“either will be undefined or will refer to something outside the domain [of moral theory proper]. Most moral philosophers have done the latter. The basic evaluative concept in their theories is defined in terms of something allegedly nonevaluative—typically, human nature, rationality, or the will of God. The alternative I am suggesting is to anchor moral concepts in an exemplar. Good persons are persons like that, just as gold is stuff like that. The function of an exemplar is to fix the reference of the term ‘good person’ or ‘practically wise person’ without the use of any concepts, whether descriptive or nondescriptive. An exemplar therefore allows the series of conceptual definitions to get started. The circle of conceptual definitions of the most important concepts in a moral theory—virtue, right act, duty, good outcome, and so on—is broken by an indexical reference to a paradigmatically good person.
Making the exemplar a person has an even more important advantage than its aid to theory. If all the concepts in a formal ethical theory are rooted in a person, then narratives and descriptions of that person are morally significant. It is an open question what it is about the person that makes him or her good. [….] [W]hen we say that a good person is a person like that, and we directly refer to Socrates, or to Saint Francis of Assisi, or to Mahatma Gandhi, we are implicitly leaving open the question of what properties of Socrates, Francis, or Gandhi are essential to their goodness.”