Employment law has traditionally organised itself around a single, apparently simple question: who is the employer? Identify that party and a long list of legal duties follows automatically -- minimum wage, safety standards, protection against arbitrary dismissal, contributions toward a pension. The question worked well enough for most of the twentieth century because the answer was rarely in doubt: a visible person or company hired workers, set their hours, supervised their work, and paid their wages, and any of these functions being present strongly implied the others were too. Platform-based gig work breaks this bundle apart in a way earlier disputes over employment status never quite managed, by distributing the functions of a traditional employer across several different actors, none of whom individually resembles one.

Consider a food-delivery rider working through a platform app. The platform does not set her hours; she logs in and out whenever she chooses, a freedom no traditional employer would grant. It does not require a uniform, does not summon her to a workplace, and does not forbid her from working for a rival platform the same afternoon -- every marker traditionally used to prove an employment relationship points toward independence. Yet the same app's routing algorithm assigns her deliveries with no room for negotiation, its rating system can end her ability to earn on the platform entirely if her average score drops below a threshold she cannot see calculated in real time, and its pricing model, which she cannot alter, determines what a delivery is worth before she ever decides whether to accept it. She is free in every formal sense a court once checked for, and constrained, in practice, almost as tightly as an employee ever was.

Courts asked to classify workers like her have mostly reached for tests built for an earlier era: the control test, which asks who directs how, when, and where work is done, and the economic-reality test, which asks how dependent a worker actually is on a single business for her livelihood. Both tests were designed around a world in which control was exercised by a visible human supervisor giving visible instructions, and both strain when the controlling party is an algorithm optimising a metric rather than a manager issuing a command. A platform can truthfully say no one instructed the rider to take any particular delivery, while also being the only entity capable of determining, through code no rider can inspect, which deliveries she will ever be offered and on what terms -- a form of control the old tests were never built to recognise, because it does not look like control at all until examined closely.

Some legal systems have responded by carving out a third category between employee and independent contractor -- a dependent contractor, a worker economically reliant on a single platform without being subject to the traditional markers of direct supervision, entitled to some but not all of the protections a full employee would receive. This compromise category is often criticised from both directions: platforms argue it imposes costs appropriate only to genuine employment, while worker advocates argue it grants only partial protection to a relationship that is, in every functional sense that matters, employment in substance if not in form. The criticism from both sides may itself be evidence that the category is doing something right, occupying a genuinely intermediate position rather than smuggling a fully resolved answer in under a neutral-sounding label.

What the gig economy exposes is that the old markers of employment -- fixed hours, a visible supervisor, a single workplace -- were never actually the substance of what employment law was protecting. They were proxies, reliable enough in an earlier economy that no one needed to distinguish the proxy from the thing itself: economic dependence on one party, and that party's practical power to control the terms on which a livelihood is earned. Strip away the proxies, as platform work does almost completely, and the underlying substance -- dependence, and control exercised through channels no court had previously needed to examine -- becomes visible on its own, forcing a body of law built on convenient shortcuts to identify, for the first time in a century, what it was actually trying to protect all along.