Does identity matter in legitimacy judgments on algorithmic governance?

Despite the increasing deployment of algorithmic systems in public governance, these technologies face a persistent legitimacy deficit when compared to human decision-makers. Existing research has primarily focused on procedural features, transparency, and output performance to explain this phenomenon. Yet, the affective and identity-based mechanisms that underlie legitimacy judgments remain underexplored. This paper addresses this gap by integrating theories of ecological rationality and affective polarization into the study of algorithmic legitimacy. Using a preregistered survey experiment conducted in Finland (N = 2040), we examine how partisan identities and group-based moral intuitions shape public evaluations of automated asylum authorization systems. We show that legitimacy judgments are strongly conditioned by affective polarization: design features that violate in-group moral expectations trigger sharply negative evaluations -- what we term ``moral red flags''. The results indicate that legitimacy perceptions are not uniformly distributed but vary systematically with partisan identities, highlighting the value of disaggregated analysis in legitimacy research. By demonstrating that legitimacy is not a neutral assessment but a context-sensitive, identity-laden judgment, this study offers a novel perspective on how algorithmic governance is evaluated in polarized societies.