Is Letting Someone Die Just as Bad as Killing Them?

Two Boys, Two Bathtubs, One Question

Imagine two cousins, Smith and Jones. Each stands alone with a young boy in a bathroom. In the first case, Smith shoves his young cousin’s head underwater and holds it there until the boy stops breathing. In the second, Jones slips into the bathroom planning to do the same — but finds his cousin already lying unconscious under the surface. Jones watches. He could easily pull the child up, but he chooses not to. The boy drowns.

The American philosopher James Rachels (1941–2003) invented this pair of cases in 1975. He asked: is Jones’s behavior any less awful than Smith’s? Most people feel the two acts are equally bad. And if they are equally bad, Rachels argued, then killing is not, in itself, worse than letting die. If real-life killings usually feel worse, that must be because of extra factors — like cruelty, fear, or breaking a special trust — not because of some deep moral difference between doing harm and merely allowing it.

This distinction — between doing harm and allowing harm — is called the doing/allowing distinction. Rachels wanted to show it has no real moral weight when everything else is equal. But was he right?

Are the Cases Really the Same?

Not everyone bought Rachels’s argument. The first objection: the two cases aren’t perfectly matched.

Philosopher Scott Hill (21st century) pointed out that Jones had an extra option: he could kill his cousin or let him die. Smith, who actively drowned the boy, never had the chance to simply stand by. Hill imagined a third cousin, who finds the boy already drowning and chooses to push him back under rather than rescue him. That third act, Hill argued, feels worse than both Smith’s and Jones’s — and the best explanation is that killing really is worse than letting die, other things being equal.

Another worry comes from Frances Kamm (b. 1948). She asked: would we be willing to harm Smith and Jones in the same way to bring the dead child back? Most people say it would be acceptable to kill Smith for that purpose — after all, he is the murderer — but not okay to kill Jones. If we treat them differently, Kamm suggests, our gut feeling is that their actions actually differ morally.

Then there’s a deeper logical problem. Shelly Kagan (b. 1956) argued that even if Smith’s and Jones’s acts were equally bad, that wouldn’t prove all killings and lettings die are equal. He compared it to addition: you can’t assume that changing just one factor (killing vs. letting die) always changes the moral total in the same way. Factors might interact. For instance, Kamm later argued that a single property — like “removing a barrier” — can behave differently depending on the context. So one pair of matched cases might not settle the whole debate.

The Trolley Riddle

The most famous puzzle in this debate was born from a train. In Philippa Foot’s original 1978 version, a runaway trolley is hurtling toward five people tied to the track. You can flip a switch, turning the trolley onto a side track where only one person is tied. Foot argued it’s permissible to turn the trolley: you’re killing one but saving five. But in a different case — a judge framing and executing an innocent person to stop a riot that would kill five — Foot said no. She thought the difference was that the judge would be doing harm to the one, while the trolley driver would be choosing between killing one and killing five.

Then Judith Jarvis Thomson (1929–2020) changed the story. In her 1976 version, you’re not the driver — you’re a bystander watching from a bridge. Flipping a switch now clearly pits doing harm to one against allowing harm to five. Yet many people still feel it’s okay to flip the switch. That blew a hole in the idea that doing harm is always harder to justify than allowing harm.

Later, Thomson changed her mind. She argued that if you aren’t willing to sacrifice yourself by jumping in front of the trolley, you can’t rightfully sacrifice a stranger. Not everyone agreed. But the trolley problem proved that the doing/allowing distinction isn’t a simple rule; often, our intuitions clash. Even if the distinction matters, we need to explain why it sometimes seems to disappear.

A Surprising Theory: It’s About How Many Choices You Had

Philosopher Jonathan Bennett (b. 1930) proposed a radical way to understand the doing/allowing distinction. Forget about actions vs. inactions. Bennett said an agent does harm if most of the ways she could have behaved at that moment would not have led to the harm. She allows harm if most of the ways she could have behaved would have led to the same result anyway.

If you pour salt on a slug and it dies, most of your possible movements wouldn’t have killed it — so that’s doing harm. If you don’t move a slug from a car’s path, almost anything else you do (stand still, walk away, check your phone) still leads to its death — so that’s allowing. On this view, killing is not automatically worse. What matters is how easy it was to avoid the harm. When avoiding a death requires huge effort (like holding very still when every tiny twitch would set off a bomb), letting die might be just as blameworthy as killing.

But Bennett’s account faces tricky counterexamples. Consider “Immobility 2”: a motion detector will trigger a fatal explosion if you make any movement at all. If you stay totally still, no one dies; if you so much as twitch, someone is killed. On Bennett’s analysis, if you wave your arm and set off the explosion, you count as allowing the death — because most ways you could move would set it off. That clashes with our strong intuition that waving your arm to trigger a bomb is doing harm, not merely allowing it. Cases like this show how hard it is to draw a clean line — and why many philosophers doubt any single formula can capture the distinction.

Why This Argument Won’t Go Away

You might wonder: why spend decades arguing about bathtubs and trolleys? Because the doing/allowing distinction runs through real life. In medicine, is it worse for a doctor to inject a drug that ends a suffering patient’s life than to turn off a respirator? In law, do we punish someone who fails to feed his child differently from someone who poisons them? In your own daily choices, does ignoring a friend’s cry for help make you as blameworthy as causing the hurt yourself?

If the distinction doesn’t matter, then letting someone suffer when you could easily prevent it is morally the same as reaching out and hurting them. Many people find that conclusion scary — it would demand much more from each of us. But if the distinction does matter, we need a clear account of why protecting a person from your own actions is more important than protecting them from things you merely fail to stop.

This debate also touches new technologies. When a self-driving car must decide between hitting a pedestrian or swerving into a wall, is it “doing” harm to the passenger or “allowing” harm to the pedestrian? How we answer may determine who gets protected first. The deep question Rachels put on the table — is there a fundamental moral difference between making something happen and letting it happen? — has no simple answer, but it shapes everything from courtroom verdicts to how you treat the people around you.

Think about it

1. If you see a stranger drowning and you have a life jacket you could throw, but it’s your favorite jacket and will be ruined, is your choice morally different from pushing someone into the water to save another person? Why or why not?
2. Imagine a friend asks you to lie to get them out of trouble. Is staying silent the same as telling the lie? What if silence lets an innocent person get blamed for something they didn’t do?
3. Suppose you knew that by pressing a button you could save five people from dying in a disaster, but pressing it would cause a separate accident that kills one stranger. Would you press it? What if instead you had to not press a button to save the five, but pressing it would stop the separate accident — does that change your answer?