Asymmetric social ties are dynamic, detectable, and consequential for inference

Relational asymmetry—the common phenomenon in which two people hold fundamentally different views of their shared relationship—has long been treated as a static structural property of social networks. We challenge this view, demonstrating that asymmetry is instead an active process driven by interpersonal uncertainty resolution that unfolds through continuous belief updating within dyads. Tracking an emerging social network longitudinally, we reveal that the path from asymmetry to resolution follows a striking nonlinear pattern: convergence is most likely when initial disagreements are either small enough to reconcile with minimal effort or large enough to be impossible to ignore, suggesting that extreme relational mismatches act as a powerful catalyst for corrective action. At the network level, this dyadic process produces a dynamic equilibrium—asymmetries resolve while previously aligned dyads drift apart, sustaining a remarkably stable aggregate rate of relational asymmetry over time despite constant turnover in which ties are asymmetric. The consequences of asymmetry propagate beyond the dyad. Third-party network members perceive dyad-level asymmetry and use it to make adaptive social inferences, including how likely new ties will form and how fluidly will information flow through the network. Individuals whose own social spheres contain greater relational asymmetry show diminished ability to rely on network structure when making these inferences. These findings fundamentally recast relational asymmetry from a structural artifact to a dynamic, cross-level signal—one that shapes not only how dyads evolve but how entire networks organize.