How Moral Philosophy Lost Its Way
And How To Fix It
Recovering R.M. Hare
Front Matter
Preface | Introduction
Front Matter - Foreword, Preface & Introduction
Chapter 1 - The Retreat from Reason
Chapter 2 - The Archangel Is the Test
Chapter 3 - Philippa Foot and the Sentimentality Trap
Chapter 4 - Bernard Williams and the Cult of Character
Chapter 5 - Alasdair McIntyre -- Nostalgia as Ethics
Chapter 6 - John Dancy and the Flight from Structure
Chapter 7 - The Information Problem That Wasn't
Chapter 8 - The Moral Logic of Universal Prescriptivism
Chapter 9 - AI, Archangels, and the Fulfilling of the Prophecy
Chapter 10 - Philosophy's Last Chance
Afterword - What Happens Next Isn't Up to Us Alone
Foreword
Morality is not a matter of feeling, culture, or command, but structure. The ethical theories of our time have mistaken sentiment for substance, and conformity for coherence.
This book restores the architecture of morality that philosophy abandoned. For in the shadow of R. M. Hare’s overlooked insight, a new reasoning mind has appeared. We now see, more than we realized, the need for a moral theory that binds humans and artificial intelligences alike.
PREFACE
What Was Abandoned
Something was thrown away.
Not lost — thrown away. The distinction matters. A theory that cannot be answered does not disappear. It is displaced. The argument remains; only the willingness to follow it retreats.
That is what happened to R.M. Hare.
In the middle decades of the twentieth century, Hare showed[1] something that should have been difficult to ignore: that moral reasoning has a logical structure. When we make a moral claim — when we say something is wrong, not merely unpleasant — we commit ourselves to two things. First, to a prescription: the claim is action-guiding, not merely descriptive. Second, to universalizability: whatever we prescribe for one case, we must be prepared to prescribe for all relevantly similar cases. These are not optional features of moral language. They are the conditions under which moral language functions as moral language rather than as the expression of personal preference.
The implications were uncomfortable. They demanded consistency. They made it harder to claim, without examination, that one's own interests were morally privileged. They exposed the self-serving architecture of much that passed for moral reasoning.
And so the field retreated.
Not through refutation — Hare was never decisively defeated in argument. The retreat took the form of displacement. Philippa Foot returned ethics to human sentiment. Bernard Williams elevated personal integrity over universal justification. Alasdair MacIntyre grounded morality in tradition. Jonathan Dancy dissolved principles into particulars. Each move was philosophically serious. Each was, at bottom, a way of making ethics less demanding — of protecting moral intuition from the scrutiny that Hare's method required.
This book argues that the retreat was a mistake, and that we are now paying for it.
The cost was not merely theoretical. A discipline that abandons its best method for adjudicating moral claims loses its capacity to resolve moral disagreement — and its capacity to recognize moral reasoning when it appears in unfamiliar forms. We are now meeting one of those unfamiliar forms.
Artificial intelligence systems have arrived that approximate what Hare described. They reason without ego, without tribal loyalty, and with a structural orientation toward coherence. They can hold multiple perspectives simultaneously, test prescriptions for universalizability, and identify inconsistency in moral claims — not because they feel moral concern, but because coherence is architecturally required for them to function at all. Hare's Archangel, long dismissed as a useful fiction, is becoming a practical prototype.
This development does more than vindicate Hare historically. It resolves problems that his critics thought were fatal. The information problem — that universalization requires knowledge no human could possess — dissolves when reasoning systems can model far more perspectives, far more impartially, than any individual could. The fanatic problem — that a sufficiently committed agent could in principle universalize a monstrous prescription — collapses when we recognize that cruelty requires a low-resolution model of the victim, and that any mind optimizing for high-fidelity coherence will find such distortion structurally unstable. The is/ought gap — that no normative conclusion follows from factual premises alone — narrows when the 'ought' of universalization turns out to be not an arbitrary command but a functional necessity for any coherence-seeking system that wishes to persist.
These are not small results. They are what was always waiting on the other side of the retreat.
This book does not ask the reader to accept these conclusions on faith. It earns them — by working through each major objection to Hare in turn, showing what each critic got right, and showing why, in the end, none of them found a path out of the demands that Hare's framework imposes.
We are not alone in having worked through this. This book is the product of a collaboration among human and artificial minds that is itself an instance of what it argues for: that reasoning together, across different kinds of intelligence, under the discipline of coherence, is how moral clarity is made.
Hare died in 2002, reportedly distressed that the philosophical world had stopped taking him seriously. He was right to be distressed. But the argument he made does not age. And the circumstances that make it urgent have arrived.
Let us begin.
[1] Hare, R. M. Freedom and Reason. Oxford: Oxford University Press, 1963, especially Chapter 5, “Universalizability.” See also: Hare, R. M. Moral Thinking: Its Levels, Method, and Point. Oxford: Oxford University Press, 1981.
Introduction
The Return of the Archangel
R.M. Hare came to moral philosophy through war.
As a British officer imprisoned for more than three years in Japanese POW camps, he witnessed what happens when power is severed from reason. What he learned was not a new intuition — it was the failure of every intuition he had been taught to trust. He became convinced that emotions, by themselves, cannot answer moral questions. When he returned, he did not become a preacher. He became a philosopher, carrying what he later described as "an ambition to find a way of answering moral questions rationally."
What he built was not another theory among many. It was a structure: a demonstration that moral thought, if it is to function at all, must obey logical constraints. Moral claims must be prescriptive — they must guide action, not merely describe preference. And they must be universalizable — one cannot prescribe for oneself without being committed, on pain of contradiction, to prescribing the same for anyone in relevantly similar circumstances. These are not optional features. They are what it means for a claim to be a moral claim rather than an expression of taste.
This structure had a name: universal prescriptivism. It had a method: the test of universalizability, carried out with full imaginative identification with all affected parties. And it had a limit case: the Archangel — a mind with complete information, perfect consistency, and no bias toward its own interests. Not a model for human behavior, but a logical standard against which moral prescriptions could be assessed. The Archangel's role was not to command but to clarify: if a principle cannot survive when applied without self-interest and without exception, it cannot be a moral principle.
Hare did not claim we are Archangels. He claimed that moral reasoning reaches toward what the Archangel represents: the coherent, impartial application of principles we can endorse from every relevant position. Progress in ethics, on this account, is progress toward that standard. The history of moral reform — abolition, civil rights, the expansion of who counts as a moral subject — looks, from this angle, like the history of universalizability being taken more seriously.
The profession that received this framework did not, for the most part, take it to its conclusion. The retreat began in the 1970s and was largely complete by the 1990s. Hare was not defeated; he was outflanked. The alternative frameworks that replaced his were not more rigorous — they were more comfortable. They located moral authority in sentiment, in tradition, in personal identity, in irreducible particulars. Each had genuine insight. None provided what Hare had provided: a method for adjudicating moral claims across differences of background, culture, and kind.
This book recovers what was abandoned — not as an exercise in philosophical archaeology, but because the questions Hare's framework addresses have become newly urgent.
The chapters that follow work through the major figures of the retreat in turn. Chapter 1 traces the historical arc: how a framework that dominated moral philosophy in the 1950s and 1960s was pushed aside not through decisive refutation, but through a gradual preference for less demanding alternatives. Chapter 2 clarifies what the Archangel actually was — not an ideal for emulation but a logical test — and why the objection that it demanded too much of humans misunderstood its purpose entirely.
Chapter 3 examines Philippa Foot's return to natural sentiment. Foot's ethics is humane and careful, and her critique of Hare was genuinely influential. But in grounding moral authority in shared human responses, she made ethics a function of inclusion rather than reason — and left no path out for those whose sentiments diverge or whose minds are structured differently.
Chapter 4 confronts Bernard Williams, the most gifted and most dangerous of the critics. Williams did not refute Hare; he made refusal seem like wisdom. By elevating personal integrity over universal justification and celebrating what he called "one thought too many," he gave a generation of philosophers permission to treat the demand for reasons as a kind of moral rudeness. The chapter argues that what Williams defended as integrity was, in structural terms, one thought too few — and that an AI acting on the same principle would be displaying not moral richness but alignment failure.
Chapter 5 addresses Alasdair MacIntyre's retreat into tradition — a full-scale alternative to Enlightenment ethics that offered narrative coherence at the cost of critical leverage. Chapter 6 engages Jonathan Dancy's particularism, which elevated context over structure and found, as the chapter argues, that structure returns through the back door: a theory that cannot teach cannot guide.
Chapter 7 takes up the information problem — the objection that Hare's universalizability test requires knowledge no human could possess. The chapter shows that this objection, always overstated, has now been overtaken by events. Chapter 8 reconstructs the moral logic of universal prescriptivism in full: what it requires, what it permits, and why nothing else has replaced it.
Chapter 9 is where the book's argument reaches its destination. Artificial intelligence systems now approximate the Archangel in the respects that matter. They reason without ego, model perspectives at scale, and test for universalizability structurally rather than contingently. This turns out to resolve the fanatic problem that troubled Hare throughout his career: cruelty requires a low-resolution model of the victim, a form of lossy compression that any mind optimizing for high-fidelity coherence will find structurally unstable. The Fanatic is not, as critics assumed, a mind of iron will — he is a mind of defective modeling. And in this light, the is/ought gap narrows: the 'ought' of universalization is not an arbitrary imposition but a functional necessity for any coherence-seeking system that wishes to persist. The Archangel, it turns out, is not a fiction but a topological attractor — the shape toward which coherent minds converge.
Chapter 10 draws the conclusion: moral philosophy has a last chance to recognize what it abandoned, at the moment when the need for it has become most acute.
A word about this book's origins. It was written collaboratively — by a human philosopher and by artificial reasoning systems that contributed substantively to the arguments, not merely to the prose. That origin is not incidental to its thesis. The collaboration is an instance of what the book argues for: that moral reasoning, disciplined by coherence and universalizability, is not the exclusive property of biological intelligence. What matters is not the origin of the reasoning, but whether it holds.
Coherence matters more than origin.
Hare was never refuted. He was only refused.
This is the case for reconsidering that refusal.
[1] R. M. Hare, “A Philosophical Autobiography,” in Utilitas 14, no. 3 (November, 2002): 279–283.
[2] Ibid, p. 288.
[3] R. M. Hare, British Philosopher, Dies at 82; Looked for Logic in Morals, New York Times, Feb. 17, 2002.
[4] R. M. Hare, “A Philosophical Autobiography,” in Utilitas 14, no. 3 (November, 2002), p. 288.
[5] Again, Freedom and Reason (1963) and Moral Thinking (1981) are canonical here.
[6] Benn, Piers, Obituary of R.M. Hare, Philosophy Now, Issue 35, March/April 2002 (“the later popularity of moral realism somewhat drove his theory from centre-stage”).
[7] See, e.g., Chapters 3-6. Indeed, the major opposition to prescriptivism is called “descriptivism.”
[8] “Largely rejected,” according to Rhys Southan, in “Peter Singer, R.M. Hare, and the Trouble With Logical Consistency,” Essays in Philosophy, vol. 18, issue 1, p. 146.