Uber, for example, announced that it has over a thousand experiments running on its platform at any given time. LinkedIn recently subjected 20 million users to thousands of experiments (unbeknownst to them) in a five year period.
It’s not just app clients that are affected. They are also the people who rely on these apps for gigs. Some research considerations that nearly a fifth of people in the United States used a digital platform to find a job in 2021.
How might all this experimentation affect workers?
A new paper by Hatim Rahman Kellogg’s, Tim Weiss of Imperial College London and Arvind Karunakaran of Stanford takes on this question.
The research team analyzed nearly two decades of qualitative data from QuickHire, a nickname for one of the world’s largest digital job platforms. They found that during their study period, QuickHire had gone through three “experimentation regimes,” which were characterized by distinctly different approaches to experimenting with workers—and very different responses from the workers themselves. At first, the experimentation was clear and conducted only on those who had voluntarily chosen. In the second stage, experimentation became covert, with workers participating in studies without their notification or consent. and in the final stage, experimentation was unlimited, with many experiments running simultaneously and continuously.
The researchers came to think of these successive changes in approach as an “experimental hand” that, they write, reshapes workers’ sense of autonomy “by first increasing it, then decreasing it, and finally normalizing its decrease.”
Rahman notes that the prevalence of pervasive, quiet experimentation is growing outside of digital platform settings as well. even public policy ideas can be tested on unconscious subjects in this way. And while Rahman understands the benefits of rigorous experimentation, he fears that if left unchecked, its overuse could begin to weaken the social fabric by eroding trust.
“That begs the question, do we want to be a society where experimentation is just the norm,” he says, “and people don’t necessarily have a say or an awareness of what’s being experimented on?”
Taking a long view
The paper builds on Rahman’s previous research, which also looked at employees’ experiences on digital platforms in an effort to uncover how opaque rating algorithms affect employees and their behavior at work.
Rahman says the newest project grew out of a desire shared among the study’s co-authors: to look further back in time to better understand the present.
“We’re generally curious about how we got to where we are now, where we have this limitless experimentation,” Rahman says. “It’s unlikely that it happened out of the blue.”
They decided to take a more longitudinal approach in their study of QuickHire, a digital platform where gig workers around the world can sign up to provide services such as software development, marketing and graphic design to clients. The researchers collected extensive data from the platform between 2004 (when QuickHire launched the discussion board) and 2020. This data included discussion board discussions between QuickHire employees and moderators and archival records of QuickHire’s terms of service, company announcements and blog posts. For the period between 2014 and 2020, the data also included Rahman’s observations as a registered customer and employee.
Tracing QuickHire’s history in this way revealed to the researchers how important a role experimentation seemed to play in the employee experience on the platform—and how dramatically that role shifted over time.
Based on company announcements and message board posts, the researchers found that QuickHire had begun its first employee experiments around 2007, ushering in what the researchers called a “regime of explicit experimentation.” In that phase, QuickHire offered details about the features they were testing—for example, a change to the platform’s “job search” that would update how employees and customers found each other. In its public discussion board post about this experiment, QuickHire invited workers to “beta test” the new feature and provide feedback. During this time, QuickHire even asked for employee input on which parts of the platform to experiment with. Workers could either consent or opt out of the experiments.
Not only did this experimental phase increase worker autonomy—as workers had control over their participation in the experiments and the ability to express their views on how those experiments should work—but it seemingly set a precedent for the what workers could expect in the future.
“This regime established basic employee expectations about how platform-based experimentation worked,” the authors wrote.
Starting around 2014, researchers noticed a shift. At the time, employees had accidentally discovered that QuickHire was experimenting with the design and display of employee performance metrics on their profiles—seemingly benign changes that, as one employee put it, were nevertheless “the first thing customers would notice on my profile.” The company later acknowledged that it had actually tested the scheme on a group of workers who hadn’t been notified. QuickHire followed a similar course of action when it launched another stealth experiment: the platform began sending some users automated messages threatening to suspend their accounts when communicating with customers or employees outside of the site;
These events ushered in the “covert experimentation regime,” in which QuickHire conducted experiments without workers’ consent. During this phase, researchers saw workers’ autonomy diminish as their ability to control their participation in experiments (and related job characteristics) decreased.
In 2017, something changed again: experiments on the platform were no longer so episodic or theme-focused. they lacked a clear endpoint and included a number of platform features. The researchers’ data showed that QuickHire subjected workers to simultaneous tests of search filters, pricing proposals, communication tools and the information displayed on worker profiles, among other things. Workers expressed that “all these new things” were changing “all the time” and affecting their ability to do their jobs.
Unlike previous experimentation regimes, the platform gave no recognition – belated or not – of its experimentation to workers. Rahman and his colleagues were surprised by how workers responded to this “unrestricted regime of experimentation”: quietly, without pushing back or leaving the platform en masse.
“Instead of observing a collective exit of employees, we found that employees responded with resignation,” the researchers wrote, “as they expected to be experimented upon every time they used the platform.”
Rahman says the order of phases in this experimental regime change is important to note — even though it likely wasn’t intentional on QuickHire’s part. After securing initial buy-in for employee experimentation, the platform moved on to stealth and unlimited experiments simply because it could.
“I think there’s this initial idea of wanting to start with buy-in and scaling [experimentation] in ways that are mutually beneficial,” Rahman says. “But as platforms or organizations gain advantage and market power, they no longer have the incentive to make it mutually beneficial.”
The results of previous experiments
But organizations are doing themselves a disservice by failing to consider the long-term implications of stealth and unrestricted experimentation, Rahman argues. First, the integrity of the experiments themselves can be undermined by jaded participants who expect any small change they see to be part of some larger test. They can also be a challenge to morale. The researchers found “indications of anxiety, frustration and stress” among the workers.
In their paper, the research team recommends that organizations form internal boards—or external oversight units—to oversee experimentation and ensure useful, insightful outcomes for both leaders and employees.
The movement for trial justice is not new, Rahman points out, and digital platforms like QuickHire would do well to learn from similar movements in medicine and academia.
“This is not something we should imagine,” says Rahman. “Look at medicine: you can still do randomized control trials, but you have to have very explicit informed consent from the subjects. The same thing happens with academia.”
Ultimately, it’s likely that regulators will force the issue. states are already they start passing data privacy laws giving users more control over the information that online platforms can collect about them. Rahman can imagine something similar happening around experimentation and informed consent.
At the very least, he argues, we need more transparency about the nature of these experiments. A natural starting point could be to require websites that ask users to accept cookies to be clear about the consequences of choosing or not.
But in the meantime, he says, organizations have to make a decision: “Do you want to wait and react to this? Or do you want to be at the forefront of trying to think of more mutually beneficial ways to apply experimentation?”
