The 'one thought too many' critique targets Kantian moral theory most directly, but its force extends to utilitarianism and any framework that demands agents justify their constitutive commitments through impartial reasoning. Williams's point is not that impartial reasoning is bad but that its scope is limited — it cannot reach the commitments that make an agent the specific agent she is without dissolving them in the process of examining them.
The phrase illuminates a recurring pattern in the AI discourse. Practitioners are asked to justify their attachment to craft, their resistance to commoditization, their commitment to depth. The very demand for justification reveals the framework making it. A person whose identity is constituted by craft does not need to justify the commitment; the commitment is not a preference she holds but a dimension of who she is. The demand for justification imports exactly the abstract standpoint Williams rejected — a standpoint from which the practitioner's life can be evaluated only after her specific relationship to her practice has been bracketed.
The AI transition makes the problem acute. The market, the managerial discourse, and increasingly the ethics conversation all press practitioners to articulate in impartial terms why their existing commitments should not be sacrificed to aggregate efficiency. The pressure is itself the violation. A framework that demands the practitioner defend her ground project against impartial critique has already demanded she step outside it, and stepping outside it is indistinguishable from abandoning it.
Williams's counterposition is that some commitments are properly immune to this kind of demand. Not all — there are bad ground projects, and reflection can reveal them — but the demand for general justification is not the appropriate way to examine them. The appropriate examination attends to what the commitments are, how they have shaped the life they structure, and what their abandonment would cost. This examination is thick, particular, and refuses the one thought too many.
The phrase appears in Williams's essay 'Persons, Character and Morality' (1976, collected in Moral Luck, 1981). It was an immediate philosophical flashpoint, cited in virtually every subsequent debate about the demands of impartial morality, and has since become one of the most recognizable pieces of philosophical shorthand in contemporary ethics.
Against impartial justification of partiality. The demand that agents justify their preferential love for their own — spouses, children, projects — imports a standpoint that dissolves what the justification was supposed to defend.
Reveals the abstraction problem. The 'thought too many' exposes the morality system's characteristic move: demanding that agents adopt perspectives so general they no longer contain the particular commitments in question.
Applies to professional identity. When practitioners are asked to justify in impartial terms their attachment to their practice, the demand already presupposes the abstraction the justification would complete.
Protects the immediate. Some responses — saving a wife, refusing to commoditize a craft — are immediate expressions of who the agent is, not conclusions of reasoning, and the immediacy is evidence of integrity, not failure.
Sets a limit on moral theory. Moral theory properly examines many things; it does not properly examine the constitutive commitments of agents through the demand for general justification.