The rise of high-speed internet in the early 2000s transformed the global labor market for high-skilled labor. As email, instant messaging, and voice over internet calling (e.g., Skype) became accessible to more people, geographic distances posed fewer logistical restrictions for the job search and recruitment process, particularly for US-based organizations looking to hire workers located overseas. In the early 2000s, for example, a computer scientist in Singapore could more easily work remotely for US-based organizations because they could communicate on project requirements and necessary changes that arose during a project. Such communication, enabled by emerging technologies, is essential for complex, knowledge-intensive work.

In retrospect, it is not surprising that the early 2000s saw a spike in organizations using these technological advancements to outsource knowledge work to countries with lower labor costs. Researchers note that beginning in the 1980s, shareholders increasingly viewed employees as expensive costs rather than valuable resources who helped organizations grow. Organizations that minimized labor costs were handsomely rewarded by the stock market—and continue to be. What better way to minimize these costs than to outsource complex, unpredictable, knowledge-intensive projects to qualified workers in countries where wages were significantly lower? Organizations frequently hired these workers as “independent contractors,” an employment classification that meant they did not have to provide healthcare, retirement, or other similar benefits afforded full-time employees. Organizations could also let go of independent contractors without incurring major costs, such as providing severance pay. The relative ease and low liability with which organizations could hire people as independent contractors reflects the weak legal and institutional regulation surrounding independent contractors; compared to full-time employees, independent contractors have attracted comparatively little attention from lawmakers in the United States and as a result have limited legal protections.

Jerry Davis, an organizational sociologist who studies how corporations have changed in the past few decades, refers to the outsourcing of operations to lower-wage workers as the “Nikefication” of organizations, referencing Nike’s infamous outsourcing of its shoe production to lower-wage overseas workers. Shareholders celebrated Nike because this outsourcing lowered its labor costs. Notably, “Nikefication” largely evaded existing legal oversight and costs. As some evidence that the technologies in the early 2000s contributed to the Nikefication of knowledge work, the Bureau of Labor Statistics reported that the “computer and electronic products” industry shed 750,000 net jobs in the United States between 2000 and 2011.

Although the spread of new and improved technology decreases the costs of outsourcing work globally, handing over knowledge-intensive work to a third party is never straightforward and rarely predictable. Moreover, knowledge-intensive work often involves intellectual property rights. Organizations go to great lengths to ensure workers do not share their intellectual property, including requiring workers to sign non-disclosure agreements and restricting workers’ ability to join competitors. These measures are not easily enforced across geographic borders that have different legal standards and enforcement mechanisms. It is particularly difficult for organizations to ensure that outsourced workers are not sharing their intellectual property with local competitors.

Outsourced workers, in turn, have to be careful when determining whether they should work with organizations across geographic boundaries. Organizations could fail to pay promised wages or hire workers to complete unethical work, leaving workers with little recourse, especially if they were hired as independent contractors. If US-based full-time employees have difficulty navigating a legal system that can seem nothing short of an intricate labyrinth, outsourced workers in foreign countries stand no chance of holding organizations accountable if they shirk their commitments towards workers. In short, outsourcing knowledge-intensive work faces a significant trust gap between both workers and organizations.

An often-overlooked component enabling the rise of online platforms, including TalentFinder is their terms of service (ToS). Talent-Finder and other online platforms use elaborate ToS or user agreements to establish a relationship with workers and clients. Such agreements fundamentally transform the relationship that platform organizations have with workers. In this setting, before either actor could register for the platform, both had to agree to TalentFinder’s ToS. In this agreement, TalentFinder first stated its purpose for establishing the platform: “The Site is a marketplace where Clients and Freelancers can identify each other and advertise, buy, and sell Freelancer Services online. Subject to the Terms of Service, TalentFinder provides the Site Services to Users” (emphasis added).

Using “marketplace” to describe its purpose evokes a certain type of imagery: a marketplace suggests a public area in which people buy and sell goods but which is organized and operated by a third party. This third party sets up and maintains the trading area, designates where sellers can locate their goods, sets the trading hours, advertises the marketplace, and takes fees for its services. The marketplace is not responsible for who meets with each other, what goods are sold, how people interact with each other, or other similar types of interactions. Like a third party in a traditional marketplace, TalentFinder’s shift from a traditional staffing firm to an online marketplace was facilitated by presenting itself as having a similar marketplace relationship with its users. TalentFinder’s ToS went on to state,

TalentFinder merely makes the Site and Site Services available to enable Freelancers and Clients to find and transact directly with each other. TalentFinder does not introduce Freelancers to Clients, find Projects for Freelancers, or find Freelancers for Clients. . . . Users are responsible for evaluating and determining the suitability of any Project, Client or Freelancer on their own. If Users decide to enter into a Service Contract, the Service Contract is directly between the Users and TalentFinder is not a party to that Service Contract. (emphasis added)

TalentFinder’s ToS carefully established that it was “merely” a facilitator for workers and clients to find each other. Its reference to people using the platform to find work as “freelancers” rather than workers emphasized their position that these people were freely choosing to use the platform on their own volition. TalentFinder did not specify who was notifying workers and clients of the “services” they found “through the Site and Site Services.” As I demonstrate in my book, TalentFinder used its algorithms to control who was visible and how visible workers and clients were to each other. Instead of acknowledging its role in this process, however, TalentFinder’s ToS shifted the responsibility of “evaluating and determining” whether users should work with one another to clients and workers themselves.

TalentFinder’s ToS went on to specify workers’ and clients’ relationship with each other, should they choose to work together:

You acknowledge, agree, and understand that TalentFinder is not a party to any Service Contract, that the formation of a Service Contract between Users will not, under any circumstance, create an employment or other service relationship between TalentFinder and any Freelancer or a partnership or joint venture between TalentFinder and any User.

Like many traditional marketplaces, TalentFinder specified that they had no part or liability when workers and clients found each other on the platform and decided to work with one another. Each actor was free to enter into any additional agreements so long as they did not change TalentFinder’s role and relationship with clients and workers. In specifying workers’ relationship with clients, its ToS further established that TalentFinder would have no explicit role in determining which projects workers accepted, when people worked, or how much they charged clients for their services:

You also acknowledge, agree, and understand that Freelancers are solely responsible for determining, and have the sole right to determine, which Projects to accept; the time, place, manner, and means of providing any Freelancer Services; the type of services they provide; and the price they charge for their services or how that pricing is determined or set.

Before workers were even approved to work on TalentFinder, workers permitted TalentFinder to share any initial information TalentFinder collected with third party “affiliates”:

To the extent permitted by applicable law, you also grant to TalentFinder and our successors and Affiliates a royalty-free, sub-licensable, transferable, perpetual, irrevocable, non-exclusive, worldwide license to use, reproduce, modify, publish, list information regarding, edit, translate, distribute, publicly perform, publicly display, and make derivative works of all such User Content and your name, voice, and/or likeness as contained in your User Content . . . for use in connection with the Site and TalentFinder’s, our successors’ and Affiliates’ businesses

As a result, by registering to use TalentFinder, users consented to allow TalentFinder and all its affiliates to market, profit from, and essentially do whatever they would like with the data TalentFinder collected from workers. Workers had no control over or insight into which data were being shared. In registering for the platform, workers shared information about their location, skills, previous experience, email address, phone number, and billing information. TalentFinder also had detailed information about workers’ platform usage and success. All this information was subject to be shared per this provision. Further, once such data were shared with these other parties, workers had no control over how these other organizations used these data. In essence, workers were forming an amorphous, multiplex relationship with an unknown number of organizations.

Emerging studies highlight how organizations can exploit such amorphous relationships at workers’ expense. These additional organizations can sell, trade, and profit from the data they gather. TalentFinder absolved itself and its affiliates from any potential harm that may come with the use of the data:

ADDITIONALLY, IN NO EVENT WILL TALENTFINDER, OUR AFFILIATES, OUR LICENSORS, OR OUR THIRD-PARTY SERVICE PROVIDERS BE LIABLE FOR ANY SPECIAL, CONSEQUENTIAL, INCIDENTAL, PUNITIVE, EXEMPLARY, OR INDIRECT COSTS . . .

Thus, almost any entity connected with TalentFinder could harvest an individual’s data in any way without the individual’s consent. The individual also gave up their right to hold any of these entities liable. As such, conceptualizing workers’ relationships in online labor markets as triadic (i.e., worker, TalentFinder, and client) is inadequate. If TalentFinder did not share any of a worker’s information with third parties, then the relationship could be characterized as triadic. However, TalentFinder did share workers’ data with other parties and affiliates, making workers connected (unwittingly or not) to far more than just TalentFinder and the clients they worked for on TalentFinder. It is thus more apt to characterize the employment relationship as amorphous, multiplex, and ill-defined. Although this characterization is analytically elusive, there is no way of knowing how many entities workers were actually entering into an agreement with.

Thus, before clients and workers even had an account, they “agreed” to TalentFinder’s role as a marketplace in which they had no formal relationship with TalentFinder. This arrangement limited TalentFinder’s formal liability to workers and clients using its platform but also provided them with control over the platform, which was essential for its ability to transition from a staffing firm to a global marketplace for labor.

Excerpted and adapted with permission from Inside the Invisible Cage: How Algorithms Control Workers by Hatim Rahman, published by University of California Press. © 2024 by Hatim Rahman.

Hatim A. Rahman is an award-winning assistant professor at Northwestern University’s Kellogg School of Management and the author of Inside the Invisible Cage: How Algorithms Control Workers.