Although certainly his contributions to metaphysics, epistemology and the history of philosophy continue to repay close attention, speaking now of, say, papers on personal identity or the book on Descartes, Williams's reputation rightly remains most closely tied to ethics, specifically his critique(s) of moral theory and his work in moral psychology. Such contributions, however, and no one seems more sensitive to this than Williams himself, may strike some readers as fragmentary, decentralized, ad hoc. The foregoing chapters have tried to combat this impression, assuming it even merits combating, in two ways. The first is by pointing up certain themes or, perhaps better, guiding principles on display in almost all Williams's writing. Three in number, spelled out in Chapter 1, these principles reflect his constant push, especially in ethics, for psychology adequate to moral phenomena, history adequate to contemporary values and philosophy adequate to the contingency of life. The second is by sketching, in connection with Shame and Necessity, at least the outline of a positive account of ethical practice, based largely on the internalization of norms and the necessity of acting in conformity with the character or identity those norms infuse.

Still, perhaps more may be expected by way of a positive programme, and in Ethics and the Limits of Philosophy Williams may seem to oblige such expectations, at least initially, when talking up a model of ethical conviction based not on certainty, nor on decision, but on “confidence”. After dismissing the possibility of ethical certainty – that is, after dismissing, for reasons considered in Chapter 6, the possibility of epistemological authority along the lines of science, capturing an ethical world “that is there anyway” – and after dismissing the possibility that the notorious value-creating decisions of existentialism might suffice, lacking, as they do, the “aspect of passivity” morality requires, Williams concludes that hopes for ethical conviction might best be harnessed to confidence (Williams 1985: 169). In other words, where absolute knowledge and radical will fail, confidence just might succeed.