Kant Ethics

A man comes to your door. He is holding a knife, and he asks, politely, whether your friend is in the house. Your friend is in the house. You know what the man wants. Every person who has ever heard this story knows what to do, and it takes less than a second. You lie.

Immanuel Kant said you may not. He said it in print, late in his life, to a critic who had put exactly this case to him, and he did not soften it. You may not lie to the man with the knife.¹ If you tell him the truth and your friend dies, the death is not yours. If you lie, and the lie goes wrong, and your friend dies anyway, the death is yours.² Ever since, this answer has been the joke people tell about Kant. The philosopher so in love with rules that he would hand a friend to a murderer rather than bend one.

The joke is wrong, and so is Kant, and both of those are worth the trouble of seeing exactly why. Kant is wrong about the man at the door. He is right about almost everything the answer rests on, and the things it rests on are the whole of his morality. That the only thing in the world that is good without any condition is a will that does what is right because it is right.³ That there is one rule for every action. The rule is not a list of commands but a test that any rule a person lives by has to pass.⁴ That nobody can ever make an exception for himself, not once, not for a good reason. The moment anyone does, the rule was never a rule.⁵ That nobody can ever take credit for being good, because nobody can see their own motives, and neither can anyone else. And that all of this is possible for one reason only, which is that you are free.⁶⁷

Most people carry a picture of morality that is roughly the opposite of this. Being good is about the results an act gets and the kind of person someone is. Rules are rough guides, and a kind person knows when to bend them. A small lie that keeps the peace is fine. A person's kindness is theirs, and they deserve credit for it. That picture is attractive, and most of the people who hold it are decent, and it breaks in three places the moment it is pushed. If goodness is the kind of person someone is, then a lucky temperament is virtue and an unlucky one is vice. Nobody deserves praise or blame for either. If goodness is results, then two men pull a child from the river and one does it for the reward. He is exactly as good as the one who did it because she was drowning. And if a kind person may bend a rule when it seems right, every promise made to anyone lasts only as long as their judgment does. Which means it was never a promise. Everyone who holds the attractive picture refuses all three of these the moment they are on the receiving end.

Kant's morality is what comes of taking those three refusals seriously and following them all the way down. It ends at a place harder than most people are willing to stand. A morality that nobody can be seen to keep, that gives no credit, that allows no exception, and that binds everyone anyway. And the man at the door is where it finally breaks, in a way that shows what it was made of.

Kant's whole morality starts from one sentence, and the sentence is a claim most people would deny on first hearing. Nothing in the world, or outside it, is good without any condition except a good will. Not intelligence, and not courage, and not a calm temperament, not health, not wealth, and not even happiness. Every one of those is good only on a condition, and the condition is the will of the person who has it.⁸ The will is the one thing that is good on no condition at all. What follows is the reason to believe it.

There is a man who has every gift. He is intelligent, and he reads people fast. He is brave, and does not flinch when a plan goes wrong. He has the kind of calm that lets him sit across a table from someone he is about to ruin and eat his lunch. He has money, and friends who owe him favours, and he sleeps well. Now put all of that in the service of a will that wants to defraud a pensioner out of her savings. Every gift on the list has just become worse than its absence. His intelligence is what makes the scheme work. His courage is what keeps him from backing out when it might have collapsed. His calm is the thing that makes him frightening. Kant says exactly this. The coolness of a villain does not only make him more dangerous, it makes him more hateful to us than he would have been without it. A frightened, clumsy fraudster is a lesser thing than a composed one.⁹

Now the same list goes to a woman whose will is set on doing what is right. Her intelligence finds the honest path that a slower mind would miss. Her courage holds when honesty costs something. Her calm means she can be fair to someone who is shouting at her. The gifts have not changed, and the only thing that changed was what they were for. That one change turned each of them from an asset in a crime into an asset in a decent life. So the gifts were never good in themselves. They are amplifiers, and they make whatever will holds them louder.

The same goes for the gifts nobody chooses. Health, wealth, a good position, the good fortune of being born in a peaceful decade. Kant says they are good only on the same condition. A tyrant with excellent health is a longer tyranny. And the claim people balk at is that happiness is on the list too. Kant says that a reasonable observer, watching someone prosper who has no trace of a good will, feels no pleasure at that prospering.¹⁰ The happiness of a cruel man is not something anyone wishes for him. You can test this on yourself, with someone whose conduct you know to be vicious, by thinking of them content, secure and untroubled. The picture does not feel like a good thing in the world, but like something out of order. So even happiness, the thing every other morality treats as the point, is conditionally good. It is good when the person who has it deserves it, and the deserving is a fact about their will.

At this point the obvious objection arrives, and it is a strong one. What about kindness, warmth, sympathy, the person who cannot walk past someone crying without stopping. Surely that is good without any condition. Kant's answer is that warmth is temperament. It is a gift, like courage or calm, and you no more chose it than you chose your height. And like every other gift, it amplifies whatever will holds it. The sympathetic man who is drawn into a cruel cause helps the cruel cause warmly.¹¹ The warm friend who cannot bear to see you suffer is the one who tells you the comfortable lie. Then you walk into the thing that will hurt you. Sympathy is pointed by something outside itself. What points it is the will. So sympathy is on the list of conditional goods, near the top, and it stays there.

That leaves the second objection, which is the one a practical person makes. A good will that achieves nothing is worthless. The world is made of outcomes. A surgeon with a good will and shaking hands is worse than a surgeon with steady hands and no interest in you at all. If the good will accomplishes nothing, it is hard to say what it is worth. Kant faces this directly and does not blink. He asks us to imagine a good will completely unable to carry out its purpose. Through some hostility of nature, through bad luck or a stingy allotment of ability, the greatest effort of that will produces nothing.¹² It would still, he says, shine like a jewel by its own light, as a thing that has its whole value in itself.¹³ Usefulness would add nothing to that value and uselessness would take nothing away.

That sounds like a beautiful thing to say and a hard thing to defend. Kant's own defence is the corrupt villain and the observer who takes no pleasure in his prospering. The defence that follows is built on what luck can and cannot reach, and it is offered as the reason his line holds. Two people each try to pull a child out of a river. One reaches her, and the other, with the same will, the same effort, the same risk, is beaten by the current, and the child is lost. Which of them is the better person. The question feels wrong, and the reason it feels wrong is the whole point. The current is not a fact about either of them. If goodness lived in the outcome, then the river would decide who was good, and the river cannot decide anything. Everything that separates the two rescuers is luck. Luck reaches the outcome, and it reaches the surgeon's hands, the timing, the weather, the strength of the current. There is exactly one thing in the whole scene that luck cannot reach, and that is what each of them was trying to do and why. The claim under Kant's line is that this is where goodness lives, because it is the only place goodness could live and still be theirs.

The surgeon objection has not gone away. The shaking hands matter, of course they do, and Kant never says otherwise. Skill is good, and you should want the surgeon with the steady hands. But skill is on the first list, the list of conditional goods, and the fact that you would rather have it does not move it. The steady-handed surgeon who does not care about you is useful to you. That is a different thing from being good, and the proof is that his steadiness would serve a worse purpose exactly as well.

So much for the first step. Every gift you have, chosen or not, is good on a condition, and the condition is the will that holds it. Happiness is on that list, which is why the prosperity of a cruel person looks wrong to everyone who sees it. The will is the only thing that is good on no condition, because it is the only thing luck cannot decide. And that means goodness does not live in what you achieve, but in what you will, and why. If the good will is not the good it does, then something else makes it good, and nothing said so far says what.

A will is good when it does what is right because it is right, and not because of anything it wants. Kant's word for that is duty. An act from duty is one that would still be done if every wish, mood and appetite in the person pointed the other way. It is easy to state and very easy to misread. The misreading is so old that Kant answered it himself. First the cases that make the distinction visible.

Two shopkeepers, the first of them Kant's and the second the one his argument needs. Both keep a fixed price, so that a child who comes in with her mother's money pays the same as anyone else. Both are honest, in the plain sense that neither of them ever cheats a customer. The first one keeps his fixed price because it is good for business.¹⁴ A shop where a child gets cheated is a shop people stop coming to, and he knows it. He is honest the way a man is honest who has done the arithmetic. The second one keeps his fixed price because cheating the child would be wrong. He would keep the price if business fell, and he would keep it in a town with no competitors, where nobody would ever find out. The two men do the same thing every day. From outside they cannot be told apart. Kant says one of them is honest from duty and the other is honest in a way that merely agrees with duty. Only the first has what he calls moral worth.

Before the objection, one more pair, because this one is the one people remember. A man loves helping people. It warms him, it makes his day, and he goes out of his way to do it. Kant says this man is amiable, and that his helping deserves praise and encouragement. And then he says the thing that makes people angry. It deserves no moral esteem, because it is in the same class as the wish to be admired. It is an inclination that happens to point at something useful. Then he takes the same man and changes one thing. A grief has fallen on him that has killed all his warmth. He is so full of his own trouble that other people's trouble no longer reaches him. He feels nothing for them. And in that state he tears himself out of his own deadness and helps anyway. No inclination toward it at all, simply because it ought to be done. Now, Kant says, and only now, does his action have its genuine moral worth.¹⁵

A poet who read Kant put the complaint nearly everyone has as a joke in two voices. The first voice is a man of tender conscience. I gladly serve my friends, but sadly I do it with inclination, and so it gnaws at me that I am not virtuous. The second voice is the poet playing Kant. There is no other way, it answers, you must try to despise them, and then do with disgust what duty commands.¹⁶ That is the reading the joke assumes. On that reading, Kant has built a morality in which the warm person is worth less than the cold one. Helping a friend with pleasure is a moral failure. The ideal person is a man who does the right thing while hating it. If that is what Kant meant, he was wrong, and not in a small way.

It is not what he meant, and the case itself shows why. Look at what Kant did to the warm helper. He did not make him cold in order to praise coldness. He made him cold in order to remove the one thing that had been hiding the question. While the man enjoys helping, there is no way to tell whether he would help if he did not enjoy it. The enjoyment and the duty point the same way, so the act tells you nothing about which of them was steering. Grief takes the enjoyment out. Now the duty stands alone. If the man helps, you can see, as nearly as such a thing is ever seen, that duty was enough on its own. The grief is a way of seeing, and nothing in it is recommended. Kant is not telling you to be that man. He is telling you that only on that man's worst day can anybody, including the man himself, see what his helping was made of.

So the distinction runs between an act that depends on the mood and an act that does not. Warmth and coldness have nothing to do with it. The warm helper who would stop helping the day it stopped feeling good was never acting from duty, however much good he did. The warm helper who would go on helping through grief would have been acting from duty all along, though nothing showed it. The warmth was just the weather he happened to be doing it in. Kant does not ask for the warmth to be absent. He asks that the act not rest on it. He said so himself, in a footnote answering the poet's serious version of the charge, before the joke was even written. The dignity of the law, he wrote, does not demand a slavish frame of mind. A heart that is cheerful in doing its duty is the sign of a genuine disposition to it. A person who does what is right with a grudge has not yet learned to love the good.¹⁷ That is Kant, not his critic.

The text can be read the other way, and that is said once. Kant's wording is that only now, without any inclination, does the act have genuine worth. A careful reader can take that to mean worth requires the inclination to be gone. The reading here takes it as a test and not an ideal. The cheerful-heart footnote tells against the other reading, and the cases make sense only as tests.

Something subtler is left over, and it survives the answer just given. Grant that Kant is not praising coldness. He is still saying that the cold helper's worth is visible and the warm helper's is not. Does that not make the cold man better, in the only sense that can be checked. No, and the reason matters for everything that comes later. It makes the cold man's worth as visible as worth ever gets. It says nothing at all about whether the warm man has it. The warm man may be acting from duty every single day, with the sun out, and nobody will ever know, because the sun never goes in. The distinction is about what can be seen, not about who is good. That gap, between a worth that is real and a worth that can be seen, is the strangest thing in Kant's morality. It reaches all the way to your own view of yourself. For now it is enough to hold the two apart.

There is one more case, and it is the one that shows duty is not the enemy of your own life. A man in despair, worn down by misfortune, wishes for death. Most people who stay alive do so because they want to. Kant says that wanting to live, and living, has no moral worth, because there was never any duty in it, only inclination doing what inclination does. But the despairing man wishes for death and preserves his life anyway, not from fear and not from love of it, but because he ought to. His staying alive has moral worth.¹⁸ It is the grieving helper again. When the wish is gone and the act remains, you can see what was carrying it. And a little later Kant adds a line that people miss. To secure your own happiness, he says, is a duty, at least indirectly. A person crushed by need and want is in constant danger of breaking their duties.¹⁹ A person with their life in some order is not. Your own well-being is on the list.

Put the pieces together and Kant's three claims about duty come out in order. First, an act has moral worth only when it is done from duty, which means it would still be done with every inclination pointing away. Second, that worth does not come from what the act achieves, but from the rule the person was acting on when they did it.²⁰ The shopkeeper's honesty achieved the same thing on both readings, and only the rule underneath was different. Third, duty means acting out of respect for a law, a rule that holds whether or not you feel like it. Never out of a wish, even a good one.²¹ That word, law, has been doing quiet work through every case here, and it has not yet been said what the law says. Duty is duty to something. The second shopkeeper was obeying a rule the first one never consulted. The whole of the next step is finding out what that rule says.

The rule the second shopkeeper was obeying has no items in it. Ten commandments, or a hundred, would tell you what to do, and this does not. It is a single test that any rule you act on has to pass, and it goes like this.²² Act only on a rule that you could will everyone to act on. Everything Kant says about right and wrong comes out of that one sentence. It sounds too thin to carry anything, and the false promise is going to show that it carries more than it looks like it can.

Before the test, the thing it tests. Every act you do on purpose has a rule underneath it, whether you have ever said the rule out loud or not. Two people leave a tip at the same cafe. The first leaves it because the waiter looked tired and she wanted to do something for him. The second leaves it because she plans to come back and wants better service next time. Same act, same coins, same table. But if you asked each of them why, and they were honest, you would get two different rules. The first: when someone has served me well and is worn down, I give what I can, to help. The second: when I will be back, I tip, to be treated well. Kant calls that private rule your maxim, and the whole test is run on it. A maxim is the rule you are actually acting on, in the form when this, I do that, in order to get this. The one that is really steering, whatever you would recite if asked.²³

Now the test, on the case Kant used, in full. A man needs money badly. He knows he cannot pay it back, and he knows the only way to get it is to promise firmly that he will. He wants to make the promise. So the maxim is: when I am in need, I borrow and promise to repay, knowing I will not, in order to get out of trouble. Kant's test is not whether this is nice, or whether it hurts anyone, or what would happen if the man got caught. The test is a single question. What if that maxim were a universal law, a rule that everyone in need followed as a matter of course. Run it. In that world, everyone in need promises to repay while intending not to. So everyone knows that a promise from a person in need means nothing. So nobody lends on a promise. So the man's maxim, followed by everyone, produces a world in which the promise he wants to make cannot be made, because nobody would believe it. The rule eats itself before it gets to any result. Kant's words are that such a maxim could never hold as a universal law but would necessarily contradict itself.²⁴

Hold that word, contradict, because it is the whole difference between Kant and everyone else. Most people, asked why false promises are wrong, say that they hurt the lender, or that if everyone did it the world would be worse. Kant does not say either of those. His point is that the false promise only works because promises are trusted, and promises are only trusted because most people keep them. The liar is drawing on a stock of trust that other people built by not doing what he is doing. His rule needs everyone else to follow a different rule. That is what it means for a maxim to fail the test: it is a rule that works only as an exception. Universalise it and it cancels, the way a forged banknote works only while most banknotes are real.

Now take the particulars out, because the case was never about money. Any rule that works only because other people are not following it fails the same way. Cutting the queue works because the queue exists, and the queue exists because the others did not cut. Cheating on the exam works because grades mean something, and they mean something because most people did not cheat. Reading the private message works because people write private messages, and they write them because they are usually not read. Every one of these, run through the test, contradicts itself in exactly the false promise's way. And now the general form, which is the false promise with the details removed. A maxim fails when the world in which everyone follows it is a world in which following it is impossible. That is Kant's first kind of failure, and it is the hard kind. He says such maxims cannot even be thought as universal laws without contradiction.

There is a second kind of failure, and it is softer, and the difference matters. Take the maxim: when others are in trouble, I do nothing, and I ask nothing of anyone, in order to be left alone. Now universalise it. A world where nobody helps anybody. That world is perfectly possible. Nothing contradicts itself and nothing cancels. It is a colder world, but it holds together. Kant says so, and even says it might be better than a world full of talk about sympathy and no action. So why does this maxim fail. Because you cannot will it. It can exist, but you, the person willing it, will need help some day, and on that day you will want the opposite. A will that says let nobody help anybody, and also, when my turn comes, help me, is at war with itself.²⁵ Kant's phrase is that it is impossible to will that such a principle should have the universal validity of a law of nature. So the first kind of maxim cannot even be imagined as a law. The second can be imagined but cannot be willed. Two kinds of failure, and two verdicts, and Kant is careful to keep them apart. Kant calls the first kind perfect duties, and they allow no exception at all, like not making false promises.²⁶ He calls the second kind imperfect, and they leave you room, like helping. The duty is real, but how and how much is yours to judge.²⁷

The argument's skeleton, spoken once, in plain steps. One: every deliberate act has a maxim, a private rule of the form when this, I do that, in order to get this. Two: the moral question about any act is not what it achieves but whether its maxim could be a law for everyone. Three: a maxim fails that test in one of two ways. Either the world in which everyone follows it is a world in which it cannot be followed. Or that world is possible, but you could not will it without willing against yourself. Four: what fails the test is forbidden, and what passes is permitted. Among what passes, some things are required, because refusing them fails the second way.

So the rule behind the second shopkeeper was a test with one question: could I will this as a rule for everyone. The false promise fails it because the promise only works while people keep promises. So the liar's rule needs everyone else to be following a different one. That is what wrong is, for Kant: a rule that works only as an exception. Which invites the complaint that the test was rigged from the start. The test seems to work on the false promise, but it was set up to. Word the maxim differently and anything passes. A rule that can be gamed by wording is no rule at all, and if that is right, the whole structure is empty. It is half right.

The strongest objection to Kant's test is that it is empty. It has been made by the two most serious philosophers who came after him, in two different forms, and half of it is right. The half that is right names a real gap, and Kant's own next step is what fills it.

The first form comes from Hegel. Kant's example, in his second book on ethics, is a deposit. Someone hands you money to keep, then dies, and there is no record.²⁸ The maxim is: when nobody can prove I hold it, I keep it, to get richer. Kant runs the test and says the maxim would annihilate itself, because the result would be that there would be no deposits. Nobody would hand money to anyone.²⁹ Hegel's reply is short and it cuts. So what. A world with no deposits is not a contradiction. A world with no property, he writes, is no more a contradiction than a world with no human beings in it. Neither is a contradiction at all. The test only bites if you have already decided that property should exist, and deposits should exist, and people should be able to trust each other.³⁰ But that decision is the whole of morality, and the test did not produce it. It was assumed before the test was run. So the test is a machine that hands you back whatever you put in. Kant's ethics, on this reading, is the preaching of duty for duty's sake with no content of its own.³¹

The second form is Mill's, and it sharpens the same knife. Kant, he says, fails almost grotesquely to show any contradiction in everyone adopting the most outrageously immoral rules. All he shows is that the consequences of their universal adoption would be such as no one would choose to incur.³² Which means, Mill says, that when Kant runs the test he is secretly counting consequences. He is a utilitarian who will not admit it, and the talk of contradiction is a costume.

And there is a third form that any clever student finds on their own, which is the wording problem. The test is run on a maxim, and a maxim is a description of what you are doing, and descriptions can be tailored. The false promise fails when the maxim is when in need, I promise falsely. But suppose the man words it differently. When my name is Fritz, and it is a Tuesday, and I owe exactly this sum, I promise falsely. Universalise that. The world in which every Fritz does this on Tuesdays for this sum still has promising in it, because there are hardly any such Fritzes. The maxim passes. So the test gives whatever answer the description was built to give. If that is right, the test decides nothing, and the person who decides is the one choosing the words.

Three objections, and what survives them, taken in order. Against Mill first, because his is the easiest to answer on the promise and the hardest to answer anywhere else. The false promise does not fail because a world of false promises would be unpleasant. Look again at what happened when the maxim was universalised. The promise did not become costly. It became impossible, because a promise nobody believes is not a promise, the way a signature nobody accepts is not a signature. That is the act disappearing under its own rule, which is a different thing from a consequence you would rather avoid. Mill has missed the difference between a world that is worse and a world in which the thing you want to do cannot be done. The forged banknote fails because in a world of universal forgery there is nothing to forge, whatever it does to anyone's wealth. So the hard kind of failure is real, and it is not consequences in disguise. Two things are owed to Mill all the same. The softer test, the one about what you cannot will, does look at what you would want, and he is right about that much. And his real target was the most outrageously immoral rules, cruelty among them, and cruelty exploits no practice, so contradiction alone does not catch it. That case is handed to the second formula, which does.

Against Hegel, the answer is half a concession. He is right that a world with no deposits contains no contradiction. A world with no promises, no property, no trust, is a perfectly possible world. Kant's own sentence says the law would annihilate itself, and Hegel read it fairly. What saves it is reading the annihilation as inside the will of the man who wants to keep the deposit, and nowhere in the world. He wants two things at once. He wants a world in which people hand over money on trust, because that is where the deposit came from. And he wants a rule that, made general, ends that world. His maxim relies on deposits and abolishes them in one motion. Hegel is right that if you did not want deposits, the test would say nothing. But the thief wants them. He has not decided against property. He wants property to exist for everyone else and to be suspended for him, once, now. That is the same shape as the false promise, and it is the shape of every wrong the test catches. A rule that works only as an exception.

Which leaves the wording problem, and here the concession is whole. The universal law test, on its own, cannot tell you which description of your act is the honest one. Kant does not solve this, and there is no sentence in his text that does. What the test can do is refuse a maxim once you have stated it honestly. What it cannot do is make you state it honestly. The Tuesday Fritz has not defeated the test. He has lied to it, by feeding it a rule he is not actually following. He is not acting on a rule about Tuesdays. Tuesday has nothing to do with why he is promising, and he knows it. His real maxim is the one about need, and that one fails. But that answer leans on the word honestly, and the objection can press there. Who decides what the honest description is. If the answer is the agent, then the agent decides whether he passes, and the test is a mirror.

So where the argument actually stands is this. The test is not empty. It catches every rule that needs everyone else to be following a different one, and that is a real and large class of wrongs. It catches them by the shape of the will, with no counting of harms. But the test cannot, alone, fix what counts as the rule being tested. And on a crude reading it catches too much: leaving early to beat the traffic works only because others do not, and cancels if everyone does. The driver exploits nobody's trust, and the first formula alone cannot see that difference. Something else has to. And the something else is already in the cases, if you look at what the Tuesday trick actually does. It re-describes the act so that the person on the other end of it disappears. The lender is gone from the Tuesday maxim. The dead depositor is gone from the maxim about unprovable money. Every gamed description works by leaving out the one whose trust is being used. Kant's second formula is about exactly that person. Kant himself said to judge by the first formula, and to use the second only to bring it closer to feeling. The reading here reverses his order, because the first alone could not stop the Tuesday trick, and that reversal is a departure from him, made once.³³ There is someone on the other end of every maxim, and you owe them something.

Kant's second formula says: treat humanity, in your own person and in every other, always as an end and never merely as a means.³⁴ One word in that sentence carries the whole rule, and the word is merely. Drop it and the rule forbids ordinary life. Keep it and the rule says exactly what the universal law test could not: who the maxim has to be honest about.

Two taxi drivers, or rather one driver and two passengers. The first passenger gets in, gives an address, is driven there, pays, and gets out. She has used the driver. She wanted to be somewhere and he was the means of getting there, and she had no interest in him beyond that. The second passenger gets in, gives an address, is driven there, and at the end pulls a knife and takes the driver's night's earnings. He has also used the driver. Both passengers treated the man as a means to their end. That is what a taxi is. Kant does not object to the first passenger. He does not object to employment, trade, asking for directions, or any of the thousand ways people use each other every day. The difference between the two passengers is not that one used him and the other did not. It is that the first used him in a way he had agreed to, and the second in a way he could not have.

That is what merely means. Someone is treated merely as a means when they are used in a way they could not agree to. The use only works if they do not know what you are doing, or cannot refuse it. The first passenger's use of the driver has him in it as a person. He set the fare, he chose the route, he can decline the ride. The second passenger's use of him works only because he is, for a moment, an obstacle with a wallet. His agreement is not merely missing. It is impossible, because the act is one that only works without it.

Kant runs the false promise again under the new formula, and the result is sharper than the first time. The man who borrows on a promise he will not keep is using the lender as a means. So far so ordinary, since every borrower does that. But Kant says the lender, whom the man means to use for his own purposes, cannot possibly agree to the way he is being treated.³⁵ Not because he would refuse if asked. Because the whole act depends on his not being asked. A false promise can only be made to someone who does not know it is false. The moment the lender knows, there is no deal to consent to, because the deal was the lie. So the lender is being used in a way that requires him to be kept out of the decision. That is merely as a means, in the exact sense the taxi driver with the knife at his throat is.

This is why it fixes what the last formula could not. Remember the Tuesday trick. The borrower re-described his act until the lender vanished from the maxim and the test had nothing to catch. The second formula will not let him do that, because it does not ask about the rule. It asks about the lender. Is there a person on the other end of this act, and could that person agree to what is being done to them, knowing it. The lender cannot be described out of the room, because the question is about the lender. The Tuesday maxim is now irrelevant. There is still a man handing over money on a belief that was planted in him to make him hand it over. So the honest description that the first test could not force is forced by the second. The description has to include everyone the act is done to.

Three more cases, and then what they share. The salesman who tells a buyer the car has never been in an accident, when it has, is using that trust to move the man's money. The buyer cannot agree to that, because the act works only if he believes it. The friend who is kind to a man because that man might be useful later is using him. He is not using him merely, as long as the kindness is real and the use is one the man could agree to if he knew. The lover who says he wants what his partner wants, to get what he wants, has crossed the line. He crossed it at the exact point where her knowing would end it. What they share is this. An act treats a person merely as a means when it depends on their not knowing, or not being able to refuse, what is done. And when the rule it is done under is one they could not agree to. That last clause matters. A thief who is arrested cannot refuse the handcuffs. But he could agree to the rule that thieves are arrested, and Kant says the punished criminal is never treated merely as a means.³⁶ The robbed driver could not agree to any rule under which he is robbed. And two cases from before are settled by the same clause. The early driver uses none of the others, who could agree to his rule, so he passes here. That is what the first formula could not see. And cruelty, which contradiction could not catch, is caught at once, because nobody could agree to a rule of being hurt for another's pleasure.

That invites a fair reply. Consent cannot be the whole test, because people consent to terrible things. A desperate man will agree to almost anything. If the rule is only that the other person could agree, then a bad enough bargain, once agreed to, treats nobody merely as a means. That is wrong. The answer is that Kant's word is humanity, and humanity is not the person's wishes. It is their capacity to set ends for themselves at all.³⁷ The driver could set his fare, choose his route, decline the ride. A hammer can do none of those, and that is the whole difference between a person and a thing. The desperate man's agreement is real. But a bargain that trades on his desperation is using the fact that his capacity to choose has been narrowed to nothing. It works because he cannot say no, not because he says yes. So Kant's test asks something stronger than whether they said yes. It asks whether the act treats them as someone who could have said no. Kant applies the same formula to what a person does to himself, because he is a person too. The formula covers the one person you are always in a position to use, and what that means is the next step.

Using people is not wrong, and Kant never says it is. The word that does the work is merely. Someone is used merely as a means when the act depends on their not knowing, or not being able to refuse, what is done to them. And when they could not agree to it as a rule. That is why the false promise fails again, and fails harder, under this formula. And it is why the wording problem is answered. The maxim has to be honest about the person on the other end, because the second formula asks about them directly. So there are two formulas, and Kant says they are one law seen from two sides. From the side of the rule, it must hold for everyone. From the side of the person, everyone it touches must be able to agree to it.³⁸ Everyone the act touches includes one person who is always in the room, and who is the easiest of all to use. You owe that person exactly what you owe the driver.

A person can wrong himself, and it is the same wrong he does to anyone else. It is the part of Kant's ethics that almost nobody carries out of the room, because it sounds like a figure of speech. It is not one. Kant means that a person has, in his own case, someone he can use merely as a means. When that happens, the thing damaged is the same thing the false promise damages, and the knife in the taxi. Three cases show it, and the third is the one that matters most for the rest.

Before the cases, the thing that makes the whole idea look confused, and Kant raised it himself. A duty is something one person owes another. If I owe it, I am bound, and if another is owed it, he can release me. But a duty to myself would mean that I am both the one bound and the one who can release. And a person who can release himself whenever he likes is not bound at all. So a duty to oneself looks like a contradiction.³⁹ Kant's answer uses a distinction that is going to carry the whole account of freedom later. The one who is bound is a person as a creature in the world, with appetites, moods, a body that wants things. The one who binds is that same person as a lawgiver, the one who can ask what a rule for everyone would be. They are one person regarded two ways.⁴⁰ And the second cannot release the first. Releasing would mean asking whether a rule that lets me off could hold for everyone, and no such rule can. So the duty holds. The person who owes it and the person who is owed it are the same person, seen from two places.

Now the first case, and Kant put it first in each of his lists of examples.⁴¹ A man's life has become a burden. Misfortune has piled up, the prospect ahead is only more of it, and he considers ending it. His maxim, honestly stated, is: when a longer life promises more pain than pleasure, I end it, out of care for myself. Kant runs the universal law test and finds a contradiction, and it is not the obvious one. The feeling that drives the man is self-love, the drive to protect and further his own life. He is proposing to use that very drive to end the life it exists to protect. It would be a fire alarm whose standard function was to start fires.⁴² A nature built that way could not hold together, because the rule turns the thing that keeps a creature alive into the thing that kills it. Then he runs it under the second formula, and it is sharper. The man who destroys himself to escape a painful condition is using a person, himself, as a means to a bearable state. But a person is not a thing.⁴³ He cannot be used merely as a means, and that includes by himself. In his last book on ethics Kant says it as hard as it can be said. To annihilate the subject of morality in your own person is to root out morality itself from the world. So far, he adds, as it is in your power.⁴⁴ The hedge is his and it matters. He is not saying the man ends all morality. He is saying the man ends the one instance of it he was entrusted with.

What this rules, and what it does not, is easy to over-hear. Kant sets aside every consideration the question is usually argued on. Not whether it harms others, not whether it is cowardly, not whether God forbids it. He asks only whether a person is bound to his own life merely by being a person, and he says yes.⁴⁵ And he says it against the strongest view on the other side, which he gives its due. The Stoic thought the wise man could walk out of life at will, with an untroubled soul, as out of a room that is smoking.⁴⁶ Kant's reply is not contempt. This very courage, this strength to hold something higher than life, should have been the strongest reason of all to keep the being who has it.⁴⁷ The person who can stand above his own suffering has just proved he is the kind of thing that must not be thrown away. And then Kant asks questions he does not answer. Is it self-murder to throw yourself to certain death to save your country. May a man bitten by a rabid dog end himself before the madness takes him and he harms others.⁴⁸ He leaves those open, on the page, and they stay open here. The rule is clear on the case it was built for. It is not a machine for every case, and Kant knew it.

The second case is quieter and reaches more people. A person has an ability, a real one, and lets it rust. Not from malice. From comfort, from the ease of a life in which nothing is asked. Her maxim is: when developing what I could do would cost me my ease, I let it go, to enjoy myself. Run the test, and this one does not contradict itself. A world of people who let their gifts rust is a perfectly possible world.⁴⁹ There are islands, Kant says, where something like it exists.⁵⁰ So this is the second kind of failure, the kind that cannot be willed. The person willing it is a being with purposes, and every purpose she will ever have needs some capacity to serve it. To will that all capacities rust is to will that nothing she will ever want can be reached. Under the second formula it is a failure of the same kind as the first case, smaller. There is more in a person, Kant says, than what they are now. Letting it rot leaves the person alive and leaves everything they could have been dead. It is treating the person she could become as a means to the comfort of the person she is.⁵¹

The third case is the one Kant thought people needed most and the one that seems least like a duty. It is the duty not to be servile. A person who bows and scrapes, who pursues even a good end by crawling for favour, has done something wrong. So has the one who lets others tread on his rights and says nothing. The wrong is to himself. Kant's words are: be no man's lackey. Do not let your right be trampled underfoot with impunity.⁵² And he goes further than anyone expects. A person has an inner worth, he says, by which he is for sale at no price and possesses a dignity nothing can take from him.⁵³ Kneeling and prostrating oneself, he writes, even to make the veneration of heavenly things visible, is contrary to human dignity.⁵⁴ Under that sits a claim. Servility is not a weakness of character, a thing to pity. It is a lie, told about a person's own worth, to someone who benefits from believing it. The person who crawls is presenting himself as a thing with a price, and there is no such person. He is using himself merely as a means to another's favour, which is the second formula again, pointed inward. And Kant is careful, in the same pages, about the one confusion this invites. Humility before the law is a duty. No person has met the law, and every person knows it. Humility before other people is no duty at all.⁵⁵ Nobody has a claim on another's worth, and to act as if they did is to spend what was never for sale.

Now the three together, because what they share is the point. In each one a person treats himself as a means to a state. The suicide, to relief. The rusting talent, to ease. The lackey, to favour. And in each one the thing used is not the body or the mood. It is the person who could have set a different end, the one who runs the test. That is what Kant means by humanity in your own person. Not species membership. The part of a person that legislates, and that can be sold out by the part that wants. This is also why the attractive picture from the start, that morality is about how others are treated, is half of Kant and the weaker half. On his account the first person anyone can wrong is himself, and he wrongs himself the way he wrongs anyone. He acts as if a person were a thing with a price.

A duty to oneself, then, is no contradiction. The one who owes and the one who is owed are the same person regarded from two standpoints. Ending a life for relief, letting abilities rust, and crawling for favour are each the use of a person as a means to a state. The person used is the one using. The second formula covers them because it covers everyone, and everyone is someone. That was the formula turned on the one person who is always in the room. Turn it back outward, to everybody else in the room. The usual assumption is that it demands the one thing nobody can produce to order, which is feeling. It demands nothing of the kind. What it does demand is what a person owes the one on the other end, once the prohibition has been kept and nobody is being used.

The law cannot demand feeling, and Kant says why in one sentence. A duty to love is an absurdity. Love, he says, is a matter of feeling and not of willing. He cannot love because he wills to, and still less because he is told to.⁵⁶ So nothing owed to another person is a feeling, and the earlier verdict on warmth stands exactly where it was put. What is owed instead is something a person can do on the worst day of his life. And it is far harder to get out of than any feeling would be.

Start from where this leaves the decent person. The warm helper was told that his helping earns praise and encouragement and no moral esteem. That was a ruling about what makes an act worth something. It said nothing about what he is actually required to do about the people in his house, at his work, at his door. Kant wrote that down too, in his last book on ethics, and it is an argument rather than a list. It has two halves, and they pull against each other.

The first half is what the absurdity line clears the ground for. If the feeling cannot be commanded, then the thing the command reaches is the maxim. Love as a duty, Kant says, has to be thought as a maxim of benevolence, which has beneficence as its consequence.⁵⁷ With the vocabulary taken off, that says something plain. Make it your standing rule to want other people's lives to go well, and the helping follows from the rule. Then he states the duty flat. To be beneficent is every human being's duty. That means helping other human beings in need toward their happiness, according to your means, and without hoping for anything back.⁵⁸ His reason for it is one sentence long. Every human being who finds himself in need wishes that other people would help him.⁵⁹ You cannot wish that and legislate it away at the same time, which is the cold world again, run on your own future.

Notice exactly what has been asked for here. A feeling toward the person is not available on command, so it is not what the duty reaches. What the duty reaches is their ends, the things they are actually trying to do with their life. Those have to become ends of yours as well. That is the other side of a sentence from earlier. Humanity is not a person's wishes, it is their capacity to set ends for themselves at all. The prohibition took that capacity and said: do not act on them in a way that goes around it. The requirement takes the same capacity and says: what they set with it has to count in what you do. One of them is a fence and the other is a pull, and both are the same fact about a person, read twice.

Two men walk past the same doorway on the same evening. Both feel the small sick pull that the sight of someone sleeping in it produces. The first takes a different route the following night, one that misses the doorway, and the pull stops. The second keeps walking that way, and it does not stop, and some of what he does that winter comes out of it. Kant has a ruling on those two men, and it is the most easily misread thing in the book. Having compassion, he says, is not in itself a duty. Taking an active part in another person's fate is one.⁶⁰ And the feeling is one of the things that actually gets a person up and out. So he calls it an indirect duty to cultivate that feeling, and to use it as a means to the taking part.⁶¹ So it is a duty, he says, not to pass by the places where the poor are, but to seek them out. And it is a duty not to flee the sickrooms and the debtors' prisons in order to escape a fellow-feeling you cannot ward off.⁶² Hold that against the earlier verdict, because it can be heard as a reversal of it and it is nothing of the kind. Nobody is being ordered to feel. The feeling is still not what gives an act its worth, which is what was said before and is not being taken back. The feeling is a tool, and the duty is to keep the tool in working order rather than to blunt it on purpose. The first man is not being judged for the weakness of his fellow-feeling. He is being judged for the route.

Now the second half, and it pulls the other way. Beside the duty of love Kant sets a duty of respect, and what that duty asks for is not a feeling either. It is only a maxim, he says, limiting our own self-esteem by the dignity of humanity in another person. And it is really only negative, which is to say: do not exalt yourself above others.⁶³⁶⁴ He builds the pair as two forces working on the same body. By the principle of mutual love, people are directed to come constantly nearer to one another. By the respect they owe one another, they are directed to keep themselves at a distance.⁶⁵ Love moves you toward a person and respect stops you at a line. Every real difficulty in this part of his ethics is a difficulty about where the line runs.

Which is why helping is the dangerous act and not the safe one. Kant works it out on the plainest case there is. We recognise that we are obliged to be beneficent toward a poor man. But the favour, he says, contains a dependence of his welfare on my generosity, and that lowers him. So a second duty arrives inside the first. It is a duty to spare the man that humiliation, and to preserve for him his respect for himself. You do it by a bearing that presents what you gave as a thing that was simply owed, or as a small service of love.⁶⁶ The mechanism is worth reading slowly, because it is exact. The help is real, and the man is genuinely better off, and the transaction has still told him where he now stands. The giver who tells the story afterwards has taken back with one hand what he handed over with the other.

The same pressure produces the sentence that closes the argument. I cannot do good to anyone according to my ideas of happiness, Kant writes, but only according to his. Children and the badly disturbed are the exceptions he names on that page.⁶⁷ He is making a claim about who owns the standard. It is what respect does to beneficence once the two are held together. The managing version of help runs along the lines the helper would want for himself. It uses the other person's situation as material for an idea of how a life should go, and the idea is not theirs. The person helped ends up living inside somebody else's plan for them. Run that as a rule, which is the only way the test ever runs. A rule that others may settle what my life is for, while I can still say what it is for, is one nobody could agree to. The rule that others act for me when my capacity to choose has been narrowed to nothing is one anybody could. So the addict and the man who cannot be reached are not a hole in this. They are inside the second clause, and the clause closes again when the capacity comes back. And none of this obliges a man to serve whatever end another person names. What he owes is that their ends count in his, and how much he gives, and to whom, stays his.

The case where the two forces tear is the one everybody is actually living in. Friendship, considered in its perfection, is Kant's phrase for it, and he defines it as the union of two persons through equal mutual love and respect.⁶⁸ Equal is the word carrying the weight, and he knows it, because nobody can measure whether the two are in balance between two people. So in the next breath he takes the definition back. Friendship, he says, is a mere idea, unattainable in practice. It is also a duty set by reason to strive toward, and an honourable duty rather than a common one.⁶⁹ And under it he puts a rule that sounds cold until the day you have needed it. Even the best friends should not make themselves too familiar with each other.⁷⁰ Love says come nearer, and respect says there is a distance you do not cross. Both run on the same person at the same time, which is why the closest relationship is the hardest one to get right.

Then he names the weight, in his own voice, at the one place it is heaviest. It is also a great burden, he writes, to feel oneself chained to another's fate and laden with another's need.⁷¹ He is not conceding there that a life of duty is a load a person carries anyway. He is arguing that friendship cannot be built on need. A friendship held together by favours puts one of the two a step lower, and lays that weight on the other. But the weight is on the page, in his words, and it is the weight the standing complaint against his ethics is made of. That is the honest version of what is being asked of a person, and it is the only version worth answering.

The argument, as plain steps. One: a feeling cannot be commanded, so no feeling is ever owed to anybody. Two: what can be commanded is the maxim, so what is owed is that other people's ends count inside yours, with the helping as the consequence. Three: a second duty runs against the first, limiting your own self-esteem by the dignity of the person in front of you. Four: because the two run together, help that humiliates has broken the rule it thought it was obeying. So has help delivered on your own idea of a good life. Five: at the closest range both forces run at full strength. What comes out there is an ideal nobody reaches, and a weight Kant names out loud. So the law reaches into the distance between you and a friend, and into the way a coin is handed across. It tells a man he may take no credit for what he gave, and tells him in the same breath that he must give. A law that reaches this far is claiming an authority no ruler ever had. Who gave it. And why does it bind, when nobody is watching and nobody could ever enforce it.

Every morality before Kant, on his account, borrowed its authority from somewhere outside the person it was binding. From God, from the wish to be happy, from a feeling in the chest that some things are wrong. And every one of those, he says, can only ever say: do this, if you want that. A law which binds without an if has exactly one possible source, and that source is the will it binds.⁷² Kant's word for this is autonomy, and he calls it the supreme principle of morality.⁷³

The three borrowed authorities go in turn, and the same thing happens to each. First, happiness. Say the reason to keep promises is that a life of kept promises goes better. Then the rule is: keep promises, if you want your life to go well. And now the man from the false promise has an answer. His life is going badly, the promise will fix it, and by the rule's own logic he should break it. A morality built on happiness has no reply to the person for whom wrong pays. It can only say that it usually does not, which is true, and which he already knew.

Second, feeling. The reason not to cheat the child in the shop, on this view, is that it would feel wrong, that a decent person's stomach turns. Then the rule is: do not cheat, if your stomach turns. And the man whose stomach does not turn is released. Worse, the man whose stomach turns at the wrong things, at strangers, at people of another kind, is licensed. Feeling varies from person to person and from year to year. A law that varies that way is a report on the person's wiring. Kant does not deny that decent people feel these things. He denies that the feeling is what makes the thing wrong.

Third, and this is the one that matters most, God. On this view the reason to keep the law is that God commands it. This looks like the one authority that has no if, since God is not a mood or a bargain. But Kant asks a question that is hard to unask once it is heard. How does anyone know the command is good. A voice tells a man to do something. Before he obeys, he has to decide that the voice is God's and not some other thing. He decides that by asking whether what it commands is worthy of God. He is holding the command up against an idea of goodness he already has, and checking it. Kant says this in the plainest possible place. Even the Holy One of the Gospels, he writes, must first be compared with our ideal of moral perfection before we can recognise him as such.⁷⁴ Before anyone can say that this is the good man, he has to know already what good is. So the idea of goodness was not delivered by the command. It was the thing he used to accept the command. A believer can answer that this shows where the idea of goodness comes from, and not where the obligation does. The reply on Kant's behalf is that a command nobody could recognise as good could bind only as a threat, never as a duty. And a threat is an if. The authority was inside all along.

That is the argument, and its skeleton with the details removed is short. Any outside authority, to bind a person, has to be recognised by him as an authority, and recognising it means judging it. The standard he judges it by is not itself outside. So there is no outside authority that does not, at the last step, run through a person's own judgment of what is good. The law that judges every command was never given from outside. It is the one already in use to accept or refuse everything else.⁷⁵

Kant's name for the alternative is heteronomy, which means a law from elsewhere, and he puts every earlier ethics under it. The mark of heteronomy is the if. If you want to be happy, if you want to feel clean, if you want to please God. Every one of those makes the law depend on a wish. A law that depends on a wish is only as strong as the wish, and wishes change. A person under heteronomy is always one changed wish away from being released. Under autonomy, there is no wish to change. The law is what a person's own reason says when it asks what a rule for everyone must be, and it serves nothing he wants.

Here is where it becomes hard to see. A law I give myself sounds like no law at all. If I make it, I can unmake it. A rule that can be cancelled by deciding to is a preference with good posture. And a whole tradition of critics has said that Kant's self-legislation is either an outside law in disguise or no law.⁷⁶ The answer turns on what the self in self-legislation is. It is the part of a person that runs the test from earlier, the part that asks whether a rule could hold for everyone. That part wants nothing for itself, and nobody chose to have it, and nobody can switch it off. When a person tries to make an exception for himself, it is the thing that notices. So the law is his in the sense that it is his own reason speaking. It is not his in the sense that he could vote it down. In one respect it is like the answer to a sum: his, produced by him, and not up to him. In the other respect it is nothing like a sum. A sum is a fact you find in the world, and a fact binds nobody. The law is not found anywhere. It is the form his own willing takes whenever it asks what could hold for everyone. It binds because it is you, willing, and not you, wishing. Kant will later split the word will in two: the will that gives the law, and the choice that takes it up or refuses it.⁷⁷ The law's source is the first, what it binds is the second, and both are the same person.

Kant draws the picture of what this law is for. The world he asks for is one in which every person is at the same time a maker of the law and bound by it. Every one of them asks what rules could hold for all, and every one of them lives by the answer. Nobody is only a subject. He calls it a kingdom of ends. In it nobody is used merely as a means, since everyone is at once a lawgiver and someone the law protects.⁷⁸ No actual society looks like this. It is what a maxim is being tested against when the test asks whether it could hold for everyone. The question is whether it could be a law in that kingdom.

And in that kingdom, Kant says, everything has either a price or a dignity. Whatever has a price can be replaced by something else that is worth the same. A tool, a service, a skill, a pleasure. Whatever is above all price, and so admits of no equivalent, has dignity.⁷⁹ The only thing on earth with dignity in that sense is a being that can give itself the law: a person. That is the ground under the second formula, said in one line. A person cannot be used merely as a means because a person is not the kind of thing that has an equivalent. There is nothing that could be given in exchange for them that would make the trade fair.

Happiness, feeling and command all bind with an if, and an if is a door the wrongdoer walks out through. Even a command from God has to pass a person's own judgment of what is good before he can accept it as God's. So that judgment was the authority all along. The law that binds without an if is the one his own reason gives when it asks what could hold for everyone. Kant's name for that is autonomy. It is his because it is his own willing, and not up to him because it is willing and not wishing. But that has just made a very large promise. It says there is a part of a person that gives the law and can act on it against every wish he has. Nothing said so far shows there is. Everything a person does is caused, like everything else in the world, and a caused act is not a given law. A law a person gives himself is possible only if he can give it. Whether he can is the question Kant himself thought was the hardest of all.

Nobody can prove that they are free. Nobody can act without presupposing that they are. And that presupposition, Kant says, is all that morality needs. Those are three claims, and each is harder than it sounds. The first one Kant learned the hard way, by trying the proof and failing, and saying so. The second is the centre of his whole account of what a person is. The third is what lets the argument stand without the proof.

First, the problem, felt properly. Everything that happens in the world has a cause. A stone falls because it was dropped, and a leaf turns because the light moved. A person is a thing in the world. A body, a brain, the mood he woke up in, the thing he is about to say. All of it has causes running back through his childhood into the weather of centuries before he was born. If that is the whole story, then the second shopkeeper's because was a cause like any other. His honesty was produced in him by the chain, the way the leaf's turning was produced in the leaf. And then the law he gave himself was not given. It was delivered, by everything that made him, and he was the place it arrived. On that picture, there is nothing to praise and nothing to blame. Everything said so far was a description of how some machines happen to run.

Kant does not deny one word of the causal story. Most people expect him to. He does not say there is a gap in the chain where a free choice slips in. He holds that everything in the world of experience has a cause and could in principle be predicted. That includes every human action, as an event. What he says instead uses an idea from his account of knowledge, and the ethics needs it. The world of experience is the world as it appears to a being built the way people are. Space, time, cause and effect are the form experience takes. They are how things show up to a mind, and not necessarily how things are apart from being shown. The red of an apple is not in the apple and not in the light; it is made, every time, by the eye that meets it. So there are two ways of regarding anything, including a person. As it appears to a being built the way people are, in time, under causes. And as it is in itself, about which almost nothing can be known. Now Kant's move. When a person regards himself as a thing in the world, an object among objects, he falls under causes, and there is no freedom. But when he regards himself as the one deciding what to do, he is not looking at himself from outside. He is standing in a second place, the place from which the question what should I do is asked. And from that place, the things that would look like causes from outside look like reasons. From outside, the shopkeeper's honesty was produced by his upbringing. From inside, when he decides, his upbringing is not what he consults. He consults whether it is right, and the same fact, seen from the two places, is a cause and a reason. Kant's phrase is that a rational being has two standpoints from which it can regard itself. One under the laws of nature, one under the law of reason.⁸⁰ The duty to the self already used them, when the one who owes and the one who is owed turned out to be one person.

This is hard, and the case that makes it concrete is Kant's own. A man tells you that his lust is irresistible. When the object and the opportunity are in front of him, he says, he cannot hold back. Kant's reply is to build a gallows outside the house where the opportunity waits. The man would be hanged the moment he had taken it. Would he still find his lust irresistible.⁸¹ The man would not take long to answer. So the irresistible turned out to be resistible, given a large enough reason on the other side. Now the second half, and this is the one the argument turns on. The same man is ordered by his ruler to give false testimony against an honest person, whom the ruler wants destroyed. The alternative is that same gallows, at once, for him. Would he refuse.⁸² Kant does not claim to know what the man would do, and neither does the man. But Kant says the man will admit without hesitation that it would be possible for him to refuse. He cannot say he will, but he can say he could. And this is Kant's sentence, the one the whole account of freedom rests on. He judges that he can do a certain thing because he is conscious that he ought. And he recognises that he is free, a fact which but for the moral law he would never have known.⁸³

Look at what just happened, because it is the reverse of what anyone expects. The natural order is: first show that I am free, then it makes sense to say I ought. Kant runs it backwards. The ought comes first, as something a person finds already in him, undeniable, when the gallows and the lie are put side by side. He did not reason his way to it. He noticed it, the way a person notices pain. And once it is noticed, the freedom follows from it, because a demand that could not possibly be met would not be a demand. Nobody is commanded to fly. If the ought is real, the can is real. You can, because you ought. Kant's own words for the relation come to this: freedom makes the law possible, and the law is how you find out you are free.⁸⁴

Now a concession, once. In his first book on ethics, Kant tried to do better than this. He tried to prove that any being that reasons must be free, from the mere fact that it reasons. Then he would get the moral law out of the freedom. He himself suspected a circle in it.⁸⁵ You assume you are free in order to get the law, and you get the law in order to show you are free. He offered the two standpoints as the way out, and whether they are is disputed. The reading here sides with the readers who say they are not, as a proof.⁸⁶ The two standpoints stay. What fails is the step from standing in the second place to being bound by the law, which is the very thing that needed proving. Kant ended that book by admitting what it could not explain: how pure reason can move a person at all. We do not comprehend the necessity of the moral law, he wrote, but we do comprehend its incomprehensibility. And that, he said, is all that can fairly be asked.⁸⁷ In the next book he did not repeat the attempt. He called the consciousness of the law a fact of reason. A thing you cannot reason your way to from anything earlier, that forces itself on you all the same. That is not a proof, and Kant knew the difference.⁸⁸ It is a report.

So what does that give, and what does it not give. It does not give a demonstration that anyone is free in the way a physicist would accept. Kant says that a being that cannot act except under the idea of freedom is, for that very reason, really free in a practical respect.⁸⁹ The qualifier is his. Practically free. Free in the only sense that matters for the one deciding. What it gives is this. There is no standpoint from which a person can deliberate and at the same time regard his deliberation as nothing but the chain running. Try deciding whether to keep the deposit while holding that whatever you conclude was already fixed, and that your deciding is a noise the fixing makes. The decision does not happen. A determinist can answer that deciding needs only not knowing the result, not believing it open. Kant grants that from outside an act could be predicted. His point is narrower. The one deciding cannot take that prediction as his reason, because taking it as a reason is already deciding. Nobody can occupy both standpoints in the same act. And the moral law arrives in the place a person occupies when he acts.

The modern version of it arrived with an experiment attached. In the early 1980s a neuroscientist wired volunteers up and asked them to flex a wrist whenever they felt like it. They were to note the position of a moving dot at the instant they first felt the urge. The brain's preparation for the movement began, on average, about half a second before the muscle moved. The felt urge came about a fifth of a second before the movement. So the brain had started the act some three or four hundred milliseconds before the person knew they were going to do it.⁹⁰ This has been told for forty years as the experiment that ended free will. The decision was made before the person decided, and the self that decides is a spectator arriving late.

Three things about this, and the first is that Kant already agreed with it. He holds that everything in the world of appearance is caused, and a wrist flex is in the world of appearance. A brain readying a movement before awareness reports it is exactly what his account predicts about anything regarded from outside. The finding does not touch the second standpoint, because it was made entirely from the first. The second thing is that the experimenter himself did not think it ended free will. He found that people could veto a movement they had prepared, in the last tenth or two of a second. He wrote that free will would not initiate a voluntary act but could control whether the act takes place.⁹¹ Whether that rescue works is disputed, and he conceded he was proposing it rather than showing it.⁹² The third thing is the one that matters, and it came thirty years later. A team re-examined what the rising brain signal actually was. Their model was that the brain's background activity drifts up and down at random all the time. A movement happens when the drift happens to cross a threshold. Line up hundreds of movements by the moment of the movement and average them. The random drifts before each one will look, in the average, like a smooth ramp that began long before. On their account the conclusion that the brain decided early is unfounded, because before about two hundred milliseconds no decision had been made at all.⁹³ That is a model, and it is contested, and later work that predicts a choice seconds ahead from brain scans is not answered by it. None of it vindicates Kant, and none of it needs to. His claim was never that the scan would find a gap.

So Kant grants that every act, seen from outside, has a cause, and does not ask for a gap. He holds that a person acting stands in a second place, where causes are reasons, and can decide nothing except from that place. The moral law is found there, not proved there. The gallows case shows how it is found: the man learns he could refuse from the fact that he ought to. His attempt to prove freedom from reason alone does not go through. In his next book he replaced it with the report of a fact. And the experiment that was supposed to have ended the matter measures the first standpoint and says nothing about the second. So freedom is not something a person knows about himself the way he knows his height. It is something he presupposes every time he decides, and can never step outside to check. And the warm shopkeeper has been waiting since the beginning to ask what that does to him. If nobody can step outside to check, nobody can ever know that he used it.

Nobody can ever know, of any act they have done, that they did it from duty. Not the biggest sacrifice of a life, and not the smallest kindness. Kant says this in plain words, and it follows from everything before it. It means that moral worth is real and invisible, and that credit for being good is not available to anyone. Most people hear it as an insult, and it is the opposite.

The warm helper on his good days could not be seen to have worth, because the sun never went in. That was about what others could see. Now turn the question inward. Can he. He helps someone, and it feels good, and he asks himself honestly whether he would have done it if it had not felt good. He would like to say yes. But he has only ever done it when it felt good, so he is guessing about a day that never came. And the guess is being made by the same person whose warmth is in question, who would very much like the answer to be yes. Kant's sentence is that it is absolutely impossible to make out by experience, with complete certainty, a single case of it. A single case in which an action that agreed with duty rested solely on moral grounds.⁹⁴ Not one, with certainty, ever. He goes on: even the strictest self-examination can never get entirely behind the hidden springs of what we do.⁹⁵ And when we look closely at our finest acts, he says, we everywhere come upon the dear self. Always jutting out, and always what they had in view.⁹⁶⁹⁷

The dear self. That phrase is the centre of the matter, and it should be taken exactly. Kant is not claiming that everyone is selfish. When a person inspects his own motives, there is always a candidate motive that is about him. It stands right next to the one that is about the duty. And he cannot tell from inside which one was doing the pulling. He gave the money, and he also wanted to be the kind of person who gives. He told the truth, and he also could not have borne the look on her face if he had lied. He kept the promise, and he also knew the other man would find out. In every case the duty was there, and the self was there beside it, and the act came out the same either way. Which one moved him is a fact. It is just not a fact he has access to.

This is not pessimism about people, and the difference matters. A pessimist says: a person's motives are worse than they look. Kant says: a person's motives are hidden from him, in both directions. The warm helper cannot prove he acted from duty. He also cannot prove he did not. Late in his life Kant wrote that the depths of the human heart are unfathomable.⁹⁸ And in the same pages, that the first command of all duties to yourself is: know yourself.⁹⁹ Not because you can, all the way down. Because the attempt is the only honest posture toward a heart you cannot see the bottom of.

There is a finding from another field that comes close to testing this, and it is the kind of thing that could have proved Kant wrong. If people could read their own reasons, then the reasons they give for a moral judgment would be the reasons that produced it. A psychologist reviewed a decade of studies that put that to the test, on people's judgments of what others had done. His conclusion, in his own summary, is that moral reasoning is rarely the direct cause of moral judgment. The judgment comes first, fast, and the reasoning arrives afterward.¹⁰⁰ It works, in his comparison, like a lawyer defending a client rather than a judge weighing a case.¹⁰¹ And he named the illusion this produces the wag-the-dog illusion: the reasoning believes it did the deciding.¹⁰² Two limits on that, both his. It is about judging, not about acting, so it reaches Kant's claim about motives only by analogy. And it describes how judgments are made, not how they ought to be, so it cannot show there is no duty.¹⁰³ What it shows is narrower and, for Kant, exactly right. The story a person tells himself about why he judged as he did is composed after the fact, by an advocate. That is the dear self, measured.

So the credit is gone, and this is what that does. It takes away the question that the person who tries to be good is quietly tormented by. Did I do it for the right reason. Kant's answer is that nobody can know, not even the person asking, and that the question was never his to settle. That is no licence. He is still bound, and the law did not go anywhere. But the audit is over, because the auditor cannot see the books. What is left is the only thing that was ever in a person's power. Act, from a rule that could be willed for everyone, and leave the verdict where it lies, which is nowhere anyone can reach.

One thing this does not permit, and it should be said before the relief settles. The audit closed because nobody can see the books, and not because the books stopped mattering. Know yourself was the first command, and it still stands after the verdict has been withdrawn. What is forbidden is not failing to see the bottom of your own heart, which is guaranteed. It is the manoeuvre a person makes instead, which is to tell himself the story that settles the question in his favour. That story is the dear self's, and it always has one ready. The composing is not in anybody's power to stop, and the lawyer will always file something. What a man can decline to do is believe it.

The strongest objection to all of this comes from the very thing that made the good will attractive at the start. The good will was the one thing luck could not touch, and the river could not reach it. But whether a person has a good will at all looks like luck too. The second shopkeeper was raised by honest people in a town where honesty was expected. The warm helper was born warm. The man in the fraud was born cold, or made cold, by things done to him before he could choose anything. If temperament is luck, and upbringing is luck, and the will is shaped by both, then Kant's fortress against luck has luck inside the walls. He kept luck out of the outcome and let it in through the character.

Kant's answer is the most demanding thing in his ethics. The will, for Kant, is nothing like a trait. A trait is a thing a person has, the way he has a temper or a talent, installed in him and listable. The will is what a person does now, in this act, from the standpoint where causes are reasons. Temperament is on the first list, with courage and calm. It is a gift, and it amplifies, and it decides nothing. The cold man and the warm man both stand, right now, before the same question, with the same law in them. The coldness is the weather one of them has to act in. Kant's picture of what a person's goodness or badness consists in is an ordering, and not a larger share of good impulses. Everyone has self-love and everyone has the law. The only question is which of the two is placed above the other when a person decides what to do. A bad person, on his account, is one who has put self-love first and the law second. The law is then obeyed only when it costs nothing, and the impulses may be no worse than anyone's. That ordering is a choice, and no luck makes it for him. Kant keeps two words apart here, and they are needed. The will that gives the law is neither free nor unfree, because it is the law, the way a compass does not choose but points.¹⁰⁴ The choice that takes the law up, or puts self-love above it, is the hand that follows the compass or not. Only a hand can be free.¹⁰⁵ It is where both goodness and evil live.

Which is why Kant says something about change that nobody expects from him. A man who talks this way about duty might be expected to counsel gradual improvement, habit by habit, like a person learning an instrument. He says the opposite. As long as the order underneath is wrong, as long as self-love sits above the law, gradual reform changes nothing that matters. It produces a person whose acts are more lawful and whose ordering is exactly what it was.¹⁰⁶ What is needed is what he calls a revolution in the disposition. A single decision, unchangeable in Kant's word, that reverses the order: the law first, self-love second, from now on. He compares it to being born again, and he means the comparison.¹⁰⁷ After it, the gradual work begins, and it is needed.¹⁰⁸ But the work cannot substitute for the reversal, because the work is done by the person who has not yet reversed. That is what answers the luck objection. The reversal is not a trait either. It is available to the cold man exactly as it is to the warm one, because it is not made out of warmth. It is made from the second standpoint, the one a person decides from. So luck reaches temperament and history, and Kant concedes every bit of that. What it does not reach is the one act that the law itself calls for, the placing of the law first. No history can do that for a person, and none can do it to him.

What is hard about this answer needs saying, because the objection does not simply lose. A person whose whole life has taught them that the world is cruel has a longer distance to the reversal than a person who was loved. Kant does not deny that, and nothing in the argument makes the distances equal. That the capacity for the reversal was given is true, and it is luck. What is not luck is whether it is used. And Kant admits that nobody can be sure, even of themselves, that the reversal has happened, which is the hidden motive again. What the argument denies is that the distance is the verdict. It denies that anyone is finished, and it denies that anyone is banked. The good person cannot take credit, because they cannot see whether they are good. The bad person cannot be written off, because the reversal is never behind them. That is what it means for the will to be outside luck: not that a history was fair, but that it is not the last word.

So: nobody can see their own motives with certainty, ever, and Kant says so. The science of how people explain their own judgments agrees with him about the mechanism while saying nothing about the law. So moral worth is real and unseen, and credit is not something anyone can claim. That ends the audit, not the duty. Whether a person has a good will is not a trait luck installed but an ordering he chooses. Once as a reversal, and then shown, never banked, in every act after. And now the case that has been waiting at the door since the very first minute. If no exception is ever yours to make, the law comes first and self-love second, and nothing can overrule it. Then the man with the knife is standing on your step, and he has asked his question.

Kant's answer to the man with the knife is wrong. His argument for it is far stronger than the joke people tell about it. And it fails at the case where he would not let the state of the world change what the law asks. Three things follow: the argument as Kant made it, the exact point of failure, and the right answer built from his own tools.

First, how it started, because the setup matters. A French political writer, in 1797, was arguing against people who reasoned from pure principles down to politics without regard for consequences. He wrote that the principle that telling the truth is a duty, taken absolutely and alone, would make any society impossible. And he said that a German philosopher had gone further still. If assassins asked you whether the friend they were hunting had taken refuge in your house, it would be a crime to lie to them.¹⁰⁹ He did not name the philosopher. Kant read it, and answered in print, and opened his answer by admitting that he had said this somewhere, though he could no longer recall where.¹¹⁰ He then defended it. He was an old man. Nobody made him.

Now his argument, which almost nobody gives whole. Kant begins by narrowing what he is talking about. A lie, for this argument, is an intentionally untrue declaration to another person. And he says he is treating it as a question of right, which means the law a court can enforce, and that word will matter. A declaration, in that vocabulary, is a statement the other person is entitled to rely on.¹¹¹ Not every false sentence is one of those, on any sensible reading of the word. A line in a play is not a declaration, and neither is a bluff at cards in a game everyone agreed to play. A statement made to someone who asks a question at the door, and who takes the answer as an answer, is. Then the claim. Such a lie, Kant says, always wrongs someone. Even when it wrongs no particular person it wrongs humanity in general, because it makes the source of right unusable.¹¹² Take that slowly. Every contract, every promise, every court, every agreement between people rests on declarations being taken as true. Kant writes that truthfulness is a duty that must be regarded as the basis of all duties founded on contract.¹¹³ Not one duty among others. The one under all of them. A lie, even a kind one, is a withdrawal from that basis. A rule that permits it is a rule that says the basis holds except when I judge otherwise. Which is the false promise again, spending trust that other people built.

And then Kant does something people forget, which is to run the case forward. You lie. You tell the man your friend is not in the house. But unknown to you, your friend has slipped out the back, and in the street he meets the murderer, and dies. Kant says that in that case you can rightly be accused as the author of his death. Whereas if you had told the truth, and the murderer had gone in, and the friend had died, you would have done nothing wrong, whatever followed.¹¹⁴ The reasoning under that is this. The truthful man does what the law requires, and what the world then does with it is the world's doing. The liar has taken the outcome into his own hands. He has decided that he knows how this will go, and staked another person's life on his guess. And the moment it goes wrong, the wrongness is his, because he was the one who left the rule to manage the world himself. So the liar has not only broken a rule. He has made himself responsible for everything the lie touches, and he cannot see everything the lie touches. That, Kant concludes, is why truthfulness is a sacred and unconditional command of reason, limited by no expediency.¹¹⁵

Before the failure, what it gets right, which is a great deal. It is right that the comfortable answer, lie when it helps, is not a rule. It is a permission to be the judge in your own case. Every person who has ever been lied to for their own good knows what that permission costs. It is right that a world of well-meant lies is not a nicer world. It is a world where every answer has to be checked against what the speaker thought was good for the hearer. Which is to say a world with no answers. And it is right that the liar takes on the outcome. The friend who slips out the back stands for every lie told to manage a situation the liar cannot see the whole of. Kant is being consistent, all the way down, and consistency is the thing that made the rest of his morality worth anything.

The failure, and the argument for it comes from inside Kant's own two formulas, not from outside. First the honest maxim: when someone is trying to use my answer to commit a murder, I give a false answer, to stop the murder. Nothing in it says when it suits me. The Tuesday trick is waiting, so the difference comes first. Tuesday was a feature that steered nothing and made the lender vanish from the maxim. The murder is the feature that is actually steering, and it puts the man at the door into the maxim with what he is doing named. That is what the second formula demanded of every maxim: the person on the other end is in it. Now run the universal law test. A world in which everyone lies to murderers at the door. Does the lie still work in that world. It does, and this is the point a careful reader of Kant made in 1986, within one limit. The murderer, to ask his question at all, has to be pretending to be an ordinary caller with an ordinary question. He cannot announce himself as a murderer and expect an answer. So he is already relying on your not knowing what he is.¹¹⁶ Which means that in the world where everyone lies to known murderers, he still expects the truth, because he does not think you know. The lie does not cancel itself. The rule passes the first test. The false promise failed because the promise disappears in the world of false promises. The lie to the murderer survives in the world of lies to murderers, because the murderer's own deception keeps the door open. The limit is this. It holds only while he does not know that you know. A murderer who knows you have seen through him expects a lie, and there the maxim fails the way the false promise failed. But there nothing you say is a declaration any more, because he is relying on nothing you say. No lie, in Kant's own sense, can be told to him, and you may say what you like. So the permission holds on both sides of the limit, by two routes.

Now the second formula, because it looks as if it convicts. A lie works only if the other cannot see it, so he cannot agree to this lie, and that seems to make him merely a means. But the test was never whether he could agree to the act. It was whether he could agree to the rule, the way the arrested thief could agree to the rule that thieves are arrested. A rule that false answers go to those using the answer to kill is one any person could agree to. The murderer included, as a person and not as a murderer. He is not being used when his question is turned aside. The householder, in turning it aside, refuses to be used, and the formula was written to let him. So both formulas, run on the honest maxim, permit the lie. Why, then, did Kant get the opposite answer. Because in that essay he ran neither formula. He argued from right, from what a court and a contract need. A court cannot see who is at your door. So a rule a court relies on cannot carry a clause about him, or the court cannot run. On that ground the answer is that truthfulness makes no distinction between persons, and on that ground it is right. Then, in his ethics, he gave the same answer without argument and with no murderer in sight. The case was argued only where the answer is strongest, and never where a person of good will actually stands.

Now the ruling, in the terms of the kingdom of ends, the world Kant said the formulas were for. The murderer has stepped out of that world. Not out of humanity. He still has dignity, nobody may torture him, and nobody may lie to him for fun. But the specific thing you owed him, a true declaration he could rely on, was owed inside the everyday practice of people asking and answering. He is at that moment using that practice as a weapon. The same careful reader put it as a gap. What Kant's ethics needs, and what he did not write, is a set of principles for dealing with someone who has left the order.¹¹⁷ And she fills it with two levels rather than one exception.¹¹⁸ On the first, the householder still owes the man with the knife everything a person is owed. He may not torture him, or hurt him for sport. On the second, he is dealing with a man who has turned asking and answering into a weapon. There the world the formulas picture is a thing to work back toward, and not a room he is standing in. Evil circumstances is not a judgment he gets to pass on the night. It is a fact his honest maxim already names, that this man is using the answer to kill, and anyone at the door can check it. That is what the second formula put into the maxim, and what the universal law test then let through. No appeal to consequences enters anywhere in it. The lie is permitted, not because the friend matters more than the rule, but because the rule, honestly stated, allows it.

What that does and does not concede. It does not concede that a good reason is an exception. The person who lies to the murderer because a life is worth more than a rule has made himself the judge. Kant is right that this cannot be the ground, because the same judge will find other cases. The exception has to come from the rule, or it is not an exception, it is the rule's end. What is conceded is that Kant, in this one essay, argued the case on the ground of right and let that settle it. So the wrong to trust in general outweighed a concrete person about to be killed by the very man whose trust was being protected. That is the failure, and it is not softness toward rules. It is a rule of the court applied where only a person of good will was standing.

One more thing about the text, because a verdict has been passed on it and it should be allowed to answer. Kant does not say, anywhere in that essay, that the friend's life does not matter. He says the liar becomes responsible for the death. And in his last book on ethics he says the lie is the greatest violation of a person's duty to himself. In lying, a person throws away and, as it were, annihilates his own dignity.¹¹⁹ So the thing Kant is protecting at the door is not only the murderer's claim to the truth and not only the institution of trust. It is the liar. He thinks the person who lies has done something to himself that no rescue can undo. The ruling here does not deny that lying costs the liar something. It denies that the cost is the same when the man you are lying to has come to kill. Kant's text does answer that: truthfulness makes no distinction between persons.¹²⁰ And that is the failure in one line, because it treats the man with the knife as a person asking.

Kant's argument, then, is that a lie is a withdrawal from the basis of all trust, and that the liar takes the outcome onto himself. Both of those are true. His conclusion is wrong, because both his formulas, run honestly on the actual maxim, permit the lie. The right answer on Kant's own terms is an exception that comes from the rule, not from a judgment of the moment. And Kant missed it because he argued from the law of the court, where who is at the door cannot enter. He never argued it anywhere else. That refusal raises a question the whole account has so far walked past. Kant's rule allowed no exception. But Kant himself wrote that most duties leave you room, and that the lie is a different kind of wrong from a theft.¹²¹ How much the law asks, and where it stops asking, is what has to be settled next.

Kant's law allows no exception, and most of what it asks leaves room. Both of those are true at once. The second makes the first bearable, and the first makes the second honest. There are two kinds of duty, and there are two kinds of law, and the murderer at the door sat on the seam between them. Getting the seam right is what stops Kant's ethics from being the machine its critics describe.

The two kinds of duty were named when the false promise and the cold world failed in two different ways, the perfect and the imperfect. A perfect duty, Kant says in a footnote, is one that admits no exception in favour of inclination.¹²² Do not make the false promise, do not keep the deposit, do not destroy yourself. These say exactly what not to do, and there is no room in them. An imperfect duty is the other kind. Help people. Develop what you could do. These say what to aim at and not how, or how much, or when, or for whom. And this is the sentence that changes what the whole law feels like. A wide duty, which is Kant's other word for an imperfect one, is not a permission to make exceptions to the maxim of your actions. It is a permission to limit one maxim of duty by another. His example is general love of your neighbour limited by love of your parents.¹²³ A person is bound to help, and not bound to help everyone equally. That is one duty giving another its shape, and the room is inside the law.

Take that back to the man at the door, because it is the first thing that makes his case different from the ordinary hard case. The duty not to lie is perfect. The duty to protect your friend is imperfect. Kant says two duties cannot truly conflict, and that where they seem to, the stronger ground of obligation wins.¹²⁴ A person who says, I lied because I owed my friend more than I owed the truth, has not shown a stronger ground. He has felt one, and called it a ruling. The answer given at the door did not do that. It found that the perfect duty, honestly stated, was never about the man with the knife. That is the only kind of answer Kant's structure allows.

The second distinction is the one almost nobody reading the lying essay knows is there. Two men pay their taxes in full. One pays because the fine for not paying is worse. The other pays because it is owed. The tax office cannot tell them apart and does not care. That is the first kind of law, which Kant calls the law of right: the coercible law, what a court can make you do. Only the second man has done what ethics asks, and that is the second kind, where the duty itself has to be the motive. Kant's words for the two are legality and morality, and everything said about the good will was about the second.¹²⁵ The coercible law, he writes, comes with an authorisation to coerce.¹²⁶ A person can be made to keep a contract, and cannot be made to keep it from duty. And a line that shows how far the two come apart. Another person is free, on the coercible law, even if in my heart I would gladly trample his freedom. All that is asked is that my outward act leaves it alone. The wish is ethics' business.¹²⁷ The coercible law asks only that I not act on it.

That split answers something the no-credit claim leaves standing. If nobody can see a motive, then nobody can be blamed either, and a morality that cannot blame is doing no work. Kant's answer is in what a court is allowed to do with a man. Judicial punishment, he writes, can never be inflicted merely as a means to promote some other good, for the criminal himself or for society. It must in every case be imposed on him only because he has committed a crime.¹²⁸ He must first be found punishable, before any thought is given to drawing some use out of the punishment.¹²⁹ The penal law, Kant says, is a categorical imperative.¹³⁰ Suppose a college of physicians proposed taking a condemned man for dangerous experiments, in exchange for his life. Kant gives them one line. Justice ceases to be justice if it can be bought for any price whatsoever.¹³¹ That is the second formula run in public, by the test already laid down. The punished criminal is never treated merely as a means, and the man punished to frighten other people is. And the asymmetry the whole account needed falls out of it. The deed is out where anybody can see it, with the intention in it, and a court grades both. What no court and no man can see is the incentive behind the intention. Whether duty moved him, or the wish to come off well, is the part that stays dark. So blame reaches the deed a man meant to do, and credit never reaches the will underneath. Nobody is excused by the part nobody can see.

This is why it matters for the door. The essay on lying is a text about the coercible law, and Kant says so in it. What the split adds is the question it leaves standing. Legality asks what a court can require of anyone at any door, and the essay answers it. Morality asks what this man, with that man in front of him, does from duty. The essay never puts that question, and the two answers were never going to be the same, because only the second can see who is asking. In his last book on ethics Kant did separate the two kinds of lie. There the lie is a wrong to yourself that need harm no one else.¹³² And he repeated the answer there without the case.

And in that last book, on the page where he discusses lying, Kant did something that the caricature of him cannot survive. He asked the everyday questions and left most of them open. Can an untruth from mere politeness, the your most obedient servant people signed their letters with, be counted a lie, when nobody is deceived by it.¹³³ An author asks a reader, how do you like my work. The reader could answer with a joke, but who always has his wit to hand, and the slightest hesitation is already an insult. So may the reader say what he does not think.¹³⁴ Kant asks, and does not rule. The rigid man of the joke would have ruled. He asked one more, and this one he answered. A householder tells his servant to say he is not at home if a certain person calls. The servant does, and the master slips out and commits a serious crime that the guard sent to arrest him would otherwise have prevented. On whom does the guilt fall.¹³⁵ On the master, plainly. And also, Kant says, on the servant, who violated a duty to himself by a lie. Its consequences are now imputed to him by his own conscience. Read that answer beside the murderer. The servant lied, at his master's order, and a crime followed that he could not have foreseen, and Kant gives the guilt to him too. It is the same shape as the friend who slips out the back, in a smaller room. Kant did not soften with age. What he did was admit, in the same paragraph, that the polite untruth and the reader's kind word were questions he could not close.

The last piece says how much the law asks of a person's money and time. That is where most people feel it. Beneficence is a duty, and a wide one. How far should you push it, Kant asks, and he answers in one sentence. Not so far that you would in the end need the beneficence of others yourself.¹³⁶ That is the whole of the ceiling, and it sits lower than the saints put it. Nobody is required to give until he is the next man at the door.

Then, on the same page, he says a thing nobody expects from the philosopher of duty. The capacity to do good depends on wealth, and wealth, he writes, is for the most part a result of injustice. Governments favour some people, and the inequality that follows is what makes charity necessary in the first place.¹³⁷ So the rich man's giving, Kant says, is not the generosity he takes it for. It is closer to a repayment.

So what does the law ask, taken whole. It forbids a small number of things absolutely. The false promise, the lie to anyone asking as a person, the contempt for a person as a thing, including yourself. It commands a larger number of things widely: help, develop, be grateful, do not exalt yourself above others. In the second class there is room, and the room is not a permission to skip. It is the space in which one duty shapes another, and the space is where a life is actually lived. And in the first class, the absolute one, the rule is narrower than its reputation. It reaches only as far as an honestly stated maxim reaches, and the everyday cases are ones Kant himself left open.

There is one more thing to say about how the law is heard, because Kant put it beside the hard cases and it belongs with them. He describes conscience as an inner court. The power that watches over the laws in a person, he says, is not something that person makes for himself. It is built into his being. It follows him like his shadow when he thinks to escape.¹³⁸ In utmost depravity a person may bring himself to pay it no heed, but he cannot avoid hearing it. And he adds a strange clause: an erring conscience is an absurdity.¹³⁹ A person can be wrong about what his duty is, and that is a mistake of judgment, and people make it. But he cannot be wrong about whether he has compared his act to what he took his duty to be.¹⁴⁰ That comparison is the only thing conscience does, and it either happened or it did not. Which is why, in the servant's case, Kant gives the guilt to the servant's own conscience and not to a court. No court could see it. The servant can.

So duties come in two kinds, and the wide ones leave room, and the room is one duty shaping another and never an exception. A perfect duty is never outweighed by a feeling, so the friend never outweighs the truth. Law comes in two kinds, and the lying essay argued the coercible kind and left the ethical kind unargued. And Kant, in his last word on lying, left the polite untruth open and gave the servant's guilt to his conscience. All of which leaves the person who has done every bit of this standing exactly where they started, with nothing to show for it. The law asked, they answered, and the world did not notice. Kant's answer to what virtue gets you is the least comforting thing he wrote.

Virtue does not make a person happy, and Kant says so without flinching. The world does not pay out in proportion to worth, and nothing in the world could. What he offers instead is not a reward. It is a hope a person is entitled to, and the ethics stands whether or not it is taken. Kant himself wrote otherwise more than once, and that has to be faced.

The ache has been building since the jewel that shines with no result. Through the credit nobody can claim, to the friend Kant's answer did not save. A person who has followed this far and is honest will ask the question the whole tradition before Kant asked. If being good gets me nothing, why be good. Kant's first answer is the one already given, and it is not the one people want. The law does not bind with an if, and a reward is an if. So there cannot be a reward attached to the law without the law ceasing to be a law. That is the account itself, and no gap in it.

But Kant does not stop there, and what he says next is stranger than anything in the ethics so far. He agrees with the questioner about the ache. Virtue, for Kant, means being worthy of happiness. He says it is the supreme good, the condition of everything else being worth wanting. But it is not the whole good.¹⁴¹ A world in which a person deserves happiness, needs it, and does not have it is a world out of order.¹⁴² A perfectly rational will with the power to arrange things would not choose it. So the complete good, the thing reason cannot help wanting, is virtue together with happiness, distributed exactly in proportion to worth.¹⁴³ And then he says the thing that everyone before him had been too polite to say straight. Nature does not do this. Nothing in nature does this. The honest die poor and the fraud sleeps well. There is no mechanism anywhere in the causal order that will ever line the two columns up.

The move is easy to hear as something it is not. You are commanded, by the law, to promote the highest good. A command to promote what is impossible is empty. So the highest good must be possible.¹⁴⁴ But it is not possible in nature. So it must be possible through something outside nature that could make worth and happiness meet, and Kant's name for that is God.¹⁴⁵ And the law asks for a will fully in line with it, which Kant calls holiness. No person's character is that in any finite stretch of life. So if that conformity is required at all, the stretch cannot be finite, and Kant's name for that is immortality.¹⁴⁶ His name for the kind of claim they are is a postulate. A postulate is a proposition that cannot be proved but that attaches, inseparably, to a law that binds you unconditionally.¹⁴⁷ Nobody can know it. A person is entitled to hold it, because the law that binds him would be commanding the impossible without it.

Read from outside, this is religion coming back in through the side door. Kant spent a whole book showing that no argument gets anyone to God. Then, in the ethics, he needs God to make the sums come out, and he lets himself have him. If the ethics needs God after all, then the law was never self-standing, and the autonomy of the last steps was a story. The answer is in Kant's own text, and it is precise. The moral necessity of assuming God, he writes, is subjective, a need, and not objective, not itself a duty. There can be no duty to believe that a thing exists.¹⁴⁸ The duty is to promote the highest good. That duty rests on the moral law alone, which is certain by itself, whatever is assumed about God or a life to come.¹⁴⁹ In the first lines of his book on religion he says it flatly. Morality, for its own sake, by no means needs religion, and by virtue of pure practical reason it is sufficient to itself. It needs neither the idea of a being above you to recognise your duty, nor any incentive other than the law itself to do it.¹⁵⁰

So the order is this, and the order is the whole point. The law comes first and stands on its own. The postulates come after, as what a will bound by that law is permitted to hope about the world. A person who refuses the hope has lost nothing of the duty. A person who takes the hope has not gained a reason to be good, since they had that already. What they have gained is permission to stop believing that the universe is finally, structurally, on the side of the fraud. Kant calls the ground of that permission a need, and not a wish. The need of someone commanded to promote something who cannot see how it could ever come about.

Where this leaves the ethics, since a ruling is owed. Nothing in the good will, the test, the second formula, autonomy or freedom leaned on God at any point. The door was not settled by him either. But Kant wrote a sentence that cuts the other way, and it should be quoted against the reading given here. If the highest good is impossible, he says, then the law that commands us to promote it aims at empty ends and is itself false.¹⁵¹ That makes the law's truth hang on the hope. The reply turns on what the ought gives. When the gallows stood beside the lie, the ought gave you the can, and the can was yours: you can act as the law asks. It never gave you that the world will answer.¹⁵² Kant ran the same step one line past what it carries: from you can promote the highest good, to the highest good can be reached. That line is where this reading stops him. He held both, the self-sufficient law and the hope the law is owed, and said in the same preface that morality leads to religion.¹⁵³ The reading here keeps the law self-standing, because everything it asked was asked before God was mentioned. And it takes the postulates as what Kant also called them: a need and not a duty, a hope and not a premise. What should not be softened is the thing under the postulates, which is Kant's refusal to pretend. He looked at the question every ethics has to face, whether goodness pays, and said no. The world does not pay. You may hope that something beyond the world does, and you may not build the law on it. And the hope changes nothing about what is asked of you tomorrow morning.

So virtue is worthiness to be happy, and the world does not honour it, and Kant says so. The complete good would be worth and happiness lined up, which nature cannot do, and reason may therefore hope for what nature cannot. That hope is a need, not a premise, whatever Kant's harder sentence says. So there is no payment, and there is no applause, and there is not even the private certainty of having done it well. A person who has taken all of that in still has to get up and do something in the morning. What is left in him that could move him.

Something has to do the lifting. The question has been asked twice here and walked away from twice. An act has worth, this account has insisted, only when it would be done with every inclination pointing the other way. And an inclination is the only thing that has ever got anybody out of a chair. So on the day the grief has killed the warmth, and the man tears himself out of his own deadness and helps, what moved him. Kant has an answer, and it is the strangest thing he wrote about the inside of a person. There is exactly one feeling in the whole human inventory that reason itself makes, and it arrives as the feeling of being humbled. That feeling is what respect for the law is made of. Respect owed to another person was a maxim, a rule about not raising yourself above him. This is the other thing the word carries, which is what the law does to you from the inside. Kant says it is the only thing that moves a good will at all. Fellow-feeling still gets a man out of his house, and it was never what gave the act its worth.

One limit before any of that, because it is a limit and not a delay. Kant never explains how a law can move a person, and he says outright that it cannot be explained. Reason would overstep every boundary it has, he writes, if it tried to say how pure reason can be practical. That admission was made at the end of his first book on ethics and nothing here takes it back.¹⁵⁴ What follows describes what the law does to a person when it arrives, which is a smaller job than explaining it.

Start with what it arrives into. All the wants a person has, taken together, Kant treats as one thing turned toward the self, and he splits that one thing in two. The first is the plain drive to make yourself happy, the self-love that a person either puts above the law or under it.¹⁵⁵ The second is self-conceit, and it is a milder thing than selfishness. Call it the satisfaction of thinking well of yourself. The quiet standing assumption, carried around all day by decent people, that you are broadly fine. Reason does two different things to these, and the difference is the whole of what follows. To self-love it does very little. It restricts it, and holds it to what can agree with the law. Wanting to be happy was never a crime and nothing asks anyone to stop. The second one it does not restrict at all. Kant's word is that the moral law strikes it down.

Whatever strikes down our self-conceit in our own judgment, he writes, humiliates. So the moral law unavoidably humiliates every human being.¹⁵⁶ He is describing the feeling of standing in front of a demand a person knows he has not met and cannot claim to have met. Everybody has had the feeling and almost nobody has a word for it. It is what happens when you say the generous thing and hear yourself enjoying how it sounded. It is the small drop in the stomach when a rule you would have applied to anybody else comes around to you. Nothing has accused you. Nobody has found anything out. The demand simply stood there being what it is, and something in you went quiet. And Kant says that this humiliation is exactly where respect for the law comes from. The humbling is the respect itself, felt from the inside, at the moment it happens. The law is respected because it does not flatter. It is the one thing in you that has never once told you that you were fine.

Now the claim that makes this more than a mood. Every other feeling a person has is received. Fear, warmth, disgust, the lift of being admired, the pull toward a particular face. Each came from a temperament nobody chose and a history nobody arranged. This one has a different cause. Nothing happened to the man. A demand was simply true of him, and the feeling came from that and from nothing else. Kant's words for it are a feeling produced by an intellectual ground. It is the only feeling, he says, we know completely in advance of any experience.¹⁵⁷ Meaning you can say it is there before you have met the man. Its cause is the law, and the law is in anybody who can ask what rule they are acting on. Kant then says the thing the whole account has been waiting for. Respect for the moral law is the sole moral incentive, and the only one that is beyond doubt. And he sharpens it until it stops sounding like a motive at all.¹⁵⁸ Respect for the law, he writes, is not an incentive to morality. It is morality itself, considered from the inside, as what moves.¹⁵⁹

That buys something the argument needed and did not have. The ordering was supposed to be nobody's luck. Self-love above the law or under it, and the single reversal that turns the order over. Both were said to be available to the cold man exactly as to the warm one. A patient objector can grant every word of that and still push. Fine, he says, the ordering is a choice. But being the sort of person a law can reach is a temperament like any other. Some people plainly have it and some plainly do not. If the capacity to be moved by a rule was installed by a childhood, then luck is back inside the walls by a slower road. This is the reply. The one thing that moves a will toward the law is made by the law itself. It arrives in anyone the law can address, and the law addresses everybody. How loud it is in a man has a history, and a life can dull his ear for it or attention can sharpen it. What no history does is take it to zero. A cruel upbringing can leave a man with no warmth to help from. It cannot leave him outside the reach of the one feeling that has no source in his wiring.

Here is the hardest thing anyone says against this, put at full strength. A feeling is a feeling. If the law moves a person by making him feel something, inclination has come back in through the one door this account held shut. The grieving helper gets out of the chair because the humiliation is uncomfortable and doing the thing relieves it. On that reading he is a man managing a mood, and Kant's whole distinction collapses into a subtler kind of wanting. The answer is a limit, not a victory, and it is worth being exact about which. The feeling is an effect of the law and not the reason for obeying it. That is why Kant refuses to call it an incentive to morality. Nobody obeys in order to stop feeling it. The feeling arrives because he already stands under the law, and what it shows is that the law is there in him. It never shows that the law won the act. Say what that settles and what it does not. It does not show that any person has ever been moved that way rather than by the relief of thinking well of himself. Nothing shows that, ever, and the reason is the dear self, which has not gone anywhere. What it settles is narrower and it is what was missing. There is a candidate motive whose source is not the person's temperament, and that is the only thing the luck objection needed answered.

Two sentences in this account now look as if they contradict, and the tension has to be said out loud rather than left to be noticed. On one side, the footnote. A heart that is cheerful in doing its duty is the sign of a genuine disposition.¹⁶⁰ A person who obeys with a grudge has not yet learned to love the good. On the other, the law that unavoidably humiliates everybody. They are about two different objects. To a creature like us the law arrives as a constraint, so it humbles before it does anything else. Loving it outright is a limit nobody reaches, and the cheerfulness Kant asks for is the kind that comes from not hating what is asked. What a person is cheerful about is the doing: the meal cooked, the debt paid, the friend sat with. The humbling is what a person feels toward a standard he stands under. The cheerfulness is what he feels toward the act he is actually doing. A man can be flattened by the demand at four in the morning and glad at the sink the next evening. Nothing in him has changed position.

Which leaves the hardest thing Kant says about you, and it is his and not an inference from him. The propensity to put self-love above the law is in every human being. It is in the species, the best included. Kant says he can spare himself the formal proof, given the mass of screaming examples experience puts before our eyes in what human beings do.¹⁶¹ He is careful about what kind of claim that is. It is not a defect of the equipment, because a defect would excuse. It has to be sought in a free choice, he writes, and therefore it can be imputed, and therefore it is morally evil. It is always self-incurred. Innate, in his phrase, and drawn upon ourselves by ourselves. Then the sting, which is the reason this matters to somebody who is trying. A man whose order is reversed may still behave impeccably. His actions, Kant says, may turn out as lawful as if they had sprung from genuine principles.¹⁶² The reversed man is very often the one who has never done a thing anybody could point to.

So the honest posture toward yourself is the assumption that your own order is the reversed one. That is not the audit reopening, and the difference is exact. The audit was a court asking whether something already done was done from duty, and it closed because the auditor cannot see the books. This asks nothing about a finished act and returns no verdict on one. It is a working assumption about the next hour, and it is the same refusal as declining to compose the flattering story. Assume the law is not yet first in you, and the assumption costs you nothing you could have had anyway, since certainty was never on offer. What it buys is the one posture under which a person keeps checking the rule he is about to act on. The motive underneath stays out of sight and is not what is being checked. The alternative assumption, that the order in you is already the right way up, is the one thing that would stop the checking.

There is one incentive, then, and it is a feeling, and its cause is not anybody's wiring. It arrives as humiliation and it is called respect, and Kant says it is the whole of what moves a good will. The mechanism stays unexplained, and Kant said it would. The propensity to put the law second is universal, and it is still yours. A life run that way can look flawless from every angle, including your own. What remains are the people who have read all of it and still think it is wrong, and their case has not been put yet.

The strongest living objections to Kant are not about the murderer. They are two. That he asks you to think one thought too many, so that a person who lived by his law would be a worse person. And that moral judgment is simply not made the way he says it is, and a science has now shown that. The first is answered, and the second is half conceded, and the half that is conceded does not touch the law. Two older objections come after them, and the answers to those were given long before they were asked.

The first objection has a case, and the case is famous for a reason. A man is at the water when two people are drowning, and he can save one. One is his wife. He saves his wife. Now ask what was going through his mind as he went in. On the picture the objection attributes to Kant, the man's thought, fully spelled out, has two parts. That it was his wife, and that in situations of this kind it is permissible to save one's wife. The philosopher who made this case famous wrote that this construction gives the man one thought too many. His wife might have hoped that his motivating thought was simply that it was her.¹⁶³ Not that it was her, and that the rule allowed it. The charge under the case is larger than the case. It is that Kant's morality stands between a person and the people he loves. That a man who checks his love against a law for everyone has already stopped loving in the way that makes the love worth anything. That the shopkeeper who gives correct change because it is right is admirable. But the husband who saves his wife because it is right is a kind of monster. And that a whole life run this way, with the law consulted before every attachment, is not a moral life but a thin one.

The answer turns on one distinction. The distinction is between a test on rules and a procedure in the moment. Kant's law is the first. It says what your maxims may be. It does not say that you must recite the maxim, or run the test, or think about the law at all, while you are acting. The husband's maxim, if anyone ever wrote it down, is: when the person I love is in mortal danger, I save her. That maxim passes. It passes so easily that nobody would think to run it. And a person whose maxims all pass has no reason to consult the law at the water. He has one thought, it is his wife, and he goes. What the law does is not sit in his head at that moment. It sits in the maxims he already has. It shows up only where a maxim would fail. Which is to say only when what he wants to do could not be a rule for everyone. At the water there is no such thing. The man who stops to check is not the Kantian. He is someone who has misunderstood what the law was for. Kant described the law as a test on what maxims may be, and not as a thought to be had at the water.

What survives of the objection is worth keeping, and its deeper form should be conceded. The wife's claim on him does rest, on Kant's account, on a permission, whether or not he ever thinks it. This account holds that is no defect. A love that could not pass the test is the love that saves your wife from a wet coat at the cost of a stranger's life. And the picture of a person perpetually auditing himself is real, people live it, and it is not a good life. Kant named that weight himself, the burden of feeling chained to another's fate.¹⁶⁴ But the checking this morality asks for is checking of rules, and rules are examined away from the water. A person who has understood that does not hold his love up against the law while he is going in. He has one thought, and the law he lives under is what made that one thought enough.

The second objection has a laboratory. It begins before Kant, with the philosopher who wrote that reason is, and ought only to be, the slave of the passions. Morality, he said, is more properly felt than judged of.¹⁶⁵ On this view the law that Kant found in himself was a feeling with good grammar. And two and a half centuries later the view acquired data. People were shown a case: a runaway trolley will kill five unless a switch is thrown that diverts it onto one. Most say throw the switch. Then the same five, and the only way to save them is to push a large stranger off a footbridge into the trolley's path.¹⁶⁶ Most say no. The arithmetic is the same. What differs is that one act is done with your hands to a person in front of you. When the second case is judged, the parts of the brain associated with emotion light up.¹⁶⁷ The psychologist from the hidden motive, the one with the lawyer, gave people a story. A brother and sister sleep together once, using two forms of contraception, tell nobody, and feel closer afterwards. People said it was wrong, and were asked why, and gave reasons, and every reason was taken away.¹⁶⁸ No harm, no risk, no one hurt. And they did not change their minds. They stuttered, laughed, said something like I cannot explain it, I just know it is wrong. The researcher called this moral dumbfounding. His account of it is that the judgment comes first, from intuition, and the reasoning arrives afterward to defend it.¹⁶⁹

So the objection, whole. Kant says moral judgment is the work of reason applying a law. The evidence says moral judgment is a fast feeling, with reason arriving later as a lawyer. If that is how people actually judge, then Kant's law is not the source of anyone's morality. It is a story told afterward about a verdict the gut already gave.

The answer has two halves, and the first is a concession. The evidence is right about the mechanism, and Kant would not have been surprised, since the dear self was already in his account.¹⁷⁰ So let it be conceded whole. Most moral judgment, most of the time, is not made the way Kant describes. The second half is the one that matters. Kant's claim was never about how judgments are made. It was about what justifies them. And the people who did the experiments said so themselves, in print. The trolley researcher wrote that his conclusion was descriptive rather than prescriptive. He did not claim to have shown any action to be right or wrong.¹⁷¹ The other said the same of the lawyer.¹⁷² The trolley researcher later argued a stronger form, that Kant-shaped judgments are the rationalised alarms, and are debunked by being so. A philosopher who examined that form concluded that the neuroscientific results are beside the point for any question about what we ought to do.¹⁷³ They tell you which judgments come from feeling. They cannot tell you whether the feeling was right. To do that, you need to ask whether the act could be a rule for everyone. Or whether the stranger on the bridge was being used merely as a means. Which is to say, you need Kant's question, and the scan cannot ask it.¹⁷⁴

There is a harder version of the objection that this does not fully answer. If nobody ever acts from duty, if the mechanism is all there is, then Kant's morality describes no one. A law nobody has ever followed is a strange thing to call a law. Kant's reply is that it need describe no one in order to bind everyone. The law is no report on human conduct. If it were, it would be a law from elsewhere, a description of how people happen to be wired. It is what a will would have to be like to be good. That no will has ever fully been like that is no objection. And it is not the whole reply, because the objection said the law describes nobody, and what does move a person has already been said here. There is one thing that moves a will toward the law, and its source lies in nobody's temperament.¹⁷⁵ That the order in a person runs the other way is true of everybody, and it is still that person's own doing. It is what he meant when he said that the law humiliates. A standard that everyone already met would not be a standard.

Two older objections remain, and the answers are already built. One was made by a philosopher in 1958. She argued that the whole vocabulary of moral obligation, the special ought that binds without an if, was a survival. A survival from an earlier conception of ethics that no longer generally survives.¹⁷⁶ A law conception of ethics, she wrote, is not possible unless you believe in God as a law-giver.¹⁷⁷ Kant's idea of legislating for oneself she called absurd, as absurd as calling each decision a man makes a vote that passes by one to nothing.¹⁷⁸ A century earlier a different critic had said the same thing in a fiercer register. Kant had borrowed the imperative form of ethics from theology without examination, and an absolute obligation was a contradiction in terms.¹⁷⁹ The answer was given when the law was traced to its source, and it needs only naming. The law is the form your own willing takes, and it is not a vote, because it is not up to you. Both critics assume that binding force can only come from a will above you. Kant's whole argument is that a will above you binds only after you have judged it, and the thing that judges is the law. What is conceded to them is that Kant's word, legislating, is a picture, and the thing it pictures is willing and not voting.

The other is that a morality with no exceptions breaks on every hard case, and the murderer is only the first. The answer is the ruling from the door, in one line: exceptions come from the rule, not from you. In evil circumstances the world the formulas picture becomes a goal to work toward, and the demand that a maxim hold for everyone never softens. What the rule never contains is the clause Kant refused: unless I judge otherwise. That clause is the one every hard case is secretly asking for.

So the one-thought-too-many objection is conceded in its deep form and answered in its crude one. The feeling-first evidence is conceded whole and changes nothing. What is standing and what has fallen can now be said in one breath.

What survives of Kant is the shape of a moral demand, and the shape is this. A rule either holds for everyone or it is not a rule. Every person on the other end of an act can refuse to be merely used, and that refusal is not something anyone gets to overrule. Freedom holds, not because anyone proved it, but because nobody can decide anything without standing where the demand is made. And nobody gets credit, because the one place worth lives is the one place nobody can see. What falls is smaller than people think. The answer at the door, as Kant gave it. The claim that the first formula alone can fix what it tests. The door needed the second to say who was in the maxim. The postulates, if anyone tries to make them more than a hope, as Kant himself did. That is the ruling. What follows is what it does when it is carried out of the argument and into the small hours of an ordinary life.

The small lie first, because it is the one everyone tells. Not the polite formula at the foot of a letter, which Kant left open and stays open. The lie told to manage someone: the diagnosis softened, the money trouble hidden, the evening smoothed by a fact withheld from the person it belongs to. The attractive picture said this was fine, and Kant's argument says something more exact than that it is wrong. It says the small lie is the whole of morality failing at one point. Not a small failure. The whole thing, small. Because the rule under it is: when someone cannot be trusted with a truth about their own life, I decide it for them. The person spared is being managed, kindly, by someone who has decided they cannot handle their own life, and they could not agree to that. So the small lie is the false promise and the taxi driver with the knife, reduced to a size where nobody feels it. That is not a reason to be cruel. The truth can be told with every gentleness the situation permits. It is a reason to notice that the gentleness was never the problem, and the management was.

Then the private exception. The one nobody knows about, that has been earned, that this once is fine. Kant's argument does not say that exceptions are wrong. It says that there is exactly one kind of exception that nobody may make. The kind that comes from a judgment that this case is different. The murderer showed the shape: an exception from the rule is the rule applied, and an exception from the person is the clause Kant refused. The private exception is the second kind, every time, and the dear self is the one doing the judging, always with a good reason.

Then the credit. The good thing a person did, and wanted to have been his. The ruling here sounds cruel and turns out to be the kindest thing in the account. He cannot have it, and the reason was the dear self: whether it was his is a fact he has no access to. So the question behind every kind act ever done, did I do it for the right reason, is dismissed. Dismissed for lack of a court that could hear it. What is left when the credit goes is the act, and the rule under it, and the next act. That is all Kant ever said anyone had. It is more than most people think they have, because it is the only part that luck cannot reach.¹⁸⁰

And then the thing under all of it, which is freedom. Most people carry an idea of freedom as a feeling, or a fact about the brain, or a gap in the chain. Kant's freedom is none of those. It is the standpoint a person cannot leave when he decides.¹⁸¹ He will never catch himself being free. There is no experiment for it. The humbling the law produces in him is a feeling of the demand, and not a glimpse of the freedom underneath. What there is, is the gallows, and the plain knowledge that he could refuse. That is the strangest sentence in the account and it is the one to keep. You are free because you are bound.

So, the whole argument, one last time, in its own order and without the scaffolding. The only thing good without condition is a will that does what is right because it is right. A will does that when it acts from a rule it could will for everyone. The people on the other end of that rule must be able to agree to it, which means never merely used. And agreeing is not all they are owed: their ends have a claim on yours, and their worth holds you off from managing them. The rule binds because a person's own reason gives it, not because anything outside him does. It can give it because he is free, a freedom he cannot prove and cannot act without. He cannot see whether he has obeyed it, so he cannot take credit. It allows no exception that comes from him.¹⁸² And it does not make him happy, and it does not pretend to. Kant got the case at the door wrong. The reading given here has also parted from him three times. On the order of his formulas, on his proof of freedom, and on the sentence that hung the law on a hope. On the law itself he was right.

Where it leaves you is not a comfortable place, and Kant never said it was. A demand that applies to everyone and has never been fully met by anyone. A worth that is real and that no one, including you, will ever see. A freedom you cannot prove and cannot put down. And an ordinary evening, with a small lie ready on your tongue, and a rule that says everyone or no one. Nobody can be seen to keep it. It binds everyone anyway. That is the whole of it. The man is still at the door, and you know what the rule says, and you know what you are going to do.