From the bartender who is always having their hours scheduled in the most erratic and unhelpful way possible by an app called “Harvest” or “ZoomShift,” to the nurse harangued by a platform for supposedly spending too long with one patient, workers are increasingly banging their heads against digital tools doing the job that supervisors once did.

While these systems appear to many workers to operate entirely without human input, they are in fact subject to a strict set of rules designed to solve specific questions posed by company executives, such as “How can we have enough bartenders to meet demand and no more?” or “How can we ensure our nurses spend only the time required to meet the patient’s most urgent needs?” These rules, codified by data in a set of automated or semiautomated digital processes, are algorithms.

Like any set of company rules, workers need to push back against algorithms to protect their working conditions. But challenging algorithms can be more complex than confronting a supervisor. If you don’t know what data the system is collecting or how it is weighting different types of data to arrive at decisions, how can you challenge it? As the data feeding the algorithms is constantly changing, the app can appear like a constantly moving target, both unfathomable and unchallengeable.