Gig workers increasingly rely on unofficial tools and peer networks to navigate opaque, algorithmic labor systems. This paper investigates the case of Avalon—a paid batch-filtering app used by Instacart shoppers—and its affiliated Telegram community. Drawing on a two-year mixed-methods study including a survey (N=178), interviews (N=20), and 51,764 Telegram messages, we examine how workers resist platform constraints through technological and relational practices. While prior research often frames such resistance as adversarial or economically driven, we apply a feminist HCI lens to highlight care, consent, and infrastructuring as central to workers’ strategies. We show how Instacart’s majority-female workforce builds informal systems to reveal hidden information, protect one another, and maintain dignity in precarious conditions. Our findings contribute empirical insights into gendered algorithmic labor, theoretical extensions of feminist infrastructuring, and design implications for worker-centered platforms that reflect relational labor values. We argue for platforms that honor refusal, transparency, and collective agency from below.
ACM CHI Conference on Human Factors in Computing Systems