2.1 Segments of the Crowd Industry

The term crowdsourcing was coined by Howe (2006) as a portmanteau of the words “crowd” and “outsourcing” (Leimeister & Zogaj, 2013). Howe defines crowdsourcing as an outsourcing of activities traditionally performed by dedicated actors—usually internal staff—to an unspecified and preferably large group of people via an open call (Howe, 2010). The World Bank coined the term “online outsourcing” for the same phenomenon and, like Howe, defined it as a contractual relationship between foreign workers and clients for the provision of services or the execution of tasks via internet-based marketplaces or platforms (World Bank, 2015). The “crowd” to which the activity is outsourced does not necessarily have to be outside the company. Internal crowdsourcing, in which the workforce acts as a crowd, is also widespread, such as at IBM and Daimler (Öhrler & Spies, 2015; Schäfer, 2015). However, broader changes in industrial relations have resulted in external crowdsourcing. Cappa et al. (2019) show, for example, that the announcement of a crowdsourcing campaign positively affects the expectations of a firm’s future profits as measured by its stock market performance.

With external crowdsourcing, a company (the crowdsourcer) posts tasks or task packages on an internet platform and calls on the crowdworkers to complete the tasks. The requirements are described so specifically that they can be completed by any properly trained internet user without further consultation with the client. The activities outsourced in this way are diverse. The World Bank distinguishes between microwork and online freelancing in crowdsourcing (World Bank, 2015). Microwork breaks projects down into microtasks that can be completed in seconds or minutes. On the part of the crowdworkers, only basic mathematical and/or reading skills are required to complete these tasks. For example, tasks are about labeling images, editing text, or categorizing data and products. As Fig. 2.1 shows, the work process for microtasks is highly standardized and meant to minimize direct communication between employees and customers.

Fig. 2.1
A block diagram of the work process for microtasks comprises search, accept task, labor, and payments. Two divergences of rejection and acceptance of multiple tasks at once, are provided.

Work process for microtasks. Dotted lines show possible divergences from the standard work process and might be platform-specific (Source: Hornuf & Vrankar, 2022)

Online freelancing is the outsourcing of professional services to crowdworkers, who usually must have relevant technical or professional qualifications. The tasks are often larger projects that are completed over a longer period of time—several days, weeks or even months. Examples of these tasks are the creation of designs or program codes, or the solving of mathematical or scientific problems (Däubler & Klebe, 2015; Leimeister & Zogaj, 2013; Risak, 2015). Figure 2.2 shows the work process for online freelancing, which is slightly less standardized and encourages communication between crowdworkers and crowdsourcers.

For many crowdworkers, microtasks could represent a gap-filling activity that is carried out between other activities and which pays a relatively low wage (Teevan, 2016; Newlands & Lutz, 2021). Some scholars have therefore criticized the working conditions in crowdworking and have described them as precarious (Kittur et al., 2013; Schriner & Oerther, 2014; Hara et al., 2018; Whiting et al., 2019). In investigating 105 mean hourly wages in crowdwork that were reported in 22 different studies, Hornuf and Vrankar (2022) have found evidence that working on microtasks results in wages ranging from $3.78 to $5.55 per hour on average. Online freelancers earn $4.87 to $20.88 per hour on average, which is up to three times more than microtask workers. However, when factoring in unpaid work, such as searching for tasks and communicating with requesters, wages of online freelancers tend to be much more similar to the wages of microwork.

Another distinction scholars often make is whether platforms restrict activities to a specific region or whether they can in principle be carried out from anywhere in the world (Rani et al., 2021). This definition relates somewhat to the distinction between microwork and online freelancing. While microwork can often be performed from anywhere in the world, online freelancers are often locked into a specific region, for example when it comes to delivering food or groceries. Activities that are limited to specific regions are also often referred to as “gig work” (Heiling & Kuba, 2017).

Fig. 2.2
A block diagram of the work process for microtasks comprises search, offer, bargain, acceptance, labor, and payments. 4 divergences, 3 for rejection and 1 for acceptance of multiple tasks at once, are provided.

Work process in online freelancing. Dotted lines show possible divergences from the standard work process and might be platform-specific (Source: Hornuf & Vrankar, 2022)

Crowdsourcers typically use an internet platform to advertise their tasks. The platform is sometimes operated by the crowdsourcers themselves, such as at IBM (Klebe & Neugebauer, 2014), but most commercial providers such as Amazon Mechanical Turk, Clickworker, Freelancer, Innosabi, TopC or Twago operate platforms that can be used by anyone. The business of these platforms has grown rapidly in recent years. According to information in the literature, Freelancer alone has over 14.5 million users working on 7.2 million projects (Däubler & Klebe, 2015). The German platform Twago lists more than 225,000 registered experts with more than 80,000 advertised projects with an order volume of over 400 million EUR on its website (Twago, 2023). Well known clients who advertise tasks via the platform include AOL, BMW, Deutsche Telekom, Facebook, Google, Honda, Intel, Manhattan Cosmetics, Microsoft, NSA, Panasonic, Postbank and Walt Disney (Klebe & Neugebauer, 2014; Däubler & Klebe, 2015; van Delden, 2014). For 2013, the World Bank estimated that the crowdsourcing industry had global sales of 2 billion U.S. dollars, and considered a global sales volume of 15 to 25 billion U.S. dollars to be conceivable in 2020 (World Bank, 2015). It is difficult to say whether these estimates actually bore out, as there are hardly any current and reliable figures for this market. However, if one looks at the development of individual market players, these aggregate figures do not seem exaggerated.

Another taxonomy of platforms was introduced by Boudreau and Lakhani (2013) and relates to the specific work processes to complete the tasks and their remuneration. Their first category are crowd labor markets and often relate to external crowdsourcing. The activities are most similar to what the World Bank defines as microwork. Well known examples of such platforms are oDesk, Clickworker and Amazon Mechanical Turk. This particular category of crowdwork is considered by some scholars to be an extreme form of Taylorism (Kittur et al., 2013; Aloisi, 2015), defined as dividing a large, intellectually demanding task into many small tasks, all of which can be completed with minimal mental effort.

The second category Boudreau and Lakhani (2013) define are crowd contests, which are competitions in which participants submit their work. Again, the interaction with crowdsourcers is kept to a minimum. Typical tasks include designing a logo or a web page or solving a company’s problem. Well known examples of platforms include 99designs, DesignCrowd, GoPillar, Hatchwise, HYVE, and Topcoder. The remuneration of crowdworkers depends on the crowdsourcer, which ranks the work that has been submitted. In some cases, a worker may not receive payment, despite completing the work.

Crowd complementors offer products, software, or services within an ecosystem built and maintained by a company and thus generate value for the company, as well as for users in that specific ecosystem. They often receive a fixed fee for their contribution that has been ex ante defined by the platform and is charged by the company providing the ecosystem or platform. Typical tasks in this third category of crowdsourcing include developing an app, recording a video, or uploading a song or photo. Well known examples of platforms are the Google Play Store, iTunes, Soundcloud, and YouTube.

The fourth and final category of crowdsourcing is collaborative community platforms, which often involve innovation contests among regular employees of a company or users of a product who receive no additional compensation for their activities on the platform. Collaborative communities are typically dedicated to a greater purpose. Activities are often unpaid and performed as a hobby, which instead of money pays off in terms of experience or recognition in the respective community. Typical tasks include developing open-source software, translating, and helping other users on the same platform. Well known examples of platforms are Apache, Translate, and Wikipedia.

2.2 The Chinese, German, and U.S. Markets

In China, crowdsourcing is commonly known as the Witkey model. The term Witkey is short for “key of wisdom” and was coined in 2005 by Liu Feng, a researcher at the Chinese Academy of Social Sciences (CASS) (Liu, 2008). In the same year, Liu created the Witkey internet platform, which aims to leverage CASS’s expert resources and scientific achievements to address the technological problems faced by companies (Liu, 2008). In building the platform’s website, Liu established an area on the internet where problems can be solved through online platforms and the solvers are paid similarly to freelancers. Liu defined the Witkey model as a “new Internet model in which human knowledge, wisdom, experience and skills are converted into real income through the Internet” (Liu, 2008).

Crowdworkers are also often referred to as witkey in China. They are “people who convert their wisdom, knowledge, ability and experience into real income by solving problems in the field of science, technology, work, life and learning on the Internet, and thus make their knowledge, wisdom, experience and skills economically valuable” (Liu, 2008). The term “crowdsourcing” as it is used in the Chinese context not only refers to the Witkey model, but also to the tasks on the platforms that can be completed offline. A typical example of offline crowdsourcing is food delivery platforms such as Meituan. In principle any Chinese citizen can register on the platform and become a delivery driver. Because the crowdsourcing platforms defined in this book exclude those that only deal with offline tasks, only the Witkey platforms in China are considered in the legal and empirical analysis, and hence fall under the definition of crowdsourcing used in the present book.

In October 2019, we found 145 Chinese crowdsourcing platforms that fall under this definition. Most of these platforms remain active, while around 28% of the initially active services had ceased to be available by May 2022. Originally, Chinese crowdsourcing platforms were mainly crowd contest platforms (Huang & Wang, 2015). Today, all four types of crowdsourcing platforms that have been classified by Boudreau and Lakhani (2013) exist. Even before the terms crowdsourcing or witkey arose, crowdsourcing platforms existed in China. For example, K68 was founded in 2003 as allegedly the first Chinese crowdsourcing platform (CCTV, 2006). Since the establishment of the K68 platform, diverse tasks and projects that can be done through the internet have been published there.Footnote 1 In 2022, the platform is still active with around 2.8 million users, and around 28,000 companies have used it so far as a crowdsourcing platform to find crowdworkers to carry out relevant tasks.Footnote 2 Many different types of tasks are currently published on the K68 platform, such as graphic design, architectural or decorative design, translation, new product testing, and naming (for example, a company, a product or a baby). The platform falls under both categories: crowd labor market and crowd contest platform.

Since 2005, Chinese crowdsourcing platforms have experienced rapid development. According to the 2010 China Witkey Industrial White Paper (iResearch, 2010), the total registered users of Chinese crowdsourcing platforms already exceeded 20 million in 2010, when the cumulative transaction volume exceeded 300 million CNY, or 39 million EUR.Footnote 3 Among crowdsourcing platforms, the Zhubajie platform, which was founded in 2005, ranks first in terms of the number of registered users and cumulative transaction volume. In 2013, the number of registered users of this crowdsourcing platform reached over ten million.Footnote 4 It was reported that the platform’s annual transaction volume reached 7.5 billion CNY in 2015, amounting to over 80% market share (PEdaily.cn, 2016).Footnote 5 There are now around 28 million registered users of the Zhubajie crowdsourcing platform.Footnote 6

According to the 2010 China Witkey Industrial White Paper (iResearch, 2010), a survey conducted in 2010 among 355 crowdworkers shows that 17% of the respondents did not earn anything on the crowdsourcing platforms; 31.5% earned less than 100 CNY a month on the platformsFootnote 7; more than 50% earned more than 100 CNY a month; and only 3% earned more than 2000 CNY.Footnote 8 By 2017 there were around 30 million Chinese crowdworkers serving approximately 190,000 companies and individuals worldwide, generating a total business turnover of 5 billion CNY, or approximately 700 million USD (Huo et al., 2017). While more recent data is difficult to obtain, given the rapid development of Chinese crowdsourcing platforms and the strong growth of registered users over the past decade, we can expect that the income of crowdworkers in China’s crowdsourcing market has increased significantly.

For Germany, Serfling (2018) defines crowdworkers as natural persons who earn at least part of their income by performing paid work via internet platforms or smartphone apps, carried out online or offline. In his definition, he explicitly excludes work that takes place within a company, that is, work that could be described as internal crowdsourcing. In terms of employment status, crowdworkers can be self-employed, full-time or part-time employees for another company, but also non-employed people such as students or pensioners. In Germany, too, a large number of activities can be subsumed under the term crowdwork. These include microtasks with short processing times, which are carried out via such platforms as Clickworker, Streetspotr and Testbirds. But there is also crowdsourcing in the form of competitions for design jobs, which can, for example, be carried out on the jovoto platform, as well as innovative and complex problem-solving tasks aimed at highly qualified professionals, such as on the twago platform (Nierling et al., 2020).

The most ambitious project providing comparative data for crowdworking activities in the European Union is the COLLEEM survey, an international research project of the European Commission’s Joint Research Council. The first pilot wave of the survey was completed in 2017 and gathered a total of 32,389 responses from 14 member states. The survey found that 10.4% of the adult population in Germany have been involved in crowdworking activities (Pesole et al., 2018). In 2018, the German Federal Ministry of Labor commissioned another study, which was conducted by Serfling (2018) and estimates that 4.8% of the German electorate engages in crowdworking activities. A year later, Serfling (2019) estimated that 4.0% of the German population are active and 2.3% past crowdworkers. In other publications, estimates of the extent of crowdwork range from 0.27% of German-speaking adults (Bonin & Rinne, 2017) to 12% of the total German adult population (Huws et al., 2016). Pongratz and Bormann (2017) estimated a projected number of up to 300,000 active German crowdworkers in 2017 (see Fig. 2.3 for an overview).

More recently, Mrass et al. (2020) took a different approach and surveyed 21 crowdworking platforms with headquarters or at least a physical presence in Germany. They found that the average number of registered crowdworkers on these platforms is 93,909 people. The crowdworking platforms themselves estimated that in 2017 there were 1,162,059 crowdworkers in Germany, which was probably an overestimate given that inactive users were most likely also taken into account by the responding platforms. Mrass et al. (2020) also provide information about the size of the platforms. The average crowdworking platforms employed 23 people. Sales increased sharply from 2015 to 2016, with an average growth of 90%. The surveyed crowdworking platforms themselves estimated the total revenues of the German crowdworking platforms in 2016 at 45 million EUR. The total revenue generated by crowdworking platforms with a physical location in Germany amounted to 203 million EUR.

Fig. 2.3
A table presents the percentage of crowd workers, the definition of the population, survey dates, and data collection method, sampling method, and sample size for 5 different studies.

Estimations of the German crowdworking market. Source: Nierling et al. (2020) and own additions

Empirical evidence in Germany shows that crowdworkers are unlikely to be part of the labor force and that Germans often pursue crowdsourcing as a secondary activity (Vandaele, 2018). However, crowdsourcing is increasing in the traditional working population (Nierling et al., 2020). According to Krzywdzinski and Gerber (2020), two-thirds of crowdworkers worked a maximum of 10 hours per week on a platform, suggesting that crowdwork was, if at all, a secondary activity for them. According to the The Online Labour Index, an endeavor carried out by the Oxford Internet Institute (Stephany et al., 2021), which provides global statistics on the gig economy, the majority (37.1%) of global online freelance labor demand was located in the U.S., but only 5.7% of the global online freelance labor supply. In Germany, global online freelance labor demand was 2.3% of demand and 0.6% of supply (Online Labour Index, 2020). In the United States, 35% of those surveyed worked more than 20 hours per week on the platforms, which means that platform work was more likely the main source of income for crowdworkers. Around 40% of respondents in the U.S. performed platform work as a part-time job of 10 hours or less per week. According to Krzywdzinski and Gerber (2020), only 16% of crowdworkers in Germany reported that their platform income accounted for more than 50% of their total income. In the United States, the figure was just under 33%. This difference between Germany and the United States applied to crowdworkers on both microtask and online freelancing platforms.

Based on a sample of 1131 crowdworkers on 60 German and U.S. platforms, Krzywdzinski and Gerber (2020) find that slightly more crowdworkers in Germany are male and that they have a higher level of education than the general population. On microtask platforms, most crowdworkers are full-time or self-employed if they have a college degree. If crowdworkers only have a high school diploma, they are most often employed full-time or currently completing their university education. On platforms that require more creativity and professional knowledge, the majority of crowdworkers are self-employed, whether they have a high school diploma or college degree. Activities requiring professional knowledge are carried out significantly less often by students. Serfling (2018) reports that nearly two-thirds of crowdworkers are paid, and about 14% receive some form of voucher as remuneration. For a stratified sample of Germans over the age of 18, the average crowdworker earns 808 EUR per week, with a strong variation between microtasks and more complex tasks. While 40% of crowdworkers earn more than 1000 EUR per week, a quarter earn less than 25 EUR per week and around a third less than 100 EUR per week. An earlier study by Leimeister et al. (2016) finds that micro workers earn on average 144 EUR per month; on design platforms it is 660 EUR.

According to Belletti et al. (2021), around 8% of German companies in the information economy already use crowdwork; 6% in manufacturing. Around half of the companies see future access to specialized skills as a possible goal of using crowdwork. The usage rate of crowdsourcing has almost doubled since 2014 and, according to the companies, will continue to grow. Part of the increasing use of crowdsourcing is due to COVID-19-related adjustments in work organization and an increased need for specialists in IT and related areas (Erdsiek, 2021).

Hoang et al. (2020) examine a sample of 4579 U.S. adults who provided comprehensive information about their online earning activities. Overall, 24% of the respondents participated in some form of platform work. While most of them (20%) engaged in online sales activities, overall 9% participated in some form of online labor platform, and some participated in both. Online labor platforms include activities such as “‘rideshare driving’, ‘delivery’, ‘online tasks’ (e.g., coding, data entry, and taking surveys), ‘house/laundry cleaning’, or ‘other platform work’ (e.g., babysitting, mystery shopping, and legal services)” (Hoang et al., 2020, p. 687). The authors also find that slightly more women participated in platform work and that 41% of respondents who participated in platform work were aged between 18 and 29. The mean annual income (29,400 USD) was significantly lower for those who worked on some form of online labor platform than for those who did not participate in any platform work (34,700 USD) and those who participated in online selling (40,400 USD). They also find that those living in the South of the United States are more than twice as likely to participate in some form of online labor platforms as those in the Northeast. Interestingly, citizenship status was significant, with non-U.S. citizens being less than half as likely as U.S. citizens to participate in online selling activities. Examining time spent on the platform, Cantarella and Strozzi (2021) find that when only paid activities are considered, crowdworkers in the U.S. work 10 hours less than traditional workers per week. When considering the time spent on unpaid activities, it is only 3 hours less. However, most crowdworkers would actually like to work more in crowdwork or other forms of employment.

2.3 General Market Trends

The basic business model of crowdsourcing has hardly changed over the last decade. Howe’s (2010) definition of the phenomenon still applies. However, in some sub-segments the speed with which the activities are to be carried out has increased. A general trend, which is particularly pronounced in China, is the move away from stationary computers and towards the use of mobile devices (Kemp, 2022). This development makes the work even more flexible and the possible execution of tasks faster than ever before. A striking example is the German company Gorillas, which was founded in 2020 and promised to deliver groceries and other supermarket goods at the same prices as in the supermarket. The goods were ordered via an app and delivered by around 14,000 bicycle couriers (Eckardt, 2022). There was a delivery price of 1.80 euros, but no minimum order value, and delivery to the customer took place within 10 minutes after receipt of the order. Although in December 2022 the Turkish competitor Getir bought Gorillas for 1.2 billion EUR, it remains to be seen whether such business models will continue to proliferate.

Just like with grocery delivery, the performance of crowdworkers will have to increase even with microtasks—which is the main focus of this book—in order to keep up with, for example, AI-based smart assistants like ChatGPT. Recent research has shown that chatbots could quickly catch up with even knowledge-intensive activities that would otherwise be reserved for academics, such as translation work, data analysis, research synthesis, statistical modelling, or defining specific phenomena like crowdfunding (Wenzlaff & Späth, 2022). While these systems are not yet perfect, the speed at which they learn is impressive, and the question therefore arises as to whether microtasks will not be completely replaced by chatbots and other intelligent AI-based systems, or whether crowdworkers will only serve to train these systems (Kittur et al., 2013). This development is contradicted by the fact that, due to international refugee movements and armed conflicts, there is an increasing demand for simple online activities that can be carried out from anywhere in the world without extensive training. An example of this is the increased demand for these tasks from Ukrainian women due to the armed conflict with Russia, which has severely impacted the traditional labor market in Ukraine (Bondarenko, 2022a; Bondarenko, 2022b).