for Human Dignity, Social Justice and
S1867299X00005742.
Sustainable Development: General Survey concerning the Social Protection Floors
Palsson, G., Helgason, A., 1998. Schooling and skipperhood: the development of
Recommendation, Vol. 2012. International Labour Office.
dexterity. Am. Anthropol. 100 (4), 908–923. https://doi.org/10.1525/
Eisenhardt, K.M., 1989. Agency theory: an assessment and review. Acad. Manage. Rev.
aa.1998.100.4.908.
14 (1), 57–74. https://doi.org/10.2307/258191.
Pasmore, W., Winby, S., Mohrman, S.A., Vanasse, R., 2019. Reflections: sociotechnical
Eurofound, 2018. Employment and working conditions of selected types of platform
systems design and organization change. J. Change Manage. 19 (2), 67–85. https://
work (9789289717489,9289717483). Retrieved from Luxembourg.
doi.org/10.1080/14697017.2018.1553761.
Fleming, P., Rhodes, C., Yu, K.-H., 2019. On why Uber has not taken over the world.
Pesole, A., Brancati, U., Fernández-Macías, E., Biagi, F., González Vázquez, I., 2018.
Econ. Soc. 48 (4), 488–509. https://doi.org/10.1080/03085147.2019.1685744.
Platform Workers in Europe. EUR29275 EN, Luxembourg: Publications Office of the
European Union. doi: 10.2760/742789.
11
Pujadas, R., Curto-Millet, D., 2019. From Matchmaking to Boundary Making: Thinking Sheridan, T.B., 2017. Musings on models and the genius of Jens Rasmussen. Appl. Ergon.
Infrastructures and Decentring Digital Platforms in the Sharing Economy, Vol. 62. 59 (Pt B), 598–601. https://doi.org/10.1016/j.apergo.2015.10.015.
Emerald Publishing Limited, pp. 273-286. Smircich, L., 1983. Concepts of culture and organizational analysis. Adm. Sci. Q. 28 (3),
Quinlan, M., 2015. The effects of non-standard forms of employment on worker health 339–358.
and safety: ILO Geneva, Switzerland. Tjora, A.H., 2019. Qualitative Research as Stepwise-Deductive Induction, Vol. 26.
Rasmussen, J., 1997. Risk management in a dynamic society: a modelling problem. Saf. Routledge, Abingdon, Oxon.
Sci. 27 (2-3), 183–213. Tran, M., Sokas, R.K., 2017. The gig economy and contingent work: an occupational
Reason, J., 1997. Managing the Risks of Organizational Accidents. Ashgate, Aldershot. health assessment. J. Occup. Environ. Med. 59 (4), e63–e66. https://doi.org/
Saldaña, J., 2016. The Coding Manual for Qualitative Researchers, 3rd ed. Sage, London. 10.1097/JOM.0000000000000977.
Schein, E.H., 1985. Organizational Culture and Leadership. Jossey-Bass, San Francisco. Waterson, P., Le Coze, J.-C., Andersen, H.B., 2017. Recurring themes in the legacy of
Schmidt, F.A., 2017. Digital labour markets in the platform economy: Mapping the Jens Rasmussen. Appl. Ergon. 59 (Pt B), 471–482. https://doi.org/10.1016/j.
political challenges of crowd work and gig work. Friedrich-Ebert-Stiftung, Bonn. apergo.2016.10.002.
Schulman, P.R., 2020. Organizational structure and safety culture: Conceptual and Weil, D., 2014. The Fissured Workplace. Harvard University Press.
practical challenges. Saf. Sci. 126, 104669. https://doi.org/10.1016/j. Winby, S., Mohrman, S.A., 2018. Digital sociotechnical system design. J. Appl. Behav.
ssci.2020.104669. Sci. 54 (4), 399–423. https://doi.org/10.1177/0021886318781581.
Selznick, P., 2014. The Organizational Weapon: A Study of Bolshevik Strategy and Yin, R.K., 2010. Qualitative Research from Start to Finish. Guilford Publications, New
Tactics, Vol. 18. Quid Pro Books. York.
12

Safety Science: Marie Nilsen, Trond Kongsvik, Petter Grytten Almklov

Uploaded by

Copyright:

Available Formats

Safety Science: Marie Nilsen, Trond Kongsvik, Petter Grytten Almklov

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Safety Science: Marie Nilsen, Trond Kongsvik, Petter Grytten Almklov

Uploaded by

Copyright:

Available Formats

Safety Science 148 (2022) 105644

Contents lists available at ScienceDirect

Splintered structures and workers without a workplace: How should safety

1. Introduction inadequacies in training and safety management, psychosocial risks,

Fig. 3. Data collection - interviews and observations.

Fig. 4. Work process in food delivery.

3.3.3. Performance pressure

You might also like