People sometimes exhibit a costly preference for humans relative to algorithms, which is often defined as a domain-general algorithm aversion. I propose it is instead driven by biased evaluations of self and other humans, which occurs more narrowly in domains where identity is threatened and...