Content uploaded by Florian Alexander Schmidt
Author content
All content in this area was uploaded by Florian Alexander Schmidt on Mar 12, 2017
Content may be subject to copyright.
good society –
social democracy
# 2017 plus
Florian A. Schmidt
Digital Labour Markets in the
Platform Economy
Mapping the Political Challenges of Crowd Work
and Gig Work
What is a Good Society? For us this includes social justice, environmental sustainability, an innovative
and successful economy and an active participatory democracy. The Good Society is supported by the
fundamental values of freedom, justice and solidarity. We need new ideas and concepts to ensure that
the Good Society will become reality. For these reasons the Friedrich-Ebert-Stiftung is developing specific
policy recommendations for the coming years. The focus rests on the following topics:
– A debate about the fundamental values: freedom, justice and solidarity;
– Democracy and democratic participation;
– New growth and a proactive economic and financial policy;
– Decent work and social progress.
The Good Society does not simply evolve; it has to be continuously shaped by all of us. For this project the
Friedrich Ebert Stiftung uses its international network with the intention to combine German, European and
international perspectives. With numerous publications and events between 2015 and 2017 the Friedrich-
Ebert-Stiftung will concentrate on the task of outlining the way to a Good Society.
For more information on the project:
www.fes-2017plus.de
Friedrich-Ebert-Stiftung
The Friedrich-Ebert-Stiftung (FES) is the oldest political foundation in Germany with a rich tradition dating
back to its foundation in 1925. Today, it remains loyal to the legacy of its namesake and campaigns for
the core ideas and values of social democracy: freedom, justice and solidarity. It has a close connection
to social democracy and free trade unions.
FES promotes the advancement of social democracy, in particular by:
– political educational work to strengthen civil society;
– think tanks;
– international cooperation with our international network of offices in more than 100 countries:
– support for talented young people;
– maintaining the collective memory of social democracy with archives, libraries and more.
About the author
Dr. phil Florian A. Schmidt is a design researcher and journalist. In his doctoral thesis at the Royal College
of Art in London he analysed the methods of digital labour platforms for creative crowdwork.
For the German labour union IG Metall he co-developed the website FairCrowdWork.org.
Person responsible in the FES for this publication
Dr. Robert Philipps, head of the FES working group on small and medium-sized enterprises and of the
discussion group on consumer policy, Department of Economic and Social Policy.
A PROJECT BY THE
FRIEDRICH-EBERT-STIFTUNG
2015 – 2017
good society –
social democracy
# 2017 plus
FRIEDRICH-EBERT-STIFTUNG
PREFACE
1 INTRODUCTION
2 ABSTRACT
3 ANALYSIS: LABOUR MARKETS IN THE PLATFORM ECONOMY
3.1. Platforms for Web-based Services (Cloud Work)
3.1.1 Freelance Marketplaces
3.1.2 Microtasking Crowd Work
3.1.3 Contest-based Creative Crowd Work
3.2. Gig work (Location-based Digital Labour)
3.2.1 Accommodation
3.2.2 Transportation and Delivery Services
3.2.3 Household Services and Personal Services
4 OUTLOOK
List of Figures
Bibliography
Florian A. Schmidt
Digital Labour Markets in the
Platform Economy
Mapping the Political Challenges of Crowd Work
and Gig Work
2
3
5
9
14
14
15
16
18
19
20
22
23
26
26
FRIEDRICH-EBERT-STIFTUNG
2
FRIEDRICH-EBERT-STIFTUNG
The so-called "sharing economy" is a much-debated topic.
Uber, Airbnb, Helpling and many other platform-based busi-
ness models want to "disrupt" industries that they portray as
"ossified" and user-unfriendly. The contenders claim to cre-
ate new services that are more flexible and cost-efficient.
And indeed, the new platforms, which serve as an interme-
diary between supply and demand, have many valuable con-
tributions to offer: they provide access to goods and ser-
vices across the world within seconds; they lower transaction
costs as well as expenditure for the allocation of resources;
they enable a multitude of new services; and they are a driv-
ing force for economic innovation.
Nevertheless, the new platforms are increasingly beset by
criticism. Usually they rely on a workforce of inde-pendent
contractors, who work on their own account and at their own
risk, for low wages and without social security. Neither the
platform providers nor their clients take on the role and re-
sponsibilities of an employer. Labour laws, worker protection,
health and safety regulations, quality of work and social se-
curity contributions mostly fall to the responsibility of the
independent contractors alone, who are also not entitled to
the kind of workers’ participation common in other sectors.
The clients of the platforms essentially gain access to an
on-demand workforce, while the independent contractors
who provide the labour are subject to precarious working
conditions.
Against this background, the Friedrich-Ebert-Stiftung (FES)
decided to have a closer look at platform-based digital busi-
ness models and their implications for the economy and for
society. While digital platforms have come to play a role in
many branches of the economy, the present publication is
focussed on the socio-politically most contested ones, namely
digital labour platforms. The study explains the basic mech-
anisms of three-sided digital labour markets and compares
its variants and subcategories. It also explains the specific
features and challenges of the different categories and pro-
poses starting points for political measures.
We wish you an interesting read!
DR. ROBERT PHILIPPS
Head of FES’s SME working group and consumer policy
discussion group
Department of Economic and Social Policy
PREFACE
3
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
ogisms entirely. Instead, this study takes the free-floating
terminology and organises it along the lines of what is most
pertinent for a political debate and the question of potential
regulatory measures. In particular, the study distinguishes
between "cloud work", "crowd work" and "gig work" as the
three most important categories of digital labour. In order to
understand the shifting labour landscape and take an active
role in designing the future of work, it is furthermore nec-
essary to look at these phenomena not in isolation but in the
context of other platform-based business models, and also
to recognise them as just the latest digital stage in a long
on-going development towards more flexible, temporary
and tentative forms of labour, with analogue predecessors
in outsourcing companies and temporary employment
agencies.
The new digital labour markets claim to be flexible, lean,
and cost-efficient, for both their clients and their independ-
ent contractors. And cloud work, crowd work and gig work
do indeed offer more and more people an attractive alterna-
tive to conventional full-time employment, a self-determined
way of working – when, where, how, for whom and on what-
ever they want. However, this new flexibility often goes hand
in hand with precarious working conditions and undermines
hard-won legal and social standards of good work. The new
platform-based business models portray themselves as the
future of work and political terms such as "Arbeit 4.0" (liter-
ally: "Work 4.0", a term used frequently by German politicians)
support this air of progressiveness. Nonetheless, with regard
to workers’ rights and social security, it seems that the new
platforms instead represent a regression to the times of the
early Industrial Revolution. This leads to the question of what
can and should be done at the policy level. To what extent
can regulatory measures protect workers’ rights from being
further diminished? How can it be ensured that the profits
(or rents) made in the platform economy do not exclusively
benefit venture capitalists and platform providers, but also
those who do the actual work and, more importantly, bear
the brunt of the entrepreneurial risk: in other words, the
cloud workers, crowd workers and gig workers? And how
can the on-demand digital labour model of the platform
economy be prevented from causing damage to the public
The so-called "sharing economy" is gaining momentum. As
of 2016, Airbnb is valued at US$25.5 billion, while Uber is
valued at US$62.5 billion. The two companies, which are pre-
sented as engaged in the "sharing of underutilised assets" –
the commercial brokering of accommodation and of transpor-
tation, respectively – are now among the most valuable start-
ups on the market. Their massive accumulation of venture
capital is driven by the investors’ hope for new forms of value
creation through the "disruption" of existing business models,
which are often portrayed as ossified, overregulated and inef-
ficient. In contrast to what the term "sharing economy" sug-
gests, however, the large digital platforms in this area are
not based primarily on the sharing of common goods but on
the commercial coordination of various services offered by
private individuals. This development gives employers access
to a huge on-demand-workforce and is leading to a shift in
the structure of labour markets. The emerging business mod-
els of what is best described as the "platform economy" rely
on private individuals who, as independent contractors, carry
out small jobs in their free time; an army of more or less pre-
carious workers who can be hired or fired in an instant. Lit-
erally thousands of digital platforms for the commercial co-
ordination of digital labour have emerged in recent years.
However, at this point it is still uncertain how many of them
are economically viable in the long run, and to what extent
the new types of jobs will replace more conventional forms
of employment.
Unfortunately, the discourse on platform-based digital
labour often suffers from inconsistencies in the use of termi-
nology and confusion in the categorisation of different plat-
form types. In order to address at the appropriate level the
multiple challenges our labour markets are faced with, it is
important to differentiate between the new business mod-
els and to use a terminology that reflects this differentiation.
The problem is not only a confusion of the different methods
used by the digital labour platforms, but also the fact that
the language used to describe them is dominated by mar-
keting terms. People in the field commonly speak of "Turk-
ers", "HITs", "awards" and the "cloud" instead of independ-
ent contractors, jobs, payment and someone else’s data
centre. However, it doesn’t seem practical to avoid the neol-
1
INTRODUCTION
4
FRIEDRICH-EBERT-STIFTUNG
good? After all, it is society that has to carry the social costs
of all the precarious workers in the long run.
The quickly evolving platform-based reorganisation of
work comes with a whole set of opportunities and risks. First,
this demands a precise terminology and analysis of its func-
tionalities. Secondly, a broad public debate is needed about
what values society deems worth protecting and what consti-
tutes decent work. Finally, it is the role of government to en-
force the agreed-upon values through regulations and en-
sure that labour laws also apply in the digital realm.
However, due to the structure of the platforms, this will be
a serious challenge. The present publication offers first and
foremost a categorisation of the different types of commercial
digital labour platforms. It also discusses the particular char-
acteristics and challenges of the different categories and
maps how to tackle them politically.
5
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
The taxonomy suggested here is as follows: if the task is not
location-based and can be done remotely via the internet, it
is cloud work. If the task is not given to a specific individual
but to an undefined group of people online, it is crowd work.
If the task is further subdivided into tiny units for piecemeal
work, each paid for with an equally tiny amount of money, it
is microtasking crowd work. If in contrary the task cannot be
subdivided but is solved in a redundant fashion, in parallel,
by an entire crowd, while in the end only one result is used
and paid for, it is contest-based crowd work. However, when
a task has to be done at a specific location and time, by a
specific person that is responsible for task, it is gig work.
These location-based services are further differentiated by
the degree of personal involvement necessary and the de-
gree of opportunities and risks they entail for the independ-
ent contractor. As a result, one arrives at the following six
basic types of digital labour platform.
Cloud work (web-based digital labour)
1. freelance marketplaces
2. microtasking crowd work
3. contest-based creative crowd work
Gig work (location-based digital labour)
4. accommodation
5. transportation and delivery services (gig work)
6. household services and personal services (gig work)
In practice, there are many hybrid platforms and further sub-
categories. The categorisation suggested here is as simple as
possible and as differentiated as necessary. The order of the
six platform types listed above roughly represents the his-
torical succession in which they have emerged. Within the
first of three categories, a substantial market consolidation
has already happened and a lot of research has already been
done. The last three categories are more recent and espe-
cially the last one is still in the making and, as yet, not well
understood. More research is needed, especially on the new-
er types of digital labour platforms, and a more refined
differentiation of the types might become necessary in the
future.
This policy paper offers a model for the categorisation of
digital labour platforms, commercial providers of an on-de-
mand workforce that consists mainly of private individuals
trying to generate an additional income. Commons-based
peer production and non-profit projects such as Wikipedia,
OpenStreetMap and CouchSurfing, which are based on actu-
al sharing, are excluded from this analysis. Academically, they
have to be dealt with separately, and politically, they should
be supported as non-commercial alternatives. Commercial
retail platforms such as eBay and Amazon, app-stores, search
engines, social networks and straightforward B2B-platforms
are also beyond the scope of this study. However, the digital
labour platforms are analysed as part of the much larger
platform economy. The platform economy in general can be
characterised as follows: it consists of online marketplaces
that involve at least three parties. The platform provider ser-
ves as an intermediary that coordinates the supply and de-
mand sides of the other two parties. This role as intermedi-
ary allows the platform provider to shift most of the costs,
risks and liabilities to the other two parties. Typically, the
platform provider does not have to cover the cost of labour
or the means of production. The platform provider offers an
entirely virtual service (just an app or a website) and can
thus grow exponentially, without having to face production
costs growing proportionally as well (very low marginal
costs). The platform provider is also the only one of the three
parties that has full access to and control over the data, pro-
cesses and rules of the platform. The particular software ar-
chitecture of the digital platform causes a systemic infor-
mation asymmetry and, through that, a power asymmetry.
Driven by venture capital and network effects, the platform
economy is prone to foster the emergence of monopolies
or at least oligarchies.
Two questions are essential for the categorisation of
digital labour platforms: are the services and tasks coordi-
nated via the platform bound to a specific location? And
are these services and tasks bound to a specific person?
Both aspects have far-reaching implications for how the
platforms operate, the situation of the independent con-
tractors, the legal framework that applies and potential
regulatory measures.
2
ABSTRACT
6
FRIEDRICH-EBERT-STIFTUNG
Figure 1
Categorisation of digital labour markets
in the platform economy 1/2
lmportant factors across all platform types are the emergence of monopolies, network effects, biased terms of service, lack of transparency,
permanent tracking and rating of user behaviour, and lack of data protection – all of which have problematic consequences for digital labour
platforms in particular.
* Because of its many structural similarities Airbnb is treated here as part of gig work, even though the role of labour is secondary on this
particular platform.
Source: ow n research.
money
goods
commercial
digital platforms
in general
services
(digital labour)
cloud work
(web -base d)
gig work
(location-based)
crowdfunding
tangible, for sale
datlng
tangible, for rent
social media
intangible, for sale
news
intangible, for rent
search
reviews
Amazon.com
eBay
etsy
communication
entertainment
information
leihdirwas.de
App-Store
iTu nes
Spotify
Netflix
Upwork
Amazon MTurk
99designs
Uber
Airbnb
Helpling
Indiegogo
Klickstarter
Tinder
Facebook
Youtube
Google News
Google Search
Yel p
Airbnb*
7
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
Figure 2
Categorisation of digital labour markets
in the platform economy 2/2
To differentiale between digital labour platforms one has to ask: is the work bound to a specific place? And is the werk bound to a specific individual?
lf it can be done from everywhere, it is cloud work . lf it can be done by anyone and is given to an unspecific group, it is crowd werk. Freelance market-
places are therefore cloud werk but not crowd work. lf the werk has to be done at a specific location and is given to one selec ted individual, it is
gig work . Local microtasking is the only form of gig work given to a crowd.
Source: ow n research.
tasks given to
selected individuals
tasks given to
selected individuals
freelance
marketplaces
micro tasking
crowd work
contest-based
creative crowd work
accommodation
transportation
deliver y
household services
local microtasking
commercial
digital labour
platforms
cloud work
(web -base d)
gig work
(location-based)
Upwork
Freelancer.com
Uber
Lyf t
Lieferando
Instacart
Tas k ra b bi t
Helpling
Kaufmich.com
App-Jobber
Streetspotr
Clickworker
Amazon MTurk
Crowdflower
99designs
Jovoto
Quirky
Airbnb
tasks given to
crowd
tasks given to
crowd
8
FRIEDRICH-EBERT-STIFTUNG
In all six categories we see political challenges with regard to
issues such as privacy, data protection, labour laws, fair pay
and the mechanisms of "algorithmic management" (the auto-
mated rating and tracking of independent contractors). The
digital labour platforms for services that are not bound to a
specific location (cloud work) – and of those especially the
two forms of crowd work – are particularly difficult to regu-
late because it is not always clear which national legal stand-
ards apply if all three groups of stakeholders reside in differ-
ent countries; this is a tricky question, especially in relation
to the minimum wage. It is even questionable whether
crowd work in its core sense is at all structurally compatible
with a minimum wage or if regulatory measures with that
goal would inevitably cause crowd work platforms to be
transformed into freelance marketplaces, which would in turn
be characterised by a much higher degree of worker sur-
veillance.
Platforms for the outsourcing of location-based tasks (gig
work) have turned out to be particularly disruptive because
they affect a larger percentage of the workforce and much
more capital in the form of physical assets is involved. On
these platforms the risk of work accidents and potential harm
to people and property is, of course, more pressing than on
the web-based counterparts. As a consequence, the question
of workers’ compensation and liability insurance becomes im-
portant here. Furthermore, a lot more sensitive personal data
are collected by the location-based services, as the gig work-
ers (and, in the case of Uber, sometimes even the clients) are
tracked via their smartphones. At the same time, the three
types of gig work platforms operate on the level of cities and
are much more visible than the web-based services. The gig
work platforms clearly fall under the local legal framework;
hence regulations are much more easily accomplished in these
three groups and are already quite advanced in many jurisdic-
tions. Also, the self-organisation of independent contractors,
as well as the development of more socially spirited non-prof-
its and platform cooperatives seem to be more promising for
location-based services.
9
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
Platform-based business models have permeated many
areas of society and commerce, from retailers of physical
goods, via streaming services for music, film and video, to
dating sites and apps. Digital platforms make the lives of
millions of people easier and we can hardly imagine mod-
ern life without them any longer. Because they are so use-
ful, important and omnipresent, a critical analysis of how
they function is important. All the more so now that also la-
bour markets are increasingly organised via digital platforms.
Because these platforms operate on an international level,
they pose a serious challenge for legislation, which usually
operates only at a national level. The organisation of labour
is a crucial question for society, and even though the shift
of jobs from regular employment towards platform-based
models for precarious self-employment is still in its early
stages, the political implications of this shift can hardly be
overestimated.
Since around 2005, digital platforms have become in-
creasingly important, causing "disruptive" change in many
industries. They pose a serious challenge, not only for estab-
lished businesses, but also for the social state and its welfare
systems. The potentially destructive force of the new plat-
forms is rooted in the fact that they can be used to circum-
vent national laws for consumer protection, workers’ rights,
minimum wage regulations and social security contributions.
Since the web-based platforms for cloud work and crowd
work have evolved into the smartphone-based platforms for
local gig work, the disruptive change is now becoming visi-
ble also in the physical world.
Digital platforms for the outsourcing of labour are of par-
ticular relevance because, on one hand, they allow for more
flexible sources of income beyond conventional employment,
while, on the other hand, they seem to be fostering a new
class of precarious workers, a so-called "Cybertariat" (Huws
2003; Strube 2015). However, it is important not to look at
the labour platforms in isolation but to see them as part of
the larger platform economy. Therefore the analysis at hand
first describes the functionality and structure of the platform
economy in general (see Figure 1) before it focuses on the
categorisation of digital labour platforms in particular (see
Figure 2). The categorisation suggested here is a tree typolo-
gy that makes it possible to locate the specific opportunities
and risks of certain branches in the platform economy, as well
as particular points at which to tackle these structures with
political measures. In what follows these political measures
will be discussed briefly. However the study understands itself
mainly as a contribution to structuring the problem and its
terminology.
The term "platform" proves to be particularly useful in
this context because it points to the crucial structural simi-
larity of various new digital business models and methods.
It focuses on the mechanics of the infrastructure in the back-
ground and is less tainted by ideology or marketing. Else-
where, also the term "platform capitalism" is used (Kenney
2014; Lobo 2014; Schmidt 2015), but in order to establish a
more neutral vocabulary, the term "platform economy" is
preferred in this study.
Commercial or Commons-based?
In order to categorise the vast amount of digital platforms,
the first distinction that has to be made is between commer-
cial and non-profit commons-based platforms. On platforms
for commons-based peer production, collaboration is more
important than competition and the fruits of labour are free-
ly shared with everybody, including people outside the plat-
form. Such platforms are part of the commons and it is cru-
cial to distinguish them from the majority of platforms in the
so-called "sharing economy" that in fact sail under false col-
ours and pretend to be about sharing, while actually being
about rent extraction or wage labour. In contrast to commer-
cial platforms, the roles and interests of platform providers
and users are not strictly separated within commons-based
peer production. Users who are engaged in commons-based
peer production projects can gradually gain influence over
the structure of their platform and have a say in the rules
that coordinate the collaboration between the different stake-
holders (see Kelty 2008). Platforms such as Wikipedia, Open-
StreetMap and CouchSurfing should therefore be politically
protected and supported as an important alternative to com-
mercial platforms. However, they are beyond the scope of
this study.
3
ANALYSIS: LABOUR MARKETS
IN THE PLATFORM ECONOMY
10
FRIEDRICH-EBERT-STIFTUNG
Three-sidedness and Power Asymmetry
Economists define the structures under discussion here as
two-sided markets or multi-sided platforms (Hagiu/Wright
2015). This means that there are always at least two other
parties between which the platform-provider functions as
intermediary. Thus, in these systems there are always three
groups of stakeholders. In order to emphasise this crucial
aspect, the study at hand speaks of three-sided platforms.
The platform owners provide the infrastructure that medi-
ates between supply and demand provided by the other
two parties. When analysing a particular platform, one has
to look closely at whether the platform provider facilitates
the exchange between the other two stakeholders merely
on a technical level – therefore serving as nothing more
than a software company or infrastructure provider, as these
companies often claim – or if they actually control the inter-
action between the other two parties, as is often the case
with digital labour platforms. In the latter case, the question
is whether these platforms effectively operate as temporary
employment companies. This is relevant to the employment
status of the workers and to the question of whether they
might have been misclassified as independent contractors,
while in fact being employees.
Typically, the software behind the commercial platforms
runs in rented data centres ("the cloud") and has three sides
of access. The users are divided into two opposing groups
for supply and for demand, and both groups see different
and very limited front-ends of the platform: small windows
on the data and the processes of the system. The platform
providers, however, have access to a back-end that gives
them a comprehensive big-data overview of all the interac-
tions between the two user groups, and they furthermore
have the power to influence the exchange between the oth-
er two and potentially do this in real-time. The platform pro-
viders control who sees what and when, what interactions
between the other two are possible and under what condi-
tions, and they wield this control technically, legally and via
the design of the interface. Therefore, the typical platform is
characterised by a systemic information and power asym-
metry in favour of the platform providers. This structural im-
balance in the architecture of the system could be coun-
tered only by decentralisation; a change that seems feasible
for gig work but much less so for cloud work and crowd
work.
The three-sidedness is also important because it allows
the platform-providers to shift entrepreneurial risks, legal lia-
bilities, the cost of labour and the means of production to the
other two parties. The platform itself is an immaterial soft-
ware product and as such it can potentially grow (or scale)
exponentially without the providers having to spend propor-
tionally more on staff or other costs of production (very low
marginal costs). Depending on the area or industry in which
the platform operates, it can often provide its service to one
of the two user groups for free, as long as one group is will-
ing to pay for the access to the other. This is the case, for ex-
ample, with social networks and search engines.
Disruption, Economies of Scale and the Rise of
Monopolies
It holds true for most platforms that the more people partici-
pate, the more useful they become for all users. These so-
called network effects foster the rise of monopolies, or at
least oligopolies, because from the perspective of the users,
it is advantageous to opt for just one search engine, one so-
Figure 3
Three-sided platform architecture
Source: ow n research.
platform provider
transparent big data backend
three-sided
intermediary
platform
supply
(independent
contractors)
demand
(platform clients)
opaque user interface / front end
opaque user interface / front end
11
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
cial network, one online retailer and one online auction house.
The result is a strong agglomeration of power in the hands
of a small number of corporations.
The tendency towards power asymmetries and the emer-
gence of oligopolies – rooted in network effects and the
centralised, three-sided software architecture – is further en-
hanced by the role of venture capital. To attract investors,
the platforms have to be "disruptive", meaning that they have
to break up an established business model or industry and
funnel its profits into the platform economy; the platform
also has to be able to "scale" indefinitely. To achieve the ex-
ponential growth expected by the investors, the marginal
costs of the product must be as low as possible, which in
turn means that the product must be mainly virtual. This al-
lows the platform providers to outsource the physical infra-
structure and operate with a comparatively small staff. A few
hundred employees are often enough to facilitate the busi-
ness exchange between millions of users and take a cut of
typically 20 to 30 per cent from every transaction between
them.
Platforms for the mediation of paid services (digital la-
bour) that are web-based and not bound to a specific loca-
tion (cloud work) make their profit mainly from the labour of
their independent contractors (even though the workers still
have to pay for their computers and access to the internet as
means of production). Platforms for the mediation of loca-
tion-based tasks and services (gig work) – especially in the
sectors of accommodation and transportation – integrate
not only the labour of their independent contractors into
their own value chain, but also their capital in the form of
cars and homes. This is partly the reason why investments in
gig work platforms, as well as the valuations of these com-
panies, are so much higher than in the crowd work sector.
Airbnb and Uber can challenge conventional companies in
the hospitality and transportation industries, respectively,
without having to own real estate for accommodation or a
fleet of taxis, without having to pay for the maintenance of
these capital goods and without being liable for the safety
of the guests, the drivers or the service personnel. Only when
seen from this perspective do the astronomical valuations of
Airbnb (US$25.5 billion) and Uber (US$62.5 billion) begin to
make sense.
The huge amounts of venture capital that the platform
companies have raised are often used to finance an aggres-
sive growth strategy that entails buying competitors and sell-
ing one’s service under value for a while, in order to reach a
critical mass of users before anybody else. The goal is a mar-
ket penetration with network effects strong enough to keep
the users on the platform even when, in order to break even,
the terms of service are eventually changed, to the disadvan-
tage of the users. Good examples are the changing privacy
settings on Facebook, now affecting 1.6 billion users, or the
changing rates for fares of Uber rides, which over time have
become significantly less favourable for the drivers in order
to benefit the platform (see Section 3.2.2).
Overreaching Terms of Service
While it is typically the case that centralised, multinational,
venture capital–funded corporations control the back-end of
a commercial digital platform, the two parties at the front-
ends are compartmentalised, fragmented and unorganised.
Hence they have to negotiate from a position of weakness
and this shows in the terms of service. The venture capital
allows the platform providers, who are operating at an inter-
national level, to risk lawsuits at a national level – for exam-
ple when being sued by workers, consumer advocates or the
government for violation of local labour laws and regulations
(Uber is the best example here). The problem of overreaching
terms of service occurs in all areas of the platform economy.
They are sprawling in terms of the sheer amount of text
(with 55,000 words, Airbnb’s terms of service have almost
the length of a novel), they are often strongly biased against
the users and they are supposed to apply to more and more
areas of life. Contracts that formerly applied only to the rela-
tionship between a software product and its individual user
now also apply to the interactions between the users, to their
private and business relationships. This continuous expansion
of the terms of service becomes particularly problematic in
the domain of digital labour platforms, where the software
licencing contracts have effectively evolved into work con-
tracts. In the case of cloud work, the situation is further com-
plicated by the fact that potentially all three parties of the
platform triangle can – and often do – reside in different
countries. The place of jurisdiction is usually the city in which
the platform provider is registered, and the terms of service
are usually written in an all-encompassing way that is sup-
posed to be binding for millions of users across the globe.
Obviously, this wholesale approach frequently collides with
the national jurisdiction of the states in which the users re-
side. A discussion of these legal disputes is beyond the scope
of this study. Instead, it must suffice to point to the work of
the German employment law experts Thomas Klebe and
Wolfgang Däubler, who have tackled these questions in de-
tail (Klebe 2014 and 2016; Däubler 2015).
Algorithmic Management through Ratings and
Tracking
It is one of the fundamental principles of the platform econ-
omy that production itself is not done by the platform pro-
vider, but by one of its two groups of users. To accomplish
this, a lot of coordination is required from the platform pro-
vider, especially in order to sift through the flood of very het-
erogeneous contributions on the supply side, and in order to
orchestrate the interactions of the users. To keep the margin-
al costs of production close to zero and ensure that the plat-
form can grow exponentially, it is imperative for the platform
providers to automate as many of these processes as possi-
ble. With only a small number of employees, they could not
possibly deal with the millions of users personally. It is at this
point where the interplay between "big data" and "algorith-
mic management" is brought into action (Lee et al. 2015);
some also call this form of control "algocracy" (Aneesh 2009;
Danaher 2016). Algorithms now do jobs previously done by
middle managers, accountants and customer service repre-
sentatives. And in the case of the digital labour platforms,
human resource management is often outsourced to the us-
ers – especially in crowd work, where the individual workers
self-assign their jobs. If the results do not match the clients’
In order to ensure informational self-determination, workers
would need a tool to monitor their big-data résumés, and the
very heterogeneous digital labour platforms would have to
agree on transferable standards or protocols, for example in
the evaluation of reliability. The alternative would be to col-
lect less data in the first place.
Gamification
Another mechanism for the automated coordination of large
groups of users, or workers, in the case of digital labour
platforms, is so-called "gamification". This tool of algorithmic
management is made possible with data from ratings and
tracking. It's a technique that allows platform providers to re-
ward favourable user behaviour by awarding virtual credit
points and by ranking the users’ performance on public lead-
erboards. The awarded points often serve as a pseudo cur-
rency within the reputation economy of one platform, but
they cannot be transferred to another. Gamification trans-
forms wage labour into a game, in an often manipulative,
behaviouristic manner. The basic principle is not new; ana-
logue precursors of gamification include military medals or
employee-of-the-month schemes. What is new is that, through
rating and tracking on digital labour platforms, even the ti-
niest actions and utterances, down to the level of single
mouse clicks, key strokes and scrolling behaviour, can be
monitored and influenced through gamification. In contem-
porary video games, such as Grand Theft Auto V, one can al-
ready get a glimpse of how this development could play out
for the workplace and the résumés of the future. Menus list
hundreds of categories with statistical data on how often,
how long, how fast and how accurate the player has solved
specific tasks. Even the most minute actions are rewarded
with "awards" and "achievements" and have their own lead-
erboards to compare the performance of different players.
This type of data not only serves to motivate gamers or
workers, respectively, but it is also a treasure trove of infor-
mation for the digital labour platforms (see Section 3.1.1.).
What’s more, the Chinese government in collaboration with
the Chinese shopping platform Alibaba is currently rolling
out a project that shows how serious and politically relevant
the role of gamification in the platform economy is. Under
the name "Sesame Credit", it has introduced a public, indi-
vidual "citizen score", based on factors such as shopping be-
haviour, credit history and the social circles of individual citi-
zens, in order to reward political obedience and publically
shame potential deviants. From 2020 onwards, China plans
to make participation in the scheme mandatory for its citi-
zens (Hatton 2015).
The Blurring Lines between Work and Play
Within the platform economy in general (beyond the dedi-
cated digital labour platforms) it is often difficult to define
what exactly counts as "work". As already mentioned, the
business models are based on letting the users take care of
the production. Is it therefore appropriate to demand wag-
es for users of social networking platforms such as Face-
book, as some activists do (see: Lanier 2014; Ptak 2013)?
And what about online games such as World of Warcraft,
expectations, the independent contractors are algorithmical-
ly rejected from future jobs, either entirely or from those
above a certain threshold of quality or pay. This is done by
blocking their account or by making certain jobs invisible to
them at the front-end of the platform interface.
For many people, Amazon and eBay were the first places
on the internet where they made business deals with strangers
and afterwards publicly rated their satisfaction with their
counterparty by awarding one to five stars. This method has
become ubiquitous and is now also used for the management
of the workforce on digital labour platforms. These ratings
create trust between users who know nothing else about
each other. They also make qualitative judgements between
humans quantitative and thus machine-readable. Amazon and
the users of its online warehouse heavily rely on the detailed
product reviews written (without compensation) by its users.
But only by reducing these judgements to five-star-ratings
can they be sorted effectively by the platform. And it is this
method that has become the standard for evaluating the
performance of crowd workers and service personnel in the
gig economy, too.
Ratings require the active participation of the users in an
act of mutual evaluation that takes place after each complet-
ed interaction. Tracking in turn refers to the passive but con-
tinuous recording and evaluation of all user interactions, even
very small ones. With its search engine, Google has shown
how the tracking of user behaviour can be turned into a high-
ly profitable business model. In a similar fashion, the detailed
data that the platform providers continuously collect about
the performance of their workforce – the knowledge about
individual worker’s thoroughness, industriousness and error
rate – becomes an important asset; part of the capital of the
platform providers.
Thanks to smartphones, the tracking and rating of cus-
tomers, service personnel and independent contractors can
now happen on the spot, face to face and in real time (Dzieza
2015). People assess each other’s performance in the physi-
cal world immediately through actively rating the other
(Holland 2016). And the platform providers can expand the
tracking of the individual worker’s efficiency on the platform
itself to tracking their movements in space. On digital labour
platforms, the aggregated ratings of workers de facto be-
come their employment reference, while the constant track-
ing of their performance can amount to a fully automated
curriculum vitae – a personal big-data sheet. This develop-
ment brings up a number of tricky questions regarding the
fairness and accuracy of these evaluations and it challenges
the right to informational self-determination. Who should be
allowed to access these big-data résumés? And is it – from
the workers’ perspective – worth striving for the possibility
to make the personal big-data sheets transferrable from one
digital labour platform to another, to not lose their hard-
earned reputation when jumping platforms? Or would that
be disadvantageous, because it would create the pressure to
fully reveal one’s complete data set when looking for a new
platform provider or employer, even if the data might contain
unfair or faulty evaluations? On digital labour platforms such
as Amazon Mechanical Turk it has been a known problem
for years that there is no proper dispute resolution policy if
workers think either man or machine has rated them unfairly.
12
FRIEDRICH-EBERT-STIFTUNG
cannot hold a full-time job because of personal health is-
sues or because they have to take care of a family member;
and, in the case of cloud work, also for people who either
live in regions without jobs, or choose to work as "digital
nomads" from abroad while travelling. There were (Txteag-
le) and are (Samasource) even attempts to hire people in re-
fugee camps in the Global South as translators, asking them
to translate texts line by line via text messages on mobile
phones.
Digital labour’s enormous flexibility is partly enabled by
the Tayloristic breakdown of what were once occupations
into their smallest possible components. Jobs become pro-
jects, then gigs and tasks and eventually microtasks. The
units of time and payment are broken down into seconds
and cents and the independent contractors switch from one
client to another with ever-higher frequency. The fine granu-
larity of tasks has the effect that both groups are willing to
take more risks with regard to the likelihood of getting paid
and the quality of the results respectively, because when
one in a succession of microtasks goes wrong for either
side, the damage that such an individual incident can cause
is almost insignificant. The aggrieved party has merely lost a
tiny amount of money or time. In the aggregate, however,
these losses become a problem, especially when the uncer-
tainty of getting paid for work already done becomes the
new norm. The tiny values in dispute also have the effect
that workers on digital labour platforms are usually not will-
ing to go to court to sue the other party for compensation
(or the platform providers for their legally questionable
terms of service).
The question is, how established standards of labour law
and social security can be sustained if what constitutes a
job is cut into smaller and smaller tasks with uncertain pay?
What is the legal status of people working under these con-
ditions? Almost all platforms for digital labour state in their
terms of service that the workers are independent contrac-
tors, as well as that, because they are "self-employed", it is
also up to them to take care of all social security contribu-
tions. But is that a realistic description of those cases in which,
although the clients might change from minute to minute,
the independent contractors work continuously for the same
platform provider, which in turn exerts strong influence over
how exactly the work must be done and what is paid for it?
Here the question is whether the independent contractors
are in fact misclassified employees of the platform. So far
there have been a number of class action lawsuits, mainly in
the United States, in which crowd workers and gig workers
have sued their platform providers in order to retroactively
demand the minimum wage they would have been entitled
to as regular employees. In the context of crowd work, there
was a prominent lawsuit against CrowdFlower and, in the
context of gig work, against Uber (see Cherry 2016; Seiner
2017). For the platform providers these class action lawsuits
pose an imminent threat to their business model, but to
date they have been able to resolve them through multi-mil-
lion dollar settlements. That also means that the legal situa-
tion remains unresolved.
13
where in endless hours of piecework, amateur players, side
by side with professional "goldfarmers", produce and accu-
mulate virtual goods of tremendous real-world exchange val-
ue? How are we to deal with the fact that exactly the same
action can be recreational play for one person and precarious
work for another? This is not a trivial or purely academic
question, given that through gamification work is actively
being transformed into a game on some platforms. It's a
challenge for regulatory measures in this field to distinguish
digital labour platforms disguised as games from games
that are misappropriated as precarious workplaces.
Nevertheless, there are two fundamental differences be-
tween the indistinct forms of work in the context of user-gen-
erated content in the wider platform economy and those on
dedicated labour platforms: the production of user-generat-
ed content is typically self-initiated and intrinsically motivat-
ed. Writing blog posts and product reviews, uploading pho-
tos to Instagram or videos to YouTube, is all obviously based
on a lot of work. But this work is done without assignment
or brief, without any deadlines or specific demands by a third
party defining what should be produced, when and how.
The same is true for all the content that is produced as a
by-product of communication on social networking sites.
And because production of user-generated content is usually
not reimbursed (at best only indirectly through advertising),
there is not a new class of precarious workers emerging in
this area, which is the concern (and to some extent already
the case) with the dedicated digital labour platforms. The
question of regulatory measures is therefore more pertinent
in the case of those platforms that explicitly provide work to
independent contractors, work that is not just perceived as a
hobby or as spare-time occupation, but as a job that is done
according to the demands of an employer or contracting en-
tity and with the intent to realise a profit.
Flexibilisation and Atomisation of Labour
The most important promise that digital labour platforms
make to their workforce, as well as to their clients is flexibility.
Independent contractors are available "on-demand" as a
"contingent workforce"; they are hired for specific tasks only
and are dismissed as soon as the job is done.1 In return, the
independent contractors can work whenever, however, for
whomever and as much or little as they want, as long as there
are enough suitable tasks available and there are not too
many competitors trying to do the same jobs. The entry bar-
riers for cloud work, crowd work and gig work are extremely
low, so that even marginalised groups can potentially find
work immediately; the only prerequisites are that they accept
the terms of service and have a fast and stable internet con-
nection. This is a huge opportunity for people outside con-
ventional career paths, without certain qualifications, with
little education or work experience; but also for people who
1 In the words of Lukas Biewald, CEO of CrowdFlower: "Before the In-
ternet, it would be really difficult to find someone, sit them down for ten
minutes and get them to work for you, and then fire them after those ten
minutes. But with technology, you can actually find them, pay them the
tiny amount of money, and then get rid of them when you don’t need
them anymore" (Marvit 2014).
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
relatively well paid. In 2013 oDesk published a graph that
showed the distribution of job types on offer.4 Search engine
optimisation (SEO) was in the lead, followed by jobs in soft-
ware and web development; but also jobs in marketing, de-
sign, writing, legal services and engineering were featured in
the spectrum.
For mediating between supply and demand, freelance
marketplaces typically charge a fee of 10 to 20 per cent from
the independent contractors. For clients the service is often
free. Even though the available data on the subject are limit-
ed, it seems that the freelancers on these platforms are com-
paratively satisfied (Leimeister et al. 2016). In contrast to the
neighbouring fields of crowd work and gig work, there is no
larger debate around the potential exploitation of workers in
this area of the platform economy. It seems that there has
been little demand for political regulation so far. However,
there are two critical aspects in the functionality of freelance
marketplaces that are of great relevance for all digital labour
platforms.
First, on these outsourcing sites individual contractors
have to compete with each other globally, and through the
practice of bidding there is the danger of entering a race to
the bottom for common tasks. How cheaply one can offer a
service depends partly on one’s cost of living but more im-
portantly on one’s degree of specialisation. The more spe-
cialised a skill is the less it is in danger of a deterioration in
prices caused by global competition. For freelancers in the
Global North, it will be less and less profitable to offer servic-
es that can be done just as well via the internet by people
from the Global South. This development also affects skilful
but routine tasks, such as the analysis of medical X-ray im-
agery (Sharma 2014).
Secondly, the freelance marketplaces are characterised by
a relatively high level of surveillance. Upwork, for example,
uses a software application called "Work Diary" to allow cli-
ents to virtually look over the shoulders of their independent
contractors. Six times per hour and at random intervals, the
software takes a screenshots of the freelancers’ computer. In
this way the client can ensure that the contractors stay on
task, instead of, say, checking in on Facebook while being on
the clock. Furthermore, the Work Diary also tracks the num-
ber of mouse clicks and keystrokes and even makes webcam
photos of the independent contractors – who can, however,
refuse clients the permission to use this feature.5 In addition,
Upwork states in its terms of service: "We will share informa-
tion contained in Work Diaries with the relevant client and
with any manager or administrator of any applicable Free-
lancer Agency." It is made clear that as a freelancer, one has
little control over the data gathered on one’s work behaviour.
The extraordinary degree of freedom on digital labour plat-
forms such as Upwork is accompanied by an extraordinary
degree of control. Interestingly, industrious workers often
welcome this form of surveillance, because it allows them to
demonstrate their reliability and therefore justifies their com-
paratively high hourly rates.
4 See: https://content-static.upwork.com/blog/uploads/sites/4/2013/08/
LongtailSkillsChart.jpg.
5 See: https://www.upwork.com/legal/privacy/#work-diaries.
3.1. PLATFORMS FOR WEB-BASED SERVICES
(CLOUD WORK)
3.1.1 FREELANCE MARKETPLACES
Freelance marketplaces (sometimes also referred to as on-
line outsourcing, outsourcing marketplaces or the online
staffing industry) transfer the principle of outsourcing from
the level of companies to that of individuals. Clients can find
independent contractors abroad via these platforms and the
latter can in turn bid for the advertised jobs. In principle, all
three parties in the platform triangle can be based in differ-
ent countries across the world, which, as mentioned earlier,
is a tricky complication with regard to the applicable legal
jurisdiction.
Upwork, one of the largest platform providers in this area,
explicitly advertises its service within the framework of the
lifestyle choice of becoming a "digital nomad", a creative,
well-educated online worker travelling the world, able to
earn money at the beach or from the side of the pool. All it
takes is a laptop and a fast internet connection.2 This type of
digital labour platform always falls into the category of cloud
work, but it is typically not crowd work.3 The important dif-
ference is that on the freelance marketplaces’ clients hand-
pick independent contractors based on their skills; the pay-
ment is negotiated individually in the end; and only one
person is eventually going to do the job. Freelance market-
places have millions of independent contractors as users and
huge revenues and have been in existence for over a decade:
eLance was founded as early as 1999, oDesk in 2003 and
Freelancer.com in 2009. In 2013, the former two merged into
eLance-oDesk, and since 2015 they have traded under the
name "Upwork". The Silicon Valley–based company now
claims to have 9 million registered freelancers, 4 million cli-
ents and a turnover of US$1 billion per year. After merging
with several smaller providers, Freelancer.com, Upwork’s big-
gest competitor, now claims to have 20 million registered
workers, who so far have finished 9 million jobs (which also
means that the majority of registered contractors never got a
job via the platform). It must be noted that the user numbers
published by platform providers are generally not very relia-
ble; the platforms typically publicise only the total number
of people who have ever registered, in order to appear larg-
er than their competitors. The number of active users is al-
ways much smaller and follows a "long tail" or Pareto distri-
bution – only a small number of "power-users" (between 1
and 10 per cent) accomplishes the majority of all jobs on the
platform. Most users who create an account are active only
sporadically or not at all. In order to evaluate the size of a
platform, revenue figures or the number of finished jobs are
much more significant.
The types of jobs mediated via freelance marketplaces
are very heterogeneous, but in contrast to microtasking
(which will be discussed the next section) the tasks are rela-
tively complex, demanding, specialised, technical and often
2 See: https://www.upwork.com/blog/category/digital-nomads/.
3 Although some freelance marketplaces also offer crowd work as an
alternative mode to outsource work.
14
FRIEDRICH-EBERT-STIFTUNG
Amazon played a key role in the development of modern
microtasking. Originally, the company was looking for a way
to synchronise or remove redundant entries in the catalogue
of its online warehouse. This, too, is a task that humans can
solve much better than computers and Amazon has started
to pay unskilled people small amounts for this work. In 2005,
the company made its new outsourcing method available to
external clients and called the service the "Amazon Mechani-
cal Turk" (MTurk). The company is tight-lipped when it comes
to MTurk, but it is assumed that about half a million people
work on the platform. The workers come from various coun-
tries, but for a few years only people from India and the Uni-
ted States could apply, and they form the majority of the
so-called "Turkers". Crowdworkers from other countries are
not paid with money but in Amazon vouchers. Since 2016,
the platform has accepted workers from Germany and other
countries again.
In comparison with its competitors, MTurk is fairly small.
With about 700,000 registered crowdworkers, the German
company Clickworker.com, founded in Essen in 2005, has
the same order of magnitude. But the Silicon Valley–based
company CrowdFlower, financed with US$28 million in ven-
ture capital, has about 5 million crowdworkers, coordinated
by just a hundred employees.7 Nevertheless, most studies
on microtasking so far have focussed on MTurk, which is also
why the following examples are mostly from Amazon’s plat-
form.
Amazon describes its form of crowdwork as "Human In-
telligence Tasks" or HITs. The principle is sometimes also re-
ferred to as "humans-as-a-service", following the wording of
similar offerings such as "software-as-a-service". Just as cli-
ents can rent data storage or processing power from Ama-
zon, they can also rent human brainpower. Tellingly, the
name "Mechanical Turk" is a reference to a historical eight-
eenth-century chess robot hoax, which involved a human
hiding in the apparatus, pretending to be a smart machine.
The point is that Amazon allows its clients to address the
crowdworkers as if they were merely machine parts, not
with their real name but as anonymous, numbered process-
ing units. And being dehumanised by the platform is indeed
the grievance most often articulated by the so-called "Turk-
ers" (see: Irani 2013, 2015). In 2015 a group of Turkers even
wrote an open letter to Amazon CEO Jeff Bezos to remind
him that crowdworkers are flesh and blood humans who
want to be treated with fairness and respect (Salehi et al.
2015 ).
From the clients’ perspective, the invisibility of the work-
ers in microtasking is not a bug but a feature. This cannot
be altered without a significant loss in efficiency and it is an
important difference from the freelance marketplaces, where
the clients handpick the workers and then want to virtually
look over their shoulders. In microtasking, the units of work
and the reimbursements are so small that it would neither
be practical nor economically feasible to deal with contrac-
tors on an individual level. Instead, the workforce is dealt
with in the aggregate – as a crowd – which leads to the
next points of criticism regarding this specific form of digi-
7 See: https://www.crunchbase.com/organization/crowdflower#/entity.
3.1.2 MICROTASKING (CROWDWORK)
There are essentially two basic types of commercial, paid
crowd work: microtasking crowd work and contest-based
creative crowd work (see Section 3.1.3). Microtasking is prob-
ably best circumscribed in terms of "cognitive piecework", a
phrase introduced by crowd work researcher Lilly Irani; micro-
tasking pioneer Luis von Ahn calls it "human computation".
Microtasking involves masses of tiny, repetitive tasks that are
distributed across a large and unspecified group of crowd
workers. The workers self-assign to the tasks and are as-
sumed to be generally unskilled (for that task) and therefore
interchangeable.6 The processing of the tasks is automated as
much as possible; the organisation resembles that of a con-
veyor belt production line and clients as well as crowd work-
ers often remain anonymous to each other. Many of the tasks
revolve around data processing problems that can best be
solved by human cognition, but the results of which can be
evaluated and assembled by a computer. The principles of
microtasking have a lot in common with the methods once
described in Frederick Taylor’s Scientific Management.
The organisation of labour as microtasking forms an in-
terdependent relationship with automation. Many of the tasks
in question are likely to be automated in the near future.
Crowd work researcher Mary L. Gray, of Microsoft Research,
describes microtasking as the "last mile of automation"; it
concerns the residual tasks of larger data processing opera-
tions that unskilled humans can still solve more cheaply and
with a lower error rate than computers. It is likely that stricter
regulations regarding the labour conditions on microtasking
platforms would accelerate the trend towards automation of
these tasks. Crowdworkers in the field of microtasking are
already training the machines that are supposed to replace
them. In turn, progress in machine learning is heavily de-
pendent on crowdworkers to create original data sets that
the machines can learn from.
Other typical forms of microtasking include: the valida-
tion of existing databases, for example of company address-
es; the digitisation of business cards (now already mostly
automated); the transcription of audio recordings; the writ-
ing of product descriptions; sentiment analysis; and, last but
not least, content moderation. All the user-generated con-
tent uploaded to social media sites has to be checked – at
least if it was flagged by other users – to see whether it is
harmless and in accordance with the terms of service. It de-
mands human value judgement to recognise real violence,
hate speech, pornography or simply female nipples (in the
case of Facebook) to distinguish them from depictions ac-
cepted by the platform and to censor transgressions. A lot
of this work is done in countries such as the Philippines, by
crowdworkers who, because of the shocking content that
people upload (for example, IS propaganda videos of be-
headings), sometimes even develop post-traumatic stress
disorders (Chen 2014).
6 The dif ferent platform types for digital labour are described here
deliberately with a broad brush to make visible their prototypical
char-acteristics. In reality, there are many hybrid forms of platform and
great diversity among the people who work there. Crowdworkers are
often highly educated.
15
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
indeed, the lack of trust is to some extent justified. In both
groups of users one finds actors that are unreliable – work-
ers that overestimate their capabilities when self-assigning
to a task and clients that provide faulty task descriptions that
inevitably cause errors in the process. In addition to these
accidental failures, there are also cases of deliberate fraud
on both sides. The economic and technical feasibility of
crowd work depends on the workers not being handpicked
and individually controlled. If the clients were legally obliged
to pay for each and every result produced under these con-
ditions – as fairness towards honest workers would demand
– fraudsters and dabblers would take advantage of this and
inevitably force the clients into stricter pre-selection, more
surveillance and human quality control of the end-results.
Maybe, for regulatory purposes, this is the right way to go,
but it would no longer be crowd work if the workers were
individually selected in advance. And if a channel of com-
munication for disputing rejected tasks was mandatory, the
work would become so expensive that the process would
be feasible only for bigger, more valuable tasks.
Finally, we should mention that there are a number of
platforms that at first sight do not look like microtasking, be-
cause the jobs are much bigger, for example testing soft-
ware and writing bug reports (for example, Testbirds). Be-
cause a lot of this work is done via smartphone apps, it is
sometimes referred to as "mobile crowdwork". But because
it is still outsourced en masse and the results can be evalu-
ated objectively and automatically, they are considered here
as part of microtasking. Furthermore, there are platforms
for location-based microtasking (for example, Streetspotr),
where the job is for example to take photos of how prod-
ucts are displayed in stores. These jobs are location-based,
like gig work, but they are bound neither to a specific per-
son nor to a specific time when they have to be accom-
plished, and are thus best treated as a form of microtasking
crowd work.
3.1.3 CONTEST-BASED CREATIVE CROWD WORK
The largest digital labour platforms that organise work in the
form of contests can be found in the domain of graphic de-
sign; more specifically, logo design. Since 2008, a number of
these platforms have emerged, the largest of which is 99designs
from Sidney, Australia. As of 2016, about 1.3 million registe-
red users contribute solutions to design tasks on 99designs.
On average, they upload a new design to the platform every
2.5 seconds, which is about 1 million designs per month. In
total, the platform has paid out 125 million euros in half a
million contests (as of March 2016). Thus, the average amount
of money paid per contest is 250 euros. However, on avera-
ge the crowd hands in 100 designs per contest, which brings
down the average pay per design to 2.50 euros, far below
the minimum wage.
The method of using creativity contests to organise work
is by no means restricted to logo design. It covers a spec-
trum that also includes more complex tasks, such as web de-
sign, the development of marketing campaigns and open
innovation projects for large companies (Jovoto), the concep-
tion and design of new physical products (Quirky), and even
vehicle design (Local Motors). Because of the structural simi-
tal labour: very low and uncertain payment, and no orderly
conflict resolution for the workers in the case of unfair treat-
ment and unpaid wages. Here, the consequences of algo-
rithmic management come into full effect. Microtaskers don’t
have a boss who assigns them to specific tasks, controls the
process and approves the results. Instead, the workers self-
assign and everything else is done automatically; for exam-
ple, by letting the computer compare the results of five dif-
ferent workers who have completed the same task. If the
result of one person differs from those of the other four, it
is considered to be wrong and the worker will not get paid.
It does happen though, that a worker hands in a result that
differs from the others because he or she has done the task
more meticulously. If the task is rejected, there is no person
on the other side that the worker could complain to about
the mistreatment. Such a channel of communication would
be important for the workers, but in relation to the size of
the tasks, it would be too expensive for the client (see Kit-
tur et al. 2013). Amazon therefore lays down in its terms of
service that the client does not have to pay for rejected re-
sults, but is allowed to use them anyway. Critics regard this
as an invitation to wage theft (see: Scholz 2015). It has grave
consequences for the ratings of the independent contractor
if results are rejected, no matter whether by an algorithm
or by the client directly. The worker will be automatically ex-
cluded from future jobs if their overall ratings fall under a
certain threshold; this is sometimes referred to as being fired
by algorithm.
Prototypical crowd work (including contest-based crea-
tive crowd work, see next section) is characterised by a mu-
tual lack of responsibility. In principle, crowdworkers have the
freedom to self-assign to any job without qualifications; they
can quit working in the middle of a task, without having to
answer to anyone for their decisions or their results. In turn,
the clients are not responsible to answer questions from the
workers or to guarantee payment for work that is done un-
der these conditions. The lack of responsibility on the client’s
side is often criticised, but when advocating minimum wag-
es for crowdworkers, one has to take into consideration that
both sides have very few obligations. If regulations forced
clients to pay minimum amounts of money – either for the
time workers invest or per task – they would indirectly be
forced to control more strictly who is allowed to work on a
task in the first place. The clients would have to demand
previous qualifications and monitor the work process and
the results more strictly, to ensure that they actually get what
they are paying for. Regulations aiming for a minimum wage
would therefore very likely force the crowd-work platforms
to become more like freelance marketplaces. The quid pro
quo of the extraordinary degree of freedom that crowdwork-
ers currently enjoy is uncertain and low wages. If labour un-
ions or the government want to improve the working con-
ditions, while maintaining this high degree of freedom, as well
as the low entry barriers, they will have to decide what is the
lesser evil: uncertain and low pay (as on MTurk) or total sur-
veillance of the work process (as in Upwork’s Worker Diary).
The mutual lack of responsibility in crowd work is accom-
panied by a mutual lack of trust. This is especially pronounced
in microtasking, where the clients and workers are typically
anonymous to each other (though not to the platform). And
16
FRIEDRICH-EBERT-STIFTUNG
ture of creative and innovative work is that the results have
to be new, which means that the client does not know in ad-
vance what the sought-after solution is supposed to look like.
The quality of an idea can be independent of the amount of
time that a creative worker invests in it. A brilliant idea can
come from a flash of inspiration, as well as from weeks of
hard work. But the latter is no guarantee of a good idea.
Therefore, it is impractical for creative work to establish a
form of reimbursement based on the amount of time invest-
ed. Slow workers would be paid much better than ingenious
ones. But it is equally impractical to guarantee a payment for
every idea that is handed in, because then the client would
have to pay even for the worst ideas that the crowd comes
up with. If regulators want to leave intact the core principles
of what a crowd is (open to everyone, no prior qualifications
needed), while also pushing for a minimum wage (be it per
time spent or per solution provided), the clients would inevi-
tably have to pay for many inferior solutions. As with micro-
tasking, such political intervention would probably cause cre-
ativity contests structurally to become more like freelance
marketplaces, with pre-selection of workers and surveillance
of the work process.
As mentioned above, creative tasks do not lend them-
selves to being subdivided and automated, but they are es-
pecially well suited for outsourcing via a crowd contest. This
has a number of reasons: creative work is held in relatively
high social esteem, especially if compared with the type of
jobs available in microtasking. Many people experience crea-
tive work as intrinsically rewarding, see it as a dream job and
become passionate amateurs. This group of creative crowd-
workers do not expect to get paid properly, even if they pro-
vide solutions of professional quality. In addition, many crea-
tive crowdworkers endure the low and precarious pay on the
platforms because they hope to professionalise themselves
in the process. While the unskilled, piecemeal work of micro-
tasking is reminiscent of labour on a conveyor belt, contest-
based creative work resembles unpaid or severely underpaid
internships in the creative industries. Microtasking is not a
profession and has no career to offer. Participation in contest-
based crowd work, however, is driven by the hope of enter-
ing a fulfilling line of work, of learning skills with value out-
side the platform. For each individual this possibility exists,
but for the majority of the crowd it is unlikely. By definition,
only a few can stand out from the crowd; everybody else is
interchangeable.
All this leads to an interesting inversion with regard to
the visibility of crowdworkers. In microtasking, they suffer
from the fact that they are not perceived as individuals and
practically remain invisible. In contest-based creative crowd-
work, by contrast, they have to invest a lot of their personali-
ty in building a personal reputation and a public portfolio.
The work they have done in previous contests is very visible
and can easily be judged by clients and colleagues. This high
visibility entices many to invest more time, effort and person-
ality into the design projects than would be economically
reasonable, given that the chances of eventually getting paid
are slim. In contrast to microtasking, it is important for crea-
tive crowdworkers to create a personal connection to the cli-
ent and to come across as friendly, attentive, service-orient-
ed, even servile – but also as innovative thinkers able to
larities, also platforms for research and development tasks
outside the immediate design domain, such as InnoCentive
(founded as early as 2001) are considered here as part of con-
test-based creative crowd work.
In contrast to freelance marketplaces and microtasking
platforms, the clients who use contest-based creative crowd
work are in search for the best possible solution out of a very
heterogeneous pool of possible solutions developed by the
crowd specifically for that client. Typically, only one solution
is needed, selected and paid for – all others are discarded.
Therefore, the amount of work done redundantly, in vain and
without pay is very high – and this is also the most heavily
criticised aspect of this platform type. If not chosen by the
client, the participating creatives typically keep the copyright
to their solutions, but because the ideas and concepts are
custom-made for the client, they become useless if discard-
ed in the contest. For the clients, however, also the discarded
solutions contribute to the decision-making process and thus
provide a value that they don’t have to pay for. Some of the
platforms even explicitly advertise their service with the prom-
ise to provide clients with free labour. The logo design plat-
form Zenlayout.com, for example, advertises its service with
the claim: "Run a Logo Design Contest. Hire 700 Designers.
Pay One."
Contest-based creative crowd work is understood here as
a subcategory of cloud work and crowd work, and as part of
the value chain in commercial product development. This ex-
cludes creativity contests conducted to support and award
creative talent for its own sake and also contests conducted
merely for marketing purposes – for example when a com-
pany conducts a one-off amateur painting contest among its
customers, where the actual results are only of secondary in-
terest to the company. It also excludes the common practice
of professional pitches, in which a number of professionals
compete by handing in proposals in order to get the commis-
sion for a bigger project. In contest-based creative crowd
work the participants have to hand in finished designs. Be-
cause the work is done completely in advance, this form of
labour is sometimes also referred to as speculative or "spec
work". Some companies also conduct in-house crowd con-
tests, or organise them as a one-off event without an exter-
nal platform as an intermediary. For these borderline cases
there seems to be no immediate need for political action or
regulations because they do not create a new class of pre-
carious cloud workers and thus have comparatively few con-
sequences for the labour market in general.
Contest-based crowd work on commercial digital labour
platforms is a different matter. Such platforms have evolved
into an industry that is systematically and continuously out-
sourcing work hitherto done by regularly paid professionals
to a "standing army" of crowd workers, for whom it has made
the possibility of fair payment into a gamble.
The choice of the contest as prime method for organising
creative crowd work is by no means arbitrary; it is the imme-
diate consequence of the type of task. Creative tasks can
typically not be subdivided into microtasks and an algorithm
cannot do the evaluation of the results. The results are nei-
ther right nor wrong, but are on a spectrum of better or less
well suited. Their value cannot be quantified objectively, but
is often subject to the client’s individual taste. The core fea-
17
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
exchange value for the design that they upload. The oppor-
tunities and risks on the largest platform for contest-based
creative crowdwork are very unevenly distributed, and 99de-
signs is not an exception.
There are, however, a few smaller platforms for contest-
based creative work that are less unfair, paying out much
higher sums and involving the crowd in the decision about
who wins the contest. Most exemplary here is the Berlin-
based platform Jovoto, which has about 80,000 registered
designers working on marketing campaigns and new prod-
ucts, often for big brands and international corporations (see:
Schmidt 2015; Al-Ani 2015). But even platforms explicitly
dedicated to fairness cannot solve the fundamental problem
of the creativity contests: the extraordinary amount of redun-
dant unpaid labour.
In contrast to platforms such as 99designs, Jovoto’s em-
ployees have a lot of design expertise and are very much in-
volved in consultations with the clients and in the guidance of
the crowd and the design process. By orchestrating the pro-
duction process, they provide a better service to both sides
and contribute a lot more to the creation of good solutions.
However, by doing that much more, they also make the prob-
lematic model of contest-based labour accessible to high-pro-
file clients with complex, high-value design projects. A higher
percentage of formerly well-paid design jobs can therefore
be transformed into precarious crowd work. In other words,
while 99designs disrupts the lower strata of the design busi-
ness, offering a discount solution to jobs that were already
done by relatively badly paid independent contractors, Jovoto
is disrupting a higher segment of the market, by enabling cor-
porations such as Coca Cola, Deutsche Bank or Beiersdorf to
make use of contest-based creative crowd work. On Jovoto,
the crowdworkers have a much better status than on 99de-
signs, but they are still in a significantly more precarious work
situation than designers working in conventional agencies
and for whom getting paid is not a lottery.
3.2 GIG WORK (LOCATION-BASED DIGITAL
LABOUR)
Platforms for location-based digital labour have become pos-
sible only through the widespread dissemination of smart-
phones with GPS trackers. They are the prerequisite to orche-
strate work that is not done somewhere "in the cloud" but
on the go and at specific locations in the city. These relatively
new gig work platforms are typically backed by much higher
sums of venture capital than the cloud work platforms and
their impact is more visible and potentially larger, especially
in the gig work subcategories of accommodation and trans-
portation, which involve a lot of capital and affect many jobs.
At the same time, regulatory measures can be put into acti-
on much faster and more effectively than in cloud work,
because for gig work this is typically done at the political level
of individual cities.
With the exception of location-based microtasking, which
due to its size is negligible (see Figure 2), gig work is always
bound to a specific person who has to show up on time to
do the job. The smartphones play an important role, not only
for allocating jobs and people in space, but also for choosing
provide unique ideas. For the success of creative crowdwork-
ers, the display of a winning personality is almost as impor-
tant as the results. The working conditions are therefore
comparatively emotional and the frustration is high if, despite
the high level of engagement, the win-ratio is too low.
In parallel to this engaged and virtuous, but economically
risky way of working, it is also a common strategy among
creative crowd workers, especially on the logo platforms, to
cut corners and reduce the time invested in a contest by more
or less blatant plagiarism, for example by taking existing
graphics from elsewhere on the internet and altering them
only slightly. For the clients, this type of copyright infringe-
ment is hard to discover and they are at risk of unwittingly
rewarding and then using a design that was in fact stolen.
If the clients find out too late, the damages for them can be
high. The terms of service of the platforms in this sector al-
ways clearly state that it is the designer who is liable, but
for a defrauded client it can be difficult to exercise that right
if the fraudster is based abroad.
The problem of plagiarism frequently also leads to con-
flicts among the participating designers. Those who try to
contribute original ideas often report stolen ideas to the client
– thus, also the policing of copyright infringements is effec-
tively outsourced to the crowd. In general, it can be said that
the extreme competition between the individuals in the cre-
ative crowd can cause a toxic work climate. All this contributes
to the fact that the creative crowdworkers are comparatively
unhappy with the working conditions and often feel unfairly
treated or even exploited (see Leimeister et al. 2016).
When analysing and evaluating specific platforms for
creative crowd work, one has to look closely at the extent to
which the platform provider is moderating the design pro-
cess in relation to how high the fee for its role as intermedi-
ary is. 99designs, for example, takes a fee of 35 to 50 per
cent of the client’s money, without making this transparent
to either the client or the designer.8 On average, 99design
takes a commission of 40 per cent, which is very high, not
only in comparison with freelance marketplaces, which only
charge about half as much. 99designs does not contribute to
the design process and, according to its terms of service, is
not liable for any outcome whatsoever. The terms also state
that when a designer finds a new client through a contest, all
future communication with that client and especially all fu-
ture commissions have to run through the platform, which
will continue to charge a fee on all ensuing transactions. The
only way out for the designer is to pay an "opt-out-fee" of
US$2,500 to the platform.9 This example shows how platforms
instrumentalise the structural power asymmetry to take ad-
vantage of the crowdworkers. The latter have a chance of
1 in 100 to "win" a payment of 250 euros for work that they
have custom-made in advance for the client. They carry all
the legal risks and on top of that have to accept an opt-out
fee that is about a thousand times higher than the average
8 While the crowd work plat form 99designs actively obfuscates the
commission it takes, the gig work platform Helpling openly lists its fees in
its terms of service. The degree of transparency differs greatly from one
platform to the next.
9 See: 99designs terms of use, section 4, "Exclusivity and Non-Circum-
vention" (as of November 2016): https://en.99designs.de/legal/terms-of-use
18
FRIEDRICH-EBERT-STIFTUNG
living space that cities need for regular long-term renting,
and that they avoid paying taxes for commercial short-term
accommodation. It is disputed how large the percentage of
professional hosts on Airbnb actually is in total. Data journal-
ism projects counter the official data published by the com-
pany with their own statistics. In the United States, this work
is done by the project Inside Airbnb (by Murray Cox), and in
Germany the website Airbnb vs. Berlin (by Studio Karat) of-
fers similar insights. The German project was able to show
that in 2015, the ten most active Airbnb hosts in Berlin to-
gether offered 281 flats. One individual even had 44 proper-
ties listed. Altogether, Airbnb lists about 12,000 flats per day
in Berlin alone, which is 0.4 per cent of all the flats in the city.
Ten per cent of these flats were provided by hosts with sev-
eral properties. However, Airbnb tries to downplay this as-
pect. In Berlin and New York, where the platform is very
popular, the authorities came to the conclusion that the com-
mercial hosting of multiple flats is unlawful. New York Attor-
ney General Eric T. Schneiderman estimated in an official
report that about two-thirds of the New York listings in 2014
were illegal (Schneiderman 2014). The report put political
pressure on Airbnb to release the actual data and in Decem-
ber 2015 the platform complied, although it only released
the listings of one specific day. Murray Cox of Inside Airbnb
and the sharing economy critic Tom Slee were able to prove,
however, that the company had purged its platform from
illegal listings shortly before the data snapshot. By doing so,
Airbnb decreased the problematic listings from 18.6 per cent
to 10.3 per cent (Cox/Slee 2016). The removal of hosts offer-
ing multiple flats happened only in New York and only for
this one occasion. Since then, the listings have slowly returned
to the original situation. This leaves little reason to believe
that the company had a real interest in reversing the trend
towards more commercial hosts. Airbnb later even admitted
that it had purged its New York data, and that 38 per cent of
its revenues in that city stem from a relatively small group of
hosts holding multiple flats with the sole purpose of renting
them out via Airbnb (Newcomer 2016).
In Berlin, the situation is similar. In May 2016 the transi-
tion period of a new law (passed in 2014) ended, which pro-
hibits the misuse of flats for short-term rentals. Now, hosts
need a special licence from the district authority if they want
to professionally offer tourist accommodation. Infringements
of this law can be punished with a fine of up to 100,000 eu-
ros. According to investigations by the German newspaper
Die Zeit (in cooperation with Airbnb vs. Berlin), the company
temporarily deleted commercial hosts in Berlin, too: the num-
ber of entire flats listed for rent dropped from 11,000 in Feb-
ruary 2016 to 6,700 in March 2016; the number of listings by
commercials hosts with multiple flats even dropped by 50
per cent down to 1,000 flats (Die Zeit 2016).
The two examples of New York and Berlin show a num-
ber of important aspects: the location-based services of the
platform economy have a stronger, more immediate but also
more visible local impact on cities than cloud work or crowd
work. At the same time, local politicians can regulate the plat-
forms much more effectively at this regional level. The ideal
of the platform providers, to have one universal terms of ser-
vice document that simply overrules or ignores the local ju-
risdictions, is met with strong opposition. This forces plat-
a particular independent contractor or client, respectively,
based on a profile with a real name and a set of ratings and
reviews derived from previous transactions. Compared with
cloud work, gig work demands more commitment from the
workers. The platforms are more personal, people know much
better who they are dealing with and they meet each other
in person. The qualities of good service personnel, such as
friendliness, punctuality, cultivated behaviour, well groomed
appearance and so on influence how users rate each other.
Even if the client does not really care who exactly is doing
the job, it is always one individual who is responsible for deli-
vering the expected results. Because gig work takes place in
the physical world, there are a lot more personal risks invol-
ved than in cloud work, where occupational accidents, traffic
accidents, theft and damage to property are of no real con-
cern (Rohrbeck 2016). In order to counter these increased ris-
ks and ensure trust between users, the identity of gig wor-
kers is checked more thoroughly in advance by the platforms,
for example by demanding a criminal record certificate. The
importance of five-star ratings and favourable mutual reviews
is more pronounced and the tracking of users by the plat-
forms providers is extended into physical space. The high de-
gree of surveillance serves the safety of the users, but it also
creates a treasure trove of very personal data. What that me-
ans was vividly illustrated when in 2012 the transportation
app Uber analysed the movement profiles of its clients to
estimate how many of them were probably having an affair.
Uber called the late-night taxi rides in question "Rides of
Glory" and published statistics for different North American
cities. However, when this sparked outrage the company
quickly withdrew this telling insight into its big data capaci-
ties.10
3.2.1 ACCOMMODATION
Airbnb, founded in 2008 and based in Silicon Valley, is the
best-known platform for the listing of accommodation by
private individuals, and with about two million listed proper-
ties in 34,000 cities all over the world, it is also the largest.
Two of its biggest competitors, Wimdu and 9Flats, merged
in October 2016 and are now based in Berlin – together they
list about half a million flats. The trend towards short-term
renting of private flats instead of booking hotels is a global,
as well as a very urban phenomenon. Most listings are in
cities and neighbourhoods popular with tourists. According
to how Airbnb portraits itself, the service primarily revolves
around organising short-term, intermediate lodging in spare
rooms or in entire flats, while the owners, who usually live
there, are away. Airbnb emphasises the private, non-commer-
cial atmosphere of other people’s homes, which is reflected
in the company’s name, a portmanteau of "air mattress" and
"bed and breakfast". But over the years, a professional mar-
ket has emerged, in which commercial hosts hold several
flats especially for the purpose of renting them out via Airbnb.
This development is the most criticised aspect of Airbnb
and similar platforms. The concern is that they take away
10 See: http://www.nytimes.com/2014/12/08/opinion/we-cant-trust-uber.
html; Uber withdrew the original post but it is archived here: ht tps://web.
archive.org/web/20140828024924/http://blog.uber.com/ridesofglory.
19
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
Uber launched a succession of new products (UberX, Uber-
GO, UberPOP, UberPOOL), which the company called "ride-
sharing services" instead of taxicabs or limos, in order to
evade legal regulations for professional transportation servic-
es. Uber promised its drivers that they can make US$40 per
hour, significantly more than minimum wage, and according
to various interviews with Uber drivers, this was also realistic,
at least for a while. The part-time drivers were happy with
the additional income, invested in their cars and became full-
time drivers. For many people who had lost their job after
the financial crash of 2008, driving for Uber seemed to offer
a new start. But after the introduction of the "ride sharing"
concept in 2012, the number of drivers grew exponentially
to 160,000 (Hall/Kruger 2015). Since then, no official num-
bers have been published, but it is estimated that as of 2016
Uber has about 450,000 active drivers per month.
The satisfaction of the drivers has since declined, while
public protests against Uber have increased (Lazzaro 2006).
The reason for this is not only that there are now too many
drivers, but also that Uber is using the mechanisms of the
platform economy described above to shift more and more
of the risks and costs of the business to the side of the inde-
pendent contractors – partly by using discount pricing to
keep the upper hand in its competition with Lyft. From the
already comparatively low fares that passengers have to pay
(up to 45 per cent less than for a regular taxi), Uber takes a
fee of 25 per cent for each ride. After having paid the 25
per cent fee to Uber, drivers still have to pay for all other
costs: the lease or credit for their car, repairs, petrol, taxes
and insurance. And they also have to absorb the cost of
driving around idly, in phases where there is less demand
or too many drivers on the streets. Because of the many var-
iables that influence the price for a ride – depending on
when, where, with what kind of car and under what sub-
brand of Uber the drivers offer their service – there is no
simple answer to what they actually earn on average.
Uber also dynamically defines in real-time how much a
ride costs, based on the traffic situation in particular parts of
the city. If there is a lot of traffic, due to peak hour or some
other local event, Uber defines this as "surge" and raises the
price. The drivers can then see the affected area on their
map and the factor by which the regular price is multiplied.
On occasions such as New Year’s Eve, during rainstorms or
after popular local sport events or concerts, when many
people want to get home at the same time, the price for a
ride can be up to eight times higher than usual.
The most detailed calculations regarding average earn-
ings on Uber available, as of 2016, are based on the data of
over a million rides. According to these numbers, the average
hourly wage, after deduction of all fees and maintenance
costs, was US$13.17 in Denver, US$10.75 in Houston and only
US$8.77 in Detroit – far below the US$40.00 Uber used to
promise its drivers (O’Donnovan/Singer-Vine 2016). On aver-
age, these drivers already earned less than the local minimum
wage, before, in January 2016, Uber introduced an additional
price cut, arguing that this was necessary to absorb a decline
in bookings due to bad winter weather.11 Regular taxicab
11 See: https://newsroom.uber.com/beating-the-winter-slump-price-
cuts-for-riders-and-guaranteed-earnings-for-drivers/.
forms such as Airbnb to adapt their service to political pressure
at least locally. At the same time, it also becomes clear that
regulators cannot blindly trust the data provided by the plat-
forms, and that independent data journalism projects serve
as an important corrective. The two examples also show that
platforms actually do change their policy if they are legally
obliged to. In the particular case of Airbnb, it was certainly
helpful that the platform projects an image of itself as being
the hub for private hosts who are primarily interested in shar-
ing underused resources without making a business out of it,
which is exactly what the legal framework in New York and
Berlin allows for. Even though the platform makes a substan-
tial part of its revenue from commercial, professional hosts,
it cannot openly fight for this without damaging its homely
sharing economy image.
3.2.2 TRANSPORTATION AND DELIVERY
SERVICES
Transportation
The public perception of platform-based taxi services is very
much informed by Uber. The company was founded in 2009
and reached a valuation of US$62.5 billion in 2016. The name
"Uber" has become shorthand for the disruptive power of the
platform company, for advocates as well as for critics. New
apps and services for gig work are commonly described or
advertised with the formula "the Uber for XYZ" and "uberi-
sation" is now a word for transforming an old business mo-
del accordingly. Its immediate competitor Lyft was founded
in 2012 and reached a valuation of US$5.5 billion in 2016. In
2015, General Motors (who also holds shares in Lyft) bought
Uber’s other competitor, Sidecar, and transformed the com-
pany into a delivery service for food (now discontinued). And
the Norwegian company Haxi is pushing into the European
market for private hire vehicles.
Seen from the passengers’ perspective, Uber offers a ser-
vice superior to that of conventional taxicabs. With just a
touch on the smartphone, they can summon a car, see who
their driver will be and where they are at the moment. They
can see a representation of the car approach in real-time on
their screen and they can give the driver a call to coordinate
details. In comparison with regular cabs, the passengers per-
ceive the service as more personal, friendly, reliable and also
as cheaper. No cash is needed for the transaction and the
passengers also feel safer because they know that Uber is
tracking all rides via GPS. That way, parents can have an Uber-
driver pick up their kids from school and always see exactly
where they are.
Also from the drivers’ perspective, Uber was seen as a
blessing, at least at the beginning. As with all other forms of
digital labour, people enjoy the flexibility and autonomy to
work whenever and how much they like. It is the prime moti-
vation for drivers to sign up with Uber and Lyft (in the United
States, many drivers use both platforms in parallel). Originally,
Uber was a service for luxury limousines with professional
drivers, regularly licensed and insured (UberBLACK). Howev-
er, propelled by the competition with Lyft, the service evolved
into a platform that enabled anybody with a driving license
and a private car to become a freelance chauffeur. After 2012,
20
FRIEDRICH-EBERT-STIFTUNG
the financial punishments in Europe and the US litigation for
now. Nevertheless, the class action lawsuits could evolve into
a serious threat to the business model, not only for Uber but
also for similar companies in the platform economy (Kessler
2015 ).
Delivery Services
To some extent, the digital labour platforms for transpor-
tation blend into those for delivering goods from supermar-
kets and meals from restaurants. With its services UberRUSH
and UberEATS, the leading "ride sharing" company is also
pushing into these markets. As of 2016, this area of the gig
economy is very dynamic and a lot of new companies emerge
only to disappear again because a competitor buys them.
Parallel to North American gig work platforms for deliveries
– such as Spoon Rocket, Yelp Eat24, DoorDash, Instacart and
Postmates – also a number of German companies have
sprung up. The market-listed German start-up incubator Rock-
et Internet has invested a lot of venture capital in compet-
ing delivery services. The Berlin-based companies Hello Fresh
(delivers via UPS, so it is not gig work) and Delivery Hero be-
long to the German start-up "unicorns", with valuations of
US$2.9 billion and US$3.1 billion, respectively. Other competi-
tors are MyLorry, Pizza.de, Foodpanda, Lieferando (which is
one of eleven subsidiaries of the Dutch company Takeaway.
com), Deliveroo and Foodora (which is a subsidiary of Deliv-
ery Hero). There are more, but a comprehensive list is beyond
the scope of this study and as yet there is little reliable data
for most of these companies. Therefore, only the basic princi-
ples are described in what follows, based on the example of
Deliveroo.
An investment banker founded the platform for restaurant
deliveries in 2013 in London. As of August 2016, the compa-
ny had raised a total of US$473 million in venture capital and
reached a valuation of about US$1 billion; the service is avail-
able in 84 cities across twelve countries and 20,000 self-em-
ployed cyclists deliver food from more than 16,000 restau-
rants; from November 2015 to August 2016 the company
grew by 400 per cent (Fegor/Murgia 2016). As in all other ar-
eas of the platform economy, Deliveroo mediates between
supply and demand. It allows restaurants that previously had
no such service to have their food delivered to their custom-
ers. For this, Deliveroo charges a fee of 30 per cent from the
restaurants, which, despite the high fees, hope to make more
profit due to the increased outreach. Trade organisations in
this sector, such as the German Hotel and Catering Associa-
tion, are concerned about the emergence of monopolies. It is
feared that customer loyalty will no longer be with the res-
taurant but with the delivery platform, which will, in the me-
dium term, be able to dictate the conditions and margins, as
already happened with Uber (Zacharakis 2016).
Contrary to the work conditions at Uber, the independent
contractors working for Deliveroo have fixed three-hour shifts,
in which they have to be available on demand in a certain
district. In the United Kingdom, Deliveroo pays them £7 per
hour plus £1 for each delivery; in Germany it is 7.50 euros,
plus 1 euro per delivery. The delivery is typically done by bike,
which means that the high initial investment of buying or
leasing a car, as well as the costs for fuel are not a problem
drivers do not necessarily earn more, but they also do not
carry that much risk. The investments of the independent
contractors in a car of their own (and one of a certain stand-
ard, defined by Uber), can hardly be paid back when the av-
erage income through ride-sharing drops below minimum
wage.12
Nevertheless, a new sub-prime credit market is emerging
here, not unlike that of the 2008 housing bubble. Uber struck
a deal worth over US$1 billion with investment bank Gold-
man Sachs to be able to lend money – via Uber’s subsidiary
Xchange Leasing – to drivers whose credit score is too low
for normal credit (Smith 2016). But if poor drivers have to go
into debt to keep up on the platform, it becomes clear that
Uber’s claim to be just a ride-sharing service in the sharing
economy is false. Also the promised flexibility in the working
hours is untenable under these conditions. Those who go
into debt to pay for a car cannot afford to be picky about
when and where to drive, they need a maximum occupancy
rate.
In addition to that, drivers also carry the financial risk of
having car accidents, because, at least in the United States,
most independent contractors of Uber and Lyft only have in-
surance for the private use of their cars. Which is why in case
of an accident, they have to hide from their insurance com-
pany that they were using their vehicle for commercial rides.
Even though Uber used to officially refer to its drivers as
"partners", it is obvious that they are not treated as equals.
They have to negotiate from a position of weakness and are
the only factor left in Uber’s calculation where the company
can cut costs to gain a financial advantage over its competi-
tors. And according to Travis Kalanick, CEO of Uber, the work-
ers are to be replaced as soon as possible by self-driving cars
anyway. Together with Lyft and Google, Uber is part of the
"Self-Driving Coalition", a lobbying group for autonomous ve-
hicles, and is already testing the technology in a research and
development project based in Pittsburgh (Crook 2016).
Across the world, Uber is involved in hundreds of law-
suits, and its services have become illegal in many cities and
even entire countries – for example in Spain, France, Belgium,
the Netherlands, Hungary and Germany.13 In a lot of these
cases, legal bans were the result of massive protests by local
cab drivers, and often Uber tried to simply ignore the rulings.
In France, this went so far that two top executives of Uber
were taken into custody and had to pay fines of 20,000 and
30,000 euros, respectively. The company itself had to pay a
fine of 800,000 euros plus court costs for the infringements
(Scott 2016). In the United States the company is facing in-
creasing resistance, too: its drivers are unionising and filing
class action lawsuits for misclassification as independent con-
tractors. In one of these lawsuits, filed by 380 drivers from
California and Massachusetts, Uber reached a settlement by
paying the drivers US$100 million. Because of this litigation,
the drivers continue to be classified as independent contrac-
tors (Isaac/Schreiber 2016). Shortly afterwards, however,
Uber was able to get an investment of US$3.5 billion from
Saudi Arabia. With that kind of backing, Uber can shrug off
12 See: www.NotCoolUber.com.
13 Wikipedia has a comprehensive list of Uber ’s legal status across the
world: https://en.wikipedia.org/wiki/Legal_status_of_Uber's_service.
21
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
Competitor Homejoy.com had to deal with four such law-
suits, which deterred investors and ultimately led to the clo-
sure of the platform in 2015 (Deamicis 2015).
here and the maintenance costs are much less of an issue.
However, the risk of getting hurt in an accident is compara-
tively high, given that the cyclists spend most of their work-
ing hours in traffic. It is the independent contractors’ respon-
sibility to have health insurance, accident insurance and third-
party liability insurance. However, according to an interview
with a deliverer from Berlin, many of them cut these costs
at least partially and cycle at their own risk (Lehmann 2016).
3.2.3 HOUSEHOLD SERVICES AND PERSONAL
SERVICES
The final category of the gig economy, again described here
only briefly, involves services that are provided by indepen-
dent contractors in the homes of their clients. Because they
are set in private environments, the factors of trust, quality
and continuity (the same person showing up to do the job)
play a much greater role than is the case with delivery ser-
vices. Again, there are as yet no academic studies about this
area of gig work, only newspaper reports and interviews
with platform providers and independent contractors. It is
quite possible that this category will have to be further sub-
divided in the future. Currently, the most well-known plat-
forms in this area, at least in Germany, primarily provide clea-
ning services; examples are Helpling and Book A Tiger, both
founded in 2014. American platforms such as TaskRabbit
(founded in 2008) and Handy.com (founded in 2012) also of-
fer errands, repairs, Ikea furniture assembly and the like –
they are generalists, promising to free busy customers from
all possible household chores. For all these household ser-
vices, the client’s level of trust towards the gig workers must
be high enough to give them unsupervised access to their
private home and the two parties do not necessarily have to
spend time together.
However, there are also personal services such as home
care and nursing, babysitting and, last but not least, prostitu-
tion (Kauf-mich.com, 2009; Olalah.com, 2015), which are about
human exchange, time spent together, and which are now
mediated by platform-based business models as well. For this
development, however, it is still early days.
There are two important and interconnected limiting fac-
tors for the success of platform-based household services
and personal services: people who have to do cleaning jobs
as self-employed and exchangeable gig workers are (or are
at least perceived to be) less reliable and deliver inferior quali-
ty, if left unsupervised, than clients and platform providers
expect. The platforms therefore try to monitor and control
the way the work is done – the personal appearance of the
gig workers, their clothes, their schedule. All of this makes
the platforms for cleaning services particularly vulnerable to
getting sued for misclassification of the workers. Both the le-
gal pressure and the striving for quality assurance have al-
ready led several platforms in this sector, in Germany as well
as in the United States, to switch back to conventional em-
ployment or at least to pay and train the workers better, and
generally invest in a more reliable, stable staff. In 2014,
Handy.com was sued for misclassification of employees as
independent contractors. As so often, the legal case was re-
ferred to a court of arbitration, but the class action law suits
remain a threat to the business (see: Seiner 2016; Said 2014).
22
FRIEDRICH-EBERT-STIFTUNG
23
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
vent beneficial and necessary innovation in the service sector
out of fear of change. Established organisational forms and
business models must not automatically fall under govern-
ment protection and a modern society also needs unregulat-
ed spaces to test and develop innovative concepts, for exam-
ple to organise supply and demand or the division of labour
more efficiently. However, when a technology or service
leaves the sandbox stage and becomes so widespread that
it affects the lives of millions of people, the government
must put the interests of society before the business inter-
ests of individual entrepreneurs. Regulation is necessary to
protect social security, workers’ rights and consumer rights
also in the digital realm. There is no reason why hard-won
labour laws should not apply, merely because the work is
organised via a platform.
The platform model offers many advantages over conven-
tional forms of organisation, but it is necessary to distribute
the entrepreneurial opportunities and risks, as well as the so-
cial costs, more fairly. Without regulation, it has become clear
over recent years that mainly the platform providers and to
some extent their clients benefit from the new opportunities,
while they shift the entrepreneurial risks to the contractors
and leave the social costs, such as threatening old-age pov-
erty, to society.
The question therefore is: how can the development of
digital labour platforms be influenced positively in terms of
fair and socially acceptable working conditions? First, it needs
more research, a consistent categorisation and taxonomy of
the different platform-based outsourcing mechanisms and
more robust data on their usage. The platforms already have
this data, but often treat it as a business secret. It is still un-
clear, for example, which digital labour platforms are already
profitable and which ones so far are still only burning the ven-
ture capital of their investors. Reliable data is also important
to differentiate active workers from those who have only
registered once and never put in any hours, to calculate aver-
age hourly wages and to find out how many people have
turned cloud work or gig work into a full-time job (according
to all estimates, only a small minority). The existing studies
are mostly based on interviews with individual crowd work-
ers and gig workers; in addition, more quantitative studies
The future of work will be shaped by a number of techno-
logy-driven developments: automation, platform-based
outsourcing of services to self-employed individuals, the di-
vision of formerly secure jobs into ever smaller and precarious
tasks, and the constant big-data tracking of the work process.
These trends are not happening in isolation, but are mutually
dependent.
Automation threatens to destroy jobs, for example by re-
placing professional drivers with self-driving cars, and by in-
troducing robots that can work hand in hand with humans on
the assembly line; but it also creates numerous new jobs (ar-
guably sometimes of lower quality), for example in the micro-
tasking sector, where humans continue to be needed for the
creation of reliable data sets that form the basis for automa-
tion. An often-cited study by researchers from Oxford Uni-
versity analysed the likelihood of specific jobs being auto-
mated (Frey/Osborn 2013). It is advisable to use a similar
approach to assess the likelihood of specific occupational
fields to be disrupted by or transformed into cloud work,
crowd work or gig work. Over the past ten years, we have
seen a rapid growth of these new forms of labour, but there
is a lot of disagreement among experts about whether this
development will continue to accelerate or reach a plateau
soon. As with automation, the change is not yet happening
across the board; only some areas are particularly prone to it.
It is important to look closely at these emerging digital la-
bour markets, from the perspective of both research and
politics, to become familiar with their mechanics and devel-
op ways to fix them, where necessary.
As of 2016, the new forms of digital labour only affect a
small percentage of the labour market and only rarely take
the function of a full-time job. Not every job can be out-
sourced to the crowd. But the basic principles of the platform
economy do indeed have the potential to fundamentally dis-
rupt the way work is distributed in society.
The already successfully implemented regulatory initia-
tives in the area of gig work show, however, that governments
can actively influence the development, at least of the loca-
tion-based digital labour markets for gig work, and that a
deterministic attitude towards technology-driven disruption
is out of place. At the same time, it is important not to pre-
4
OUTLOOK
24
FRIEDRICH-EBERT-STIFTUNG
in various platform-specific forums, they usually revolve
around how to get the best out of the difficult working
conditions individually; they are less about workers’ partici-
pation, collective bargaining and improvement of the digital
labour model in general. A crowd, in the physical world as
well as online, can potentially bundle the forces of all partici-
pants. The group is then stronger than the sum of its parts.
The individuals in the working crowd, however, are typically
in total competition with each other, thereby weakening
each other instead of acting together and improving their
joint negotiating power with the clients and, more impor-
tantly, the platform providers.
Having said that, over the past two years, a new move-
ment has been taking shape under the name "Platform Co-
operativism", which is not trying to negotiate with platform
owners but aspires to run its own platforms. Initiated and
promoted by German-born digital labour expert and activist
Trebor Scholz, professor at the New School in New York, the
movement advocates a new platform type based on cooper-
ative ownership (Scholz 2014, 2016). With the revival of this
old approach and its application to new forms of labour, the
crowd workers and gig workers can regain control over their
working conditions. By building and owning the platforms
themselves, they can design working conditions from the
bottom up, which are continuously determined by workers’
participation instead of investors’ expectations of exponen-
tial growth and profit or economic-rent maximisation. It is
highly questionable whether comparatively small and local
coop-platforms can compete economically with exploitative
competitors operating on a global scale. However, as with
"organic" and "fair trade" labels, the activists could foster a
willingness in their clients to consciously pay a little more for
a service that is guaranteed to have substantially better, more
ethical, production conditions.
Also outside the cooperative model, on conventional dig-
ital labour platforms, it would be important for the clients to
have more reliable information about the working conditions.
Not only positive fair trade labels, but also warnings, similar
to those on advertising for alcohol or gambling, could be
useful to give larger companies an incentive to stay away
from exploitative platforms as part of their corporate social
responsibility efforts. In the same vein, efforts at self-regula-
tion by the platforms should be supported. In 2015, the Ger-
man crowd work platform Testbirds published a "Code of
Conduct – a guideline for a prosperous and fair cooperation
between companies, clients and crowdworkers".14 The docu-
ment, signed also by the management of the platforms Click-
worker and Streetspotr, addresses a lot of the grievances of-
ten brought forward by crowd workers and takes a stand for
fair payment and open and transparent communication be-
tween the different stakeholders. If nothing else, the docu-
ment proves that there is also a willingness on the side of
the platform owners to counter the negative image of the
industry and to improve conditions. Such advances can have
an important signalling effect, at least for certain market
segments, to reverse the downside spiral in terms of quality
and price.
14 See: http://www.crowdsourcing-code.com/.
are needed, as well as data journalistic watchdog projects
(similar to Inside Airbnb and Airbnb vs. Berlin) to counter the
official and often misleadingly selective data published by
the platforms.
In general, there is a need for more transparency regard-
ing the processes on the platforms, the terms of service
agreements, the mediation fees incurred and the liability
rules. The terms of service agreements should be presented
in a form that allows users to make informed decisions about
the conditions under which they want to work. The platform
companies have outstanding capabilities in the field of us-
er-friendly interface design. They should be obliged to use
these skills to create navigable surfaces for the terms of ser-
vice agreements, with more options than just an all-encom-
passing "agree" button.
Perhaps it will also require independent, trustworthy digi-
tal labour organisations, similar to consumer advocate institu-
tions, to test and evaluate the working conditions on the vari-
ous digital labour platforms, and warn workers, independent
contractors and clients of particularly problematic clauses in
the terms of service agreements. The watchdog project Fair-
CrowdWork.org organised by Germany’s largest trade union,
IG Metall, is already a step in this direction. On this site, trade
union legal experts offer assessments and warnings regarding
the terms of service agreements of numerous digital labour
platforms. However, given that there are hundreds of relevant
(and thousands of marginal) digital labour platforms, with
constantly changing terms of service and jurisdictions spread
across the world, it is hardly possible to keep up with the
workload of this legal evaluation in real-time.
The evaluation of working conditions on digital labour
platforms is easier, at least in principle, because the workers
themselves can do this; in other words, it can be crowd-
sourced. This is also a feature of FairCrowdWork.org, inspired
by the tool Turkopticon. Developed by Lilly Irani and Six Sil-
berman, Turkopticon is a successful approach to reverse the
information asymmetry on Amazon Mechanical Turk and
give the crowdworkers the opportunity to evaluate their
clients – not just the other way around (Irani and Silberman
2013 & 2014). Turkopticon is a tailor-made external add-on, a
very useful hack, to make one particular platform a bit fairer.
For a meta-evaluation website such as FairCrowdWork.org,
however, the challenge is to find enough workers from across
many different platforms to obtain meaningful and reliable
assessments.
Cross-platform mobilisation and organisation of crowd-
workers turns out to be difficult. It seems that many crowd-
workers, even if they are dissatisfied with the working condi-
tions or the remuneration on a specific platform, have little
interest in either self-organisation or representation of their
interests by trade unions. Plans for potential regulation of digi-
tal labour platforms by the state are typically met with great
scepticism. For the majority of the crowd, work on the plat-
forms is temporary and sporadic anyway, a small side-job,
not worth fighting for. For them it is much easier and more
promising to simply search for a new platform with better
working conditions. Professional full-time crowdworkers, on
the other hand, fear that regulation of the platforms would
not improve their jobs, but destroy them. Although there are
forms of self-organisation by crowdworkers and gig workers
25
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
In addition, it must be ensured that people are not involunta-
rily pushed into these precarious new working conditions, for
example because conventional companies that do pay for
social security, safety and training of their workers can no
longer compete with the cheaper platform-based companies
that have found a way to route around any social costs of la-
bour. Disruption must not be an end in itself. The new struc-
tures must be measured against their social compatibility
and, if necessary, regulated by law in order not to harm the
public good in the long run.
At the same time, it must be remembered that, within the
spectrum of different platform types outlined in this discussi-
on paper, it is the crowd workers in particular who will, in all
likelihood, be subjected to wage dumping, at least to a large
extent, simply because individuals are interchangeable in the
crowd. In an open crowd that is recruited from a global pool
of workers without pre-selection based on qualifications, it is
not possible to secure adequate pay for each individual,
because a potentially high proportion of unqualified and un-
motivated workers must be absorbed in the overall balance
– either by paying only small amounts to everybody, or by
paying only the very best contributors properly, leaving all
others empty handed. The demand for a minimum wage for
global cloud work also faces the problem of what particular
countries take as a benchmark for fair pay, when workers and
clients can come from anywhere. As we have seen before in
earlier phases of outsourcing, certain jobs will inevitably be
shifted to developing countries still with a low cost of living,
but relatively high levels of education and good knowledge
of English. Countries such as India, Indonesia and the Philip-
pines already play an important role here.
Regulatory efforts have significantly better prospects in
the area of gig work. On one hand, because clients and con-
tractors are subject to the same local jurisdiction, on the oth-
er hand, because claims for misclassification of employees as
independent contractors are particularly valid in this area of
the platform economy. In addition, gig work often disrupts
service sectors that are already well organised when it comes
to minimum wage and worker protection (in contrast to data
processing and design jobs common in crowd work). For
household services, be it cleaning or repair work, there is also
the argument that governments could potentially benefit
from the transformation towards platform-based business
models, because this now makes visible and taxable areas of
the economy in which clandestine employment has tradition-
ally been endemic.
In the household services sector, one phenomenon is be-
coming particularly clear, which ultimately applies to all forms
of digital labour: when the quality of the results and the trust
between client and customer become more important than a
low price, the platform model quickly reaches its limits. In those
cases, it pays to invest in individual workers, train them and
bind them to the employer with fair working conditions and
real career prospects. This also makes it highly unlikely that
the entire labour market will eventually dissolve into micro-
tasking. Nonetheless, ten years after the emergence of the
first crowd work platforms, it can be said with some certainty
that this is not just a temporary phenomenon. A new low-
wage sector for digitally mediated labour has been estab-
lished and it will continue to exist and grow.
It is the job of governments to ensure the continuity of
the social security systems in order to be able to take care of
those who are no longer able to do so themselves. The great
challenge therefore is to generate social security contributions
even from microtasks, to oblige platform providers and clients,
as the two parties that benefit the most from the contingent
on-demand workforce, to at least partly carry the social costs
or oblige them to pay so much that the independent con-
tractors can themselves cover the costs for things like health
insurance, and are also obliged to do so.
26
FRIEDRICH-EBERT-STIFTUNG
List of Figures
6
7
10
Figure 1
Categorisation of digital labour markets in the platform
economy 1/2
Figure 2
Categorisation of digital labour markets in the platform
economy 2/2
Figure 3
Three-sided platform architecture
Bibliography
Ahn, Luis von 2005: Human Computation, Pittsburgh.
Al-Ani, Ayad; Stumpp, Stefan 2015: Motivationen und Durchsetzung von
Interessen auf kommerziellen Plattformen: Ergebnisse einer Umfrage un-
ter Kreativ- und IT-Crowdworkern, HIIG Discussion Paper Series 2015 (5),
http://www.ayad-al-ani.com/pdf/SSRN-id2699065.pdf (13.8.2016).
Al-Ani, Ayad; Stumpp, Stefan; Schildhauer, Thomas 2014: Crowd-Studie
2014: Die Crowd als Partner der deutschen Wir tschaft, HIIG Discussion
Paper Series 2014 (2), http://papers.ssrn.com/sol3/papers.cfm?abstract_
id =24370 07 (13.8. 2016).
Aneesh, Aneesh 2009: Global Labor: Algocratic Modes of Organization,
in Sociological Theory 27 (4), Milwaukee.
Araujo, Ricardo M. 2013: 99designs: An Analysis of Creative Competition
in Crowdsourced Design, in: First AAAI Conference on Human Computa-
tion and Crowdsourcing.
Asher-Schapiro, Avi 2014: Against Sharing, in: Jacobin Mag, 19.9.2014,
https://www.jacobinmag.com/2014/09/against-sharing/ (13.8.2016).
Balestier, Courtney 2016: The Supermarket Must Die: App-Fueled Services
Can Kill It, in Wired, 14.4.16.
Benner, Christiane 2015: Crowdwork – zurück in die Zukunft?: Perspek ti-
ven digitaler Arbeit , Frank furt am Main.
Botsman, Rachel 2012: Welcome to the New Reputation Economy, in:
Wired UK, 20.8.2012, http://www.wired.co.uk/article/welcome-to-the-
new-reputation-economy (13.8.2016).
Brabham, Daren C. 2013: Crowdsourcing, MIT Press, Cambridge.
Brabham, Daren C. 2012: The Myth of Amateur Crowds, in: Information,
Communication & Society 15 (3), pp. 394– 410.
Chen, Adrian 2014: The Laborers Who Keep Dick Pics and Beheadings Out
of Your Facebook Feed, in: Wired, 23.10.14.
Chen, Michelle 2015: Uber Wins a Battle With New York, Now It’s War, in:
The Nation, 24.7.2015, https://www.thenation.com/article/uber-wins-
battle-with-new-york-now-its-war/ (13.8.2016).
Cherry, Miriam A. 2012: The Gamification of Work, in: Hofstra Law Review
40 (4), Hempstead, New York.
Cherry, Miriam A. 2016: Beyond Misclassification: The Digital Transforma-
tion of Work, in: Comparative Labor Law & Policy Journal, Champaign.
Coldwell, Will 2016: Airbnb: From Homesharing Cool to Commercial Giant,
in: The Guardian, 18.3.2016.
Cox, Murray; Slee, Tom 2016: How Airbnb’s Data Hid the Facts in New
York City, Report vom 10.2.2016, http://whimsley.s3.amazonaws.com/
wordpress/wp-content/uploads/2016/02/how-airbnbs-data-hid-the-
facts-in-new-york-city.pdf (13.8.2016).
Crook, Jordan: Uber Confirms it ’s Testing Self-driving Cars in Pittsburgh,
19.5.2016, http://techcrunch.com/2016/05/19/uber-confirms-its-testing-
self-driving-cars-in-pittsburgh/ (13.8.2016).
Däubler, Wolfgang 2015: Crowdworker – Schutz auch außerhalb des
Arbeitsrechts?, in: Benner, Christiane (Hrsg.): Crowdwork – zurück in die
Zukunft?, Frankfurt am Main.
Danaher, John 2016: The Threat of Algocrac y: Reality, Resistance and
Accommodation, in: Philosophy & Technology 2016, pp. 1–24.
Davidson, Adam 2016: Managed by Q’s „Good Jobs“ Gamble, in: The New
York Times, 25.2.2016.
Dayen, David 2016: Uber Spends $100 Million to Save its Business Model,
But It May Have Just Doomed It, in: Alternet, 27.4.2014.
Deamicis, Carmel 2015: Homejoy Shuts Down Af ter Battling Worker
Classification Lawsuits, in: Recode, 17.7.2015, www.recode.net/2015/7/17/
11614814/cleaning-services-startup-homejoy-shuts-down-after-battling-
worker (13 .8. 2016).
Deutschkron, Shoshana 2013: The Rise of Specialists Online: Growing
Oppor tunity for a Long Tail of Skills, Upwork Blog, ht tps://www.upwork.
com/blog/2013/08/1billion-odeskskillslongtail/ (31.5.2016).
Die Zeit 2015: Frankreich: Uber-Manager in Paris festgenommen, 29.7.2015,
http://www.zeit.de/mobilitaet/2015-06/uberpop-frankreich-festnahme-
fahrdienst-p rote st (13. 8 .2016).
27
DIGITAL LABOUR MARKETS IN THE PLATFORM ECONOMY
Kelty, Christopher M. 2008: Two Bits: The Cultural Significance of Free
Software, Durham.
Kenney, Mar tin 2014: Rethinking Labor (and Capital) in the Era of the
Cloud, Vortrag auf dem BRIE-ETLA Annual Meeting, Helsinki, 29.8.2014.
Kessler, Sarah 2014: Pixel And Dimed: On (Not) Getting By In The Gig Eco-
nomy, in: Fast Company, 18.3.2014.
Kessler, Sarah 2015: The Domestic Workers Alliance Creates New Frame-
work For Improving Gig Economy Jobs, in: Fast Company, 6.10.2015.
Kessler, Sarah 2015: The Gig Economy Won’t Last Because It’s Being Sued
to Death, in: Fast Company, 17.2.2015.
Khaleeli, Homa 2016: The Truth About Working for Deliveroo, Uber and
the On-demand Economy, in: The Guardian, 15.6.2016, https://www.the-
guardian.com/money/2016/jun/15/he-truth-about-working-for-deliveroo-
uber-and-the-on-demand-economy (13.8.2016).
Kingsley, Sara C.; Gray, Mary L.; Suri, Siddhar th 2014: Monopsony and the
Crowd: Labor for Lemons?, Microsoft Research, New York.
Kittur, Aniket; Nickerson, Jeffrey V.; Bernstein, Michael S.; Gerber, Eliza-
beth M.; Shaw, Aaron; Zimmerman, John; Lease, Mat thew; Horton, John J.
2013: The Future of Crowd Work, San Antonio, CSCW Conference Paper,
San Antonio.
Klebe, Thomas; Neugebauer, Julia 2014: Crowdsourcing für eine Handvoll
Dollar, in: Arbeit und Recht, 2014 (1), 62. Jg., pp. 4-7.
Knight, Sam 2016: How Uber Conquered London, in: The Guardian,
27. 4.2016 .
Kosner, Anthony Wing 2015: Google Cabs and Uber Bots Will Challenge
Jobs Below The API ’, in: Forbes, 4.2.2015.
Kreider, Tim 2013: Slaves of the Internet, Unite!, in: New York Times,
26 .10 . 2013 .
Lee, Min Kyung; Kusbit, Daniel; Metsky, Evan; Dabbish, Laura 2015: Working
with Machines: The Impact of Algorithmic and Data-Driven Management
on Human Workers, in: Proceedings of the 33rd Annual ACM Conference,
pp. 1.603-1.612, CHI ’15, New York.
Lazzaro, Sage 2016: Uber Drivers Plan Boycot t After Fare Cuts Slash Their
Earnings to Below Minimum Wage, in: Observer, 19.1.2016, http://obser-
ver.com/2016/01/uber-drivers-plan-boycott-after-fare-cuts-slash-their-ear-
nings-to-below-minimum-wage/ (13.8.2016).
Lehmann, Hendrik 2016: „Deliveroo kürzt uns einfach das Gehalt“, in:
Tagesspiegel, 23.5.2016, http://digitalpresent.tagesspiegel.de/deliveroo-
kuerzt-uns-einfach-das-gehalt (13.8.2016).
Leimeister, Jan Marco et al. 2016: Crowd Work im Netz, Arbeitspapier der
Hans-Böckler-Stiftung.
Leimeister, Jan Marco; Zogaj, Shkodran; Dur ward, David 2015: New Forms
of Employment and IT – Crowdsourcing, in: 4th Conference for the Regu-
lating for Decent Work Network, Genf.
Leong, Nancy 2014: The Sharing Economy has a Race Problem, in: Salon,
2.11. 2014 .
Lobo, Sascha 2014: Auf dem Weg in die Dumpinghölle: Sharing Economy
wie bei Uber ist Plattform-Kapitalismus, in: Spiegel Online, 3.9.2014.
Martin, David; Hanrahan, Benjamin V.; O’Neill, Jacki; Gupta, Neha 2014:
„Being a Turker“. In Proceedings of the 17th ACM Conference on Compu-
ter Supported Cooperative Work & Social Computing, pp 224–235.
Marvit, Moshe Z. 2014: How Crowdworkers Bec ame the Ghosts in the Di-
gital Machine, in: The Nation, 4.2.2014.
Morozov, Evgeny 2016: The State Has Lost Control: Tech Firms Now Run
Western Politic s, in: The Guardian, 27.3 2016.
Motoyama, Marti; McCo, Damon; Levchenko, Kirill; Savage, Stefan; Voelker,
Geoffrey M. 2011: Dirty Jobs: The Role of Freelance Labor in Web Service
Abuse, in: Proceedings of the 20th USENIX Conference on Security, 14–14.
SEC’11. Berkeley, CA, USA: USENIX Association, 2011.
Newcomer, Eric 2016: Airbnb Says It Removed 1,500 Listings in New York
Before Data Release, in: Bloomberg, 25.2.2016, http://www.bloomberg.
com/news/articles/2016-02-25/airbnb-says-it-removed-1-500-listings-in-
new-york-before-data-release (13.8.2016).
O’Donovan, Caroline; Singer-Vine, Jeremy 2016: How Much Uber Drivers
Actually Make Per Hour, in: BuzzFeed News, 23.6 .2016, https://www.buzz-
feed.com/carolineodonovan/internal-uber-driver-pay-numbers (13.8.2016).
OECD 2015: OECD Employment Outlook 2015, Paris.
Die Zeit 2016: Ferienwohnungen: Airbnb kündigt Vermietern in Berlin,
27.4.2016, http://www.zeit.de/wirtschaft/unternehmen/2016-04/airbnb-
berlin-ferienwohnungen-vermieten-zweckentfremdung-gesetz (13.8.2016).
Dzieza, Josh 2015: The Rating Game: How Uber and its Peers Turned us
Into Horrible Bosses, in: The Verge, 28.10.15.
European Agency for Safety and Health at Work 2015: A Review on the Fu-
ture of Work: Online Labour Exchanges or „Crowdsourcing“: Implications
for Occupational Safety and Health, https://osha.europa.eu/de/
tools-and-publications/publications/future-work-crowdsourcing (13.8.2016).
Eyeka 2016: The State of Crowdsourcing in 2016, Paris.
Fish, Adam; Srinivasan, Ramesh 2012: Digital Labor Is the New Killer App,
in: New Media & Society 14 (1), pp. 137–152.
Frey, Carl Benedik t; Osborne, Michael A . 2013: The Future of Employment:
How Susceptible Are Jobs to Computerisation, Oxford.
Giles, Jim 2011: Getting the Job Done, With a Silicon Boss, in: New Scien-
tist 209 (2798), pp. 20–21.
Gray, Mary L.; Ali, Syed Shoaib; Suri, Siddharth; Kulkarn, Deepti 2016: The
Crowd is a Collaborative Network, in: Proceedings of the 19th ACM Con-
ference on Computer-Supported Cooperative Work & Social Computing,
134–147, CSCW ’16, New York.
Greenbaum, David 2014: Humans And Computers Will Come Together For
Middle Work, in: TechCrunch 12.7.2014, https://techcrunch.com/2014/07/12/
humans-and-computers-will-come-together-for-middle-work/ (13.8.2016).
Hack, Günter 2014: Internetkultur: Der Aufstieg des Datenproletariats, in:
Die Zeit, 10.9.2014, http://www.zeit.de/kultur/2014-09/daten-proletari-
at- interne t (13.8 . 2016 ).
Hagiu, Andrei; Wright, Julian 2015: Multi-Sided Platforms, in: Harvard
Business School, 1.11.2015.
Hall, Jonathan V.; Krueger, Alan B. 2015: An Analysis of the Labor Market
for Uber’s Driver-Partners in the United States, Princeton.
Halzack, Sarah 2014: Elance-oDesk Flings Open the Doors to a Massive
Digital Workforce, in: The Washington Post 13.6.2014, https://www.was-
hingtonpost.com/business/freelancers-from-around-the-world-offer-soft-
ware-developing-skills-remotely/2014/06/13/f5088c54-efe7-11e3-bf76-
447a5d f6411f_ story.h tm l (13.8. 2016).
Hamburger Institut für Sozialforschung 2015: Von Maschinen und Men-
schen: Arbeit im digitalen Kapitalismus, Bd. 36, Hamburg.
Harman, Greg 2014: The Sharing Economy Is Not as Open as You Might
Think, in: The Guardian, 12.11.14.
Harris, Christopher G . 2011: Dirty Deeds Done Dirt Cheap: A Darker Side
to Crowdsourcing, in: IEEE Third International Conference on Social Com-
puting (SocialCom’11).
Hatton, Celia 2015: China „Social Credit “: Beijing Sets Up Huge System,
26.10.2015, http://www.bbc.com/news/world-asia-china-34592186
(13.8. 2016).
Hodson, Hal 2013: Crowdsourcing Grows Up as Online Workers Unite, in:
New Scientist, Feb. 2013.
Holland, Martin 2016: Peeple: App zum Bewer ten von Menschen trotz
heftiger Kritik verfügbar, 8.3.2016, http://www.heise.de/newsticker/mel-
dung/Peeple-App-zum-Bewerten-von-Menschen-trotz-heftiger-Kritik-
verfuegbar-3130254.html (13.8.2016).
Hunt, Elle 2016: Peeple, the „Yelp for People“ Review App, Launches in
North America on Monday, in: The Guardian, 7.3.2016.
Huws, Ursula 2015: iCapitalism and the Cyber tariat, in: Monthly Review 66 (8).
Isaac, Mike; Scheiber, Noam 2016: Uber Settles Cases With Concessions,
but Drivers Stay Freelancers, in: New York Times, 21.4.2016, http://www.
nytimes.com/2016/04/22/technology/uber-settles-cases-with-concessi-
ons-but-drivers-stay-freelancers.html (13.8.2016).
Irani, Lilly C. 2015: Justice for „Data Janitors“, in: Public Book s, 15.1.2015,
http://www.publicbooks.org/nonfiction/justice-for-data-janitors (13.8.2016).
Irani, Lilly C.; Silberman, M. Six 2014: From Critical Design to Critical Infra-
struc ture: Lessons from Turkopticon, in: Interac tions 21 (4), pp. 32–35.
Irani, Lilly C., Silberman, M. Six 2013: Turkopticon: Interrupting Worker Invi-
sibility in Amazon Mechanical Turk, in: Proceedings of the SIGCHI Confe-
rence, pp. 611–620, CHI ’13, New York.
Kasperkevic, Jana 2016: Airbnb Purged More than 1.000 New York Lis-
tings to Rig Survey – Repor t, in: The Guardian, 10.2.2016.
28
FRIEDRICH-EBERT-STIFTUNG
Staun, Harald 2013: Shareconomy: Der Terror des Teilens, in Frankfurter
Allgemeine Zeitung, 22.12.2013.
Stone, Brad 2012: My Life as a TaskRabbit, in: BusinessWeek, 13.9.2012.
Strube, Sebastian 2014: Crowdwork: Vom Entstehen der digitalen Arbei-
terklasse, Radiobeitrag, Zündfunk, Bayerischer Rundfunk, 16.3.2014.
Sundararajan, Arun 2015: The „Gig Economy“: is Coming. What Will it
Mean for Work?, in: The Guardian, 26.7.2015.
Surowiecki, James 2015: Gigs with Benefits, in: The New Yorker, 6.7.2015.
Swarns, Rachel L. 2014: Freelancers in the „Gig Economy“ Find a Mix of
Freedom and Uncer tainty, in: The New York Times, 9.2.2014.
Tanz, Jason 2014: How Airbnb and Lyft Finally Got Americans to Trust
Each Other, in: Wired, 23.4.2014.
Taylor, Astra 2013: The People’s Platform: Taking Back Power and Culture
in the Digital Age, London.
Taylor, Colleen, Ha Anthony 2013: TaskRabbit Confirms Layoffs As It Rea-
ligns to Focus on Mobile And Enterprise, in: TechCrunch, 8.7.2013.
The Economist 2014: Should Digital Monopolies Be Broken Up?,
29.11.2014, http://www.economist.com/news/leaders/21635000-europe-
an-moves-against-google-are-about-protecting-companies-not-consu-
mers-should-digital (13.8.2016).
The Economist 2013: The Workforce in the Cloud, 1.6.2013, http://ww w.
economist.com/news/business/21578658-talent-exchanges-web-are-star-
ting-transform-world-work-workforce (13.8.2016).
The Economist 2015: There’s an App for That, 3.1.2015, http://www.eco-
nomist.com/news/briefing/21637355-freelance-workers-available-mo-
ments-notice-will-reshape-nature-companies-and (13.8.2016).
Thompson, Derek 2013: Airbnb CEO Brian Chesky on Building a Company
and Starting a „Sharing“ Revolution, in: The Atlantic, 13.8.2013.
Tortorici, Dayna 2013: More Smiles? More Money, in: n+117 (2013).
Tufekci, Zeynep; King, Brayden 2014: We Can’t Trust Uber, in: The New
York Times, 7.12.2014.
Weber, Lauren 2015: What if There Were a New Type of Worker? Depen-
dent Contractor, in: Wall Street Journal, 28.1.2015.
Weber, Lauren, Silverman, Rachel E. 2015: On-demand Workers: „We Are
Not Robots“, in: Wall Street Journal, 27.1.2015.
Westhale, July 2014: Gig Economy: How Technology Is Changing the
Workforce, in: Launchable, 16.4.2014.
Wong, Jamie 2012: The Rise of the Micro-Entrepreneurship Economy, in:
Co.Exist, 29.5.2012.
Wong, Julia Carrie 2016: Airbnb: How US Civil Rights Laws Allow Racial
Discrimination on the Site, in: The Guardian, 6.5.2016.
Zacharakis, Zacharias 2016: Foodora: Die pinkfarbene Verführung, in: Die
Zeit, 7.6.2016, http://www.zeit.de/wirtschaft/2016-06/foodora-liefer-
dienst-essen-geschaeftsmodell (13.8.2016).
O’Hear, Steve 2016: Rocket Internet’s Helpling Mops Up Another $45M
For Its On-demand Home Cleaning Service, in: TechCrunch 25.3.2015.
Papsdor f, Christian 2009: Wie Sur fen zu Arbeit wird: Crowdsourcing im
Web 2.0, Frankfurt/Main.
Pasquale, Frank; Vaidhyanathan, Siva 2015: Uber and the Lawlessness of
„Sharing Economy“ Corporates, in: The Guardian, 28.7.2015.
Popper, Ben 2015: Airbnb’s Worst Problems Are Confirmed by Its Own
Data, in: The Verge, 4.12.2015.
Ptak, Laurel 2013: Wages For Facebook, http://wagesforfacebook.com/
(31.5. 2016).
Rogers, Brishen 2015: The Social Costs of Uber, in: The University of Chi-
cago Law Review, https://lawreview.uchicago.edu/sites/lawreview.uchica-
go.edu/files/uploads/Dialogue/Rogers_Dialogue.pdf (13.8.2016).
Rohrbeck, Felix 2016: „Meine Wohnung wurde zerstört “, in: Die Zeit,
21.1.2016, http://www.zeit.de/2016/04/wimdu-berlin-apartments-vanda-
lismus (13.8. 2016).
Rutkin, Aviva 2014: Off the Clock , on the Record: Wearable Tech Lets Boss
Track Your Work, Rest and Play, in: New Scientist, 20.10.2014.
Salehi, Niloufar; Irani, Lilly C.; Bernstein, Michael S.; Alkhatib, Ali; Ogbe,
Eva; Click happier, Kristy Milland 2015: We Are Dynamo: Overcoming Stal-
ling and Friction in Collective Action for Crowd Workers, in: Association
for Computing Machinery, Seoul 2015.
Said, Carolyn 2014: Handy.com Housecleaners’ Lawsuit Could Rock
On-demand Companies, in: SFGate, 13.11.2014, http://ww w.sfgate.com/
business/article/Handy-com-housecleaners-lawsuit-could-rock-5891672.
php (13. 8 .2016).
Schechner, Sam 2015: Two Uber Executives Indicted in France, in: Wall
Street Journal, 30.6.2015.
Schmidt, Florian Alexander 2013: The Good, the Bad and the Ugly: Why
Crowdsourcing Needs Ethics, in: Third International Conference on Cloud
and Green Computing (CGC), pp. 531–35, 2013.
Schmidt, Florian Alexander 2015: The Design of Creative Crowdwork:
From Tools for Empowerment to Platform Capitalism, London.
Schneiderman, Eric T. 2014: Airbnb and the Cit y, Repor t, Büro des New
York State Attorney General, New York.
Scholz, Trebor 2016: Platform Cooperativism: Challenging the Corporate
Sharing Economy (dt. Wie wir uns die Sharing Economy zurückholen kön-
nen), Rosa Luxemburg Foundation, New York.
Scholz, Trebor 2015: Think Outside the Boss. Public Seminar, http://www.
publicseminar.org/2015/04/think-outside-the-boss/#.V6-W165rX-m
(13.8. 2016).
Scholz, Trebor 2014: Platform Cooperativism vs. the Sharing Economy,
Medium, https://medium.com/@trebors/platform-cooperati-
vism-vs-the-sharing-economy-2ea737f1b5ad#.moyp7dbht (13.8.2016).
Scholz, Trebor 2013: Digital Labor: The Internet as Playground and Factory,
Ne w Yor k .
Scott, Mark 2016: Uber and Its Executives Are Fined in France, in: New
York Times, 9.6.2016, http://www.nytimes.com/2016/06/10/technology/
uber-and-its-executives-fined-in-france.html (13.8.2016).
Seiner, Joseph A. 2017: Tailoring Class Actions to the On-demand Eco-
nomy, 77 Ohio State Law Journal, (i. E.).
Sharma, Dinesh C. 2014: Indiens IT-Industrie, 25.3. 2014, http://www.bpb.
de/internationales/asien/indien/189895/indiens-it-industrie (13.8.2016).
Shead, Sam 2015: London Startup Deliveroo Has Raised $100 Million for Its
Restaurant Delivery Service, in: Business Insider Deutschland, 24.11.2015,
http://www.businessinsider.de/london-startup-deliveroo-has-raised-100-
million-for-its-restaurant-delivery-service-2015-11 (13.8.2016).
Shet, Vinay 2014: Street View and reCAPTCHA Technology Just Got
Smarter, in: Google Online Security Blog, 16.4. 2014.
Singer, Natasha 2014: In the Sharing Economy, Workers Find Both Free-
dom and Uncertainty, in: The New York Times, 16.8.2014.
Smith, Lindsey J. 2016: Wall Street Loans Uber $1 Billion to Offer Subprime
Auto Leases, in: The Verge, 3.6.2016, http://ww w.theverge.com/2016/
6/3/11852940/uber-subprime-auto-loans- drivers-xchange (13.8.2016).
Smith, Rebecca, Leberstein, Sarah 2015: Rights on Demand: Ensuring
Workplace Standards and Worker Security in the On-demand Economy,
in: National Employment Law Projec t, September 2015.
Imprint:
© 2017
Friedrich-Ebert-Stiftung
Publisher: Division for Economic and Social Policy
Godesberger Allee 149 / D-53175 Bonn
Fax 0228 883 9205, www.fes.de/wiso
Orders/contact: wiso-news@fes.de
The views expressed in this publication are not necessarily
those of the Friedrich Ebert Stiftung The commercial exploitation
of the media published by the FES is allowed only with the
written permission of the FES.
ISBN: 978-3-95861-745-2
Cover photo: ©Photographee.eu – fotolia
Design: www.stetzer.net
Layout: www.zumweissenroessl.de
Printing: www.bub-bonn.de
www.fes-2017plus.de
