Appearance
13.6 — The Hard Cases, Worked
Five cases. None of them resolve cleanly, and that is the point.
The purpose is not to give you answers. It is to show what the three frameworks do when they collide, so that you can recognise the shape of a real dilemma rather than reaching for the first argument that arrives.
Case one: the trolley
A runaway trolley will kill five people. You can divert it to a side track where it will kill one.
Almost everybody says divert.
Now the variant. You are on a bridge above the track, next to a large man. Pushing him off will stop the trolley and kill him, and save the five.
Almost nobody says push.
And the arithmetic is identical.
What it shows
The variants isolate what people are actually responding to, and the differences are informative.
Using somebody as a means. In the first case the death is a side effect of diverting; in the second, the man's body is the instrument. Kant's second formulation (10.9) tracks the intuition exactly.
Physical directness. Pushing feels different from pulling a lever, and this is a fact about human psychology rather than about the ethics — and it is worth knowing, because it means your moral intuitions are sensitive to distance and physical contact in ways that have no justification.
And doing versus allowing. Most people distinguish these strongly, and philosophers argue about whether the distinction can be defended.
What it is good for
Not for answering. You will not meet a trolley.
For diagnosing. The variants show you that your judgement is tracking something other than the outcome, and identifying what it is tracking is the exercise.
And for one real application: the intuition that distance and directness reduce responsibility is very strong, and it is almost certainly wrong. The harm you cause at a distance through a system is not less harm.
Case two: the lie that protects
Somebody's parent is dying and asks you, directly, whether they are going to die.
The family has asked you not to say.
The frameworks
Consequences: depends entirely on what the person actually wants and what the knowledge would do, and you do not know.
Kant: deceiving them treats them as a means to the family's comfort. And the second formulation is decisive: could you tell them what you are doing and why? I am lying to you because your children asked me to. No.
Care ethics: the response should fit this person, and the question is what they are actually asking. Frequently a dying person asking that question is not requesting information; they are checking whether you will be honest with them.
Virtue: what does an honest and compassionate person do?
Where this book lands
The strongest consideration is that the information is theirs.
A person has a claim on the facts about their own body and their own remaining time, and the family's discomfort does not override it. And the practical evidence, which Volume V covers, is that dying people usually know, and that the pretence isolates them at the point where isolation costs most.
And the correct move is usually not a binary. What are you thinking? What would you want to know? Which is honest, does not withhold, and lets them set the pace.
Case three: the whistle
You discover that your employer is doing something that harms people. Reporting it will very likely end your career, and you have dependents.
The frameworks
Consequences: the harm prevented against the harm to your family, and the probability that reporting achieves anything, which is frequently low.
Duty: you did not cause it. Whether there is a positive duty to prevent harm you did not create is exactly the contested area (13.3's negative and positive rights).
Virtue: courage is a virtue and so is responsibility to dependents, and here they conflict directly.
What actually helps
Three observations that do more than the theory.
The choice is presented as binary and rarely is. Internal escalation, documentation, an anonymous route, a regulator, a professional body, or leaving and then speaking — the options between silence and self-immolation are usually several and are usually not considered.
The cost is real and the estimates are systematically wrong in both directions. People overestimate the chance that speaking will fix it and underestimate how long the personal cost lasts.
And the question that clarifies it: how severe is the harm, and is anybody else in a position to stop it? A person who is the only one who knows is in a different position from one of forty people who know.
3.9's rule applies: decide the line in advance. In the moment you will have a reason, and so did everybody in that hall.
Case four: the obligation to strangers
A child is drowning in a shallow pond in front of you. You will ruin your clothes. Obviously you save them.
Peter Singer's argument: distance is not morally relevant. A child dying of a preventable disease elsewhere is the same case, and the cost to you is comparable. Therefore you are obliged to give until giving further would cost you something of comparable moral significance.
Which is a valid argument from premises most people accept, and its conclusion is enormously demanding, and almost nobody acts on it.
The responses
Reject the conclusion and find the flawed premise, which is what most people do and few do explicitly. The candidates: distance does matter because relationships generate obligations (13.5); the drowning child is a single determinate case and global poverty is not; and a duty that consumes a life cannot be a duty.
Accept it and fail to act on it, which is the honest position of a great many people including this book's.
Or accept a bounded version. A fixed proportion, given effectively, treated as an obligation rather than as generosity.
Where this book lands
The argument is not refuted, and the fully demanding conclusion is not livable, and the bounded version is the defensible compromise and everybody who adopts it knows it is a compromise.
And the most useful part of Singer's argument is not the conclusion but the premise: distance is not morally relevant. That premise alone, taken seriously, changes how you read almost every decision about supply chains, about outsourced harm, and about who your society treats as far away.
Case five: two obligations
Your elderly parent needs care that will consume years. Your child needs you now, in the years that will not come again. You cannot do both properly.
And this is the most common genuine dilemma in an ordinary life, and none of the three theories addresses it well.
Consequences: incommensurable. You cannot compare a parent's last years with a child's childhood in any unit.
Duty: both are duties, and no rule ranks them.
Virtue: what does a good person do? A good person is in agony.
What is actually true about it
Three things, and they are more useful than a resolution.
There is no option without loss. Any framework that produces a clean answer has hidden the cost. And the harm done by pretending otherwise is that people torture themselves for not having found the right answer, when there was none.
The correct question is not "which is right" but "what can be sustained." A plan that is morally optimal and collapses in eight months is worse than a plan that is imperfect and holds.
And the person deciding is also a person. Every framework in this Part treats the agent as a calculator; care ethics is the only one that notices they have a body and a limit (13.5). A carer who is destroyed helps nobody, and this is not an excuse, it is arithmetic.
What the five cases show
Four findings.
Your intuitions track things that have no justification — physical distance, directness, and whether the harm is done or allowed. Knowing that does not remove the intuitions and it does let you check them.
The frameworks disagree, and the disagreement is the information. When all three point the same way, act. When they diverge, you are in a genuine dilemma and you should stop looking for the argument that settles it.
Most presented binaries are not binaries. Case three is the clearest and it applies everywhere.
And some cases have no answer without loss. The mature response is to choose, accept the cost, and not spend the following years litigating it.
What to carry from this page
When your judgement changes between two cases with identical arithmetic, find out what it is actually tracking.
Could you tell them exactly what you are doing and why? It settles more cases than any other single question in this Part.
Look for the third option before accepting the binary.
Distance is not morally relevant, and taking that seriously is uncomfortable and correct.
And a plan that is optimal and collapses is worse than one that is imperfect and holds.
Next: 13.7 — why good people say nothing, which is the failure this whole volume keeps returning to.