Brightly clad delivery drivers, darting in and out of restaurants and traversing the city on bicycles, mopeds, or in cars, have become a persistent sight across urban areas. Their routines persist even when heavy rain or oppressive heat empties the sidewalks. These delivery drivers are the most visible representatives of a growing segment of the labor market: gig workers, who have no stable employment contract but get paid for each task they complete.

The recent rise of gig work has been driven by technological advances, like smartphones and the Internet, which allow for just-in-time coordination. Yet the work arrangement itself is not a novel phenomenon. In early industrial England, handloom workers were typically paid per piece, dockworkers in the nineteenth and early-twentieth centuries were hired on a daily basis, and day laborers have been common in agriculture since ancient times. In fact, from a historical perspective, the formal employment exemplified by unionized work in the post-war era is the aberration. Yet such formal employment has become the norm throughout wealthy countries. Our social safety nets have been built around this form of work, and it has long served as the focal point of philosophical debates about workplace justice.

Gig work introduces a distinct concern: uncertainty about income and free time. In a traditional job, employers absorb much of the risk of fluctuating demand. Employees work and receive a salary according to a set schedule, regardless of the number of customers. This creates predictability: You know when you will work and how much you will earn. Gig work inverts this arrangement. Because workers are paid only when they complete a task, they bear the risk of fluctuating demand: If there are no customers, then there are no gigs and no pay.

In work with Jakob Moggia, we argue that this risk shift is a defining feature of the gig economy: Businesses shift the risk of fluctuating demand onto workers. We distinguish five mechanisms employers draw on: short-term contracts, flexible work hours, flexible remuneration, short-term scheduling, and a reduction in insurance coverage. In the limit, these mechanisms allow a firm to purchase exactly the amount of labor it needs at precisely the moment it needs it. The platform-based work of delivery drivers is one example of how this risk shift is instituted. Zero-hour contracts, whereby employers do not guarantee employees any work in advance, are another.

This risk shift is normatively significant because the resulting instability of gig work can erode what we might call planning agency—the capacity to form long-term goals and organize one’s life in pursuit of them. Taking out a mortgage requires knowing that you will have an income next year; enrolling in an evening course requires knowing you will have the money to pay for it and the time to attend it; and even coaching your child’s sports team at weekends becomes difficult if your schedule fluctuates constantly. The problem with gig work is thus not simply one of precarious pay. Instead, the key concern we identify is that the uncertainty inherent in gig work can inhibit our ability to shape our life according to a coherent, long-term plan—an ability that, as philosophers from John Rawls to Alasdair MacIntyre have argued, is central to human flourishing.

Importantly, however, while gig work can undermine our agency in this way, it need not have this effect. This is because, in principle, the flexibility cuts both ways: Just as employers are at liberty to purchase labor on demand, workers are at liberty to decide whether to accept an offer—neither side has to make any prior commitments. So, while the risk shift occurs regardless, it is in principle possible that gig workers are adequately compensated for taking on the additional risk—e.g., through greater flexibility or higher hourly earnings.

More recently, focusing on platform-based work, I have argued that a person’s outside options are critical to evaluating whether some gig work arrangement is objectionable. Contrast two cases: a student who occasionally drives for a ridesharing app to save towards an expensive holiday and a single parent who drives for a ridesharing app to cover rent, food, and basic healthcare for their family. For the student, the arrangement might feel liberating: Without incurring obligations, they can drive when they feel like it and target peak-demand periods with high pay. For the single parent, despite enjoying the same terms, the arrangement might feel highly restrictive: Being dependent on the income, they might feel compelled to accept rides whenever any are offered and even when pay is low.

Empirical research by the sociologist Juliet Schor and her colleagues supports this diagnosis. Studying workers across different types of platforms, Schor et al. find that worker satisfaction varies greatly with how dependent workers are on the income. Those who drive or perform tasks to supplement an already sufficient income tend to value the opportunity: They can be selective. Those who rely on these platforms to make ends meet report exhaustion, powerlessness, and a pervasive sense of vulnerability.

So, the concern that the instability and unpredictability of gig work will diminish our planning agency becomes especially relevant when workers rely on this work to satisfy basic needs (of themselves or of their dependents).

This indicates that it would be too quick to simply blame employers who rely on gig workers. True, a number of reasonable concerns have been put forward about the behavior of these firms: that they utilize epistemic asymmetries to engage in a form of algorithmic domination, that they exploit workers by simultaneously denying them the benefits of employment and the benefits of self-employment, and that they deny workers the opportunity to form valuable collegial relationships. But two caveats are important here. First, some forms of gig work are organized in a decentralized manner. Second, if people choose this work, they will typically consider it their best option.

While the focus of the public debate is typically on labor platforms like Uber and DoorDash, which sell services directly to consumers, other platforms operate more like a digital marketplace: Platforms like TaskRabbit connect independent parties who enter a contract with each other. I have proposed calling the former platform companies and the latter broker platforms. Broker platforms exercise less control: Workers set their own prices, choose their clients, and decide how to perform their work. Yet, the concern remains: Dependency on the income can still render workers subject to an insecurity that undermines their planning agency.

This point connects to the second caveat. When people choose gig work, they typically do so because they consider it their best option. Now, it is of course possible that they are misled or manipulated into thinking so. But often, it plausibly is the best work available to them. In some contexts, a case can be made that gig work crowds out better job opportunities: Consider the effect of the ride-hailing industry on taxi fleets. And there is a concern that gig work imposes social costs, e.g., by freeriding on social security measures. But the basic point remains: While many employers (corporate and individual) who rely on gig work might reasonably be said to exploit the vulnerability of gig workers for their own benefit, they cannot, at least not exclusively, be blamed for creating conditions that lead individuals to consider gig work to be their best option to begin with.

This perspective on gig work indicates two possible policy responses. First, one could seek to directly improve the conditions of gig work. This could be done through regulation: Laws might require transparency in pricing algorithms, cap commission fees, or mandate that platform companies treat certain workers as employees. Or it could be done by reimagining the work’s setup: Some have attempted to turn labor platforms into worker-run co-operatives. Both approaches hold promise, but each faces practical limitations: Regulation is often a contested process, while cooperatives struggle to compete against venture-capital-backed incumbents.

Alternatively, one could seek to improve the outside options of workers. If gig workers are vulnerable to exploitation because they so urgently depend on the income, then better job-training programs, more robust social insurance programs, or an unconditional basic income would all help. The risk shift inherent in gig work becomes less concerning as workers become less dependent on the income: They suffer less uncertainty as a result and will only accept the increased risk for adequate compensation. Improved outside options would likely come at a price not just to employers but also to consumers: The costs for services like meal deliveries might rise, especially in pouring rain or blistering heat. But this, arguably, just indicates what a fairer price for these services would be.

Many thanks to Martina Valković for the invitation to contribute to the blog and for excellent editorial comments on an initial draft.