My purpose is to offer an assessment of the scientific legacy of Converse's “Belief Systems” by reviewing five productive lines of research stimulated by his authoritative analysis and unsettling conclusions. First I recount the later life history of Converse's notion of “nonattitudes,” and suggest that as important as nonattitudes are, we should be paying at least as much attention to their opposite: attitudes held with conviction. Second, I argue that the problem of insufficient information that resides at the center of Converse's analysis has not gone away, and that newly fashioned models of information processing offer only partial remedies. Third, I suggest that the concept of the “average voter” is a malicious fiction, as it blinds us to the enormous variation in political attention, interest, and knowledge that characterizes mass publics, in Converse's time as in our own. Fourth, I develop an affirmative aspect of Converse's analysis that has mostly been overlooked: namely, that if ideological reasoning is beyond most citizens’ capacity and interest, they might fall back on a simple and reasonable alternative, which I will call “group‐centrism.” And fifth, I consider the possibility that while the majority of individual citizens falls short of democratic standards, the public as a whole might do rather well.