Uber sets the prices, assigns the work, and deactivates accounts. Calling that 'independent contracting' is a governance choice, not a neutral fact.
When an Uber driver gets deactivated, they get an email. The email typically cites a policy violation or a low acceptance rate or a complaint threshold that was crossed. The driver cannot call anyone to appeal it in real time. There is a form. They can submit it. The process takes time, and during that time, the account is off. The driver cannot work. The company that sent the email does not employ them.
This is the governance structure of the gig economy, and it is worth looking at directly rather than accepting the framing that comes from the platforms themselves.
Uber, DoorDash, TaskRabbit, Fiverr, and similar platforms coordinate labor at scale using algorithms. The workers on these platforms are classified as independent contractors in most jurisdictions where the platforms operate. That classification is consistently presented by the platforms as a description of the relationship, reflecting the flexibility and autonomy the workers supposedly have. My read is that it is better understood as a governance design choice, one that has significant consequences for how risk and reward are distributed.
An independent contractor, in the traditional legal sense, controls their own work. They set their prices, accept or reject work, and bear their own costs and risks in exchange for that autonomy. What the gig platforms offer is more complicated. The platform sets the prices. The platform determines who receives work through an algorithm that is not transparent to the worker. The platform sets the conditions under which accounts are deactivated. Workers can choose their hours, which is real flexibility. Everything else is set by the platform governance structure.
Researchers who have studied Uber's labor practices, most notably Rosenblat and Stark in work they published around 2016, documented how algorithmic management on gig platforms creates a relationship that looks a great deal like employment while maintaining the legal classification of independent contracting. I am hedging on the specifics of their findings here because I have not verified their full paper against a local source, but the broad picture they describe has been widely reported and confirmed in subsequent coverage: surge pricing is set by the algorithm, acceptance rate is monitored, and opaque deactivation policies create a power asymmetry between the platform and the worker that is hard to square with the independence the classification implies. The information asymmetry runs almost entirely in one direction.
I wrote about platform governance and multi-sided markets in an earlier post, where I focused on how platform owners design rules that serve their own interests first. The gig economy is a particularly clear case of that pattern. The governance rules are embedded in the algorithm, not in a contract that workers negotiated. The platform decides who gets work, at what price, under what conditions, and what constitutes a violation sufficient to end the relationship. Those are not neutral technical parameters. They are policy choices that the platform made and encoded in software.
California tried to reclassify gig workers as employees through AB5 in 2019. The law passed. Uber, Lyft, and DoorDash spent something in the range of 200 million dollars on a ballot initiative, Proposition 22, which exempted them from the law and was passed by California voters in 2020. That figure is widely reported in press coverage of the campaign, though I am noting it as public reporting rather than a verified precise count. Proposition 22 has been challenged in court since then, and the legal situation has continued to evolve. The point is not the specific dollar figure. The point is that classification is contested, and the platforms have the resources to contest it politically and legally in ways that individual drivers do not.
The IS angle on this is worth stating directly: algorithm design is policy design. When a platform engineer writes the code that determines how surge pricing works, they are making a policy decision about how much drivers earn in high-demand periods and who bears the volatility of that demand. When the code sets an acceptance rate threshold below which accounts are flagged, that is a policy decision about what constitutes acceptable worker behavior. The code does not neutrally execute a pre-existing policy. The code is the policy. The people who write and update it are setting the terms under which workers engage with the platform.
This has implications for how IS researchers think about system design. If we accept that algorithm design is policy design, then the questions we ask about gig economy platforms are not only technical questions about efficiency and matching. They are governance questions about who holds decision rights, who bears risk, and whether the people subject to the governance rules had any voice in designing them. The workers on these platforms are numerous, but they are distributed and hard to organize, and the terms under which they work are set algorithmically rather than negotiated.
There are genuinely difficult tradeoffs here. The platforms do provide real value: flexible income, accessible work, services that did not exist before. I am not arguing that gig work is uniformly harmful. I am arguing that the governance structure is not as neutral as the "independent contractor" label implies, and that the asymmetries in the system are design choices rather than natural features of the market. Someone decided that the platform would set prices. Someone decided that acceptance rates would be monitored. Someone decided that deactivation would happen via email with no immediate appeal. Those decisions reflect a particular distribution of power, and they are worth scrutinizing as the governance choices they are.